New RF transistor for inexpensive QRP PA

[Steve Haynal]

Hi Group,

I saw that DigiKey is now selling the AFT05MS003N for $1.79. This is a new transistor introduced within the last year that appears to be a good candidate for an inexpensive QRP PA. It is speced down to 1.8MHz. In push pull, two should be able to produce 5W. What do people think?

73,

Steve

KF7O

[ZL2APV]

This one looks pretty good and compares very favourably with the PD85004 (at $US3.33). Having a common source and heat tab simplifies design and the Rthjc of 4.1 is really good in terms of heat removal. It should be very flat to 50 MHz and probably higher. I would certainly consider a pair of these for a 5 watt amplifier with its attendant 2nd harmonic advantages although at the price of a little more circuit complexity.

73, Graeme zl2apv

[John Williams]

A very interesting device. Would be fun to give it a try.

John W9JSW

--
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 Graeme,

I am considering including John's filter and possibly PA on a single larger board along with the HL V2. There are more enclosures available for larger boards, it simplifies and reduces costs, it sets us apart from the Red Pitaya and Lime SDR which do not have PA and filters. I really need to do more of the layout before making a decision though. My biggest concern with the current PA is the stronger 2nd harmonics. Also, since the opamp has balanced output, we could simplify things by going directly into a balanced amplifier. If someone where to test and build a push/pull PA, I'd be very happy and glad to include it in the first V2.0s.

73,

Steve

KF7O

[Steve Haynal]

Hi John,

Thanks for the boards! I received them in the mail today, and they give me a much better feel.

As I mentioned earlier, I am considering 1 single board with HL2.0 + filters. I can simply reuse most of your layout for the filters. I've been considering 10cm by 15cm boards as that is a common size for enclosures and good PCB price at Elecrow. For example, see 4004H-6N which I can buy at my local Fry's. This size may be too small to also have the PA on board, but the case is shallow enough that the PA could live on a small 5 cm by 5 cm board and heat sink the transistors on the other side. There are less connectors involved with just putting the PA on an extra board. The amp stages are where I expect the most experimentation anyway. What do you think?

Regarding this transistor, it may need an odd voltage, but we already have to have a regulator for the opamp design and could share that.

73,

Steve

KF7O

[ZL2APV]

Hi Steve,

I seem to have so much on the boil at present but I would like to get a push pull amp built up as I believe that is ultimately the best way to go. I have an experimental one built from a pair of switching fets in the same packege but with the drain connected to the tab and with its associated heat transfer problems. I did not expect it to be flat right right out to 30 MHz - and it wasn't but it did give a beautiful clean 5 watts out on 20 metres until it thermally ran away on a sustained output. I had no thermal compensation. This result and the excellent work done by Claudio is encouraging me to make a serious effort.

I agree that using the push pull output of the v1.42 board is a good idea but it really forces you to use a separate Tx and Rx chain in that case. Jim has got the output so flat as it stands that another transformer in the path if we stay with the switched outputs as it is at present would not really matter and the cost is negligible plus the amplifier can be tested stand alone. My current on air radio using a HiQSDR board goes from the board's +20 single ended output to a pair of push pull 2N5109's which have a choke dc feed to the collectors and are capacitor coupled to a pair of RH16HHF1's in line with what you are suggesting but just 1 step further up the chain. This works well.

I wonder if you are looking at a couple of Pereguine switches for switching between radio or VNA outputs or do you feel that that should be added as a user outboard connection? Same goes for Ant1/Ant2 switch as Quisk provides for this facility already and I can easily add it to ghpsdr3-alex if it is needed.

John's filter is looking really good. You are probably not aware that Glenn VK3PE is building the whole set (Rx and Tx + 5W amp) and is currently testing them. I am waiting a few bits and pieces and will do the same. We will publish the results and a build help will be added to the wiki. My intention is to follow this with the 5 watt pp amplifier and John has already started his push pull version so hopefully by the time we get that done, you will have your v2 and software and the radio software guys will have their stuff going too.

73, Graeme zl2apv

[James Ahlstrom]

This is an interesting device, and it is always nice to see a new RF transistor, especially one useful at HF.  I am concerned about the 7.5 volt operation.  This requires another supply.  And there is no mention about linear service in the data sheet.  Mosfets often struggle with IMD problems, especially at low supply voltages.

Jim

N2ADR

[James Ahlstrom]

Hello Steve,

As it happens, I am rethinking the 100 watt power chain at my shack.  My current setup is several separate boxes connected together, and it is messy.

An interesting device I came across is the 2SC6144.  This is a bipolar switching transistor, but manufacturers have been pushing up the speed, and this one has an Ft of 330 MHz.  It handles 50 volts, 10 amps and 25 watts, so it should be OK for 5 watts push pull.  It costs only 69 cents at Digikey.  Its real advantage is the TO-220F package.  This is the size of a regular TO-220, but the case and tab are plastic, so there is no need to insulate it from the heat sink.  Insulation is always a problem since the tab is usually the collector in non-RF transistors.  I don't really believe the 25 watt spec because they don't give the degrees/watt on the data sheet.

The straight forward RF transistor solution for 12 volts is the RD06HHF1 or the RD06HVF1 by Mitsubishi.  These are $4.25 from RFparts.  They are TO-220 RF MOSFET transistors and the tab is the grounded source, so no insulation is required.

Buying linear-service RF transistors for HF seems like an exercise in frustration.  Should we really be paying $9.95 for an MRF475 from the 1970's?  An interesting and relatively recent (2010) discussion of RF and non-RF MOSFETS is http://ludens.cl/Electron/mosfetamps/amps.html.  Start half way down the page at "MOSFETs for high power class AB linear amplifiers".

Jim

N2ADR

[James Ahlstrom]

Hello Steve and Group,

I like the idea of using boxes with card slots, but they are inconvenient if you want to bolt TO-220 transistors to the case.  So Steve's suggestion of a separate amp board make sense.  But putting the very bulky filters on the board with the Hermes Lite makes HL less useful as an instrument.  Graeme has suggested separate switched Tx/Rx outputs for VNA use.  Certainly some will want the HL board without the PA and filter.   Also, the filter would change for 6 meters.

There are many boxes available with card slots.  The box Steve mentioned is for a 4 inch board, which is 101.6 mm.  I presume that really means a standard size 100 mm wide board.  The other boxes from Context Engineering for that width are all 152 mm long, and 32, 40 and 62 mm high.  Hammond Manufacturing has boxes for 100 mm boards that are all 34 mm high, and 120, 160 and 220 mm long (and come in cool colors).  There are other manufacturers too.

Suppose we divide HL into three boards, all 100 mm wide.  There is a main board with everything from the Ethernet jack to the op amp RF output, a second board with the filters, and a third board with the power amp.  The main HL board is as short as possible, and less than 120 mm.  People who don't want the power amp put the HL board into the Hammond 120 mm long, 34 mm high box (1455L1201).  All connectors are on one side, which I will call the "back".

For people who want the amp, they probably want the filters too.  Steve has a good reason the prefer the filters on the main board;  there are a lot of wires for the relays and T/R switch that connect them.  We have several solutions.  Since we have card slots, we can switch to the longer Hammond 160 or 220 mm boxes (1455L1601, 1455L2201), and slide the filter board into the same slot as the main board.  Then standard 0.1 inch 90 degree pin headers connect all the relay wires.  PC board fingers are an alternative.  If the main HL board is (say) 90 mm long, we have 70 or 130 mm available for the filters.  The power amp can fit on the third board at the front of the box.  It is probably not in a card slot, and is bolted to the top.  It is at the front so we can access the screws.  Maybe the power amp and TO-220's could be on the filter board instead of a third board, but we would have to be lucky that the card slot position allowed for short leads when the transistors are screwed to the top.  The transistors are screwed to the top because the heat dissipation is better, and they are at the front so we can access the screws and slide the board out.

If that does not provide enough room for the filters, or if we don't want a 220 mm long box, we can switch to the Context Engineering 152 mm long and 62 mm high box (4012H-6N).  That will provide room for a second full size board plus a short board in the main board slot.  Connections are made between the short board and the main board as before.  To connect the short and full size board, we can just solder a row of wires between them and slide them out as a unit.  The power amp board (if any) is as before.

I think this provides lots of options.  We could put two HL main boards in one box.  We could bolt a small heat sink on the front top or front panel and run a few tens of watts instead of 5.  We could change to 6 meters.  We could add different boards and add more connectors to the front panel.  Transverters anyone?

Jim

N2ADR

[Steve Haynal]

Hi Jim,

Thanks for the links. Both are very interesting and helpful. In case you haven't seen these, Claudio tested the similar PD85004 here and here. He includes IMD numbers. The PD85004 has higher thermal resistance of 21 compared to the better 4.1 for the AFT05MS003N.

Is it a correct that your OPA2677 amp will run at 7.5V? Since I need 1.2V with the Cyclone IV, I am reworking the power supply. I am leaning towards one of these two three output devices:

http://www.ti.com/lit/ds/symlink/tps65265.pdf

http://www.richtek.com/assets/product_file/RT7273/DS7273-04.pdf

These devices can supply quite a bit of current, although we won't need that much and the inductors will be sized to our current requirements.

Rather than use the 3 outputs for 3.3V, 2.5V and 1.2V, I prefer to have Vamp, 3.3V and 1.2V as the current requirements for 2.5V are minimal and I will just use a small inexpensive LDO from 3.3V to 2.5V. This allows us to use a fixed voltage to your amp that is enough below a weak 12V battery to run everything, and avoids requiring an NCP1117 with jumpers and associated area. It also opens the door for using some of the LDMOS devices that run at lower voltages.

As you know, I am quite keen on the HL V2.0 being comparable in class and price to the softrock RXTX. As long as the IMD numbers are reasonable, two AFT05MS003N devices seem like a simple and inexpensive way to get a few watts out without severe heat sink requirements. They are also readily available from Digi-Key and Mouser. 

73,

Steve

KF7O

[Steve Haynal]

Hi Jim,

Thanks for your detailed discussion on this. The Hammond enclosures are a good option. Here is another option. It is less expensive and taller, but has just a single board slot. Ideally, I'd like an enclosure choice with good availability in single quantities from various suppliers, but also with the exact or similar enough model available from China.

The PCB manufacturer I use in China, and others in China, have price breaks for boards up to 10cm by 10cm in size, and then another price bucket up to 16cm by 10cm. I'd like to fit into the first bucket if it is just the HL V2.0 without filters or PA. I still think we can, even the original 8cm by 10cm dimensions with target case appear doable. I'm willing to go to the second bucket and possibly reduce the size to ~15cm by 10cm if the filters and/or PA are included on the HL V2.0 board.

My thoughts are that those not wanting to use any on board PA or Filter could just not build them on the board. I want to have optional 0.1 inch sockets in strategic spots such that alternate PA and filters (including your preamp and filters) can be installed. Even at 15cm by 10cm, these boards are pretty small. Imagine a base board that is 15cm by 10cm that supports upto to a 15cm by 10cm companion card for alternate PA and/or filters. But the trick is to keep these optional board interconnects simple, workable and inexpensive. Besides RF, I2C for port expansion and power need to be supplied. 

Some people also want an RX BPF. With the PA and filters on a main board, a small companion board could have a bank of RX BPFs.

Regarding TR switching, I would like to support option for TR switch before TX filter bank so RX can use the TX filter bank, alternate TR switch after TX filter bank so RX can have own BPFs, pe4259 bypass switches so unfiltered RX and TX power can be run to 2 SMA connectors for instrumentation/transverter use.

73,

Steve

KF7O

[ZL2APV]

Just a cautionary note about the Peregrine switching as applied to the v1.42 frontend board. When the board is in receive mode a short circuit is applied across the OPA2674/7 via pin 1 of the PE4259. In receive mode of course there is no RF applied to the Tx chain but I wondered of there is any condition where it could be i.e. full duplex or transverter mode or perhaps some other Hermes-Lite compatible software like HPSDR could do it. Also is the op-amp unconditionally stable with a short via the leakage reactance of the output transformer? So far I have not managed to induce any instability and I have given mine a pretty rough ride.

73, Graeme zl2apv

[James Ahlstrom]

Hello Steve,

On Sunday, June 26, 2016 at 3:34:15 PM UTC-4, Steve Haynal wrote:

Thanks for the links. Both are very interesting and helpful. In case you haven't seen these, Claudio tested the similar PD85004 here and here. He includes IMD numbers. The PD85004 has higher thermal resistance of 21 compared to the better 4.1 for the AFT05MS003N.

 

Please see thermal discussion on a separate post.

 

Is it a correct that your OPA2677 amp will run at 7.5V?

Yes, the OPA2677 is characterized at 12 and 5 volts, and will work with any voltage in between.  At 12 volts, the data sheet gives a 10 Vpp max output swing, and at 5 volts, 3.5 Vpp.   The large signal bandwidth decreases from 200 MHz to 100 MHz but for a smaller output swing.  When I designed for maximum output, I found the distortion increased rapidly as the output approached 1.0 volts from each rail; that is, 10 Vpp with a 12 volt supply.  So I designed for 1.5 volts from each rail, and got 7.5 Vpp from a 10.5 volt supply.  I estimated the bandwidth as 45 MHz.

For a 7.5 volt supply the output will decrease markedly and the bandwidth will decrease somewhat.  If we leave a 3 volt margin to the supply, we get 4.5 Vpp, or 11 dBm with back termination.  Since 5 watts is 37 dBm, we seem to need 26 dB gain.

I am a bit over my head here.  I am not an amp expert - that would be Claudio.  I did a paper napkin design of a bipolar amp, and tried to connect the op amp to the bases.  The problem is the low base impedance, and the low gain.  It would work if we added a 4:1 input transformer, but it would be nice to avoid another transformer.  A MOSFET amp seems more promising.  They have 50 ohm gate swamping resistors, but we could avoid those and just connect the op amp to the gates.  The swamping resistors are then the back termination resistors of the op amp.  We expect 19 dB gain from a MOSFET, still less than the required 26 dB, but maybe that is incorrect.  We are not in a 50 ohm system any more.  The op amp will provide 4.5 Vpp to the infinite gate impedance and that sounds like a lot of drive.  Of course we may need negative feedback and swamping resistors for stability, but it seems the op amp provides a lot of drive.  There should be a supply voltage between 7.5 and 11 where all this comes together for 5 watts.

Rather than use the 3 outputs for 3.3V, 2.5V and 1.2V, I prefer to have Vamp, 3.3V and 1.2V as the current requirements for 2.5V are minimal and I will just use a small inexpensive LDO from 3.3V to 2.5V.

Yes, since the op amp and power amp need high current, this is an excellent choice.  I wasn't aware of these switching regulators.

 

As you know, I am quite keen on the HL V2.0 being comparable in class and price to the softrock RXTX. As long as the IMD numbers are reasonable, two AFT05MS003N devices seem like a simple and inexpensive way to get a few watts out

Yes, I also want HL to be SoftRock done right.  But I think we should design for 5 watts out if possible.  The QRP crowd wants 5.000 watts and needs every milliwatt.  If we have 2 watts, they will just need a 2 to 5 watt amp.

I am dubious about getting 5 watts continuous, but remember that "5 watts" in amateur service means PEP for SSB and low duty cycle CW.  Digital modes often specify a 1/4 reduction in drive.  I think we should have no problem with 5 watts PEP.

Jim

N2ADR

[in3otd]

Hello,
I have tried to do a quick design for a push-pull amplifier using these devices, a first draft looks like this, if someone would like to try:

on paper it has about 20 dB gain up to and over 50 MHz and should be quite stable; in practice it won't be that good of course and the output transformer may need some special attention to obtain a good coupling. It could reach 5W (CW), no idea about the IMD3 performances, hi. It may be possible to drive the devices directly from the OPA2677 but I did not verify this, maybe in the following days.
Still, I'm not convinced of using 7.5 V devices, unless maybe we would like to use EER for the PA...

By the way, I was working on a design with two AFT09MS015N, which I hope would do 20+ W, but the thermal issues with these devices will be even worse, as they are still SMD, intended to be soldered to a GND plane on the PCB. Do not expect this design to be ready before the H-L v3, though :)

73 de Claudio, IN3OTD / DK1CG

[Takashi K]

Hi all,

This is not QRP amplifier but,..

I found an interesting amplifier, RD16HHF1 P-P , source follower 30W amp. Its IMD3 seems under -50dBc.

Supply voltage is 30V.

http://www13.plala.or.jp/ja4gii/PA15.html

Please use google translator because this page is in Japanease.

 

Has anyone ever seen this kind of amplifier ?

73 Takashi, JI1UDD

[Takashi K]

Sorry, I made a mistake.

30W is achevied by two units. one unit is 15W.

[Steve Haynal]

Hi Claudio,

Thanks for the schematic! All of these push-pull with MOSFET/LDMOS devices take roughly the same form. There appears to be just a handful of tricks - extra resistance here or there, some feedback, etc. Do you think we can come up with a generic template for these types of amplifiers that have optional components to cover the best set of tricks and tweaks? I think we should dedicate PCB space to incorporate the most versatile form of these amplifiers as we can always leave off or use 0 ohm components.

Besides 7.5V being a somewhat uncommon voltage, are there other reasons you are not convinced about using 7.5V devices?

What are you thoughts on reducing power output from a PA such as this in order to provide a low level output for test and measurement use? For example, using the HL2 as a VNA will not require considerable power, but I'd prefer not to add another low level RF output.

73,

Steve

KF7O

[in3otd]

Hello Steve,
yes, the basic schematic for these push-pull PA looks all the same, I too have thought to do a small "universal board" to be able to experiment with different devices. The board I have designed for the PD85004 push-pull can of course already be used with other SOT-89 devices.
Having a single PCB with footprints for SOT-89 (PD85004, AFT05MS003N), PLD-1.5W (AFT09MS015NT1) and TO-220 (RD16HHF1) might be possible but I didn't try to design that yet.
BTW, here is an interesting application note about heatsinking RF SMD devices.

My main concern about using 7.5 V devices is the reduced overall efficiency we will get, partly from the DC/DC used (which may be needed just for obtaining this voltage), partly from the increased currents in the devices and components w.r.t. a PA with the same power working at a higher voltage. Add to this the reduced performances of the driver OpAmp, which may complicate the design, if you want to run that also at 7.5 V.

Regarding using the PA at low power, apart from the wasted power in the devices bias, the output impedance may be far from 50 ohm, people using that output for a VNA may not like that. Need also to double check the stability when loaded by a real open or short circuit, uncommon with actual antennas but not unusual for VNAs or signal generators.
A switch to route the driver output to the main output looks better.

73 de Claudio, IN3OTD / DK1CG

[Tony Abbey]

Are you sure it is only specced for 7.5V? - the datasheet shows up to 12.5V and max VDSS is 30V, although it doesn't give any performance figures at that voltage.

Tony - G3OVH

[in3otd]

well, the manufacturer clearly specifies the device performances at 7.5 V. Then it guarantees that it can withstand a lot of mismatch at a certain power output with a 9 V supply. The maximum absolute drain voltage is 30 V, which, yes, is enough for a push-pull PA at 12 V, when matched and with ideal waveforms. The "12 volt" devices same manufacturer have a maximum drain voltage of 40 V.
So this device will likely work also at 12 V but the reliability in this condition will be unknown and there will be little margin (if any) in case the load is not matched.

73 de Claudio, IN3OTD / DK1CG

[Steve Haynal]

Hi Tony and Claudio,

I was wondering about how critical the voltage requirements are too. I guess one would have to try it and see. If I end up putting the PA on the mainboard, I am leaning towards a universal enough design and footprints that accommodates the RD16HFF1 and some of these newer low cost Airfast devices. The fallback is to use the tried and true RD16HFF1s at >= 12V. But we could still experiment and test the Airfast devices at 9V or 12V and see how they work. 9V is interesting as there are many inexpensive power supplies that can supplied the required current. I've had mixed results regarding noise with inexpensive power supplies though.

Here is an interesting document describing the Airfast family. There is also the AFT05MS004 at $2.21 in the same case but can handle a bit more power. Perhaps this would be rugged enough at 12V if the overall power is still kept at 5W.

73,

Steve

KF7O