Building a amplifier like this ground up is quite a task, and requires a lot of commitment . This started by me needing approx. 1 KW of RF power on 144Mhz for EME use. I originally built a 2M amplifier using a 4cx1000K tube. The tube failed and I had to spend $500 for a new tube. The amplifier was also big and heavy. I then started thinking of building a RX1AS style GS35b amplifier but I stumbled into a issue with drive power. I use a K3 with a X144 transverter module – so I only have 10W drive, and to drive a triode like the GS35b I would need 100w at least. I started building a MRF151 driver amp that I happened to have in the junk box. That design is based on a pallet from C&C and it’s very critical RF design. I could only get 100W out of it with about 280w DC – that I found really bad and the amplifier was not really stable.
As I said before going to a ground up design like this is a lot of commitment, I don’t think building your own stuff has anything to do with money. I always say to my self – phew I could not afford to buy a $3000 turn-key amplifier. So you start collecting parts for your design. And I think in the end I get close to the amount I had to pay for a ready built amplifier. But the joy is in the hunt for parts, looking at other designs, seeing your hard work play out. And hopefully in the end having a fully functional amplifier – that is better than anything out there. There is just so much joy using equipment that you made. Also when things break – you have so much more first hand knowledge about your equipment that you can just dive in and fix the problem. I had two main concerns for a project like this is:
- Finish the project – I could not afford to have over $1000 tied in parts that I could not use.
- Make a reliable amplifier that can run JT65 and be robust enough to transmit into high SWR without blowing the LDMOS.
For me to finish the amplifier was to overcome my personal shortcoming of lack of focus. I nearly switched over to my HF amplifier project after taking part in the CQ WPX contest. After battling with 59++ LY/OK/YL stations with my TL922 “only” outputting 800W I decided on pushing on and finishing my HF amplifier, and that could easily shelved the 2M amp project! Also being a father of 3 young kids, I need to use my time in the evenings well!
Reliability came in the form of good monitoring – mastering the REF-FWD metering I have been studying. More on that later
and I don’t plan at all just do 😉
So – I decided to go SSPA
I’m not an expert – just a hobbyist. I have a general idea what I’m doing – just showing what and how I did it – perhaps you get some ideas for your own project – please come back to me if you have some improvement ideas for me!!
I wrote down some critical design features I wanted:
- Remote operation – be able to operate and monitor the amplifier remotely
- SWR protected – ofc!
- good thermal design
- SMPS – never again big bulky power supplies.
- One unit (that came later) – light and nice
- CPU controlled.
So the hunt started – hunt for ideas – components – and everything !
here is a block diagram:
I had studied Jim Klitzing page quite well, He is like GM3SEK of the SSPA world !! – I have endless admiration of HAM’s (they are really professionals) – that put so much time and dedication into making really cool designs. I think most hams that have not built stuff like this have any idea how much time and effort goes into the work these guys do.
I looked ad a pallet kit from RFHAM, and also from Communication Concepts – then I stumbled onto the kit W6PQL is doing. I sent a email to C&C and Jim. I got an answer 30 min later from Jim – never got any answer from C&C. So the choice was easy. Jim responded to all my questions (some quite stupid!) promptly and professionally – so I ordered the kit with the LDMOS device from Jim – he suggested trying the BLF188XR – supposed to be bit more powerful (we’ll see)
Now I needed a 50v solid state power supply
I happen to work for HP in Iceland – so we have a big HP customer base – and the old P-series blade systems (now being decommissioned) by most customers use a 3KW switch mode power supply (SMPS) outputting 51V. I talked to a customer that was throwing out some of these blade chassis and got 4 power supplies and 4 blower from the BC’s. These power supplies are made by ASTEC – and are quite easy to work with. They require a short of two pins to enable. When they get mains supply – they output +5V DC.
They are quite long but “low” – so I decided to take circuit boards out of the orignal chassis and mount in another better suiting enclosure.
picture shows the underside of the amplifier. The Astec 3KW SMPS on the left side, 5/12/24V low-power supply on the right side. SSR is to control 220v into the SMPS. Small board center under the second SMPS board is a Allegro hall effect current sensor – 40mV per 1A.
The whole amplifier is controlled by a ATMEL microcontroller. I use a “teensy2++” board found at https://www.pjrc.com/teensy/ It’s a 8 bit microcontroller, running at 16Mhz. It has some built in hardware stuff like USB, I2C, 8x 10 bit ADC, PWM.
The control board is then doing:
- FWD / REF power metering – and SWR
- temperature monitoring
- FAN control – PWM
- PTT IN
- Standby control
- RX / TX sequencing
- 50V SMPS control
- BIAS control
- displaying status – and other relevant data on a LCD
- sending status via USB serial – and interpreting serial commands
- fault detect
this is probably the only thing I contribute to the project -everything else is borrowed. 😉
Although some might say using a microcontoller would make the project more complex – I beg to differ. The design is much more simple and feature rich than using a analogue control board. Few components, not a critical analogue design (that I’m quite bad at). The only drawback is slow timing on critical events. I would probably always use a analogue protection circuit if using a device that could not survive a short/open circuit for some milliseconds. Even if the CPU is runnign at 20Mhz – it’s doing a lot of stuff and will be delayed in determine SWR and cutting power/bias in case of a high SWR.
The circuit description is simple. It’s a Teensy2++ break out board for the small Atmel. It has the 16mhz xtal, USB port and some extra components. I use a ULN2803A 8 port Darlington driver to sink higher currents: relays, PWM FET etc. I made the board quite generic. Broke out most of the ADC’s using a 3 pin header (+5v, GND and ADC). LCD display is using 7 pins and the front panel switches (up/down/select). I also connect the I2c/TWI to communicate with the ASTEC SMPS and. There are two P-channel MOSFETS – they are overkill in size, but I had them in the junkbox (IRF9340) – they supply +12V BIAS and +12V for the SSR to turn on mains power for the ASTEC SMPS – for some reason I decided using the mosfet driver for that task – I could easily just drive it of the uController.
This is the control board during assembly. top right are 5x (one missing) 3 port headers of various ADC inputs, two jumpers on the right to select 12v/24v supply for the TX/RX relay (had not chosen what relay to use at the time I made the control board!) ULN2803 was missing…
Front panel – back side. Some lat modifications – Voltage divider to get the 50V down so I can measure Vdd, and a 5v voltage regulator to get a very stable 5v for the Allegro current sensor. (had some trouble with jumpy 5V supply).
Originally I was going to use small 24v fans that came with the heatsink. They had a annoying high pitched whining sound . I had a really (and I mean reeeeally) powerful fan I got from the HP blade system enclosure. I think it’s rated 2,5A @ 24v – 60W. It sounds like a jet when at full power and could probably be used to power a RC airplane 😉 – I decided to put a IRF 540 MOSFET and drive it off a PWM pin of the Teensy. That is working very well, at speeds that deliver alot air (much more than the original fans) the fan is quite silent. I can then go 2-3 steps more to deliver amazing cooling but with some noise.
I use a AD8307 log amp to convert RF to DC. The AD8307 has 90dB dynamic range so it’s very sensitive. It returns 25mV per dB and the input must not exceed 17dBm. I had studied the W6PQL combined LPF and directional coupler and decided to “improve” the design by replacing the detector diodes Jim uses with a AD8307 at the same place. This I could never get to work reliably. I have not still found out what was the problem. I got very strange readings when power increased. I suspect the close proximity with the LPF was causing this. So some work must be done to get this working, perhaps shielding the detectors from the LPF.
I looked in my junk box and found a ready made directional coupler that was designed for this power level and frequency. The coupling was 33dB. being optimistic about the power output of the amplifier I decided I would have the maximium output to be 63dBm. This would then be 63dBm – 33dB coupler- 16dB pad = 14dBm output @ 2KW.
picture of the detector – not much to see – small bugs – decoupling caps , and the 16dB Pi pad. red wire is the +5V positive feed, just lazy not to have one pad for the +5V 😉
The firmware must be calibrated – there are two modes to calibrate – one is a single point calibration. It will ask for a 50dBm output. From that it will know the correct level of the AD8307 and will assume 25mV per dB (slope). To get better accuracy two calibration points can be measured. This enables the uController to accurately calculate the slope of the AD8307. BTW – all this programming is stolen/borrowed from TF3LJ/VE2LJX and just slightly modified by me.
Now it’s just picture time:
Amplifier front – need the fish-net for the cooling input.
PA unit – temperature sensor on the left.
The business side of the amplifier. I’m not really a builder – don’t care so much about wire routing etc.. but if it works …it works 😉
Temperature sensor is a LM35 – use a coax for gnd/Vout and it’s stable even if it’s sitting on the LDMOS. I have another sensor on the heatsink
The heatsink is a high-performance heatsink I ordered from Farnell.
This picture shows a LPF that burned – google video of it: https://www.youtube.com/watch?v=9lU3I3BgmkE
and first spectrum analyzer test:
it shows 2nd harmonic @-26dBc and 3rd @-23dBc . The reading was taken at the forward output of the directional coupler, I don’t have any tap that can handle the power – and this makes the readings bit off as the coupling is not linear. I don’t have a reading behind the LPF as it burned 😉
I now ordered a LFP from F5CYS – that should do the trick.
View of the front panel display – some programming work to be done – and showing 80% efficiency ! hmm 😉
well, I will update when I have more info – but it’s looking good !
As for the firmware – when I have done bit more work – removed junk – and properly referenced where I stole the code I will publish it!!
73, Benni TF3CY