Introducing...uTMax!

[nickb]

Hi All. 

I've been away from this group for a log time. I must admit I thought it was dead. Turns out I had not been getting notifications. But that is another story. Anyway, if anyone was looking for support on uTgui and didn't get it - I'm sorry.  It's best to email me directly anyway.

The other week I had 30 x 5881 tubes to look at and it made me think. I kept hitting the current limit of 200mA so had to reduce the test range to make it work. I wanted to save the data and plots for each one and it was all very tedious. So I made a to-do list.

(1) Auto plot and data file save function

(2) Auto-increment of numbered suffix on the file name.

Well, these were both software so I blew the dust of off uTgui and added those features. Of course, this got me thinking and I added to the list.

(3) Bigger current range needed

(4) Increase the grid voltage range

I took a look at the uTracer schematic and both seemed to quite easy to do.  The HV switch transistor is rated at 500mA and had an associated  current limit circuit of around 250mA. I could make that 450mA and fly close to the margins. Should a short happen I'd probably lose the KSA1156 but it was worth it for a try until I replaced it with something beefier or put a foldback on the current limit. The measurement range also needed to change. This I what I did:

I added a 3.3 ohm 5% in parallel with R38 to increase the limit to 450mA. 

I added an 18 ohm 1% in parallel with R45 to double the measurement range.

Added a 400mA range to the GUI.

This does compromise the low end anode current resolution a little but I think that I can live with. The droop on the boost capacitor is increased but since the actual voltage is measured anyway, the error was not too much to worry about. I could always compensate in software if I really wanted to. 

I didn't see much virtue in doing the same for the screen supply.

The grid voltage range is set by that gain of the op amp IC3 as 10, making the range equal that gain times the 5V supply = 50V. In practice this is restricted to 48V.  I changed the gain to 12 by giving a 60V range (58V in practice). There is plenty of margin on the output of IC3 to make this work

Added a 121K 1% resistor across each of R13 and R1.

Added a 60V range option to the GUI

And so uTMax was born.

The grid voltage range  could be made even higher but the negative rail would have to go higher to say 50V by changing R3. A zener would have to be added to drop the extra 10V going in to the LM337 to keep that within it's limits and T7 and T6 would need to change. It didn't seem worth the effort.

I keep meaning to integrate Derk's excellent SPICE model extraction into uTgui, but I never have found the time. One day..

Nick Barton

[nickb]

Been chatting with Ronald and I'd like to make it clear that this is just an experiment and not something you can replicate. You would need the special GUI in any case. Just so you know, quite a bit of of work needs to be done to make this idea into something that would be reliable and with better accuracy. 

Nick B

[Ihor]

I thought it might be interesting to update this thread mainly because I noticed recently there there are lots of new features implemented in uTMax. I myself tried on only long time ago, but I just saw that the author also (re)wrote completely the original Ronald's firmware (for the PIC controller) which can be easily replaced and has lots of interesting features. 

http://bmamps.com/store/index.php?main_page=product_info&cPath=109&products_id=440

This is the citation from the website:

"The new firmware has the following enhancements.

• Dynamically controlled boost pulse widths to allow the the boost voltages to be set more accurately at lower voltages.
• All measurements are over-sampled to lower noise and improve accuracy.
• The heater voltage accuracy* is improved by changing the PWM frequency to 2KHz. 
• Grid voltage 12bit resolution (4095 steps) upgrade from original 10 bit (1023 steps)
• Calibration data saved to chip for portability
• V3 introduces firmware that can be updated from the uTmax GUI"  

I was wondering if someone used it and can share the experience. I think that grid 12bit resolution is interesting but I wonder if someone found it useful in practice. Most of us use external heater supplies, so that addition is probably not so critical, but about the the first two improvements in the list, I thought it was possible to do with original firmware as well, that oversampling part, and changing the pulse width is also done in Ronald's firmware (at least he describes that issue and solution in his weblog).   

There is also a very nice design of the tube pin switching board in the section "A Pin Switch Matrix Board" (see https://bmamps.com/v01/home/techie-corner/utracer-utmax/ ) which is rather an overkill but I remember some users on the forum were asking for that and also in Ronald' gallery there are some similar implementations. Now it is possible to easily build it yourself!

Cheers, 

Ihor

 

[Ihor]

just to add, this is from Ronald's blog: https://www.dos4ever.com/uTracerlog/tubetester2.html

Section: 4. Fully interrupt based boost converter control (Part I).

"Additionally, in the previous section it was concluded that in order to accurately control the output voltage of the anode and screen boost converters in the low-voltage range, the boost converter pulse needs to be reduced to 5 us for voltages below 50 V. So to summarize the requirements for the new boost converter control firmware:

• Completely controlled from the interrupt service routine, with minimal overhead for the main program.
• Boost converter pulse length 5 us or 20 us depending on output voltage of the boost converter."

[Davo]

Damn, i haven’t spoken that guy since a year and half, and i was blown away from the interface that he made back then..

Only the resize windows was cool.

He’s a smart guy, and know his stuff tho

I just saw all the improvements! 

Holy Shit!!

It dusn’t take long and the utracer is a Roe-Pruefer lol

Verstuurd vanaf mijn iPhone

Op 12 jan. 2023 om 12:35 heeft Ihor <iste...@gmail.com> het volgende geschreven:

I thought it might be interesting to update this thread mainly because I noticed recently there there are lots of new features implemented in uTMax. I myself tried on only long time ago, but I just saw that the author also (re)wrote completely the original Ronald's firmware (for the PIC controller) which can be easily replaced and has lots of interesting features. 

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/42f51c62-a551-4d59-90d3-5f9c7c186865n%40googlegroups.com.

[Davo]

I guess it’s a matter of time and everyone has an automatic tube tester lol

With nice shiny curves …

and i guess as with with an option to save all relevant data etc. to an pdf

I also like the option for adding a display 

Or displays.. a nice future would be also

To use it as tester without a computer 

Just for a quick test at any place

I saw loads off extra futures  i’m quite stunned tho!

Verstuurd vanaf mijn iPhone

Op 12 jan. 2023 om 13:04 heeft Davo <djda...@gmail.com> het volgende geschreven:

Damn, i haven’t spoken that guy since a year and half, and i was blown away from the interface that he made back then..

[Ihor]

and i guess as with with an option to save all relevant data etc. to an pdf

I also like the option for adding a display 

Or displays.. a nice future would be also

To use it as tester without a computer 

Just for a quick test at any place

I saw loads off extra futures  i’m quite stunned tho!

Verstuurd vanaf mijn iPhone

... I just cannot self-advertise after that :))) You already (for 2 years) can run your uTracer without a computer, with a small OLED screen in the middle of nowhere and even using your iPhone only (or any possible embedded system like raspberry pi or anything else): 

https://boffin.nl/wp/utracer-and-esp32/ and 

https://boffin.nl/wp/

 

[Davo]

True Ihor lol

Without a phone or computer etc.

Like the traditional tester do…

It would be a nice addon in my opinion,  haha but who am i…

Just for the sake of having it all…

But that matrix option would be coolest utracer gadget all time…

I don’t think Ronald would be pleased that he re-wrote the whole pic.

But on the other hand this guy is taking the whole utracer project to a higher level

Only the 10 to 12bit upgrade is amazing 

He&Ronald should work together, same as you Ihor & Martin, you guys are utracer hero’s also.. and everyone else i forget to mention 

Verstuurd vanaf mijn iPhone

Op 12 jan. 2023 om 15:25 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/a3a57881-e30a-4ed4-9c12-e5b212aae4edn%40googlegroups.com.

[Big Josh]

The FUtracer also took the project to another level, and uses that superior firmware. The original uT firmware is buggy and prone to issues. At least that was my experience, especially with testing tubes like EL34s.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/61DD5F2F-B4B6-4981-8250-BE0D92351E0A%40gmail.com.

[Ihor]

The FUtracer also took the project to another level, and uses that superior firmware. The original uT firmware is buggy and prone to issues. At least that was my experience, especially with testing tubes like EL34s.

I was not aware of FUtracer, looks like an interesting project. It also uses uTMax as the main software, so probably that firmware (PIC) is the same as one I mentioned here before (uTMax can deal with both versions, and the new one can actually be placed in the standard uTracer3+ board). The design looks very similar, but no schematics yet and prices, also specs are not clear, if it si closer to uTrcaer3 or something between uT 3 and 6. . 

 

[Martin Manning]

I really never had any issues with the original uTracer firmware causing oscillations or other problems. When I did see oscillations, I could always fix it by adding a ferrite to the grid or plate lead. There were some hardware and UI shortcomings early on. I bent Ronald's ear about them, and eventually they were resolved. Ihor's double acquisition is a great improvement as regards hitting the target voltages on quick test points and for getting more accurate curves, plus the UI just looks nicer.

[Ihor]

I also have not had any problems with the oscillation but I can imagine that it can happen. On the FUtracer website there is also a comment where to place resistor in order to deal with oscillations, which means that they were observed there as well. 

I see that the developer of uTMax already read this thread and changed the description from the yesterday's one to 

• Dynamically controlled boost pulse widths to allow the the boost voltages to be set more accurately at lower voltages. The original utracer firmware used only 5us and 20us widths."

taking into account those 5us and 20us values.

That uTMax's improvements log also says that "The heater voltage accuracy* is improved by changing the PWM frequency to 2KHz." but Ronalds blog says that his heater uses 19.5kHz for PWM. 

In both cases, "dynamically controlled" and 2kHz vs 19.5kHz does not tell to the user what it means in practice for his measurements. 

12bit vs 10bit for the grid voltage means that one can set that voltage in steps of 0.012V and in the second case with 0.05V.... and most of us typically work with -2,..,-25v, so for the tubes that in any case have accuracy in terms of currents  of +/- 10-15%, then dependent on the voltage of the heater, and then dying with time emission, all those things are nice on a paper but might be too much. At the same time it is always good to have the most accurate measurement device and by digging and analyzing other designs to improve the current state of the art. 

I also like the improvement of the Vg voltage for small values (from 0 to -0.5V) described on the uTMax website and implemented in FUtracer, which is also easy to implement in the original uTracer3, but again it should be useful for very limited number of people I would say. First you need to work on designs that have an operation point at -0.5V of Vg (or even more positive), which are quite rare. Second of all, in almost all books it says that the grid voltages even at around -0,1,..., -0.2V will cause grid currents.... so calibrating Vg for -0.1V for the perfect match does not mean that it will be there during the measurements, or otherwise it would be interesting to see such experiments that confirm the usefulness of such improvements in practice. 

 

[Big Josh]

" On the FUtracer website there is also a comment where to place resistor in order to deal with oscillations, which means that they were observed there as well. " 

I think it refers to "IF" you have an issue? I've personally never had an issue on the FUtracer hardware with oscillations. Only on the uTracer hardware. It's harder to get oscillations with the tube sockets being on a PCB.

You make too many assumptions, and lots of theoretical guessing without knowing. We can't speak to anybody else's use cases. For you, maybe you don't need higher grid accuracy, that's fine. Others do. I don't really either, but so what? I'm not everybody. My amp grid voltages are often close to 0, so I always test around there with small variations. It's nice to be able to reliably and repeatedly get higher resolution results. 

The other thing you're forgetting is how painfully slow the utracer is with Ronald's GUI and firmware. utMax with the vastly improved firmware is significantly faster. It's like racing a bullet train vs. a wheelchair. ;)

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/a0f9f356-7475-4814-bbab-abd92d5b512en%40googlegroups.com.

[Ihor]

I attached a few images just to the other also know what we are talking about. 

Indeed, I can imagine the oscillations are not there but I would say because there is a well-done PCB with he tube sockets that prevents that. Looking at the PCBs of uTracer and FUtracer (the other image), the PCBs are almost identical (I do not even know how it is possible to "Copyright" that new PCB comparing to the green on, but Ok) with that extra trimpot and usb connection (but probably a few more mods) and slight "rotation" of some components (transistors). What I mean is that the main PCBs in both cases most likely not prone to oscillations. At the same time, for me, and many others that PCB with sockets would pose some limitations. It only design for tubes like EL34, 12A**, EF80, EL84, so it is not only those but "like", still there are lots of tubes which one has to connect with wires. ..... The other solution now to make. uTracer not to oscillate is to build such PCB and connect to uTracer.. I could not find any details on FUtracer website how the oscillations are dealt with on the main PCB. 

Abou the grid voltages, many of us already build Ronald's Grid Loupe https://boffin.nl/wp/utracer/ and PCBs are easily available. Making it is about 10 minutes and it maps the 0,..-50 range to +10,..-5V, which gives you really good calibration around zero, and even more, with 10bit resolution over the range of 15 volts the stepping is great (0.015V). As a bonus it can measure grid currents.

The acquisition speed is indeed important but probably for those who are using it more "industrial" settings testing many tubes. I can only say for myself in this case, and in my settings when I test a tube (30 points, 5-7 curves) is it probably a minute or two. If that would take 10 minutes, I would not complain either, because I do it from time to time, and I do not own more than 100 of tubes. At the same time, it is interesting to know about those speedups quantitatively. If the speedup is 10 times that's great, if it is about 2 times, from the practical point of view it is the same. I do not see any forum for FUtracer, but for promoting that clone it would be nice to put move comparison videos, for example comparing the acquisition speed for the same settings (and other show off cases that show how bad uTracer is), so the buyers are immediately convinced with the improvements.

       

[Davo]

Love this comment 

“ Ronald's GUI and firmware. utMax with the vastly improved firmware is significantly faster. It's like racing a bullet train vs. a wheelchair. ;)”  

But the improved grid voltages, thus this means that the grid loupe addon is no more needed?

As far as i know a 470r or 1k grid resistor is something standard on a tube amp to avoid oscillation, but those ferrite beads are normally the way to go to deal with those nasty oscillations.

But that relay matrix addon is far-on the most coolest gadget you can get!

Just select a preset a there ya go, quite fool proof, sometimes things can get nasty with wrong connections

If properly designed, no flame bow’s and oscillations are easy to solve

But yeah, maybe better not to romanticize things too much haha 

Op 14 jan. 2023 om 03:20 heeft Big Josh <shif...@gmail.com> het volgende geschreven:



To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/CABi0_%2BP78nt26gRpsaLar61HX7bHQkCzNvN9xfWGRah5SoD5Cw%40mail.gmail.com.

[Davo]

Dus that FUtracer has copyright?

It’s almost a complete knock off…

Yeah sum minor things.. 

but looking at the whole spectrum, just an upgrade… 

The heart is the uTracer concept with a fancy purple finish and a fancy sticker on the pic…

As it comes to copyright, you gotta have a damn good lawyer…

I’ve spend the half of my life in the music biz & things get quite complicated when someone copies ur song and changes the arrangement & chords a bit put sum extra effects init etc… kinda walking on the line

Somehow I really like the project, but also i have mixed feelings about it..

I hope he somehow contacted Ronald.

They should work together tho!

It ain’t easy to reproduce the whole board inc. upgraded firmware..

The board ain’t that hard for me, but as it comes to programming pffff

It’s just like learning a new language 

You can’t learn one, all the other crossover languages etc etc. Makes my head dizzy

And it’s boring also, accidently touching a charged 450v capacitor from time to time gives you the feeling that you live haha

And keep my focus 200%

Op 14 jan. 2023 om 12:06 heeft Davo <djda...@gmail.com> het volgende geschreven:

Love this comment 

[Martin Manning]

In early 2015 I compared Ia-Va traces at 0V Vg for a 12AX7 where the grid voltage was first provided by the uTracer and then fixed at zero by tying the grid to the cathode. I found that the traces didn't agree very closely, and brought it to Ronald's attention. It was known that the uTracer couldn't really set Vg at 0V (the upper limit is actually about -50mV) due to the saturation voltage of the grid supply output transistor, but hat wasn’t enough to explain the difference in the traces. For some reason the voltage was not being set accurately as it approached zero. 

Ronald found that the grid bias was in fact falling below the set voltage during the 1 ms measurement pulse, and the cause was a rise in the ground reference for the PIC’s PWM signal due to heavy current draw from the LED’s in the anode and screen voltage optocouplers. This problem existed on the uT3, and was enough to cause low emission readings on tubes like 12AX7 where the standard test point is at -2V Vg. It became much more noticeable in the uTracer 3+ configuration, (400V upgrade), where the optocoupler current is almost doubled. The fix is the 10x larger capacitors in the Salen-Key LPF in the grid supply circuit, which prevents the grid voltage dip from appearing until after the measurement pulse is over.

So, no, you don't really need the Low/+ Vg loupe unless you want to get a true 0V Vg, or positive Vg, or you wan to measure grid current in triodes.

[D.A.R Achterberg]

Ah okay, i have the gridloupe also and noticed that with the NOS 12ax7,ecc83 tube,  that the graphs were quite off…

But in never dive into it what caused it,

Because the gridloupe was there, but it was the dip lol Makes sence now

Btw why do people ask so much money for those ecc83 Philips/mullard tubes? They can be real bugger tubes (heater flash) which can be solved quite easy with a resistor on the heater supply.. 

Op 14 jan. 2023 om 12:41 heeft Martin Manning <mman...@fuse.net> het volgende geschreven:

In early 2015 I compared Ia-Va traces at 0V Vg for a 12AX7 where the grid voltage was first provided by the uTracer and then fixed at zero by tying the grid to the cathode. I found that the traces didn't agree very closely, and brought it to Ronald's attention. It was known that the uTracer couldn't really set Vg at 0V (the upper limit is actually about -50mV) due to the saturation voltage of the grid supply output transistor, but hat wasn’t enough to explain the difference in the traces. For some reason the voltage was not being set accurately as it approached zero. 

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/3eaa6582-3ab1-451c-a130-eb857f7e8dean%40googlegroups.com.

[Martin Manning]

"Dus that FUtracer has copyright?

It’s almost a complete knock off…

Yeah sum minor things.. 

but looking at the whole spectrum, just an upgrade… 

The heart is the uTracer concept with a fancy purple finish and a fancy sticker on the pic…"

Yes, FUtracer pirated Ronald's clever hardware design and developed some new firmware and software. The motivation is spelled out quite clearly in the name ;^) After uT3 and 3+ were working reasonably well, Ronald got immersed in the uT6 project and left the uTracer3 behind. I believe he was approached by the China-based eTracer guy about collaborating on a better tracer, but Ronald wanted to keep doing his own thing. So now there is eTracer as a competitor.

[Davo]

Yeah i heard about that e-tracer story…

Quite sad though the e-tracer has quite more potential, Ronald is a genius as it comes engineering circuits, but his skills with programming are quite basic, leave that to the new generation, combine the old with the new generation and you have a winning team!

There’s a old saying from the Roman Empire: Divide and conquer

But maybe Ronald dusn’t want to work together? He dusn’t like social media also, so that can make things quite complicated without skype or similar 

And for the sake of the utracer group he should add a standard basic forum pluggin to his server though…

Give a few trustworthy die hards admin rights and he dusn’t have to care about it

Google groups is quite annoying and makes everything so blurry & info htf

Op 14 jan. 2023 om 13:19 heeft Martin Manning <mman...@fuse.net> het volgende geschreven:

"Dus that FUtracer has copyright?

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/912715a1-0a19-420a-8de9-88a758bf7d42n%40googlegroups.com.

[Ihor]

- ". It's like racing a bullet train vs. a wheelchair. ;)”, still having some numbers would help:) The typical routine is to put a tube set the heater on, wait 2-3 minutes (which is still very short preheating and should be longer) and then collect the curves either within 1 second (FUtracer) or 1 minute (uTracer), (I still do not think it is 10-50 times faster), so selling the speed as a main advantage is obviously not for me. If someone has FUtracer let's just compare a few tunes and make small vides, 30 points, 10 curves with 1 or 2 tubes, so then we see real tests and not "advertisement statements" 

- it is possible to make those tests even with uTracer3+ just by buying that new PIC with new firmware, which is apparently fully compatible with uT3. It still costs reasonably (14 pounds) but at the same time the one on FUtracer website is $30. I am not sure from where they are shipped, but in order to get something in EU from UK or USA I would ahve to pay 21% extra and then 7 euros more as customs/post fees, which makes it ridiculous for the chip that costs 3-4 euros and for which firmware is available and easily flushable.

- the schematic is closed for this moment as it says on the website but the design is indeed 99% as in uTracer3. In some fields similarity of 10-20% is considered as plagiarism... I also have no idea how it is possible re-copyright and start selling something commercially which is already commercial and in public domain:) 

- rewriting the whole firmware sounds doable but still would be a very complex task. I think Ronald's PIC are already protected with fuses so the firmware cannot be extracted, but if not disassembling it and changing a couple of routines would be the easiest way in this case. The other firmware is available online, and probably can be used for comparison (if Ronald wants to go that way, in Europe all software is copyrighted by default and all the rights belong to the author). Another option is just to take both PICs and compare their behaviour for some "weird" cases which only Ronald knows should work or the way they should react, and see if those are also in the new firmware. Any slightest coincidence means that the firmware was not" written from scratch by an independent hired programmer who never saw the uTracer".

- the grid loupe has no problems with the range around 0 volts because it maps the region of -30,..,-40V to that 0V of Vg, so if the other range is calibrated and linear then the 0 range with gridloupe will be perfect, without any mods to that current mirror transistors or capacitors. Again the set and measured Vg around 0V with and without an actual tube are two different worlds, in one case the load to Vg voltage source is infinite, and in the second case the grid current flows which means that the loading of Vg is there and sagging is also there. There are no grid stoppers in A2 or AB2 class amplifiers.      

[nickb]

I'd like to refocus this thread as it seems to have strayed into the weeds.

UTmax

First, the uTmax GUI. I originally did this for my own needs and also as an exercise to help the linux guys. From the start I wanted to generate SPICE models.  This was all long before Derk came up with his. In due course I replaced my own models with Derk's as he is a much better mathematician than I'll ever be. Unashamedly it did (and does) what I want it to do.  I've taken a lot of input from others along the way and hope that it's useful to a wide range of users. The fact that doesn't take up your entire screen yet and you can get fullscreen and multiple plots are strong features.  If you don't like something it's open source so you can do as you please. Many have.

Firmware
This was written from scratch and was a considerable effort. I do ask a nominal fee just to cover my costs that I feel is very reasonable. Local taxes and / or fees are hardly my responsibility. Please don't complain about that to me.

The benefit of the dynamic pulse widths is to give more accurate initial voltage when below 50V or so. If you overshoot you have to discharge the caps and it all takes time. It just seemed better to me to try to get it as close to right from the get go.

The problem with the 19.5KHz heater supply is series inductance. The impedance of that inductance is proportional to the frequency so lowering the frequency to 2KHz virtually eliminates the problem. There remains a small drop due the Rds(on) of the FET and that is easily compensated for.

If some have found it faster then that is great but that was never a goal. Also, it's does put extra effort into getting the voltage accurate which might slow things. I've never done a comparison, not least as my original utracer chip left the earth some time as a result of an unfortunate probe slip.  I really would not expect big difference as the hardware is really the limiting factor.

The oversampling is a significant benefit as it's done very quickly right on the chip.

The 12 bit grid is a very nice thing to have.  The grid control is linear so errors are more problematic at low voltage end. Also, with the range expanded to 100V I think it's now almost a requirement. V3.03 will support for the 85V grid supply will be out soon. I tested that feature the other day and it work but I have to back test a bit before making a release.

From V3.00 it's upgradable in the field. You have to use utmax to do that. There were a few early adopters with who do not have that feature and if they wish to reach out to me I'm sure we can work something out. 

Pin Switch Board.

Calling this overkill seem like a discourteous thing to say. It's a tool and serves a purpose and a huge step up from manual patch plugs and switches.  I reckon to spend a hour a week changing switches on my tracer.  There is great frustration and time wasted when an error is made. At an hour saved per week it will pay for itself in no time. It was on the back burner for many months as I just couldn't source the chips. I just completed it's first test today and an looking forward to getting it into box and trying it out in earnest.  It won't be long now.

Utmax has been updated to support it but not yet released. All in good time.

I love the sound of the relays https://photos.app.goo.gl/SFaEGrhrpowKtRGE6

I did kick around all solid state methods of doing this but relays were simply the easy and reliable way to go.

Onward and upward..
Nick

[Davo]

Hey Ihor, since you have the skills to program, as a member of the utracer community, i rather see you making upgrades to the firmware as someone outside our community though!
Maybe you and Ronald can make a colab?

I must say that those nice curves are looking really awsome and those rainbow colors make’s ‘m so easy to separate,

But that relay matrix is the real boss lol

I don’t like messy patch cables, but it’s the best way to have less oscillations m, but it’s a challenge to make an oscillations free switch matrix.

A relay matrix is the fastest and fool proof system..

I’m a noob when it comes to programming. So pic’s systems are a no go for me, otherwise i had a project running in kicad haha

Op 14 jan. 2023 om 15:52 heeft 'nickb' via uTracer <utr...@googlegroups.com> het volgende geschreven:

 I'd like to refocus this thread as it seems to have strayed into the weeds.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/8c3bd3bd-d969-4a7e-b734-cc6b47252aedn%40googlegroups.com.

[Ihor]

It takes quite some time to improve the firmware and especially to rewrite it from scratch especially because all the acquisition routines and steps are not described in Ronald's blog, so a lot of respect for that. The biggest problem would be to maintain it over the years (in general any software) taking into account that for some of use those are just really hobby projects, not connected to own business or work directly. For Ronald and uTMax developer that's a different story. 

The switching board is also cool, but indeed, we do not know anything about the oscillations, and if there are no oscillations with that board, why would anyone need FUTracer (assuming you can plug a new PIC into uTracer3, so probably only for new users). Personally I like simplicity and with so many relays as on the board, and sourcing parts from China to go cheap will sooner or later lead to failures in one of those. Then lots of precautions and self-tests should be implemented already on the board because otherwise the uTracer will be damaged. Are there any estimates of the costs, or at least which relays are used? 

For the information I also attached a photo of a working relay board which was proposed in 2018, so if someone is interested they can find the details in Ronald's Gallery and ask the authors for the PCB. 

[Davo]

Hi Nick!

Long time no speak, i know you haven’t any intentions to make money on it

All the effort you put into uTMax is just for the love of the uTracer man

You are dedicated same as Ihor & Martin you guys made the uTracer way more accessible for any computer, tabloid or phone..

Ronald is the man when it comes to engineering but he’s not a programmer tho… He can do some basic stuff, but you guys took it to next level!

All the stuff should be all open source, then stuff is getting really awsome…

I really don’t know how it’s fixed as it comes to laws, rights etc.

It would be a real bummer if some asshole rips & changes some coding and starting to make big money with the open source project..

You guys should work together and make 1 gui that fit’s all. Dunno if that’s possible?

I see different projects which all have strong features…

I can understand also that everyone wants sail on their own ship

This is just a thought

Op 14 jan. 2023 om 19:24 heeft Davo <djda...@gmail.com> het volgende geschreven:



[Davo]

I have seen that relatboard on Ronald’s site, a long time ago, but then i allready gave it up, cuz i’m a noob programmer lol

The relay board is only magic when you select a tube and all connections & settings loads from a preset..

The relay dusn’t have to be really expensive i guess.. 

as long you switch when there’s no or low vdc, switching a few hundreds vdc will destroy them fast, a diode and maybe a capacitor for each relay will do the trick

China boards aren’t that bad, you can use thicker copper layers, etc but will cost more.. and maybe use a relay socket, form quick maintenance, desolder a relay is a pain without a heater is a pain with double side boards, and could damage a board quickly 

Op 14 jan. 2023 om 20:02 heeft Davo <djda...@gmail.com> het volgende geschreven:

Hi Nick!

[Davo]

A relay board with more sockets or a smaller board with 9 outputs and use the standard socket wiring as in the manual 

Verstuurd vanaf mijn iPhone

Op 14 jan. 2023 om 20:02 heeft Davo <djda...@gmail.com> het volgende geschreven:

Hi Nick!

[Davo]

This relay board is quite cool but it’s the same as a switch,

But if you can make a preset in the software and use the matrix as switch then you have the best of both worlds

Verstuurd vanaf mijn iPhone

Op 14 jan. 2023 om 20:02 heeft Davo <djda...@gmail.com> het volgende geschreven:

Hi Nick!

[nickb]

The pin switch board can be controlled from utmax so all the pin changes happen automatically.  You could easily have it controlled from something else that had a database of pinouts but the integration with the GUI is a nice way to work.  For testing I used Python and that could be expanded to read a pinout database.

I choose two types of relay, one for current handling for the heaters and the rest for high voltage.  The board had to be four layer to get the necessary clearance to handle the voltage even though the density is low. I put ferrites and damping on all I/O as a pre-emptive strike on oscillations so there should not be a problem there.  I had ten boards made with VAT and shipping there were about £30 each.  I bet a could get 20 for almost the same money.  I'll do a BOM and cost i out when I get some time.

Nick

[Ihor]

I was thinking about Chinese quality relays and not the board :) there are 54-56 of them in those two projects, can be also costly if one would go for Japanese brands I assume. 

About the uTMax, I was and still not sure if those Calibration values that one has to use are now exactly the same as they are in the original uTracer windows software. A couple of years ago they were different for sure, because I tested that and now all the dialogs and names look like they are the same. If they are the same, I took a look at the source-code and all the conversion of set Va, Vg, Vs,... values to HEX values for the uTracer (and back) are different compared to Ronald's routines. Meaning that if one sets Va=100V it will be slightly different HEX code that is sent to the PIC. Sometimes those values will be coinciding of course. So, one will get slightly different measurements with uTMax and other software (for example the original windows software or uTracerJS). 

For example uTMax has conversions for Va Vs and Vf as 

x= (uint)(1023.0/5.0*( (v + adc_real.Vsu -Vdi +Vdar)/adc_scale.Va / calData.VaVal));

x = (uint)(1023.0/5.0*((v + adc_real.Vsu -Vdi +Vdar)/ adc_scale.Vs / calData.VsVal));

f  = 1023.0*(VfNow/adc_real.Vsu)*(VfNow/adc_real.Vsu); 

and Ronald as 

dblTemp = (1024# / 5#) * (dblAnodeR1 / (dblAnodeR1 + dblAnodeR2)) * _(dblMeasMX(I, intVa) + dblVsupSystem) * dblCalVar1

dblTemp = (1024# / 5#) * (dblScreenR1 / (dblScreenR1 + dblScreenR2)) * _(dblMeasMX(I, intVs) + dblVsupSystem) * dblCalVar2

dblTemp = 1024# * (dblMeasMX(I, intVf) * dblMeasMX(I, intVf)) / (dblVh * dblVh)

The important part here is that ration 1024/5 and 1023/5 which are different and never will be coinciding, independently what is in the rest of those variables. For example the values for the heater are for sure different, for 6.3V translates differently (and sends to the PIC) by uTMax.  

[Ihor]

"You could easily have it controlled from something else that had a database of pinouts but the integration with the GUI is a nice way to work." 

I think it is good to have it controlled by uTMax but probably to make it even more independent and useful for other uTracer users who stick to the windows software, it would be nice to have a microcontroller there with a small screen and buttons to chose the tube type. So then it would be a self-contained tube switcher with it is own updatable database of tubes. One could use a raspberry pi with a touchscreen or so for example. 

[D.A.R Achterberg]

Ihor didn’t i said that i was a noob programmer lol

Maaan that coding is like abracadabra to me haha

I see differences in the code, but what it controls??

But only Nick can tell you that…

Might be wise if you guys were at least at the same page with that though!

If people wanna use both programs or whatever that stuff gotta be same, i guess

I’m not a programmer, but that’s not rocket science…

Op 14 jan. 2023 om 20:40 heeft Ihor <iste...@gmail.com> het volgende geschreven:

I was thinking about Chinese quality relays and not the board :) there are 54-56 of them in those two projects, can be also costly if one would go for Japanese brands I assume. 

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/d7bf6f27-d3d0-4cfd-8bb6-8e84820dd115n%40googlegroups.com.

[nickb]

There are many ways to cook an egg :) If someone wants their own recipe then this is what you need to know.

The board is controlled by a a 12 character string "53abcdefgh" where a-h are the uTracer connections and the position is pins 1-10.

a-h are:

0=n/c

1= anode

2=screen

3= grid

4=cathode

5= heater A

6= heater B

I've been using miniature switches which are rated at 30K operations and not had an issue. The relays are 100K.  There may be a caveat with relays in that there are supposed to switch non zero currents to get that life.

The difference is calibration values is a non issue. You run the calibration once which take two minutes and never have to do it again. Done.

Nick

[Ihor]

The board is controlled by a a 12 character string "53abcdefgh" where a-h are the uTracer connections and the position is pins 1-10.

Nice, that would be enough to create your own addons that can control the board. 

 

I've been using miniature switches which are rated at 30K operations and not had an issue. The relays are 100K.  There may be a caveat with relays in that there are supposed to switch non zero currents to get that life.

I am really curious is that board will cause the oscillation during the real measurements or not:) In practice, it is a huge array of "wires" linking 6 pins from uTrcaer to 10 output pins, within a small PCB space, lots of capacity between the traces, especially for those short (in time, so high frequency) measurement pulses, and the disconnected traces are actually wave-guides at those frequencies.....potentially with lots of overshoots and interference, especially if the measurements are now acquired with the claimed super-speed. It may work with switching LEDs but if THAT board does not cause oscillations then all the stories about oscillations with uTracer is just a myth :) with uTracer the oscillations are still due to user-specific implementation of the connection to the tube sockets and not related to the main PCB, so inserting such switching board in between might affect stability of the setup, definitely not improving it.  

It might be useful to scan the database of the 8- and 9- pin tubes and see which pins are used for which purposes. It can happen that because of some historical reasons some pins are never used as heater pins, or for example Anode is never on some specific pin. That would reduce the number of relays quite a bit, taking into account that for the general case we have 6 (utracer pins) x 9 (tube pins) case. 

 

The difference is calibration values is a non issue. You run the calibration once which take two minutes and never have to do it again. Done.

It was not clear if those calibration values can be just copied from the uTracer's software or not. Apparently one has to do a separate calibration. Indeed, in that case the conversion formulas can be designed from scratch and compensated with a different calibration, and the measurements will be quite similar. The fact in this case is that one will never get exactly the same measurements for the set Va, Vg and so on (not just because of the tube variation/state in time) but just because the conversion functions will never be the same mathematically (and cannot be mate the same with any calibration procedure because of the form of functions). Those variations most likely will be minor, but it is not clear why not to implement the conversion like it was done by Ronald (which is openly available) but do it in a different way, but even different in a very peculiar way with going for 1024/5 instead of 1023/5 and some other uniqueness :)    

[Davo]

Hi Nick!

You are gold man!

Just like Ihor & Martin Manning.

You guys are the next gen uTracer engineers. Martin is good in making mods

Ihor made a gui to make uTracer available at any device even a phone or raspberry!

You guys should stick your heads together to make something even bigger as you both made…

The availability of Ihor’s gui & your version with all the nice addons which make tge user experience a real joy tho!

Even that nice rainbow curves! Tabs exporting data & the relay matrix!

I was waiting for that sooo long…

I’m a noob programmer, i can handle electronics but coding pic’s, software etc.

Is hard for me, maybe in the future…

A few years ago I asked Ronald how to code that, but he didn’t respond to that question, i guess many people asked him the same question lol

He’s a true electronic engineer, but a average programmer. If i look at his website the guy is stuck at html.

Browsing at his website is a nightmare lol

But that’s where you guy’s come in!

You guy’s proved ur skills

We as the uTracer family can feel ourselves as fortunate due all the improvements!

This ain’t about the money, but about the love for tubes and sharing knowledge the get better as a community tho…

Hey Nick with ur relay matrix ur just a step closer at the Roehren Tester

I Dunno if you know it?  But i guess so

This tester has 1 asset that as far as i know of all, can recognize tubes!

We all know that the markings on a tube can be so fragile, and having tubes without markings can be a real pain!

Some tubes like rectifyers, 12ax7, EL34 tubes etc are easy to recognize.

The Roehren Tester is quite an unique device and quite expensive & difficult to build for the most of the people

Op 15 jan. 2023 om 10:12 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/984a450f-4808-4e63-baa1-27552e377547n%40googlegroups.com.

[Ihor]

" It's like racing a bullet train vs. a wheelchair. ;)”  

So the actual speedup in acquisition of FUtracer compared to uTracer is ..... 2 times ..... Okay :))) As I thought, it is just lots of sales statements all over the place without any convincing side-by-side comparisons.

Here is the acquisition speed video

https://www.youtube.com/watch?v=dle7xnOKeK4

and some indications of the prices:

https://www.eevblog.com/forum/projects/project-vacuum-tube-curve-tracer-(the-futracer)/25/

[D.A.R Achterberg]

Wel the curves look nice! But the speed isn’t that important to me..

I can’t find any decent info on the pic & shipping costs..

If the pic can easily be plugged in the uTracer board, it might be worth it.

I’m really happy with Nick’s matrix option that’s the real gold for me!

The pic might be worth it, but that matrix rocks

If somehow this function can be adapted 👌🏻

 

The possibility to detect unknown tubes.. that what i call I awsome feature, we all know the tubes with no markings on it, pulling out a tube the wrong way can ruin your whole day

You should collab with Nick Ihor 1 universal gui that covers all systems, all all the nice features combined or at least all the cal settings the same

you guys are both outstanding in what you do, so a collab woud be even better..

But hey just a thought lol

Op 16 jan. 2023 om 16:11 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/b4b7ffd7-cc0a-44d2-bd4b-b18fb5ed6a08n%40googlegroups.com.

[Ihor]

I can’t find any decent info on the pic & shipping costs..

If the pic can easily be plugged in the uTracer board, it might be worth it.

Yes, it is a direct replacement. 

 

The PIC via uTmax website is 14.50 pounds: http://bmamps.com/store/index.php?main_page=product_info&cPath=109&products_id=440

The PIC via FUTracer website (exactly the same PIC) is $30

The FUTracer Kit (which is just 3 PCBs without any components and the PIC) is $200

The FUTracer Kit (3 PCBs with just a couple of SMD components soldered and the PIC) is $250

The assembled FUtracer is 996euros: https://reverb.com/item/47954748-futracer-tube-tester-and-curve-tracer-2021 

but now discounted to $619 https://www.reverb-discount.com/product/futracer-tube-tester-and-curve-tracer-2021-tubes.html

In all the cases, shipping costs and import taxes should be added.

 

[Rudie]

The Pic price shows as £14.50 with a shipping cost (to Ireland) of £12.95.

--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/04245a75-e217-4444-97f2-b76044f1cce3n%40googlegroups.com.

[Davo]

So the pic can run in a utracer3+ also

Then 14pounds ex shipping might be interesting to order… 

Op 16 jan. 2023 om 22:01 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--

[nickb]

Hi Davo,

You enthusiasim and kind words are appreciated. Thanks.

Auto detection is a nice idea.  I guess the switch board it would go some way to enabling tube detection, but would need some careful thought. I think that the heater voltage would need to be known. I say this since  to detect the other electrodes requires a hot heater and you would not want to burn out the tube under test. I've never looked at the RhoTest tester in any detail. There might be some hints if I did.

I've been wiring the new switch board in to an enclosure today. I also tested with the latest uTmax to check the control of switching which was sucessful. The BOM cost is coming out around UK£70. That assumes I buy enough parts for ten boards. There is a snag in that the relay driver chips are scarce right now.

My background is as a hardware engineer, designing boards for high speed digital signal processing and coding big FPGAs. I've been a modder of uTracer since it first came out. I had one of the earliest 300V 200mA guys and almost immediately learned that I needed 400V, 400mA and a -57V grid range a long time before Ronald did. So I designed some mods and that was the main motivation to write a GUI as I needed one to support my creation. That was all a vey long time ago and the exact timeline is a bit hazy now.

-Nick

[nickb]

Hi D.A.R,

It is not necessary for them to be the same.  They employ different calibration files in different places so you can happily install both and there is no conflict.  The only thing you could do if they were identical would be to copy the values from one calibration file to the other. I think anything involving manual effort like that stands a good chance of going wrong. I would therefore advocate doing a calibration anyway. With utmax the calibration is very quick and easy to do.

Historically the difference arises from when I wrote it back around 2010. I don't recall the uTracer being as well documented then as it is now and so I  worked it out by looking at the schematic.

--Nick

[Davo]

Hi Nick!

Ur welcome! It’s really cool to see all the improvements  you have created…

And it goes on & on…

The link i posted gave a quite good insight about the structure behind it..

So ur saying that the relay matrix could be used as a tool to recognize tubes also

The german tester is a matrix also…

That’s sum serious stuff to engineer 

Almost insane..

70pounds for the BOM sounds fair enough tho, count me in!

Btw D.A.R is my email also lol sometimes my phone switches my e-mailaccount..

Op 17 jan. 2023 om 19:04 heeft 'nickb' via uTracer <utr...@googlegroups.com> het volgende geschreven:



To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/47030c8c-3ca0-44b2-a896-6d231056fd8fn%40googlegroups.com.

[Big Josh]

The speed is significantly better, whether you care about that or not (and you can get pretty close using the utracer with Nick's firmware chip). For me, the biggest problems with the utracer were the terrible PCB quality, and the issues with the firmware. The FUtracer is significantly better on both accounts. I'm sure you'll see some comparisons as I'm not the only one who went from utracer to FUtracer.

Not sure where you got those prices for the completed versions, but they're not accurate. Also, the 2nd one is a scam site, not actually selling that futracer pictured.

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/b4b7ffd7-cc0a-44d2-bd4b-b18fb5ed6a08n%40googlegroups.com.

[Ihor]

I have no reasons not to trust you but on this website, the "developer" of FUtracer says the speedup is 2 times and shows the youtube video where it is clearly shows how "fast" FUtracer is:

https://www.eevblog.com/forum/projects/project-vacuum-tube-curve-tracer-(the-futracer)/25/

Nick's firmware and FUtracer firmware are EXACTLY the same, I checked it, byte per byte. Thus the acquisition speed of uTracer with Nick's chip and FUtracer will be EXACTLY (and not "pretty close") the same, because the hardware does not define the speed, and even more, the hardware is the same, because the changes are in the grid voltage part and not in the HV converters, those are just copy-pasted.  

Now, going back to Nick's firmware with the claimed 12bit Vg resolution. PIC16F18875 only has 10bit PWMs on board, so it is not possible to have genuine hardware 12bit PWM there. It is possible to "flip" that digital pin with some tricks to create pseudo 12bit resolution, but that is quite tricky because it is very dependent on coding of the timings, and in such cases any interrupt routine for example will break that perfectness of such 12bit pseudo PWM. 

On Wednesday, 18 January 2023 at 21:34:55 UTC+1 Josh wrote:

The speed is significantly better, whether you care about that or not (and you can get pretty close using the utracer with Nick's firmware chip). For me, the biggest problems with the utracer were the terrible PCB quality, and the issues with the firmware. The FUtracer is significantly better on both accounts. I'm sure you'll see some comparisons as I'm not the only one who went from utracer to FUtracer.

 

[Martin Manning]

Josh wrote: "For me, the biggest problems with the utracer were the terrible PCB quality, ..."

I don't understand this. My uTracer is a fairly early one, purchased in August 2013. It has a good quality two-sided board, with solder mask and silkscreen. I've removed and replaced dozens of components over nine years with no problems. It has a number of parts hanging off the bottom, but with the uT3+ board refresh, that won't be necessary for new builds.

[Big Josh]

Yes, they are exactly the same firmware. BTW- that video isn't claiming 2x the speed of utracer. It's 2x the speed of the previous version of Nick's firmware, which was already faster than the utracer IME.

I don't know about the bit depth difference with the chip, but I do know the hardware change and firmware made for a huge improvement in grid voltages between 0 and -1. Do a measured test in 10 steps between 0 and -1 and see what kind of results you get.

--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/06cc8905-9fc5-42c1-8c3a-ec0517636a3bn%40googlegroups.com.

[Big Josh]

Martin, my utracer board had traces lifted, and the vias weren't plated through. It was absolute junk. I complained about it, and gave examples of the issues, and showed him better examples of how it should look and all he did was get offended that I didn't think his creation was perfect. I had to explain to him what plated through hole vias were. 

--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/f3ac0482-dc4f-4800-85ac-c8fae23f37edn%40googlegroups.com.

[Martin Manning]

Josh wrote: Martin, my utracer board had traces lifted, and the vias weren't plated through. It was absolute junk."

Sounds like your experience predates mine as far as the PCB goes. I mentioned above that an issue was discovered with the grid bias voltage sagging, and the fix for that, ca. October 2015. There was also an earlier issue I brought to Ronalds attention in late 2013, where the anode and screen voltages were not being set correctly during the QT derivative runs. That was solved with the addition of a couple of caps to speed up switching of the anode and screen supply discharge transistors. These two hardware mods improved the performance at low Vg settings significantly, so I would say that the uT3 hardware has been performing quite well since late 2015.

[Davo]

Ronalds board is a good quality pcb…

You always have to watch out with double sided board imo

They can sometimes be a pain to desolder, it’s all about knowledge & using the right tools…

Ronalds board is well made, i don’t see the benefits to be honest…

Firmware upgrades etc that’s something else.

But hey, it’s a free world, if making a better board, makes you happy, i don’t mind though…

Grtz

Dave 

Op 19 jan. 2023 om 19:04 heeft Martin Manning <mman...@fuse.net> het volgende geschreven:

Josh wrote: Martin, my utracer board had traces lifted, and the vias weren't plated through. It was absolute junk."

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/12e0f897-d4ba-41ab-843c-823d024f2826n%40googlegroups.com.

[Big Josh]

It sounds like your experience was better than mine, that's not too surprising. However, running a test for accuracy is the only way to truly judge the grid accuracy now. I don't doubt it's improved over time.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/12e0f897-d4ba-41ab-843c-823d024f2826n%40googlegroups.com.

[Ihor]

Just to demystify and summarise a few things for those who are not into the field. 

So far we figured out that the FUTracer is better because 1) it has better via's on on the PCB, 2) super-accurate with 12bit grid resolution 3) super-fast. 

First of all, so far I have not seen any confirmation for any of those claims, compared to Ronald's blog where each step, each experiment or trick is described, tested and pictured. No disrespect to Josh and Nick, it is just information for others. 

1) the board looks nice and of really good quality. I myself had no problems with the standard one, The board layout has nothing to with the claim that there are no more oscillations. It is an extension board and probably tricks in the firmware with the oversampling that might make it less prone to oscillations. In any case, oscillations are only due to poor implementation of wiring to the tube by a user. For many of us it never happened. I have not resoldered anything but I would not expect any problems with the quality of the traces. I deal a lot with vintage computers, and fro example many boards on Sinclair ZX Spectrum 48k were made from so called SRBP materials....in simple words it is a laminated paper/plastic, two sided. Those are the worst you can probably imagine but those 40 year boards survive desoldering of large chips multiple times without problems. It is a matter of skills and tools. 

2) 

a)the mentioned table with measured Vg around 0,..,-1 has nothing to do with the 12bit resolution, the people should not be confused. Those are independent things! The nicer matching around zero is because of the compensation of one opamp with a potentiomenter, which can be easily added to existing uTrcaer3.  

b) the 12 bit resolution is not a real hardware thing but a software implementation on the PIC. That PIC16F18875 has only 10bit PWMs (I worked with them as well). Even more, the 10 or 12bit is the resolution in the setting of PWM width and not the voltage as in a typical DAC. The PWM is translated to the grid voltage with extra 2 opamps. So, 12 bit is a nice creative move of Nick but just because one can, it is not shown in practice that it actually does anything useful. The easiest experiment to prove that is a 10 minute work: just put a swipe for Vg from -2V to -1V for example with the step of 0.005V and with an oscilloscope show us that uTracer produces steps every 0.05V and FUtracer's steps are 0.0125V. So each step of uTracer is magically split into 4 ... And try it with and without a tube, I would not expect any visible differences in practice (with a tube), but it still would be interesting to see what it actually does. 

  

c) there was a claim that 12 bit is necessary because FUtracer will go to -85V Vg instead of -50V of uTracers. in the region of -40 or -50V the accuracy is still 0.05V, ... is it 0.1% already.... who really cares if the measurements were done with -30V, -35, -40 or -30.01V, -35.01V, -40.01V, .. those grid voltages will be way off in practice due to other practical reasons. We are not even talking here about the PWM'ed heating which is also claimed to be improved, but for sure if used, will "crash" all those improvements in 12bit grid settings because of variation of emission in time.  

d) all the arguments of high resolution and precision are useful only in the range of 0,...,-0.5V of Vg...but how many people need that region (it is not even a question, because it is probably 1% of users). If we talk about audio, there are just a couple of tubes that are useful in that range. 

3) we are still waiting for videos with speed comparison of uTracer and FUTracer. At least two people on this forum have them. So measuring any tube of your choice and acquiring 10 curves with 30 points in the full range of 400V would immediately clarify and confirm all the claims. If we can only do it with FUTracre, amy of us can repeat that experiment with our uTracers.  

I think all this will stimulate nice discussions and things will only get improved for both camps!:)

[Ihor]

I found some a bit of time to test my uTracer setup and also to try one way of accelerating the acquisition. As you can see my setup is full of long wires and no oscillations. 

I tried KT88 with 5 grid voltages and 20 points per curve and it takes about 1 minute to acquire, see the video:

https://photos.app.goo.gl/9EHrP1BDBMvzyKvg8

Then I modified uTracerJS so the curves are actually not acquired sequentially one by one but in parallel, so all grid voltages are applied for the same Va, every step of Va, in such a way the boost converters do not have to be charged and discharged all the time and a lot of time is saved. Again vide where the speed up is about 2 times:

https://photos.app.goo.gl/8BQmmUMqj4roFgZF6

The visualization is different because the software right now is not adopted for this, so every point is just colored differently, but you get the picture.

And there 5 curves with 50 points per curve acquired in such a new way within 1.5 minute:

https://photos.app.goo.gl/BJ3nr44N9YbRv9Y9A

So, now we can easily acquire 10 or 100 curves in parallel, more or less within the same time span as one or two curves sequentially, as it is implemented now in all our softwares. If I find time I might implement that feature properly and release it in the future, but it will require quite some changes in my case. In such a way FUtracer can be made even faster, now we are talking about bullettrains :)    

[nickb]

That's great news.  The same thought about the method had occured to me a long time ago but I've always been too busy to try it.  Now i have a bit more incentive ;)

--Nick

[Davo]

Yeah that was sum straight-up info Ihor!

With real proof, nice!

it’s sometimes hard for me to understand that whole programming thing. But if i think i building blocks i can understand.

The positive thing is that only we can get real bullet trains to join forces!

This is something wel all love.

Better 1 big community instead of 10 different tracers community’s based on the same concept.. 

To bad ronald dusn’t has a forum…

Google groups is okay better than nothing, a forum would make things much more easy.. But not on diyaudio or other mainstream forums… where there are so many users and you will have to thru 70 pages of crap & irrelevant info

Op 20 jan. 2023 om 10:41 heeft Ihor <iste...@gmail.com> het volgende geschreven:

I found some a bit of time to test my uTracer setup and also to try one way of accelerating the acquisition. As you can see my setup is full of long wires and no oscillations. 

<PXL_20230120_090304594.jpg>

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/a98f74c9-29d1-4854-a8e8-bc2b0a799236n%40googlegroups.com.

<PXL_20230120_090304594.jpg>

[Ihor]

Yes, looks like a working approach:) I just released it in the last version of uTracerJS but here are some more serious tests:

20 points on the curve and 11 Vg values. 

uTracerJS Old way, acquisition time 2:24 min 

http://files.smal.ws/v/uLystU

 

uTracer New way, acquisition time 50 sec

http://files.smal.ws/v/gvDJPX

and for fair comparison Ronalds software, acqusition time 2:40 

http://files.smal.ws/v/CTuG3q

There is also some small lag in the webinterface, but it is jsut visual slowness, because the browser has problem with dealing with lots of small "messages" and updating screen not with one points per time as before but with 11. Probably I will look into that, but the speeds up are there. 

[Davo]

I noticed 1 problem with this method…

I took 2 screens at the end, and i noticed

Something about the inconsistency 

with the parallel approach you’ll see it on all curves.. 

butt overall very impressive Ihor nice!

Op 20 jan. 2023 om 13:46 heeft Ihor <iste...@gmail.com> het volgende geschreven:

Yes, looks like a working approach:) I just released it in the last version of uTracerJS but here are some more serious tests:

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/f20b7331-56e1-49a3-bb37-096bb11b9b1bn%40googlegroups.com.

[Ihor]

Indeed, I saw that as well. It looks that it is there, the difference, and it is in hardware I believe. So apparently if one sweeps nicely and slowly then pumping Va every time has a smoothing effect. If for a given Va one acquires multiple curves with Vg, Va probably is not controlled well, sagging and not re-pumping for such short measurement intervals. Probably oversampling option can be used, which is available already. Another option, if it is indeed stays problematic, one can use that Parallel Acquisition as a "quick look" to your tube curves and then after figuring out the parameters, take nice smooth curves with the Sequential approach:) 

Ronald's curves, for comparison, also do not look too smooth but also not as non-smooth as with the parallel acquisition, something in between I would say. 

 

[Martin Manning]

I wonder if the LPF in the grid bias circuit is causing this? Its settling time is deliberately slowed down to avoid the ground reference voltage rise noted above.

[D.A.R Achterberg]

My first thought it was hardware related..

I guess it must be somewhere in the resistors, some can be critical to get spot-on readings… maybe 0.5 0.1% ones, or hand picked from your own stash which can take some time haha

I made Ronald’s transistor addon and it was quite a pain to the scale index spot on haha

Temp humidity etc can mess it up also

I noticed that the next day i wanted to complete the mod.

Op 20 jan. 2023 om 14:11 heeft Ihor <iste...@gmail.com> het volgende geschreven:

Indeed, I saw that as well. It looks that it is there, the difference, and it is in hardware I believe. So apparently if one sweeps nicely and slowly then pumping Va every time has a smoothing effect. If for a given Va one acquires multiple curves with Vg, Va probably is not controlled well, sagging and not re-pumping for such short measurement intervals. Probably oversampling option can be used, which is available already. Another option, if it is indeed stays problematic, one can use that Parallel Acquisition as a "quick look" to your tube curves and then after figuring out the parameters, take nice smooth curves with the Sequential approach:) 

Ronald's curves, for comparison, also do not look too smooth but also not as non-smooth as with the parallel acquisition, something in between I would say. 

 

<utCurves.jpg>

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/322495d7-4bc8-4f3d-b22d-d61209042282n%40googlegroups.com.

<utCurves.jpg>

[Martin Manning]

To be clear, the non-smooth character you are referring to is the wave in the grid curves going from low to high anode voltage?

[Ihor]

Probably it will be a good idea for the future to compare, if more users will try and share their experiences. Theoretically there is nothing special in asking uTracer to acquire data in both ways and it should the job right. I do not think that the grig is swept too fast, but that still have to be explored. The "wave" in the horizontal direction for each point (so left and right from the set Va) is normal, Va is never accurate and we can see (actually in both cases) that those waves are present. The wave in the vertical direction is variation of Ia and in one case it is smoother visually and according to the theoretical models or averaging of real curves that curve will be smooth, so anything which is not smooth is less accurate :) 

I put those plots in one picture and I also see that the curves in one case are also a bit lower (especially for lower Vg). This is definitely not a problem of software but just hardware plus variation of the tube, plus I am not sure what. Just for  test I will to more repetitive measurements, but I am almost sure that there is no difference in two ways of acquisition and what we see, the difference, is just a random occurrence and could be totally opposite next time. Otherwise there should be consistent overshoot in Ia while swiping Vg from -15 to -5 in this case.  

 

[Tim Goldstein]

In the general topic of making traces faster the example on Ronald's site of adding a second set of inductors to the bottom of the board made a marked difference. Didn't measure the change in time but it was very apparent. Of course this is using the method of charing and discharging for each trace and not the new parallel method. Guessing with the parallel method it would have less effect from not charging the caps as many times. But it sure does charge the caps faster and is cheap and easy to do.

[Ihor]

On Friday, 20 January 2023 at 16:38:14 UTC+1 ti...@wiserltd.com wrote:

In the general topic of making traces faster the example on Ronald's site of adding a second set of inductors to the bottom of the board made a marked difference. Didn't measure the change in time but it was very apparent. Of course this is using the method of charing and discharging for each trace and not the new parallel method. Guessing with the parallel method it would have less effect from not charging 

In my uTracer I have that mode since beginning (I mean I implemented it) and I know Martin as well, so many of us are using it and all the plots that I present are with this mod, that's good to mention.   

[Martin Manning]

Here's my version, with my hardware. The smoothness is very similar, but the level is shifted just a bit, and again it's the parallel that is lower on the anode current, and the screen current is a little higher..

[Ihor]

I think we should move to another thread, but right now I am at the conclusion that the differences just because of reality and not software:))

I did more experiments and can say that the previous result is probably because the tube was not heated well between the experiment. Right now I did 4 experiments, first fast acquisition (fast1) then slow (slow1) and then after 5 minutes fast2 and slow2. As you can see from the plots there is almost no difference between fast1 and slow1 (only in a small region, left top plot) and there is also very small difference between fast2 and slow2 (top right) and the same time there is a huge difference between (fast1 and fast2) and (slow 1 and slow2) bottom row, and even in weird direction because Ia's are not just all higher or just lower but nonlinearly higher for higher values and lower for lower. So it is just life :) there is nothing in software theoretically, so let's see if we manage to point it down to hardware or is it just measurement noise and conditions.  

[Martin Manning]

PS: The acquisition time is not that much different, 2:16 for parallel and 2:43 for sequential, a 16.5% improvement.

[Martin Manning]

In a previous test I took multiple readings for a quick test and found that the uTracer hardware takes around 40 minutes to stabilize.

[Ihor]

"PS: The acquisition time is not that much different, 2:16 for parallel and 2:43 for sequential, a 16.5% improvement."

It probably depends on the currents that are measured, it looks like in your case those are way higher and probably boost converters working more to compensate, so they are still pulsing a lot. The improvements in terms of time is when one wants to take larger amounts of curves with not so many points per curve. If there are many points but only a few curves, then the improvements will not be large I suppose. In my case as you saw in the videos the improvements are at least 3 times (2:40 vs 50 seconds), so it just dependent on settings but that's to be expected. Even the standard acquisition does not take a fixed time per point, sometimes it is slower sometimes it is faster, it is also clearly visible from the video that I attached before (for example the measurements using Ronald's software). 

[Big Josh]

Agreed, nothing wrong with fruitful discussions. 

It's not just the vias, the boards in general are higher quality. Keep in mind though, that these weren't made to convert anybody from utracer, only to have (IMO) a better option. Some people have converted, and they've commented on the difference in quality. To each their own.

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/d2833b6c-e80f-4b09-986c-319238577cd5n%40googlegroups.com.

[nickb]

I'd like to respond to some points, viz

"the 12 bit resolution is not a real hardware thing but a software implementation on the PIC. That PIC16F18875 has only 10bit PWMs (I worked with them as well). Even more, the 10 or 12bit is the resolution in the setting of PWM width and not the voltage as in a typical DAC. The PWM is translated to the grid voltage with extra 2 opamps. So, 12 bit is a nice creative move of Nick but just because one can, it is not shown in practice that it actually does anything useful. The easiest experiment to prove that is a 10 minute work: just put a swipe for Vg from -2V to -1V for example with the step of 0.005V and with an oscilloscope show us that uTracer produces steps every 0.05V and FUtracer's steps are 0.0125V. So each step of uTracer is magically split into 4 ... And try it with and without a tube, I would not expect any visible differences in practice (with a tube), but it still would be interesting to see what it actually does. "

and

" PIC16F18875 only has 10bit PWMs on board, so it is not possible to have genuine hardware 12bit PWM there. It is possible to "flip" that digital pin with some tricks to create pseudo 12bit resolution, but that is quite tricky because it is very dependent on coding of the timings, and in such cases any interrupt routine for example will break that perfectness of such 12bit pseudo PWM. "

Using the term "psuedo" is an insinuation that the 12 bit feature is somehow not 12 bit. Like "pseudo-scientific" which means "not scientific" or pseudo random which is not random at all. I find that objectionable.

Consider the DAC's used in some 16 bit CD players.  They actually have 1 bit DACs. If you measure the performance, you get 16 (or more) bits after filtering that is applied. No one calls that pseudo.

The chip has a 10 bit PWM feature.  Its output is either +5V  or 0V i.e. it's also has just one bit on it's output. The width of its output is modulated and filtered to give you 10 bits of resolution. Again not called pseudo. Code 0 represents 0V, 1 is 5/1023V and so on to code 1023 which is 5V.  The crucial thing is that if you measure it you see steps of 5/1024V

The 12 bit is mapped so that we have in effect intermediate codes viz. 0, 0.25, 0.5, 0.75, and so on up to the top end of 1023, 1023.25, 1023,5 and 1023.75.  There is a quirk here in that as code 1023 is 5V we cannot get higher than that so anything above 1023 gets limited to 5V.  There is no problem here as that is the highest permitted grid voltage anyway.  If you connect a meter to the output and measure you will see steps of 5/4096V i.e. 12 bit. If none of the rest of the examples are labelled "pseudo" then there is not reason to call this pseudo either.  

Yes it's done in software and care was taken to get the timing right. Why is that being slanted in some way as inferior? If the algorithm was moved off into a piece of hardware does that suddenly make it better? Of course not.

Here are plots of 12 bit vs 10 bit taken on my tracer with a 60V grid range.

So, given that it does work, and rather well, let's move on to the statement that it isn't needed. First I should point out that the voltage numbers that were quoted are misleading as it's the grid voltage that matters. With a -100V range ( which is what we are talking about) each 10 bit step is a tad shy of 0.1V, not 0.005V mentioned which is just the filter output . Similarly misleading are the incorrect figures of 30.01V vs 30.00V. The correct numbers would be 30.0V and 30.1V. I measure a lot of tubes around 1V and just because some others don't doesn't mean it's not important or useful. BTW, the "1%" number quoted is just plucked out of thin air with no discernible justification. A delta 0.1V is 10%, the same as 1 LSB if you are using 10bits. With correct rounding it will be +/-5%. But it's much worse than that as when you are calculating gm as in a quick test it's the ratio of desired delta voltage to actual delta voltage that matters. Now that is a +/-50% error. I'm rather of the opinion that errors of such magnitude matter a great deal.

Here are a couple of charts comparing the errors on 10 vs 12bit as the delta V is changed around 1V grid voltage. Still plenty if room for improvement even at 12 bit.

"there was a claim that 12 bit is necessary because FUtracer will go to -85V Vg instead of -50V of uTracers. in the region of -40 or -50V the accuracy is still 0.05V, ... is it 0.1% already.... who really cares if the measurements were done with -30V, -35, -40 or -30.01V, -35.01V, -40.01V, .. those grid voltages will be way off in practice due to other practical reasons. We are not even talking here about the PWM'ed heating which is also claimed to be improved, but for sure if used, will "crash" all those improvements in 12bit grid settings because of variation of emission in time."

This is a straw man fallacy. The problem here is you have restricted the discussion to curve tracing of a subset of tubes and then attacked the improvements as being not relevant. That is not reality.   I  think I have demonstrated beyond doubt above that it's a worthwhile thing to have. 

Accuracy matters as you want repeatability. The emission point is void too. You just need to allow sufficient time for the heaters to stabilise. In any case you face the same issue whether you use PWM or a DC supply. The PWM is very repeatable.  If the heater temperature is stable it's not going to "crash" the improvements.  BTW the PWM IS much improved, I've measured it.

BTW As far as I know the futracer board doesn't do 100V and no-one ever said that.

I'm adding 100V support to the firmware as people have asked for it.

"Who really cares", well I do and I bet I'm not alone.

[Ihor]

Using the term "psuedo" is an insinuation that the 12 bit feature is somehow not 12 bit. Like "pseudo-scientific" which means "not scientific" or pseudo random which is not random at all. I find that objectionable.

Just to also clarify, I used the word "pseudo" related to the whole thing, "12 bit PWM". I fully agree with your description and it is actually implemented as I was hypothesising. I am not saying that it is not possible to implement 12 bit (or more) or it is completely useless. I was doubting it is practical usefulness for the current uTracer3+ (because the PIC is sold exactly for that model). As a hardware engineer one knows the difference between the hardware implementation of PWM (as a separate module that runs independently of the main loop) and software version where dealing with timings and interrupts is crucial. And going back to "pseudo" :) it is exactly as in pseudo-random generator, because the random generator on one's computer also has nothing to do with randomness (in a true "hardware sense") but at the same time produces samples which are indistinguishable from a true physical source of randomness (so it is not possible to distinguish with statistical tests and that's the best we can do). Before your post, it was also exactly the same as pseudo-science, because that one is also sounds true and good but never has proofs:) That's great that you added those comparisons and clarifications!   

 

Consider the DAC's used in some 16 bit CD players.  They actually have 1 bit DACs. If you measure the performance, you get 16 (or more) bits after filtering that is applied. No one calls that pseudo.

Indeed, there are approaches such as DSD audio with one bit. No one calls them pseudo but also calling them just a DAC is a huge generalisation and simplification as they should be specified by a type/name (segmented DAC, Delta-Sigma, etc.) those are all different approaches that lead to the same 16 or whatever bit resolution (making the user happy) via totally different hardware and software solutions. 

  

So, given that it does work, and rather well, let's move on to the statement that it isn't needed. First I should point out that the voltage numbers that were quoted are misleading as it's the grid voltage that matters. With a -100V range ( which is what we are talking about) each 10 bit step is a tad shy of 0.1V, not 0.005V mentioned which is just the filter output .

It is probably was a typo in one place but in the same post in many other places I wrote that it is 0.05 ~50V/1024. It is also clear from your plots with the grid voltage that show exactly that, that each step of Vg is about 0.05-0.07V different. 

The plots are nice indeed! I will try to measure something similar but just with and without the tubes. I assume you did it without the tubes, showing the ultimate performance because with a DUT the time stepping would be not uniform (the capacitors would be charging and that would lead to different timings between the measurements). 

 

Accuracy matters as you want repeatability. The emission point is void too. You just need to allow sufficient time for the heaters to stabilise. In any case you face the same issue whether you use PWM or a DC supply. The PWM is very repeatable.  If the heater temperature is stable it's not going to "crash" the improvements.  BTW the PWM IS much improved, I've measured it.

 You saw the plots above where I preheated a tube for 5 mins, measured once and then after 5 minutes once again... the differences are "huge" and it is not solely because of the hardware, the tube itself influences that a lot. Even if one makes a perfect measurement device, that tube then goes inside an amplifier, where the hardware setup (in terms of heater and other voltage settings) would most likely be worse than in the test setup. It is probably the same as claiming that working with tubes while building the amps requires professional DMMs with accuracy higher than 0.05% and it improves your design and sound,... it only improves the accuracy of the measurements :)   

 

[Davo]

You might be right on that Ihor!

This goes about my reach tho…

As far as ik know i saw a comment from Ronald back in the past that he didn’t had the gear the measure the heater thing exactly on the uTracer…

He looks to me as a fortunate guy to me. So the option to measure it securely wouldn’t be cheap…

Grt,

Dave

Op 25 jan. 2023 om 09:30 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/ed1f6e03-2fc7-4bf6-a3bf-ddf5dafed7e9n%40googlegroups.com.

[Ihor]

Back to reality :) I made a few measurements myself. As you can see, the simple measurements (200 points per 1V of grid voltage) with no-load show nice steps, indeed 10bit there do not even reach "10" bit so it is a bit worse than the theoretical estimation. Now the other plots show Vg when the tube is there and with some measurement where the is not conducting much (couple of mA) or quite a bit (50-90mA), it is KT88. For the images you can see what kind of "mess" is happening to the grid voltage while repeatedly doing 100 measurements with the same Vg... and this is what happens in real life and not with the debug set. The nice steps are only on "average" steps (so quantisation noise is low indeed), but with actual measurements there is a comparable extra noise

  

[Davo]

Hi Ihor,

Which analyzer software do you use?

Op 25 jan. 2023 om 10:39 heeft Ihor <iste...@gmail.com> het volgende geschreven:

Back to reality :) I made a few measurements myself. As you can see, the simple measurements (200 points per 1V of grid voltage) with no-load show nice steps, indeed 10bit there do not even reach "10" bit so it is a bit worse than the theoretical estimation. Now the other plots show Vg when the tube is there and with some measurement where the is not conducting much (couple of mA) or quite a bit (50-90mA), it is KT88. For the images you can see what kind of "mess" is happening to the grid voltage while repeatedly doing 100 measurements with the same Vg... and this is what happens in real life and not with the debug set. The nice steps are only on "average" steps (so quantisation noise is low indeed), but with actual measurements there is a comparable extra noise

<CaptureVgNoLoad.PNG>

<CaptureVgnoload1.PNG>

<CaptureVgnoload2.PNG>

<CaptureVgnoload3.PNG>

<CaptureVgnoload4.PNG>

<CaptureVgnoload5.PNG>

  

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/659593d2-e53d-451c-820e-bb85eed86fd9n%40googlegroups.com.

<CaptureVgnoload2.PNG>

<CaptureVgnoload3.PNG>

<CaptureVgnoload5.PNG>

<CaptureVgNoLoad.PNG>

<CaptureVgnoload4.PNG>

<CaptureVgnoload1.PNG>

[Ihor]

This was done with Analog Discovery 2 (14 bit- resolution per channel). 

So here is another plot where I compared the standard setting (the first image) against using the GridLoupe (two other images), which again you can build in 10 minutes and you will get accurate voltage ranges from +10 to -5V (https://www.dos4ever.com/uTracerNotebook/Notebook.html#grid3 ) so we get resolution of 15/1024V, comparable to 12bit native. As you can see from the plot, 1V range of Vg is mapped really nicely without much steps, still spiky (there are about 10 of measurement pulses per "step"). 

[Ihor]

And just to complete the story. Here my measurements of the grid voltage from -3 to 0 volts, on a uTracer3+, last time calibrated 4 years ago, with the step of 0.1V, which as we now know makes send for the uTracer3+. In that small range of -5,..,0V wi will have resolution of 0.1=0.5V but we will also have the same resolution in the range of -30, or -50V, so we can still make those nice steps of 0.1V. In any case, here are the plots of the steps. I also used the delay of 1sec to settle down all the voltages, and the steps looks quite uniform and not bad I would say. They are not perfectly uniform, but even with in the range of -1,...0V (see the red lines) one gets quite accurate voltages (compared to the user-defined). They are definitely not "so worse" as stated on the FUtracer website and specified by the table there.  

 

[Martin Manning]

I'm thinking that the FUTracer guy's experience was before the mod to the bias circuit filter. Prior to that the grid voltages were coming in high during the measurement pulse, making good 12AX7 tubes look like they were low in emission.

[nickb]

I have not seen your plots over time. I did just a test using an EL84 ,measuring the plate current over twenty minutes. The standard deviation was just 1.02% over that entire period. If you are seeing more than that then you may have a problem with your tracer.

I see only a small degradation in the 12 bit performance if I have a tube plugged in. Of course this is not the right way to do this. It really should be triggered to sample during a measurement pulse wich is quite a bit harder to set up. Here are my "12 bit with a tube" results. I feel sorry for you have those big 0.1v steps 😉

[Ihor]

I have not seen your plots over time. I did just a test using an EL84 ,measuring the plate current over twenty minutes. The standard deviation was just 1.02% over that entire period. If you are seeing more than that then you may have a problem with your tracer.

Those indeed were not plots in time, but just sample measurements every 5 minutes or so. It was really dependent on a tube. So it is possible to get 1% error but it also possible to get a lot more. The most extreme was a kt88 with a heavy heater. That plot showed that after 5 minutes some curves moved up (in current measurements) and some down(!), depending on the grid voltage. That means that there are some grid voltages there where points did not move at all. So if I do very "smart" measurements I can show an error of 0% and in other extreme cases the error will be about 10-15%. It is not really a problem with the uTracer because there is no consistent drift or any other constant bias that drifts in time, it is mostly the tube behaviour in my case.  

 

I see only a small degradation in the 12 bit performance if I have a tube plugged in. Of course this is not the right way to do this. It really should be triggered to sample during a measurement pulse wich is quite a bit harder to set up. Here are my "12 bit with a tube" results. I feel sorry for you have those big 0.1v steps 😉

The plots that you show really miss good resolution in time (the number of samples that your oscilloscope takes) because they are just averages of the reality. I used 14bit differential input with 1k samples per sec to display that, and it shows that there is way more going on on the grid while the measurement pulse is coming, there is the whole "interaction" between the tube and uTracers during the acquisition :) The steps are only nice if one uses "20xxxxxxx" command on uTracer that just sets the voltages and does not measure anything.   

Thos step-like curves indeed are not comparable to the 12 bit resolution, but it is what uTracer3 produces. No one rejects the fact that it will improve the settings of Vg and make it more accurate. I was just saying that with uTracer3 we can get 10 curves per 1V of grid voltage (in the worst case) and in the best case 20 (again with the step of 0.05-0.1V). Out of curiosity, I would like to hear about at least one application that requires acquisition of more than 10 curves per 1V of Vg, or any other application that needs setting of Vg with high (0.1%) precision, where 0.05V is not enough :) 

Using uTmax or uTracerJS one can for example acquire 3 curves per 1V and then interpolate everything in between with a very good accuracy. With uTracer's 10 curves per volt and software interpolation one can reach 12bit grid resolution easily, just with one click of a mouse (on the interactive plot in uTracerJS).  

Again, a really nice discussion because we figured out lots of thins and showed lots of examples to the users!

[nickb]

The data was captured using a lab quality DVM 5.5 digits. My 8 bit scope, even in high resolution "11 bit" isn't good enough.

If the Analog Discovery 2 ever comes back down to a sensible price I'd get one but at well over £400 it's silly money for the use I'd get.

I do feel that you seem to be missing the point.  If you are just doing curves then the higher grid resolution is not important.  However my focus here is on a quick test of Gm  where the delta Vg matters and can lead, as I have shown, to very large errors. In utmax I recently added code to take account of the finite number of bits so, as long as delta is not too small, it will do the calculation of  gm using the derived delta value. However as the magnitude of delta increases so this matters less.

With utmax interpolation is done very accurately with B-splines or, if you wish, from a derived SPICE model.  That of course takes much more time so it's no longer a quick test which is rather the point.

At the end of the day all these things are improvements and that can't be a bad thing.

[Davo]

Ah okay… the soft looks good!

Yeah i almost bought it last week haha.

But i saw a 2nd hand Vellerman PSGU250 for an unbeatable price of 90 inc. Shipping haha

So i’m going to start learning  PSGU250 to use it & maybe sell the unit with a profit and switch over to the Digilent or the Picoscope.

I dunno 4 sure but i thought they were both 12bit. and have free fft & bode plot software also!

The thing that keeps me away from the Digilent is the crappy housing, shitty probes &  looks like no attenuators & also no protection for unwanted noise from surrounding gear.

The only real plus from Digilent are the multiple inputs. If should buy the Digilent I definitely make a proper housing for it though…

Op 25 jan. 2023 om 15:35 heeft Martin Manning <mman...@fuse.net> het volgende geschreven:

I'm thinking that the FUTracer guy's experience was before the mod to the bias circuit filter. Prior to that the grid voltages were coming in high during the measurement pulse, making good 12AX7 tubes look like they were low in emission.

On Wednesday, January 25, 2023 at 6:53:55 AM UTC-5 Ihor wrote:

And just to complete the story. Here my measurements of the grid voltage from -3 to 0 volts, on a uTracer3+, last time calibrated 4 years ago, with the step of 0.1V, which as we now know makes send for the uTracer3+. In that small range of -5,..,0V wi will have resolution of 0.1=0.5V but we will also have the same resolution in the range of -30, or -50V, so we can still make those nice steps of 0.1V. In any case, here are the plots of the steps. I also used the delay of 1sec to settle down all the voltages, and the steps looks quite uniform and not bad I would say. They are not perfectly uniform, but even with in the range of -1,...0V (see the red lines) one gets quite accurate voltages (compared to the user-defined). They are definitely not "so worse" as stated on the FUtracer website and specified by the table there.  

 

--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/0951a221-229f-47d0-80dc-02557f85c123n%40googlegroups.com.

[Ihor]

The data was captured using a lab quality DVM 5.5 digits. My 8 bit scope, even in high resolution "11 bit" isn't good enough.

It is a very nice and accurate device, but DVM essentially averages while taking measurements, converting to DC, so it is not about the voltage accuracy/resolution I would dsay.. 

If the Analog Discovery 2 ever comes back down to a sensible price I'd get one but at well over £400 it's silly money for the use I'd get.

Indeed, it is currently an expensive device, which a couple of years was quite reasonably priced (about 270 euros) and luckily on a local secondhand marker there are AD2 for sale from students which can buy it with academic discounts for about 100euros and sell them after the studies for even cheaper:) I would also not pay myself the full price.  

I do feel that you seem to be missing the point.  If you are just doing curves then the higher grid resolution is not important.  However my focus here is on a quick test of Gm  where the delta Vg matters and can lead, as I have shown, to very large errors. In utmax I recently added code to take account of the finite number of bits so, as long as delta is not too small, it will do the calculation of  gm using the derived delta value. However as the magnitude of delta increases so this matters less.

Ok, this is at least a good example. Measuring gm in a quicktest setting indeed requires accurate information about what Vg was. I am not sure about the actual requirements to the accuracy in this case. Probably software trick would be to take a couple of quicktest measurements with different settings for delta_Vg and then get a nice average. .  

With utmax interpolation is done very accurately with B-splines or, if you wish, from a derived SPICE model.  That of course takes much more time so it's no longer a quick test which is rather the point.

uTracerJS also uses b-splines and modeling for spice and even with just the measurement and b-spline interpolation (without the modeling) it is possible to immediately get the gm, ra and mu in any part of the curve-space, interactively. It is even possible to see if it changes a lot or not:

http://files.smal.ws/v/E6HNZ9

[Davo]

I really love the conversation you guy have!

These are sum really good in-depths how the uTracer actually works plus the con’s & pro’s of both approaches…

Btw could the heater voltage be improved so no use an external heater supply?

Op 25 jan. 2023 om 19:35 heeft Ihor <iste...@gmail.com> het volgende geschreven:



--

You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/539e9224-84e1-460d-8e12-574e4f9b489cn%40googlegroups.com.

[nickb]

Utmax does all that too along with multiple simultaneous plot views. All the views in one go as pdf, jpg etc.. 

[nickb]

I am of the opinion that with the new firmware and the steps outlined on the website that an external supply is not needed. The firmware takes care of the external loop inductance and the inductor change solves the biggest DC resistance issue. You can compensate for any DC resistance simply by adding the IR drop ( heater current  x circuit DC resistance)  to the desired target heater voltage. That is then saved to the database so you never have to do it again.

I completely understand the feeling of confidence that having an external supply gives you.  What is nice is with the PWM power supply is that the voltage is set automatically, not that I change it very often.

There is a  hook in utmax to allow external supplies to be controlled but you need to sniff the serial data to use it.

[Tony Dorris]

Where is the information for the connection of a twin triode for q quick test? I see the connections for the plates using the plate and screen connection for the plates and I assume the grids get tied together for this test, but where exactly does the second cathode get connected? I do not see this in the manual anywhere.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/24c41ebf-641b-41fe-b555-b3cf25506a47n%40googlegroups.com.

--

Tony Dorris

(812) 266-4826

[Agus Sugiono]

I use Nick's PIC, Ronald's uTracer and your uTracerJS. That's the best combination for me :D

Regards

Agus

On Thursday, January 12, 2023 at 6:35:38 PM UTC+7 Ihor wrote:

I thought it might be interesting to update this thread mainly because I noticed recently there there are lots of new features implemented in uTMax. I myself tried on only long time ago, but I just saw that the author also (re)wrote completely the original Ronald's firmware (for the PIC controller) which can be easily replaced and has lots of interesting features. 

http://bmamps.com/store/index.php?main_page=product_info&cPath=109&products_id=440

This is the citation from the website:

"The new firmware has the following enhancements.

• Dynamically controlled boost pulse widths to allow the the boost voltages to be set more accurately at lower voltages.
• All measurements are over-sampled to lower noise and improve accuracy.
• The heater voltage accuracy* is improved by changing the PWM frequency to 2KHz. 
• Grid voltage 12bit resolution (4095 steps) upgrade from original 10 bit (1023 steps)
• Calibration data saved to chip for portability
• V3 introduces firmware that can be updated from the uTmax GUI"  

I was wondering if someone used it and can share the experience. I think that grid 12bit resolution is interesting but I wonder if someone found it useful in practice. Most of us use external heater supplies, so that addition is probably not so critical, but about the the first two improvements in the list, I thought it was possible to do with original firmware as well, that oversampling part, and changing the pulse width is also done in Ronald's firmware (at least he describes that issue and solution in his weblog).   

There is also a very nice design of the tube pin switching board in the section "A Pin Switch Matrix Board" (see https://bmamps.com/v01/home/techie-corner/utracer-utmax/ ) which is rather an overkill but I remember some users on the forum were asking for that and also in Ronald' gallery there are some similar implementations. Now it is possible to easily build it yourself!

Cheers, 

Ihor

 

[Ihor]

I use Nick's PIC, Ronald's uTracer and your uTracerJS. That's the best combination for me :D

Regards

Agus

When you use that combination you are still missing the 12bit Vgrid resolution of Nick's PIC. It is possible to implement it because uTmax is opensource and it is there, but I was not sure how many people are using new PIC. So, if there are some volunteers for testing I can take a look at it:) 

[Tim Goldstein]

I have Nick’s PIC taped to the uTracer case waiting for a good reason to drop it in and play with it. I am a uTrace neophyte but with instruction on what you want tested would be happy to do some testing

 

Thanks

Tim G

Autistic, He/Him/His

www.TimGoldstein.com

Neurodiverse Communication Specialist – Emotional Speaking

Author - Speaker - Trainer - Consultant - Vocal Coach

Top 50 Global Neurodiversity Evangelists 2023

--
You received this message because you are subscribed to a topic in the Google Groups "uTracer" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/utracer/6U_E6eS3_0o/unsubscribe.
To unsubscribe from this group and all its topics, send an email to utracer+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/156fd092-c759-43ed-af4a-9309c280fcaan%40googlegroups.com.

[nickb]

Hi Tony,

Sorry for slow response. It seems I am not getting always email notifications from the group, even though I'm subscribed.

I expect you have this figured out, but in case you haven't, here goes. The connection data for a 12AX7 with 12.6V heater, for example, is this:

So,

Anode goes to pin 1

Screen to pin 6

Grid to pins 2 and 7

Cathode to pins 3 and 8

Heater 1 to pin 4

Heater 2 to pin 5

[Ihor]

I implemented the possibility to send 12bit Vg values to Nick's PIC on uTracer3+ as it is done in uTmax so it should work for others as well. Agus already tested it but making some measurements with an oscilloscope and checking if the stepping is gone would be nice. Otherwise one would barely see the difference. To activate that mode, one has to change the "Version" in the dropdown menu as in the screenshot. The newest version is on the main website: https://boffin.nl

[nickb]

That's super. I'm sure everyone will appreciate the new feature :)

On my side I have implemented the fast sweep. It can be enabled on the "options" dialog of uTmax. I have two tracers to try it on neither of which are the same as Ihor's which has the fast charge feature whereas neither of mine do.  However I do have a fast discharge on one.

A standard sweep on a standard tracer took 153 secs

A fast sweep on a standard tracer took 67 secs

A standard sweep on a fast-discharge tracer took 89 secs

A fast sweep on a fast-discharge tracer took 55 secs

This is available on the site a 3.04k now.

https://bmamps.com/v01/home/techie-corner/utracer-utmax/

It should be noted that the settling time for the grid is about 53mS for 10bit and 71mS for 12 bit.

[Pedro Cimini]

Ihor, do you have a relay suggestion for the switching matrix? I was only able to find affordable up to 200VDC, which I think might be too low for some of the regular tubes I test...

Em sábado, 14 de janeiro de 2023 às 16:00:38 UTC-3, Ihor escreveu:

It takes quite some time to improve the firmware and especially to rewrite it from scratch especially because all the acquisition routines and steps are not described in Ronald's blog, so a lot of respect for that. The biggest problem would be to maintain it over the years (in general any software) taking into account that for some of use those are just really hobby projects, not connected to own business or work directly. For Ronald and uTMax developer that's a different story. 

The switching board is also cool, but indeed, we do not know anything about the oscillations, and if there are no oscillations with that board, why would anyone need FUTracer (assuming you can plug a new PIC into uTracer3, so probably only for new users). Personally I like simplicity and with so many relays as on the board, and sourcing parts from China to go cheap will sooner or later lead to failures in one of those. Then lots of precautions and self-tests should be implemented already on the board because otherwise the uTracer will be damaged. Are there any estimates of the costs, or at least which relays are used? 

For the information I also attached a photo of a working relay board which was proposed in 2018, so if someone is interested they can find the details in Ronald's Gallery and ask the authors for the PCB. 

[Ihor]

I have not looked into that before, but you can check Nick's solution or ask for suggestions:

https://bmamps.com/v01/matrix-pin-switch-board-for-utracer/

Ihor

[D.A.R Achterberg]

You are thinking to compicated, use rfi supressors in the in & output of the board…

What dus matter if a relay can handle the dc current especially those heater currents

Nick sells only complete finished boards, to save a-lot of time supporting all the things that can go wrong…

The guy is the next Ronald Dekker of the uTracer & taking it to the next level though..

Op 13 nov 2023 om 10:45 heeft Ihor <iste...@gmail.com> het volgende geschreven:

I have not looked into that before, but you can check Nick's solution or ask for suggestions:

--
You received this message because you are subscribed to the Google Groups "uTracer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to utracer+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/092a3414-7699-4d2d-a6ce-be7b1328e84bn%40googlegroups.com.

[Big Josh]

Nick is far superior. ;)

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/A0234675-B643-4E93-AB5F-5EA9CD948CF9%40gmail.com.

[Davo]

Yes he is! But without Ronald we wouldn’t be here at all…
Being a pioneer is complete another level then upgrading an existing product ;)