During the early part of 2019 some DATV operators on QO-100 were looking at bigger power amplifiers. Many were using the very inefficient Spectrian boards obtained through eBay, heat output from these boards is huge.
Initially a couple of amateurs that I know of tried a pallet from Ampleon. There are two varieties; one with a single LDMOS device and another with an additional driver stage, both intended for ISM use. Of the two built one is still in use and working well; the other expired fairly quickly I understand due to poor earthing during tests. (1)
Then we heard that Jim G7NTG was carrying out tests with the Ampleon transistor BLC2425M9LS250. After some smoky tests he eventually came up with a finished PA which worked very well. I won't get into Jim's story any further because it's all written up on the BATC wiki; see the references at the bottom of this page. (2)
Realising it wasn't going to be that cheap; and there was some risk of damaging an expensive transistor I decided to go ahead. Luckily Jim had several PCB’s made so I purchased one from him. Jim no longer supplies the boards but I understand the Gerber files and material specifications are available.
Now I'll be honest and say that I borrowed the mechanical layout from a German amateur DC2TH Wolfgang. Pictures of his build were published on a newsgroup; I could see he had access to a workshop with milling facilities but I wasn't going to be deterred from going ahead. The only tools available to me were simple hand tools, drill, taps files etc. (3)
An old pal I used to go shooting with; an engineer and gunsmith has a milling machine and he milled a trench in the copper for me. I supplied a paper template and showed him how I would have to solder the transistor into the trench. He had a lot of work on but I received the copper back finished in two weeks; the rest was up to me.
Along the way I took photographs; some stages were missed due to not having enough hands. One part I should have pictured was soldering the LDMOS device to the copper. There is an excellent video guide on YouTube made by another Jim (W6PQL), I copied his technique which entails the use of an electric hotplate and temperature gauge. In the video flux is applied to the foot of the transistor and a length of solder placed in the trench. Not being brave enough to guess how much solder I used the hotplate and a 25 watt solder iron to tin the trench; just enough solder to the area where the transistor foot would seat. (4)
Now a series of pictures showing various stages in my build. Later I will describe the bias arrangement and one or two other points worth a mention.
Checking threads and alignment after drilling and tapping.
Luckily there had been plenty of space to locate the PA enclosure.
Tapping the copper for sidewall fixing. Yes I know my workmate looks crap but the new one's are not as good.
Input SMA socket in and well lined up, some components soldered on the board.
Output "N" connector mounted. It was necessary to overhang the heatsink to get a plug on.
This is my hotplate used to solder the LDMOS to the copper circa 440 degrees F. Since then I used it to remove some XRF-286S from a Spectrian board. The black rectangle is exhaust paint which is where the temperature is measured. The IR thermometer doesn't like reflections.
Getting there. The RF pickup probe has since been shortened and terminated to the board ground.
The lid was cut from an old diecast box, following test and setup it was sealed with some RF gasket kindly donated by Phil G8XTW.
This was a nervy moment with power applied for the first time, this was just to set the bias at one amp...OK 999mA. More about the bias later.
The big day, unfinished mechanically but ready for big RF out. Took it to my pal Arthur G4CPE because he has test gear which works 13cms and way higher. That's him on the right.
Well it worked no problem, well apart from some sealing and firmer fixing to the lid. We tested it to just above the limit of Arthur's attenuator 158 Watts; and for a brief moment 200 Watts. Gain was 19dB with 33dBm drive and 52dBm out.
So there it is in it's home next to the dish. The dish has been changed to a 1.2 metre Gibertini since an earlier blog, Gibertini call it their 1.25 metre because it measures 1.3 x 1.2 metres.
Summary and references.
Well given the tools I had and the workshop (kitchen) facilities I'm quite pleased with the result. What else can I say about it?
Some of the passive components and the LDMOS transistor were purchased through Digikey. In the UK buy through the .uk website with free postage on purchases over £33.00. Goods are supplied from the USA and normally arrive in about 3 days, there's no additional tax. (5)
The LDMOS bias is not temperature compensated at present. I used a buck converter down to 8 volts from 28 and then a linear regulator to 5 volts. I also incorporated a crowbar circuit and a glass fuse. At the time of writing this I have just received a temperature compensated board from Patrick ON1BTE, over the coming days this will replace my temporary board.
So a great project provided you are cautious and check where all those Watts are going.
5) Digikey UK