Tuesday, October 5, 2021

EI9GQ 16 Watt Power Amp - Start to Finish

QRP is great and by far my favorite way to operate - especially since completing my scratch-built Furlough 20/40 QRP SSB and digital mode transceiver.  But as my friend Pete, N6QW says, "some days you just want to put on your big boy shoes."    For "big boy shoes" for the Furlough I chose to build the  Eamon Skelton, EI9GQ 16 watt power amplifier from his book "Building a Transceiver".  The amp uses dual Mitsubishi RD16HHF1 transistors in a push pull configuration.  This was my first attempt at anything beyond the SimpleSSB IRF-510 amp that I used in the Furlough.   EI9GQ's books is a good reference and there are several other builders who have documented their builds online.  Here's mine...

Beginning with the schematic - I used EI9GQ's specification.  The transformers gave me pause and I had to really "noodle" over their construction.    In addition to figuring out the windings and interconnections, if you read the article EI9GQ recommends torroids and shielded wired that do not seem to be available anymore.   For transformer T3 he specifies HEM3011 torroids, I substituted two FT50-43s glued together for each side of the transformer,  I also just used #26 AWG for the bifilar turns rather than the shielded audio cable that he recommended.

The next thing to do was to figure out the transformer windings and start to think about layout.  Using my primitive graphic arts skills I sketched out each transformer and carefully number the wires.  This proved very helpful when it came to construction.  Here you can see my layout and tentative parts placement on the top of the heat sink. I made the board about 1/2" longer than the heat sink in both directions to allow for overhang of the SMA connectors I intended to use.

I've been using a CNC mill to carve PCBs for Manhattan style builds for a while  and I do my layout in a free program called  Carbide Create.  It's fairly easy to use and combines the editing package with a G-code generator to produce the program for the CNC mill.  Here is my layout.

I always print an an actual size image of the board to check for part fit.  This save a lot of time and wasted PCBs if the parts don't fit.  For this build I milled the traces for the voltage regulator and bias trimmers, cutout the center square for the transistors and drilled some mounting holes.  I decided to wait until construction time to attach Manhattan pads for the components. EI9GQ used mostly u\Ugly construction with all the non-ground connections in the air.  I prefer a pad to solder to most of the time but this time I kept my options open. I ended up doing some Ugly construction and some on pads.

Once I was happy with the layout it was time to fire up the mill and get carving.  The PCB carve took about 20 minutes.  Enough time for a cup of coffee and to contemplate life.

Next up - drilling an tapping holes to attached transistors and PCB to the heat sink.  Two trips to the Depot - one for the #6/32 tap and one for the tap handle since my other one didn't clamp down on the small tap.

With the PCB attached to the heat sink it was time to start laying down parts.  I started with SMA connectors, transistors and DC bias circuits - leaving the exact placement of the transformers until I had a better feel how everything would go.  My only wiring mistake is visible in this next picture.  I had soldered the 8 volt regulator in backwards,

I glued down a few hand cut Manhattan pads and soldered up all the components on the input side of the the amp. I attached the input coils.  At this point the amp was about 90% complete - I just had to attach T2, the DC wiring for the drains and T3.  And also to maker sure I got the transformer connections right.

Amp done and ready for testinng.

And here is a demo of the recreated smoke test.  Much to my delight, after correcting the orientation of the voltage regulator that amp worked straight away.  With the transistors biased at 500mA each I measured just over 13dB flat across the entire HF amateur radio spectrum,

The next step will be to replace the IRF-510 stage of my Furlough with this amp and see how it works on the air.  As I mention at the end of the video I suspect I'll need to add another driver stage to get enough drive for the full 16 watts.

After close to two years of working on projects I'm starting to realize that Pete's adage - "if you know stuff, you can do stuff!" is really true.  It has been a steep knowledge climb for me - but I'm starting to feel it.

73 from Great Falls




  1. Bravo Dean! Nice project and excellent documentation!

    Pete N6QW

  2. Dean, a really nice project, and very easy to follow. And clever adjustment with the glued toroids. I can't wait to try this! Good job!

  3. Well done Dean. I built this same design in my latest transceiver - The Optimiser. I didn't strive for 16W though. In the end I settled for about 10-12W out on 80m but I did use a lot of filtering. Congratulations!

  4. Replies
    1. Yes indeed, Nick. The output of course will be based on the drive level - and sometimes when the noise is high 5 or 10 watts can make a difference - particularly when you begin at 5. I can tell you I would rather go from 5 to 15 than from 750 to US legal limit of 1.5< :) Once I get it hooked up to the rig I will evaluate performance and look at harmonics closely. The LPF should be good enough to knock them down.

  5. Great work, Dean

    Glenn vk3pe

  6. Where did you get the great hunk of aluminum? Your mill work looks a neat as ever. Hint: the RD16 start to wake up at 20-24 volts. You get no measurable increase in distortion products, but the gain ups a bit. Maybe not helpful when you are in POTA-ville, but when at the home QTH, you can always add a few more volts to the twins. You'll like the result I suspect. Great work as always.

    1. Running them that high destroys the safety margin of the transistor, yes you can do it, but you better have a good protection circuit and no hiccups..

  7. Thanks -

    Heatsink was one of the Marlin P. Jones specials bought a while ago for a future project....that turned out to be this. Once I confirm that all is well in Tx in the rig I may play with supply voltage - right now I am just happy that it amplifies on the bench! Baby steps.

    73 - Dean