DB9MAT MRF-101 100 W HL2 PA

[Matthew]

Hi group,

Following on from the thread regarding the DB9MAT HL2 100 W PA companion, I have made some changes to the rx antenna routing. The gateware supports this now. My changes are:

- 0 ohm links can be installed at build to allow either INTTR or EXTTR to control the PA. My logic behind this is we can leave the HL2 TX/RX relay always in TX with INTTR and switch the TX/RX relay on the PA PCB with EXTTR.

- RX signal path is routed through the HL2 dedicated rx input.

- Option to switch between receiving on the TX antenna and a dedicated RX antenna.

- 100 W RF tracks are thickened to 2.0 mm. I did some quick estimations current for 100 W and 10 deg C temperature rises for different antenna impedances and decided to thicken a little (i imagine in reality the FET would blow before track heating became a problem?).

- I have done some experiments with transmitting on 1 band and receiving on a different band at the same time (aspirations of so2r). This wasn't very successful for me, I suffered bleed over of the signal, but this thread suggests it may be improved without the HL2 TX/RX relay involved. As a result of this, I have really tried to separate between top layer and bottom layer the RX and TX signal path.

- PureSignal attenuation is on the dedicated RX antenna side. I think this routing logic will allow RX on the TX antenna and PureSignal feedback on the RX antenna input?

https://github.com/m5evt/HL2-MRF101

This is the first time I have used kicad properly. I tried it around 8 years ago and dismissed it. A lot has changed in the last 8 years. Now after a couple of days using it, I can't see how I will ever use Eagle PCB again, it even imports my Eagle projects with ease.

I've never got a 4 layer PCB made for home projects so I was a little shocked when I saw the prices (with MoQ of x5). So given the cost and the fact that this is the first time I have used kicad I would appreciate some review. It would be great if DB9MAT had an operational feedback to put into this design revision.

73 Matthew M5EVT.

[Reid Campbell]

Hi Matthew,

Just had a quick look at the circuit diagram, are you bypassing the HPF when you switch to external antenna?

Cheers

Reid
Gi8TME/Mi0BOT

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/dd789c60-14fc-4193-b079-c033a868e0c6o%40googlegroups.com.

[Matthew]

Hi Reid,

Thanks for taking the time to have a look at the schematic. Semi-deliberate choice to bypass HPF, I wanted to have 2 relays of isolation from the TX signal. Not sure how necessary this is but going for belt and braces approach.

73 Matthew M5EVT.

[Reid Campbell]

HI Matthew,

I understand now what you are trying. I would be concerned about the overload from broadcast stations but you have already got your receive only antenna, so can experiment by switching the HPF on and off with your current setup. I haven't looked at the layout; would it be possible to have the Rx only antenna's position in the receive chain configurable via 0R links. That might compromise the isolation you are looking for.

I noticed that you are using the contacts in parallel with Rx only relay. As they are not going to carry high current, would putting them in series give better isolation? Probable not, don't know - anybody have any experience of this?

How are you handling PS feed back? You seem to consider it an external input. Should it not be picked up on the board via the output coupler, with suitable attenuator?

You mentioned the cost for the prototypes, what sort of cost are you talking about?

Sorry for all the extra question but I was thinking about it a bit more over the weekend.

Cheers

Reid
Gi8TME/Mi0BOT

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/e5154058-3140-4276-99a9-0ea264311c0fo%40googlegroups.com.

[Matthew]

Hi Reid,

$83 for x5 (MoQ of 5) PCBs to ship to the UK. I would just distribute the spare PCBs. I don't think the component BoM amounts to too much. I've just purchased a Meanwell 48 V DC power supply for this.

To be honest, I don't really know very much about PS.I think Mathis added it on request. You are probably correct that the ideal way would be to route it on the board. Routing wise, this is a bit of a pain, and risks pickup routing near all the high power tracks, it might be better by coax link anyway? I'm not sure if everyone has settled on a coupling bridge design yet. I think this is an area to experiment and move any breakout PCBs onto this PA PCB in the next rev.

If I have bleed over problems, I'll be trying all sorts with cut tracks for relay tracks like you mention and treating this as an experimental PCB!

73 Matthew M5EVT.

[Reid Campbell]

Hi Matthew,

On 08/06/2020 13:00, Matthew wrote:

Hi Reid,

$83 for x5 (MoQ of 5) PCBs to ship to the UK. I would just distribute the spare PCBs. I don't think the component BoM amounts to too much. I've just purchased a Meanwell 48 V DC power supply for this.

Have you looked at JLCPCB. I done a quick check for a 100mmx160mm, 4 layer - about £35 for 5 delivered to UK

To be honest, I don't really know very much about PS.I think Mathis added it on request. You are probably correct that the ideal way would be to route it on the board. Routing wise, this is a bit of a pain, and risks pickup routing near all the high power tracks, it might be better by coax link anyway? I'm not sure if everyone has settled on a coupling bridge design yet. I think this is an area to experiment and move any breakout PCBs onto this PA PCB in the next rev.

I cloned your repo and opened up Kicad. It highlighted that the schematic might be out of date. I noticed in the layout that your attenuator for PS was connected to relay K12 and not pin 6 of of J2. It might be my end as I had Mathis's version on the system.

I'm not well up PS either but it was mentioned that it could be taken off the directional coupler. I have seen other sampling points which used capacitive divider but can't find the reference now.  That might be slight more linear as it avoids the possibility of the ferrite saturating. Maybe that's not a problem either.

Is it the blind leading the blind :-(

Cheers

Reid
Gi8TME/Mi0BOT

If I have bleed over problems, I'll be trying all sorts with cut tracks for relay tracks like you mention and treating this as an experimental PCB!

73 Matthew M5EVT.

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/6d95012e-5815-4695-8bf9-4ccc07eae791o%40googlegroups.com.

[Matthew]

Hi Reid,

Yes, it sounds like neither of us are too wise on PS!

I had the issue with schematic out of date when I cloned Mathis's repo. I presumed it was a feature of kicad files being stored in an internet repo or something similar? Would be good if someone could confirm this is a "normal" message for kicad.

The idea for moving the attenuator was so that for PS, the feedback can come through J3 (and be attenuated), and the rx signal can come through the tx antenna. If it is on J2 pin 6, the rx signal would be attenuated too. This could be changed with the addition of another relay, but we are out of spare control lines. I think it is a reasonable compromise that for PS you cannot use a dedicated RX antenna? If someone really wanted they could add their own breakout board and connect it to J3 with some switching logic.

I have looked at JLPCB. I've read some mixed reports on internet forums. I've used Elecrow in the past and found them to be very good so was going to go with this. If it was a 2 layer board I would go for the cheaper price JLPCB without a thought, but I've had problems with prototype 4 layer boards in the past with internal tracks splashed/shorted and I couldn't inspect this by eye. I wasted hours on the board before I picked this up. This board claimed to be flying probe tested. For a prototype where there are enough unknowns, it feels wise to spend the £20 extra for what some say is better quality?

I think I am about ready to order these now.

73 Matthew M5EVT.

To unsubscribe from this group and stop receiving emails from it, send an email to herme...@googlegroups.com.

[Marc OLANIE]

Hi Matt, Reid et all

 

« I presumed it was a feature of kicad files being stored in an internet repo or something similar? »

 

It seems to be a version warning. You don’t have to worry about that, KiCAD take care of this automatically

 

The « cap coupler » has been discused in a thread of the Apache mailing list as far as I can remember. It works flawlessly and offers a better « gain flatness » from HF to the lower UHF

 

 

 

>I have looked at JLPCB. I've read some mixed reports on internet forums.

 

I’m working with this company for more than 10 years without any troubles, for 2 or 4 layers boards. Quality is « more than perfect » for ham related pcb, from HF to UHF. And they probably are one of the cheapest supplier on the block.

 

I couldn’t find the time to solder Mathis’s amp yet… my pcb are still in stand-by (I’m working on the VHF transverter HL2 extension designed by Hamish OH2GAQ). I’ll probably work on it next month

 

If I had only one modification to do, I would change the relays and chose the Axicom D2n series (same kind used in the modern version of the Alex frontend).

VY 73’

Marc f6itu

 

.

[Reid Campbell]

Hi Marc,

Thanks for the clarifications. I think I hit the wrong option in Kicad and reverted to the previous version of the schematic, hence my confusion.

Matthew, it might be useful to add the capacitor sampling circuit with a ufl, so a coax is all that is needed for PS sampling?

Cheers

Reid
Gi8TME/Mi0BOT

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/BYAPR01MB5016950ECE82B29CA945446AE3850%40BYAPR01MB5016.prod.exchangelabs.com.

[Matthew]

Hi Marc.

Thanks for your input. The 3 V version of the relay series you mention seems to be hard to find and has high minimum order quantity (long lead times on the V23105A5308A201). Would you change the relays purely for an extra amp of headroom on the current rating? I always struggle comparing DC ratings for relays and translating this to RF power/current. It does feel sensible to use a relay that is proven in use for these sorts of RF powers and I imagine there is plenty of experience with Alex

73 Matthew M5EVT.

[Marc OLANIE]

Hi

I didn’t pay attention on the 3V version… good point.

For another project (using MRF300) , I was considering using the 48 V rail -as it is mandadory for this kind of transistors- and use a switching regulator to create a 12V rail.

But I admit, it’s a far more complex approach.

I’m using the capacitive divider on 2 other rigs -an Angelia based tranceiver and a Red Pitaya one. I saw on the Mouser catalog  that Johanson is running a rather agressive marketing campaign to promote it’s high Q, high voltage RF capacitors. Prices are far lower than those given by ATC. That could be a more attractive sourcing for the capacitive RF sampler and for all LPF caps on the board.

Hope it will help

BCNU

Marc f6itu

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Matthew
Envoyé : mardi 9 juin 2020 00:29
À : Hermes-Lite <herme...@googlegroups.com>
Objet : Re: DB9MAT MRF-101 100 W HL2 PA

 

Hi Marc.

 

Thanks for your input. The 3 V version of the relay series you mention seems to be hard to find and has high minimum order quantity (long lead times on the V23105A5308A201). Would you change the relays purely for an extra amp of headroom on the current rating? I always struggle comparing DC ratings for relays and translating this to RF power/current. It does feel sensible to use a relay that is proven in use for these sorts of RF powers and I imagine there is plenty of experience with Alex

 

73 Matthew M5EVT.

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/9945e0b1-8b52-46ca-b0a6-fac427fb389eo%40googlegroups.com.

 

[Probir]

Hi Mathew,

 

In MLCC Johanson  is having only 300volt  RF capacitors that’s what I can see in Mouser.

 

73 Probir VU2BQF

.........................................

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/BYAPR01MB50167ACFFA6BCCEC8417D99DE3820%40BYAPR01MB5016.prod.exchangelabs.com.

[Marc OLANIE]

Hi Probir

Did you had a look on the « kit » packages ?

https://www.mouser.fr/ProductDetail/Johanson-Dielectrics/S-500?qs=yCnrNFeXz%252Bh3v4IYXlPbUQ%3D%3D

https://www.mouser.fr/ProductDetail/Johanson-Dielectrics/S-1KV?qs=yCnrNFeXz%252BgSICPR1%2FIlnQ%3D%3D

Cheers

Marc

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Probir
Envoyé : mardi 9 juin 2020 09:42
À : herme...@googlegroups.com

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/5ED2080ABB1E4409A7508495BB5086DD%40cedalabz.

[Matthew]

Hi Marc,

I was on the website, ready to upload the gerbers and order the PCBs last night. Then I saw your message. I've now paused and thought about this, thanks for your valuable input.

There is 12 V on the PCB to/from the HL2. 12 V seems to open up a wider range of 3 A relays which could help with any obsolescence issues in the future. I think we could keep the ULN2003 and connect 12 V to the COM (pin 9)? Then move to the 12 V version of the relay you suggest. The PCB routing change for this isn't too bad. I propose the V23105A5403A201 which I notice is the same as the one you used on your Alex design.

I will also add the PS cap coupler onto the PCB. I think uFL jumper cables of the correct length might be hard to find. I propose making the new PS coupler connection vertical SMA and changing J3 to vertical SMA.

73 Matthew M5EVT.

[Marc OLANIE]

Hi Matthew

 

It’s probably a very egotist point of view, but I would never « inclued » the predistortion feedback coupler on the pcb itself.

You know, SDR is a hard drug, DDC/DUC based on the OpenHPSDR platform is even more harder and addictive. You begin with an Hermeslite, and quickly you realized that this addiction pushed you to own a collection of HiQSDR, red Pitaya, Odyssey and Hermes boards. An external coupler is probably more effective and useful if it could be used on several rigs. And this way, you will avoid all risk of internal signal coupling.

 

But as I said, it’s a very personal point of view.. I may be totally wrong

 

73’

Marc f6itu

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Matthew

Envoyé : mardi 9 juin 2020 10:52

À : Hermes-Lite <herme...@googlegroups.com>
Objet : Re: DB9MAT MRF-101 100 W HL2 PA

 

Hi Marc,

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/117baef1-8616-4581-8979-df890964d11fo%40googlegroups.com.

[Matthew]

Hi Marc,

I've just been looking through your Alex BoM for the HPF and LPF caps. I've been costing this up and the kit you link to below does look appealing (1 kV). It would be nice to know exactly what came in the kit. One would hope at least x5 of each 1206 value? And all standard pF range values?

73 Matthew M5EVT.

[in3otd]

Hello Matthew,

I also was curios to know what was included in the kit exactly but I wasn't able to find any detail even on the Johanson website.

Then I saw that Digikey has the content details at the end of the related product page, hi.

73 de Claudio, IN3OTD / DK1CG

[Matthew]

Thanks Claudio, not quite enough of the ones that I need. This kit could be great for someone to design an auto atu around.

I changed the PCB to use Axicom D2n (12 V relay) for the high power RF lines. I kept the smaller relay footprint style for the HPF and RX ant switching, but I've also changed these to be 12 V relays. I think this should open up more possible options for relays when specific ones are out of stock.

I decided to order the PCB from jlpcb. I then started looking at components, but I wish I had done this before ordering the board. 0.036 ohm current sense resistors in an 0612 footprint are hard to find without a long lead time and/or MoQ. The MAX11645 is currently listed as out of stock. Stock due late July with Farnell. The MAX11613EUA seems an ideal replacement, but with a different I2C address. I will order one of these for the prototype.

I can find most of the LPF capacitors @ 500 V in the Johanson range specified above. However, there are some values that won't be in stock until much later this year. Marc f6itu has put in a lot of hard work to specify good caps for this Alex PCB, however, these are in a 1111 footprint and not 1206. It is a bit tight under the board around these caps and I'm not sure the 1111 caps will fit (with a 1206 footprint). I am going to look at if I can re-design the LPF to work with the more commonly available Johanson caps that are in stock.

I'm a bit lost in sourcing components and re-designing the filter caps values at the moment, but I will try and update the github repo with the new PCB design one evening this week.

73 Matthew M5EVT.

[Marc OLANIE]

Hi Matthew

 

« I'm not sure the 1111 caps will fit (with a 1206 footprint). »

You’re right…. They don’t ☹. I chose the 1111 « square » format because it was compatible with both the Vishay and the ATC form factor

 

After several measurements, on these « low » frequencies, the Vishay caps gave me approximately the same performances than the high priced ATC caps, with the same voltage ratings

 

Sourcing is boring :-/ but it must be done…

73’

Marc

 

[Matthew]

Interesting, did you measure the ESR (at HF) with a VNA? Or just look for heating/RF power loss? I was thinking about buying one or two and using a VNA to investigate at HF to try and put some numbers on it because my previous non-scientific observations when I have bought some of these fancy RF caps agree with yours.

I was playing around with some capacitor values for the filters and noticed the current 40/30m filter doesn't present anything like 50 ohm at 10 MHz. I simulated this using qucs. I wonder if someone else could confirm this with LTspice?

73 Matthew M5EVT.

[Graeme Jury]

Do the caps fit standing on their side? I think that it was Wes Hayward who suggested side mounting lowered parasitics and I did this on my filters although I don't know if it made a difference.

73, Graeme ZL2APV

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/a460f8d2-0162-4d6f-961e-7aad784acbc7o%40googlegroups.com.

[Scott AK5SD]

Here is the S11 (Red, Right Scale) and S21 (Blue, Left Scale).

S11 is only 8 or 9dB at around 4MHz. There is also a fairly significant (additional 0.5dB) dip in the S21 at that frequency (I was assuming very high Q inductors so the actual results might be worse). I'm not sure how the component values were determined, but this is worse than what I would expect by just choosing the nearest standard value component.

Scott

AK5SD

[neil whiting]

I also mount my ATCs vertically where they are highly stressed.

This is recommended by ATC themselves for stripline use but applies to other situations too.

See:

https://www.rfcafe.com/references/app-notes-copyrighted/atc-select-cap-wireless.pdf     (last paragraphs)

https://atceramics.com/userFiles/uploads/pdfs/vertical_orient.pdf

73,  Neil G4BRK

To unsubscribe from this group and stop receiving emails from it, send an email to herme...@googlegroups.com.

[Marc OLANIE]

Hi

 

So far,  I didn’t had any sourcing problem. The MAX11645 is available at Mouser (I’m often mixing Mouser and Farnell orders to reach my goals :- )  )

I never measured the ESR of the Vishay caps, only the Q compared with the one of the ATC. They seems reasonably close, that was the reason why I chose this brand. I will run a new batch of measures on the fourth version of my lpf and keep you informed. My third iteration was completely messed up (too much parasitic capacitive effect with the inner layer of the pcb… that was a beginner’s mistake)

… and yes, these Vishay caps are perfect when soldered vertically

https://wiki.electrolab.fr/images/8/84/Lpf_condos2.jpg

but frankly, I doubt it could have any measurable impact below 30 or 60 MHz, or even in the VHF range. But it’s a convenient way to stack 3 caps on the same footprint when using parallel combinations to reach the exact C value

 

VY 73’

Marc f6itu

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de neil whiting
Envoyé : mardi 16 juin 2020 10:02

À : Hermes-Lite <herme...@googlegroups.com>
Objet : Re: DB9MAT MRF-101 100 W HL2 PA

 

I also mount my ATCs vertically where they are highly stressed.

To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/5c715762-ef2f-47c5-ab6b-e82318abaf15o%40googlegroups.com.

[Matthew]

Thanks for the links and interesting discussion on vertical mounting of the caps. I've done things like this in the past when I got the footprint wrong on a PCB design or didn't have the right sized component to hand. I can now pretend this was for a sound technical reason.

I finally got there with my order mixed between Mouser and Farnell, except for the MAX11645 substitution. I couldn't find the MAX11645 at Mouser (and I just check mouser.fr too https://www.mouser.fr/Search/Refine?Keyword=MAX11645). Am I missing something obvious?

73 Matthew M5EVT.

[Marc OLANIE]

Well… this is not the first time the Mouser search engine gives some all purposes obscure messages….

Try this link

https://www.mouser.fr/_/?Keyword=MAX11645EUA%2B&bws=1

hope it will work

Marc

 

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Matthew
Envoyé : mardi 16 juin 2020 16:31

À : Hermes-Lite <herme...@googlegroups.com>
Objet : Re: DB9MAT MRF-101 100 W HL2 PA

 

Thanks for the links and interesting discussion on vertical mounting of the caps. I've done things like this in the past when I got the footprint wrong on a PCB design or didn't have the right sized component to hand. I can now pretend this was for a sound technical reason.

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/e9bdc699-3d1b-48d3-a523-72b468d6f780o%40googlegroups.com.

[Matthew]

Components are still arriving in endless separate packages from Farnell and Mouser. The boards arrived today. But some of the important components won't be here until early next week. Some mechanical and thermal challenges to consider in the meantime. Plus I still need to tidy up the schematic and upload gerbers to github.

Does anyone have any idea how Mathis has connected the stacked the PCBs? I can't really see from any of the photos. I'm guessing the boards are spaced by something like 50 mm?

73 Matthew M5EVT.

[Marc OLANIE]

Hi Matthew

I think Mathis used a 150x55cm « official » aluminum box, thus implying that the space beween the two boards is 40 mm. Hope it won’t affect the Z in the receive path.

 

As I’m running 4 different projects at the same time, I didn’t finished yet the « original » version of the amp. Some components are missing (the current shunt resistor, all « power » resitors, some caps and… the output bnc) . I’ll probably send an order next week to finish it.

 

Oh, btw, I think that a SMA/BNC « dual footprint » would be nice, allowing people to chose their own kind of coax output. I really don’t trust bnc’s,  and SMA connectors can handle 100W with a serious security margin (approximately 1 kW below 100MH). steel angled connectors are far better than a BNC from a pure mechanical point of view.

Just my 2 cts

Marc f6itu

 

 

 

 

De : herme...@googlegroups.com <herme...@googlegroups.com> De la part de Matthew

Envoyé : jeudi 18 juin 2020 21:15

À : Hermes-Lite <herme...@googlegroups.com>
Objet : Re: DB9MAT MRF-101 100 W HL2 PA

 

Components are still arriving in endless separate packages from Farnell and Mouser. The boards arrived today. But some of the important components won't be here until early next week. Some mechanical and thermal challenges to consider in the meantime. Plus I still need to tidy up the schematic and upload gerbers to github.

 

Does anyone have any idea how Mathis has connected the stacked the PCBs? I can't really see from any of the photos. I'm guessing the boards are spaced by something like 50 mm?

 

73 Matthew M5EVT.

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/2f363fda-d96f-445f-a4c0-dcd4b1a8fe33o%40googlegroups.com.

[Christian Veith]

Hi Marc,

Mathis posted some pictures of his planned enclosure in the original post.

Ich was bigger as the 230/50V power supply as well as the heatsink / fan combination was also included in his box. The spacing between the HL2 and the pa is given with the two 2.54mm inline pin connectors.

Best regards

Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/BYAPR01MB501648208EF3A2858C472B97E39B0%40BYAPR01MB5016.prod.exchangelabs.com.

[Matthew]

Short update.

I have wound all the coils, populated all the filters (change 40/30m values for maybe better Z=50) and swept with my VNA to measure RL and TL. Switching filters from software/HL2. Everything is looking good so far.

I am still yet to settle on a good way to connect the 2 boards. I tried 30 mm spacers between the boards. I have male 2.54 mm headers on each board. I am using female, wire-to-wire loom but this clashes, so I've had to space the boards some more. I'm wondering about going much closer with the boards and trying some PC104 board-to-board connectors. But then I'm concerned about all the clocks from the HL2 making it into the filters/RF lines on the PA board. More experimentation required. I need to decide on this mechanical arrangement before proceeding with the PA components, as the heat sink/FET location will become the next part of the 3 dimensional problem.

73 Matthew M5EVT.

[Matthew]

I did some quick tests with the filters and the rx line from the PCB connected to the HL2. I see the same 1.8 MHz clock noise seen with the N2ADR filter board. However, as usual, toggling the FPGA-generated power supply switching clock signal removes this. I was quickly switching through the bands and noticed that around 21.020 MHz there is a spur when the FPGA-generated power supply switching clock is enabled. Perhaps SDR software should have the option for frequency selective enabling/disabling of the FPGA-generated power supply switching clock?

The RX/TX antenna switching is working as expected. Some Samtec connectors arrived today from the ESW series https://www.samtec.com/products/esw these will allow 20 mm clearance between boards.

I will install these connectors at the weekend and do some more detailed comparisons between the N2ADR PCB and this PCB for noise/spurs. I'm still skeptical with the VSWR bridge on the PA PCB is right over the HL2 clocks. I also need to experiment with the small heatsink I have sourced for the HL2 PA to check it is sufficient.

73 Matthew M5EVT.

[Steve Haynal]

Hi Matthew,

I didn't see any spurs around 21.020 with the FPGA-generated clock on or off. Maybe something is different between our setups? The top was off of mine. I've seen people cover relays with foil and then ground the foil. Maybe that can help. I never see any spurs on my HL2 that does not use the N2ADR filter board. It is also possible to have several different FPGA-generated frequencies to switch between.

73,

Steve

kf7o

[Matthew]

Sorry, I was in a rush last night when I wrote this and wasn't clear. The spur I was seeing on 15m was with the MRF-101 HL2 PA board connected and no N2ADR PCB. At the moment I have this PCB temporarily lashed up about 50 mm above the HL2 PCB. The signal routing is antenna in on MRF101 PCB -> MRF101 PA filters -> RF3 on HL2 PCB.

I also wasn't clear, I didn't mean that the VSWR bridge is above the HL2 clocks, I meant above the HL2 switching power supplies.

Yes, I have ideas for foil, or a metal sheet between the PCBs if needed.

This weekend I will do a full test of all frequencies with N2ADR, bare HL2 and with MRF101-PA PCB and post the results here.

To answer an offline message I was sent, the FPGA-generated clock on/off is set by software, linHPSDR has this option and I know Quisk does too. Not sure about others. Depending on build of HL2, some modifications are needed for older ones (I think). A search of the archives of this list should clarify this.

73 Matthew M5EVT.

[Matthew]

The kicad files for the PCB I ordered and I am building are here:

https://github.com/m5evt/HL2-MRF101

Measurements for the filters are here:

https://github.com/m5evt/HL2-MRF101/tree/master/measurements

The filter settings for SDR software are here.

I have got far enough with the build that I don't think there is anything catastrophic in my changes. I some spare PCBs. Reid, Alan, Steve - are you interested in a PCB if you cover postage?

73 Matthew M5EVT.

[Alan Hopper]

Hi Matthew,

yep, very interested in a pcb, perfect timing as I've recently been spending more time on voice and am tidying up the analog modes of Spark.

73 Alan M0NNB

[Reid Campbell]

Yes, I'll take one so I can use to check mod's to PowerSDR to support it. Let me know the costs and I can pay via phone number if your account supports it.

Cheers

Reid
Gi8TME/Mi0BOT

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/c07c7f4a-0cb3-43a5-a22a-f25e782c4e9ao%40googlegroups.com.

[Steve Haynal]

Hi Matthew,

Yes, I am interested. Please let me know shipping and handling costs.

Do you know what has happened with Matthis, the originator of this project? I haven't seen any posts from recently. He was going to send me a unit at one time.

73,

Steve

kf7o

[Marc OLANIE]

Hi Steve et All

 

>Do you know what has happened with Matthis, the originator of this project? I haven't seen any posts from recently. He was going to send me a unit at one time.

 

We had a quick Tweeter « ping » two days ago, I think he has quite a lot of professional obligations these days.

 

If your QRZ.com postal adress is ok, I can send you a board asap

73’

Marc f6itu

 

[Matthew]

All boards are accounted for. Alan, Reid, is your address on QRZ correct? I imagine it will be cheap to post via Royal Mail. Steve - I will look at the best options to post from the UK and get back to you early in the week.

I haven't heard anything from Mathis. I assumed he was busy with other things.

73 Matthew M5EVT.

[Alan Hopper]

Hi Matthew,

yep address on qrz is fine.

73 Alan M0NNB

[Reid Campbell]

Yes, I'm correct in QRZ.

Cheers

Reid
Gi8TME/Mi0BOT

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/f23eef18-c2d3-4c51-b9c3-6d0b3d2e96cfo%40googlegroups.com.

[Matthew]

 I connected up the PCB stack today using the Samtec connectors from the ESW series. See attached photos (top and stack.jpg). This results in a stack height of around 21 mm between PCBs. 20 mm hex spacers and some washers were used.

I have the RX signal routed from the TX ANT on the MRF101-HL2 PCB through the filters and into the RF3 connection on the HL2. I haven't removed J1 on the HL2 (yet). I can switch between RX antenna in (no filters and routes straight to RF3) and TX antenna for RX.

I did a fairly thorough search across HF and in particular the ham bands for noise problems. When I updated the PCB, I deliberately routed the RX antenna in away from the HL2 PCB, no I2C lines nearby etc. The switch mode power supply spurs from the HL2 PCB are much stronger with the MRF101-HL2 compared to the N2ADR board. Of note are the following. (N.B. all tests are with no antenna connected, LNA gain was fixed and auto waterfalls were turned off in linHPSDR):

15m - There is a small amount of noise in the CW section of 15 m, with and without FPGA-generated power supply switching clock. Changing to RX antenna, this disappears. FPGA-generated power supply switching clock results in some noise around 21.030 MHz. Turning this option off, more broadband noise is around 21.100 MHz. This might not be noticeable with an antenna and atmospheric noise? The attached file 15m_fpga_on_off.png shows the effect of FPGA-generated power supply switching clock (F-gpssc) on/off.

60m - F-gpssc both on and off results in noise (in different places), but none in the UK portion of the 60m band. Switching to RX antenna, no noise present. This is shown in attached 60m_rx_ant.png

Down in the medium wave section F-gpssc both on and off results in noise (in different places). Again, rx antenna produces no noise. The spurs are quite strong in both configurations, see attached mw_fpga_on_off.png. The magnitude of these spurs can be compared with N2ADR.png attached.

I did wonder if it was pickup from the coils, but it isn't easy to test this as there is no LPF bypass configuration.

In summary, I don't think anything is a show stopper. For the shortwave listener these might be a problem, but a shortwave listener won't need a 100 W PA(?). There is enough space in the stack to put a shielding plate, I did something similar with homebrew MCHF and this really reduced noise on the RX. I may consider this in the future, but I don't have the motivation for this at present.

I am nearly ready to build the PA part of the PCB, but first I need to build a 100 W dummy load. My QRP dummy load in a temporary pot of mineral oil got quite hot when I built my HR-50.

73 Matthew M5EVT.

[Matthew]

Hi Reid,

I can't seem to figure out how to contact you directly. I've just updated my qrz page with an email address. Could you drop me and email?

73 Matthew M5EVT.

[Steve Haynal]

Hi Group,

I saw this build of the DB9MAT PA by Marc: https://wiki.electrolab.fr/Projets:Lab:2020:Ampli_100W

I've received my bare PCB from Matthew. It looks very nice. I don't know when I will find time to build it. Is there anyone in the US interested in building this? I will send the PCB and pay for the parts. You can keep it for a few weeks of play before returning it to me.

73,

Steve

kf7o

[terry long]

Hello Matthew,

Nice work.  Can you post on GitHub, if you have a completed BOM with recommended manufacturing (or Mouser/Digikey) part numbers that can be used to source the components?  Thanks.  This would help those of us who might want to build up the board from scratch..

Terry

N8AB

[Matthew]

I've just quickly done a lot of copy and pasting effort to produce:

https://github.com/m5evt/HL2-MRF101/blob/master/HL2_MRF101.csv

Please double check the code matches up with the value, it is likely I have made some copy and paste mistakes. Others contributing to this BoM via a pull request would be much appreciated.

I've been side tracked by a few things recently, but this week I soldered the final components onto the PCB. I'm going to look at adding the code support in linHPSDR and installing the heatsink onto the FET.

73 Matthew M5EVT.

[Steve Haynal]

Hi Matthew,

I've added links to your fork and the original to the community projects wiki page:

https://github.com/softerhardware/Hermes-Lite2/wiki/Community-Projects

73,

Steve

kf7o

[Matthew]

I haven't really spent much time on this over the past month. I have got the MCP4561 pa bias working in linHPSDR. I'm looking at the options for reading the MAX11645 ADC (for the hl2-mrf101 temperature and bias current). I've considered i2c reads via software but it looks like doing this via a new instantiation of slow_adc verilog module shouldn't be too bad?

As far as I can tell the memory map of the MAX11645 and MAX11613 are similar enough that the slow_adc module will work. The chip addr needs moving out of the module (currently hard coded to 0x34). Then in the HL2 to PC packet RADDR 0x03 could be used to send back the hl2-mrf101 temperature and bias current. However, I'm struggling to get my head around how all the i2c bus writes are scheduled (in control.v). Some pointers/help for how to schedule a second set of ADC reads would be appreciated.

73 Matthew M5EVT.

[Steve Haynal]

Hi Matthew,

What i2c bus does the MAX11645 connect to? The same one that is currently reading the MAX11613 and found on DB17? If so, it may be possible to add a couple of extra states to read the MAX11645. I don't like the idea of instantiating another slow_adc module as that also instantiates another i2c controller which takes measurable FPGA resources. Right now there are two i2c controllers: one dedicated to reading the slow ADC only because the original protocol has fwd/rev data returned all the time, and the other is generic for read and writes of the other two i2c busses. I would like to evolve the Verilog to have only one i2c controller. I'd prefer if software reads the values when needed via a generic i2c read as it is less complexity in the FPGA where we have limited resources.

73,

Steve

kf7o

[Steve Haynal]

Hi Matthew,

I think it was late when I responded to this and don't think I did so clearly. I don't think it is a good idea to add another slow_adc instantiation as that is relatively costly in terms of FPGA resources. However, I am willing to add periodic reads of the MAX11645 to an existing finite state machine as that is typically pretty cheap in terms of FPGA resources. This can take two forms. If connected to the i2c bus on DB17 that the MAX11613 is already connected to, we can add a few states to the existing slow_adc FSM to read the other ADC. If connected to one of the other i2c busses, then I'd like to follow through with a plan to merge the slow_adc FSM with the FSM that already exists for generic i2c and add a few states to that FSM to read the MAX11645. This should require fewer precious FPGA resources in the end. 

Currently, the slow_adc reads about every 30 ms. If combined with the generic i2c FSM, we could still have it read both slow ADCs about every 30ms, but any i2c traffic from software (which is actually very rare) would take precedence and may bump a slow_adc read.

There is also the ability to just read the values from software using the generic i2c interface when required.

73,

Steve

kf7o

[Matthew]

Hi Steve,

Don't worry about the late reply, it made sense to me. I should have been able to work out myself that another slow_adc instantiation is not an efficient use of resource.

The MAX11645 is connected on DB7 (i2c bus labelled SCL2 and SDA2 on the schematic) and is on i2c ADDR 0x36. Beyond that, it seems to behave the same as the MAX11613, but with only 2 AIN. Given that people seem to be making various amplifier boards for their HL2s, I figure that a generic "external amp temperature" and "external amp current" are useful for others, especially with the i2c extender board recently posted. So I think this would be good to have in gateware, but I understand the pressure for space and the need to do more "clever" things with available logic blocks.

I have been adding code to better support i2c reading in linHPSDR. I can read EEPROM values and have working code but I still haven't got my head around more complex i2c reads and I'm using the MAX11645 as a useful exercise to do this. I've already hit a few obstacles that I've had to re-think linHPSDR code.

As far as I can tell it would take me 2 i2c read/writes to get a value from the MAX11645. A blind write isn't a good idea because I wouldn't see any potential error message (in the RADDR)? I need to space these out to avoid starving normal messages. The MAX ADC needs an ACK for each byte sent to the master. Does the i2c ACK get handled by the hl2 or does this just get passed straight through to software and the software needs to send the ACK (perhaps something related to addr 0x3d data [23])?

A couple more i2c questions that I would like to add some info on the wiki for:

- If an RQST is set, does the response come in the next "normal" scheduled hl2 to PC iq packet?
- A read (ADDR 0x3c data[31:24] = 0x7) must have C0[7] set to 1 (RQST) to allow the data that has been read to be sent to the PC? A read with no RQST results in no data sent to PC?

I would love to see some more user designed peripherals (auto atu?) linked up via i2c to the hl2/software, hence why I am interested in better supporting i2c reads.

73 Matthew M5EVT.

[Steve Haynal]

Hi Matthew,

The slow_adc is currently not accessible via software. All you have is the return values. The FSM in slow_adc.v might be helpful to understand how i2c is used to read values. One command is written and then 8 bytes (all ADC data) are read. This process repeats every time we sample.

To answer your questions, a response will be sent the next hl2 to PC IQ packet once the i2c has finished. Depending on the length of the i2c transaction and the number of radios/bandwidth, this may not be the next immediate IQ packet. Responses will only occur in one of the two 512 byte frames that are part of a hl2 to PC IQ packet. The response will replace one of the normal responses that are cycled through. Software should only issue at most one outstanding transaction with expected response. Software should also only issue a single command with expected response in one of the two 512 byte frames that are part of a PC to hl2 packet.

If the C0[7] bit is set to 1, the HL2 will always issue a response for a command. For most commands it is just the address and data of the command that was issued, and is set pretty much immediately. Only for i2c is the response possibly delayed and contains real data. If the C0[7] bit is not set, then the HL2 just continues to cycle through its standard response addresses.

73,

Steve

kf7o

[mathis.s...@gmail.com]

Hi everyone,

Sorry for being AWOL for so long. I am still alive, but the past couple of months have been crazy busy and I didn't really have the "brain capacity" left for this. That will hopefully change now though! :)

@Matthew: Could you get in touch with me off list? mathis.s...@gmail.com - I'd really like to see your improvements merged into a single 100W PA companion for the HL2!

@Steve: I shall get in touch with you soon regarding the board I wanted to send to you! If you're still interested and have the need for a board, I'd very much build one for you!

Anyways, I've got quite a bit of reading to do, a lot has happened here in the last couple of months.

73s, Mathis, DB9MAT

[Steve Haynal]

Hi Matt,

I understand what you mean. It has been busy for me too, and I haven't gotten as far on beam forming as I'd have liked to. I have one of M5EVT's board, but haven't built it up. I would like a built board, and am willing to pay for parts and assembly if someone is interested.

73,

Steve

kf7o

[mathis.s...@gmail.com]

Hey group,

I have been in contact with Matthew M5EVT and we agreed to merge his additions into the main repository, which I have done today. Furthermore, I have cleaned up the git repository, added a README and am in the process of assembling a little bit of documentation in the Wiki of my repository. Several people are building or have already built boards and I would love to get their input into the Wiki. I am trying very hard to at least somewhat complete the documentation before starting the next project I have in mind, an interface and control board to connect G4DDK Anglian and Iceni 2m and 70cm transverters to the HL2... :)

@Steve: If you don't find someone more local to you to build a board please ping me again and we'll work something out!

Best 73s,

Mathis, DB9MAT

[terry...@gmail.com]

I am intersted in possibly sourcing board a couple of boards and componts to build.  Is there a rough estimate in board and component costs for the completed AMP/LPF?

Terry

N8AB

[J P Watters]

I would also be interested in building a board or two. They would go
with the pair of HL2 build 9's that I have.

I was just sourcing 300W LPF and amplifier components.

I also have a couple of transverters, 144mhz/28mhz and 440mhz/28mhz
that I could experiment with.

..jpw J P Watters
KC9KKO
Morris, IL USA

> --
> You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/14858e94-8eb3-47a4-be6b-217cd4113ba5n%40googlegroups.com.

[Matthew]

It depends on a couple of things. If you order from JLPCB and then manage to split up the minimum 5 boards between others, it probably comes out somewhere near $20 per board (depending on tax/duties). Then somewhere around $150-200 for components depending on how good a junkbox you have?

73 Matthew M5EVT.

[Matthew]

I had a message offline asking about this PA project and I thought I would reply to the group. If a Makerfabs run was to be setup for this, someone would have to organise it. Most of the components are 0805 or bigger, so it shouldn't feel too daunting building this. Once the PCB is built, the second half the challenge them becomes mechanical/thermal management (and finding a 48 V PSU). In the first instance there should probably be a few more people building them to understand performance before anyone organised a large run.

To order a PCB, you will need the gerber files. I was recommended JLPCB and used them for this 4 layer board. Mathis has now merged my changes into the main project repository at https://github.com/mathisschmieder/HL2-MRF101. Everyone should use this as the point of reference now. However, I notice the gerbers are no longer there. Mathis - is the removal the gerbers from the repo accidental?

73 Matthew M5EVT.

[Steve Haynal]

Hi Matthew and Group,

Another option would be for someone to organize a partial build. Places like JLCPCB have a standard parts library:

https://jlcpcb.com/parts

Someone could create a BOM for all the parts with library equivalents on the HL2-MRF101 and have a small batch of boards partially assembled.

For any assembly run, I think the toroid winding will be the toughest to setup and get right.

73,

Steve

kf7o

[nekv...@gmail.com]

Matthew,

U2 on the schematic is MAX11645, while in the BOM it is MAX11613. I believe MAX11613 is correct, as the datasheet show it as a 4-channel ADC.

73 Brian K1LI

[Matthew]

My notes say this must be MAX11645 and you cannot substitute for the *43. My notes suggest I2C address 0x36. I think there is a *13 on the HL2 and 2 devices on the same address would likely be troublesome.

Sorry, I don't have time to check this all properly now, I suggest looking at datasheets for the 2 devices and checking the HL2 schematic.

73 Matthew M5EVT.

[Mathis Schmieder]

Exactly. The address unfortunately is fixed for these devices, so we need to use the 45 on the companion. Originally I wanted to use the 43, as I wasn’t aware that the I2Cs are shared. We only need two channels though, so it isn’t a problem.

--
You received this message because you are subscribed to a topic in the Google Groups "Hermes-Lite" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/hermes-lite/41jxoNJDfss/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/8c6749d4-e9e8-4326-b708-a3a2ba833cb7o%40googlegroups.com.