Ver 1.0 of the 10W PA, using RD16HHF components

[John Williams]

Comments please... Much simpler design with far less components than BS170 based design. See annotations on sheet1 for questions being posed.

John - AC9HY

[John Williams]

Here are the 2 designs that I used to derive this design...

[Sid Boyce]

Looks good.
What drive power?

I just remembered I have the K5BCQ 20W PA built up, requires 1W max input.
73 ... Sid.

On 21/12/14 15:26, John Williams wrote:

Comments please... Much simpler design with far less components than BS170 based design. See annotations on sheet1 for questions being posed.

John - AC9HY

--
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.
For more options, visit https://groups.google.com/d/optout.

-- 
Sid Boyce ... Hamradio License G3VBV, Licensed Private Pilot
Emeritus IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support
Senior Staff Specialist, Cricket Coach
Microsoft Windows Free Zone - Linux used for all Computing Tasks

[John Williams]

100mW, +20dbm. The pa-2008 amp had a drive level of 10mW and 2 stages of amplification. Hoping I can get to +37dbm or more with this design. Will know soon...

John

[Sid Boyce]

OK John,
I have a 20W PA dating back about 1980's, 1mW in and 20W out which I connected to my UHFSDR through an attenuator box and it blew the driver on the UHFSDR which went into oscillation.
The attenuator box is one I built from an article in QST I think, unused for at least 20 years so the switches may have gone bad.

I replaced the chip but that didn't work and I never got around to looking at it again.

When I powered that beast up and tested with a signal generator I accidentally got it up to 44W and had to reduce the input. Even at the 44W level the output looked clean on the scope.
73 ... Sid.

[John Williams]

Minor updates.

On Sunday, December 21, 2014 9:26:56 AM UTC-6, John Williams wrote:

[John Williams]

Have some numbers from the BOMs. All of these are excluding the PCB costs.

1W using BS170's, and full complement of bands - $53.00 more or less. $30 of this cost is relays, so min build cost is around $25.

10W using RD16HHF1's, and full complement of bands - $55.00 more or less. Same $30 cost of relays, so similar min build cost of $25 ~

I like the second design better, brings more value to the table and is much easier to assemble due to reduced parts inventory. Definitely a more robust design.

I will test this one first and if it works well, will probably abandon the BS170 design...

Comments please...

John - AC9HY

[Sid Boyce]

Hope the second one tests out OK, it's a much neater and more robust solution.
73 ... Sid.

--

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.
For more options, visit https://groups.google.com/d/optout.

[Steve Haynal]

Hi John, 

It looks good and worth a go for a test. A google of RD16HHF1 turns up many hits so you have a lot to go on. The cost looks reasonable. OshPark at $5.00/3= $1.67 per square inch for a 2-layer board is a reasonable high estimate. China can be a third of that. 

73,

Steve

KF7O

[John Williams]

First 1W PCB came out to around 15 sq inches... That is why I decided to look for a design with less parts... Here are the gerbers, take a look...

https://github.com/AC9HY/Hermes-Lite/tree/Softrock/frontend/basic%201w%20pa/gerbers

Uploaded to OSHPark and they quoted $112 for 3 boards... Also, did not like the snakeing that I had to do to take a route across the board. 4 layer would be cleaner but way too expensive. Not very good at this, thought about trying an autoroute but could not get by the error checking complaining about board outline. Gave up.  

I did an inventory and I have about 90-95% of the parts for a discrete prototype. Plan to build it after the holiday fuss is done. Will do a conventional prototype on a protoboard with the complete circuit except no filters. If happy will then go for PCBs and can send the prototype to someone else that maybe has some better test equipment.

John - AC9HY

[Steve Haynal]

Hi John,

I took a quick look at your gerber files. The design is big. You may be able to pack things in a bit more. I've found kicad's "drag trace" functionality useful to move things around and pack in.

73,

Steve

KF7O

[Glenn P]

There is a nice 20W amp here as a kit and also several LPF and BPF kits that look nice.
http://www.qsl.net/k/k5bcq/Kits/Kits.html

glenn
vk3pe

[Sid Boyce]

Hi Glenn,
I have used the Version 1 amp which I have sitting idle now and ready. It's a very nice build.
Circuit in my post of 4 days ago.

I started on the BPF kit but never got around to winding the toroids.
73 ... Sid.

--
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.
For more options, visit https://groups.google.com/d/optout.

[Glenn P]

Hi Sid
Looks like  a nice project that could suit this one.  Filters good too. Would save a lot of work on this group, to concentrate on the initial design brief.
Toroid winding is always a bit of a pain !!

glenn

[Sid Boyce]

Hi Glenn,
Definitely as it is a tried and tested design and readily available as an easy to build kit.

The filter kit I got from Kees is the one designed by WB6DHW. The one reason I never got around to winding the toroids - lots of T37 cores take an enormous amount of time and effort and several other projects were on the go and still are.
73 ... Sid.

[John Williams]

Glenn,
The beauty of this project is that you can take the basic frontend and these components, and build your own transceiver. These are good kits.

I have the BPF also. I started out planning to use this filter but got to reading about using the FET switch at high frequencies and that some that built the MOBO (uses the same switch) have observed a decrease in receive sensitivity on 17/15/21/10M.  I ran Elsie on the Peaberry BPFs and really was impressed. So I decided to build a board with those filters instead.

I have a 40/30/20 Softrock and a pre-hardrock 50W amp. I want to build a single board 10M-160M 5-10W transceiver that I can mate to the 50W amp. I am also carefully watching costs so that one can build part of the kit to be more softrock like in an entry SDR radio. Easy to build, low component count and low entry cost.

Others will use the HL and mate them to HiQSDR or Hermes components or whatever meets their needs. The opportunity for this approach is so versatile. I can't help the mainline part of the project, I just don't have the FPGA skills nor the software skills. But I do like building things and can take prior designs and adapt them to these needs. So that Is why I am doing this.

Net is that there will be no "one size fits all". There will be multiple designs and multiple routes to take.

John - AC9HY

[David Jones]

" I want to build a single board 10M-160M 5-10W transceiver that I can mate to the 50W amp."

That is actually my goal too.  Or at least the option of being able to do that.  Like you I have softrock/peaberry's that cover all bands.  But it would be nice to have an all band unit and not have to worry about the external sound card like the sotrocks.

[John Williams]

Contemplating a move to the RD06HHF part. Better match to the 5-10W goal, and a bit better pricing at the 25 or more quantity... Plus the input impedance is closer to 50 ohms... Same TO-220 form factor and same grounded tab so no insulators required for heat sink.

John - AC9HY

On Sunday, December 21, 2014 9:26:56 AM UTC-6, John Williams wrote:

[Steve Haynal]

Hi John,

The RD06HHF looks like an interesting option. I scanned through the datasheet today. 5W-10W is plenty for me, and you may be able to use less expensive/smaller toroids, relays, etc.

Thanks again for working on this even before you have a Hermes-Lite. The Hermes-Lite really really isn't complete without some sort of filters, TR switch and PA. 

73,

Steve

KF7O

[in3otd]

Hello John,
was reading the threads about PA designs, you might be interested in some measurements I did on RD16HHF1 and PD85004-based designs, details on my website at http://www.qsl.net/in3otd/ham_radio/QRPPA2008/QRPPA2008.html and http://www.qsl.net/in3otd/ham_radio/PD85004_PP_PA/PD85004_PP_PA.html ,
The QRP-PA2008 is good up to 15 W PEP, if you increase the bias a little bit, while a PD85004 push-pull can provide maybe 7 or 8 W PEP but those small SOT-89 devices are difficult to heatsink and have anyway a high internal thermal resistance, which limits the maximum power you can get from them in a linear mode.
The power figures above are for a two-tone distortion better than -30 dBc, of course you can get more power if you push them strong enough... the PD85004 are even better than that. Don't know how clean is the Hermes-Lite output, should be fairly good I think, will be a pity to spoil that with a not-so-linear PA. Or maybe there is still room for some sort of predistortion in the FPGA ??

73 de Claudio, IN3OTD

[John Williams]

Claudio,

Thank you so much. I do not have the equipment to do such good data collection.

Can you comment on the attached push-pull design where I propose that we change the input transformer to a 200ohm to 100 ohm impedance match, and drive the pair of RD16HHF1 devices with +20 dbm to hopefully attain 20 db of gain. How do you think it will perform and how will the 3rd order intermod look. What bias would you suggest for most linear performance?

Secondly, I am looking at a design with a single RD16HHF1 device, build according to the attached model. Do you think it's behavior will be any better?

The goal for gain is 5-10W so if either of these hit +17db of gain with the most linear performance and lowest IMD, those will be the key factors.

I can probably send you a sample of the 1 device amplifier for you to test if that is appropriate? I am leaning that way due to low parts count of the design...

John

--

[in3otd]

...forgot to comment on your schematic; you are missing the decoupling capacitors on the output transformer center tap.
You could maybe achieve a better distortion by lowering the feedback resistors between drain and gate and compensating the lower input impedance by increasing the gate bias resistors.
The gate bias will need some sort of temperature compensation, I do not have figures at hand for the drain current temperature dependency, but you might tentatively copy what was done in the QRP-PA2008. As any LDMOS bias circuitry achieving a good compensation over a wide range of temperatures is difficult anyway.
You might consider to have a center tap on the input transformer secondary also, might help to reduce distortion.

Edit: just saw your answer to my previous email, will comment there also.

73 de Claudio, IN3OTD

On Sunday, December 21, 2014 4:26:56 PM UTC+1, John Williams wrote:

[in3otd]

Hello,
for the push-pull PA I will expect about the same performances as the QRP-PA2008, in that design the IMD of the first stage is negligible, compared to the final push-pull stage. The bias circuitry should compensate the drain current temperature sensitivity and provide a low impedance at low frequency, so that the low-frequency IMD products generated and feed back to the input do not actually change the gate voltage and generate further HF IMD.
I expect the single device amplifier to be quite worse, even at half the maximum output power, a push-pull get rid of even harmonics, which at high power levels can mix with the fundamental again and generate IMD3, and in a single-ended stage there is no even harmonics attenuation.

I should still have some RD16HHF1 around here, I might find some time during these holidays to do some measurements; I can measure some of your circuits also if you wish, but it might take some time to do it, outside of holidays, hi.

73 de Claudio, IN3OTD (actually DK1CG at present)

[Zl2APV]

Hi Claudio et al,

wish I had seen your website earlier. I have been experimenting with FDT86256' and getting very similar results to you with the exception of gain falling off with frequency. I had come to the conclusion that the pd85004's would be a better choice for reasons of better frequency response plus source on tab vs drain on tab. Heatsinking was the greatest challenge and thermal feedback was vital. I reckon your design has hit the mark and I am very impressed with your considered approach and careful results. Congratulations on your publishing this and I am sure many will build on this or use the info as is.

73, Graeme zl2apv

[Zl2APV]

Hello list,

I thought I would share the method I used to heatsink my FDT86256's.I used a small pin grid array but any around 30 x 30 mm sink would do. The 4 corners were drilled to take a pk screw and a matching plate was cut out of 3 mm aluminium with a notch in the bottom to clear the fet. Matching holes were drilled in the corners to clear the screws. The fet was soldered to copper foil cut to match the plate so that it sat in the bottom notch and sandwiched between the sink and plate with the 4 pk screws. The bottom edge was drilled & tapped 3 mm for mounting to the pcb via small spacers (a couple of washers) in my case the foil extended beyond the heatsink for connection to the output transformer but of course it would be grounded for the PD85004'S. The heat sensing diodes in the gate bias circuit could be soldered directly to the fet tab in this case rather than clamped to the heatsink. Pin grid arrays are very efficient for their size and effective heatsinking can make or break this kind of amplifier.

There is some great work going on in all facets of this project. Look at what you have kicked off Steve and feel good about it.

73, Graeme zl2apv

[Steve Haynal]

Hi Graeme,

Can you post a picture of this? Sounds interesting.

73,

Steve

KF7O

[Zl2APV]

Hi Steve,

Sorry but I am about 1000 km from home traveling in our caravan. Been away for a month so far and will probably be another 3 weeks before back home. The amp is in my workshop so no photos possible. The copper foil I sandwich between heatsink and plate is for copper chasing or repousse, available from art & craft shops. Computing is a bit limited as using my cell phone.

73 Graeme

[in3otd]

Hello and happy new year!,
I have built and characterized your single RD16HHF1 amplifier following the 'RD16HHF1 1-60MHz 20 dB.pdf' schematic, I have only omitted the TL1 line at the input since it didn't seem to do much difference and the amplifier was easier to build without, hi.
I have put a summary of its measured performances (frequency response, IMD, etc.) at http://www.qsl.net/in3otd/ham_radio/AC9HY_RD16HHF1_PA/AC9HY_RD16HHF1_PA.html . In short, it works as expected, but IMHO its IMD performances are not exceptional, also considering the high quiescent current. The high-frequency response (around 50 MHz) could be improved, the output transformer starts to roll-off there, maybe you could tweak the output network to compensate this; I can send you the measured S-parameters of the transformer if you would like to play with that
Maybe a push-pull stage (with predriver?) might require less power and provide better IMD at high output power. As usual, the best solution depends on the design goals (parts count, cost, output power, IMD, etc.).
Unfortunately I cannot build any RD16HHF1 push-pull at the moment, as I have found just a single device in my junkbox...

73 de Claudio, IN3OTD

[Steve Haynal]

Thank you Claudio for building and testing this PA and providing some detailed analysis.

73,

Steve

KF7O