Today I've powered it up for the first time at low power from a 32V 3 amp lab PSU and with 15dBm of drive it produces 31dBm output at a bias current of 1Amp increasing the bias up to 3Amps increases the gain to 18dB.
Seems to be nicely centered on 2400MHz.
Next step is more drive and a bigger PSU / dummy load - aiming for circa 100 watts into a 1.2M dish.
Set up in the picture is a bit of a lash up for test purposes. The heatsink is surprisingly small but according to the specs should suffice as the fan push a lot of air through it - albeit noisily !
- 13cmPA_working.jpg (796.48 KiB) Viewed 1500 times
Just ran it up to 8amp Idd and 32Vdd (256W DC input) for an easy 100 watts RF out with about 2 watts of drive.
Heatsink is rated at 0.2C per watt and I measured it as 0.19C per watt ie 30 degree temp rise for 156W heat so about right - the SW shut it down when then the heat spreader reached 60C (ambient in the shack was 27.8C) - so being conservative here - at least for the time being. Looks like the cooling is working ok.
Ran it up to 200W briefly with no problems but load was only rated to 150W but had a fan on it. Will have to dig out my 1kW load !
Now waiting on the nRF905 868MHz telemetry radios to finishing the monitoring and control SW.
Not sure about what to do with the design - I have enough PCB material for about 6 more PCB's but getting them CNC machined along with the heat spreader could be a problem.
Lots of fun getting coax warm in the process !
Also found 'Aluminium Warehouse' sell copper cut to length, (I was running up an aluminium order for beams) so explored the cost of some spreader, making assumptions about what would be suitable (40x100x6mm). See attached.
I would like one of these amplifiers. How much CNC is required? We have people who can do that type of thing in Falmouth as far as I know. Do you need some encouragement ( rather than help! ) in the production?
I'm still in the testing phase and will be for a couple more days while I iron out the HW and SW bugs but so far everything is looking promising - fingers crossed.
A water cooled CPU cooler could well be an option but I decided on forced air cooling as its going to be located outside in all weathers - hence the extensive telemetry to keep an eye on it.
I got my copper from ebay it was £5.50 and measured 100x100x10mm thick.
Copper is not nice stuff to machine and it needs quite a few operations to prepare it -a fellow ham cnc'd it for me and has the data to get the slot and fixing holes in just the right locations.
He also cnc'd the Rogers pcb material - no mean feat given the tolerances required !
The controller / telemetry board could do with a proper PCB and if it all works out ok I'll likely run one up. It monitors temp, Idd, Vdd and bias voltage if anything goes out of limit it shuts down. It also controls the cooling fans. There's a spare ADC input to monitor RF power but currently I don't have a suitable detector. The bias is an active temp compensated design as required for LDMOS.
The data is passed over a two way 868MHz telemetry link which also allows remote control of the PA.
This is not a low cost project in terms of £'s per watt and there maybe more cost effective options out there - I went down this path more as a technical challenge than anything else.
I'll post pictures once it's all done.
I'm using the exact material specified in the application note - Roger Corporation RT / Duroid 6035HTC 0.03" thick and 35uM copper 1oz per square foot.
Seems to handle the power no problem as its PTFE based and ceramic loaded for heat transfer.
I suspect FR4 would go up in smoke at these power levels !