High current mode for uT 3+

[Martin Manning]

I'm considering modifying my tracer for higher current capability. I'd originally dismissed this idea due to loss of resolution when low currents are measured, but if I could switch between high- and low-current modes that would be great. The Lab Notes pages describe the changes: reduce both the hardware current limit resistor and the current sensing resistor (plus increase the 'floating battery" cap). The GUI's calibration form includes an input for the current sensing resistor values, which causes the compliance limit ranges to be adjusted. 

Increasing the hardware current limit to 700mA (by placing a 1.5Ω resistor across the standard 2.7Ω), and reducing the current sensing resistor to 6Ω (by paralleling two more 18Ω resistors with the existing one), puts the compliance limit at 600mA. This would be fine for my high-current measurements, as it would permit tracing of audio power pentodes with both the anode and the screen at the limits of the uTrace 3+ high voltage supplies.

Switching both the hardware limiting and current sensing resistors is messy, though. I'm wondering what the danger would be if the hardware limit were increased to 700mA and only the current sensing resistors were switched. I can see where the voltage across the current sensing resistor could exceed 5V, and the limit of what the ADC can use, if the compliance limit were switched off in low-current mode. 700mA across 18Ω would be 12.6V. That would only be the result of an accident, but it would be nice to know if the accident would have serious consequences.

Thanks in advance for any thoughts on this.

MPM

[Martin Manning]

PS It did occur to me that something as simple as a 5V Zener across the current sensing resistor might protect the rest of the circuit.

[Bill van Dijk]

Hi Martin,

 

I know my suggestion is more costly, but I found that many tubes able to handle that kind of current also work at much higher voltages than the 400 V limit of the 3+. I have also procured the V6 from Ronald (as a matter of fact I was one of his beta testers on that model), and I find it great for the large tubes.

One of the things we discovered along the way is that the V6 is somehow more prone to spurious oscillations, but in most cases I found them fairly easily suppressed. 1000V at 1A is bloody impressive!

 

Just a thought, Bill v

--
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/d8f4a9a7-331f-4f8c-b70f-7cf2b568aca1n%40googlegroups.com.

[Bill van Dijk]

Hi Martin,

 

The two resistors paralleled would be about 0.964 Ohm, in order to produce 5V across it would require a current of more than 5 Amp, I expect to see smoke long before that (if it even could go that high). To limit the current through the resistor at 0.6 Amp, you would need a 0,58 V zener. I don’t think they make one. Also, a zener would not protect the circuit. If the current got high enough to cause the zener to break down it would protect the resistor, the current through the rest of the circuit would increase even more as a result of the zener breakdown.

 

Bill v

 

From: utr...@googlegroups.com [mailto:utr...@googlegroups.com] On Behalf Of Martin Manning

Sent: Sunday, January 30, 2022 9:54 AM
To: uTracer <utr...@googlegroups.com>

--

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/dafb63f8-8530-432f-b526-9abf86935e3fn%40googlegroups.com.

[Martin Manning]

Thanks, Bill. I thought hard about getting a µT6, but decided against it due to the need to build an entirely new enclosure.  I know 450-500V is typical of large audio amps, but 400 is close, and really close enough, I think. Most data sheets have curves at 400V too. To me, it seems like this modification allows the 3+ to operate up to its full capability. After studying the circuit some more, I think it is already protected against an over voltage on the current sensing resistor.

[Martin Manning]

Bill v, I'm referring to the current sensing resistor, which is 18Ω standard. It needs to be reduced to 6Ω to get the compliance limit to increase to 600mA (that relationship is coded into the GUI by Ronald).

[Martin Manning]

Apparently lead inductance rules out a remote (i.e. panel mounted) hi-low switch arrangement. It does not work at all. Zero current measured in a standard calibration test using two 10k resistors. I'm now configured for 600mA, with 2x 18Ω resistors paralleled on each 18Ω current sensing resistor, and a 1.5Ω across the 2.7Ω hardware limit resistor (just because it's easier than soldering two 2.7's). That should get 700mA max instead of 750, fine for a nominal 600mA limit, I think. With this configuration I was able to make a full trace of a 6L6GC at 400V Vg2 (max anode current close to 300mA at Vg1 = 0). I also tried a few tests where low levels of current were measured (tracing 1Meg resistors, tracing a 12AX7, and a Vg Low grid current test using a 1k resistor G1-k), and still got good performance. I think I'll leave it this way for now and see how it goes. I did find that running in the standard 200mA configuration with the hardware limit increased does not seem to cause any problems or even upset the calibration, so some kind of board-level switching on just the current sensing resistors might be a possibility. 

MPM

[Martin Manning]

Some thoughts on the current capability... In my mind the µT3+ capability is more about the ability to handle the current with screen voltage up to typical levels for the tube type. Increasing anode voltage isn't going to be very interesting, just an extension of some already very straight grid lines. There could be issues out there for a given tube, but those may or may not be uncovered by a 1ms pulse. 600mA is enough to get a full trace to 0V Vg1 on a KT88/6550 (the largest audio tubes generally in use) at a typical Vg2, and replicates the curves seen in data sheets. Yesterday I went for nearly the full capability by tracing a KT88 at 350V Vg2, which approaches the 600mA compliance limit of the high-current configuration I'm running. It could be extended further, as the GUI accepts am input of 3R for the current sensing resistors, which results in a 1.2A compliance limit. 

MPM

[Bill van Dijk]

Hi Martin,

 

Very interesting. Have you done any calculations on the tank capacitors? That much current, even for a ms discharges the tank capacitors. Would adding capacity to the tank be needed at some point? How would that affect the tank charging circuits?

[Martin Manning]

Hi Bill,

There is quite a bit of discussion on the Lab Notebook pages on this mod (and it is an interesting read), but there is no mention of the reservoir capacitors or their charging circuits. See: https://www.dos4ever.com/uTracerNotebook/Notebook.html#Current1

On the voltage/current relationship of the 3+ vs. large audio tubes, the very common 6L6GC and EL34 are often operated with anode and screen voltages nearly equal at zero signal) in the range of 450-500V. The 3+ can't get there, but comparing test results to the data sheets at 400V is likely to be sufficient for assessing a tube's performance, IMO. 

MPM

[Ihor]

Looks interesting! You can probably get some idea about the discharge of caps using the quicktest menu. There you can set the desired Va and Vs and then see the "measured" Va and Vs after acquiring a measurement. There is a thread here where we were talking and comparing the sagging of Va and Vs for example for 2A3: with the set Va=250V and Ia of 40-50mA one gets measured Va=245V. So it is interesting to measure something like that for 6L6 or KT88 for Vg closer to zero, having Ia around 300-400mA and see what is 'set' and 'measured' Va. It might be not so bad but also can be also tens of volts, shifting the actual curves that you measure right now quite seriously.

Ihor

[Ihor]

This is from that other post with 2A3, so the "center" and "Meas" values

[Martin Manning]

Hi Ihor, 

I've often used QT for that purpose. Here are some tests on the KT88 in the trace above: 

Low and high bias in the middle of the Va sweep, and high bias (-5V) at low Va (below the knee, where Ig2 is high). The sag in Vg2 is the thing to look at, as the curves are supposed to be at constant Vg2. These results don't look too bad. The large sag in Va at low Va makes sense, and it should plot correctly, I believe, since the tracer records the actual Va for the plot.

MPM

[Martin Manning]

One more...  Here is a set of curves from the same KT88 combining the high-current mod and the Grid Loupe, with Vg1 from -10 to +5V. Vg2 is reduced to 300V to stay under the 600mA compliance limit. It looks like the Vg1 = +5 traces are low, so it's possible that the voltage is not really at +5 during the measurement. Ig1 would be interesting too, but there is no way to get that with the screen channel already in use. If there is a need for this sort of thing, the uT6, plus an added extension board, will be capable of Ig1 measurement for pentodes. 

MPM

[Martin Manning]

I've found a strange result tracing low value resistors with the uT3+ high-current modification described in the Notebook pages. Here I've traced two 470R resistors, separately and together. The screen channel shows lower current than the anode channel when they are traced separately (~5% at 225V), but when traced simultaneously the anode channel repeats nicely, but the screen channel shifts up to (nearly) match the the anode channel. As the value of the resistors increases, (lower current) the effect goes away.

Is there anyone else running the high-current configuration who can repeat this test? 

[Big Josh]

Just a guess, but considering the variance isn't very significant, I would assume the difference you're seeing in current is the result of a difference between stated and actual voltages. For example, both set to 400VDC, but actual recorded Va is slightly higher than Vs.

Thanks,

Josh

--
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/08a30c75-b4c4-47e5-8a0b-168731434c85n%40googlegroups.com.

[Martin Manning]

Thanks for the reply. What bothers me is the screen current shouldn't be different at all, same as the anode current. I tried the same test again with the 470R's connected directly at the GCSA terminal block, with all other wires disconnected. Same result, so it isn't caused by the lead dress in my enclosure. Note the screen voltage at each data point is a higher, 5-10V, when the current trace is lower. I'm thinking I might try replacing the transistors in the screen channel and see what that does. 

[Big Josh]

Is that voltage really higher? I don't know how his GUI reports voltage. Is it the recorded voltage (actual), or target voltage? If you do a quick test where it shows both the target and actual voltage, is the screen voltage still over compared to anode voltage?

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/346b7653-1dbf-4657-ab66-682d8c7df914n%40googlegroups.com.

[Ihor Smal]

You can try to check the .utd file that corresponds to that acquisition (by opening in the text editor) and as far as I remember (when I was reimplementing that part for the uTracerJS and the ESP32 software), it records the actual measured voltage (for Va and Vs), so not the “target” voltages from the the main menu. 

Cheers, 

Ihor

 

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/CABi0_%2BNbB4ft83%3Dx_e-E4jU2qp0i8Sk2Uq3_E0VPJemiQZT8uA%40mail.gmail.com.

[Martin Manning]

The actual test voltage is recorded at the end of the pulse. I can run a Quick Test with two resistors separately and together, but I'm pretty sure I'll see the same result. 

Thinking about how the A and S channels could be talking to each other, when testing two 470R resistors simultaneously there is less than 1k between the outputs, and that might be the difference. My plan to replace transistors is on hold for the moment because the PN5416 (original uT3) and KSP94BU (shown in the latest construction manual) at the HV outputs are both obsolete. ZTX958 looks like it might be a good replacement (and I have some), but I don't know for sure. Anybody have an idea? I guess I could give it a try, but if it doesn't work my tracer will be down (unless I reinstall the old PN5416's). Next thing to try would be the HV switches, KSA1156 (uT3+ upgrade), which I also have.

[Big Josh]

My build uses ZTX558, though it isn't a uTracer. It should work in your uTracer also.

To view this discussion on the web visit https://groups.google.com/d/msgid/utracer/a45a5e3d-0534-4195-b674-a24422e95cdbn%40googlegroups.com.

[Martin Manning]

Thanks, Josh. I've swapped the PN5416 with ZTX958, and replaced the KSA1156 (in that order). Afterward I still saw the downward shift in the screen channel current when it is the only one in use, and the increase in voltage at the measuring points. The higher voltage might just be due to the decreased current (less energy taken from the reservoir), which suggests that the current is actually lower, as opposed to its being the same, but measured lower. At least I confirm ZTX958 works for PN5416/KSP94BU.

MPM

[Martin Manning]

Yesterday I finished up by replacing the screen channel's BC558 hardware current limiting transistor with no effect on the observed screen current shift. There is one transistor left, the KSP44BU "flush valve" that drains the reservoir, but I don't have any of those on hand. If that transistor is slow, it can produce a false current reading. 

 In the meantime, can anyone try the test below with a standard uT3+ and a couple of 1k5 resistors? It's pretty simple: trace both resistors simultaneously, then tick "Save Plot" and trace each one separately. As you can see, in my result the anode repeats almost exactly, but the screen shifts down about 2% when traced separately.

[Martin Manning]

Sorry, in the above I meant to say tick "Keep Plot" so that the separate anode and screen traces will be added to the trace of the two together.

[Ihor]

I did a similar test but with 2.2kOhm resistor (only have those) and using uTracerJS, because it was easier to visualise and also I am attaching the actual .utd files here, so everyone can import them in uTracerJS and then visualise and compare. '2x-2.2k' is a plot for 2 resistors at the same time, and then '1x-2.2k-a VS 2x-2.2k' compares acquisition of 1 resistor on anode VS only anode measurements of two resistors and then '2x-2.2k VS 1x-22k-s' compares acquisition of one resistor on Screen connector VS screen measurements of two resistors simultaneously. 

Cheers, 

Ihor

  

[Martin Manning]

Thanks, Ihor. 

Just to be absolutely clear, what I did was to connect one resistor from the anode to the cathode terminal, and another from the screen to the cathode, and traced them simultaneously. 

Then, I disconnected the anode resistor, and traced the one on the screen alone. 

Then, I reconnected the anode resistor, disconnected the screen resistor, and traced the one on the anode alone. 

In my case, the anode I vs. V when traced alone looks almost exactly the same as when both resistors are traced together, while the screen drops ~2% when it is traced alone.

I think what I'm seeing in your plots is that the anode trace moved a bit (<1%) when that resistor was traced alone, and the screen trace did not. Is that right? 

In your plots, which trace is shown by the solid line and which is shown by the dashed?

[Ihor]

In my case, I did almost the same, but I di don't disconnect the "useless" resistors during the test, but just set 5V there. I am not sure if it changes much in practice, but you can also try it like that in your experiment. I can also do exactly as you did, by my curves I would say are perfectly consistent (from the engineering point of view, meaning that some small accuracies can be ignored) and that's why I do not expect to see such larger difference as in your curves. In your case, when you disconnect one of the resistors (for example the Screen one), there is still some voltage on the Screen for example, so I am not sure how that flying voltage influences the measurements on anode. In my case the Screen is kept very close to the ground (5V only) via cathode and resistor. Probably it is worth trying that. 

In the plots. the solid and dashed lines are changing meaning for different plots, so one has to look at the left and right axes. The left axis is always solid and on top it says what the measurement is (ia or Is and for 2x (two resistors) or 1x ...A (anode-connected resistor or 1x ... S (screen connected single resistor). the right axis is always dashed and also on the top you can see what kind of measurement it represents. It is actually possible to plot all the curves in excel or so using those .utd files. 

Cheers, 

Ihor

   

[Martin Manning]

Thanks, Unfortunately the uT3 GUI has limited options for the test sequence: It sweeps the anode alone or the anode and screen together. Sweeping the screen alone is not an option. The mystery remains, i.e. why is the anode channel unaffected when the screen is disconnected, and not the other way around, when the circuits are identical? If you or anyone else can run the same test as I did just above (5 posts up) I'd appreciate it :^) I'm down to replacing the last screen channel transistor, KSP44BU,   and then I'm out of ideas.

[Ihor]

This is the table with results, but the anode and screen voltages are not the same, and different for each run which is normal. 

Below you can also see the screenshot how to sweep only screen voltage with uTracerJS and keep the anode constant :) There is practically no single option or thing that windows software can do and uTracerJS cannot :) (except acquiring data with changing heater settings, which are of very limited practical interest to anyone).

Ihor

 

[Ihor]

oh, I read a bit wrongly, I thought that you are talking about uTracerJS, but in the original software you can sweep Screen voltage exactly in the same way, the test in the dropout menu is "I(Vs, Vg) with Va, Vh constant""

[Martin Manning]

Yes! Thanks for that suggestion Ihor! You have to make three runs to get the data, and you have to plot it manually to see what the change in current is at constant voltage, but there is a difference in the result, and I am seeing something similar to what you got at much higher current (see the plot below). The screen still drops a bit more than the anode when it is traced with the other channel at low voltage, but it is now down about 1% compared to the simultaneous traces (the anode only drops a few tenths of a percent). There is some unexplained crosstalk between the channels for sure, but evidently it is reduced when the other channel is operating at low voltage (and current) as opposed to being left open at high voltage. This is important since in pentodes the screen and anode currents can trade by an order of magnitude when control grid voltage and anode voltage approach zero while screen voltage remains high. I guess I can consider it a non-issue now :^)

[Ihor]

[Ihor]

I am not sure why the text did not appear in the previous message, but here it is, for the plots above :) 

I did the test with two resistors today and then removed one by one, so with "floating" Anode and Screen outputs. Here are the plots. Again the left axis is with solid line and the right one is dashed. 'test 2x 2.2k' is when 2 resistors are tested with "I(Va=Vs, Vg),...." test, and then 'test 2x 2.2k A" is only Anode resistor present and "test 2x 2.2k S" - only Screen resistor. The test test setting is also "I(Va=Vs, Vg),....". While testing both resistors, the curves are nicely matching. Just Anode resistor -  the curve matches also the Anode resistor curve of 2 resistors. Just Screen resistor - the curves have slight shift. I am not sure if it is consistent or not but that's the same behavior with the mentioned here high-current modification of uT3+. 

One think what I noticed is that while doing tests with Va=Vs and then having nothing connected to Anode or Screen contacts, really slows down the acquisition (around 2-3 times, which is still a matter of 3-5 sec). It also should be like that, because uTracer wastefully charges and discharges one capacitor via transistors (probably heating it in one or another way), it might be that the "crosstalk" is caused by something due to some processes during the extra delay time. 

[Martin Manning]

Thanks for following up. That looks similar to what I'm seeing, and it is definitely a function of the current level. I made the plot below a few days ago using progressively smaller resistors.There is a 2k2 in the series, and that's where the phenomenon becomes evident. I also noticed the increase in acquisition time, and you may be on to something there. The high-current modification is essentially just a smaller current sense resistor value.