Ok, it made RF before I broke it. Not a serious break, one of the trimmers failed and I cannot set the bias to the right hand bank. Going to be a bit of a job to pull it apart and remove the dead trimmer, thinking about it now, I really should have used SMD parts for everything that I could on the board to make fixing things like this easy. Will see how i go tomorrow when i dissemble it LOL
Ok I have made wholesale changes to the schematic. Changed the voltage divider biasing to a more conventional biasing schema, adding in some feedback, which i am a bit unsure of, its probably not totally necessary for 40 and 20m band amplifiers, but for higher using IRF510’s its needed, I can always not populate those footprints when i build it. The resistors i had from source to ground were removed, those were meant to be swamp resistors on the gate to tame the lower bands, but they are probably not really needed and finally the biggest change has been to the output transformer, in which the single turn winding is going to be copper tubes with VCC to the drains delivered via the center tap on the single turn winding.
I have also been thinking about ground paths and return loops and all that other stuff that is over my head. But, I think something i am going to do is change the SMD bias components to through hole and move the bias traces to the top of the board, this will give me an almost unbroken ground on the bottom layer. Which can only been a good thing I guess. And I also need to do some research on the rf power handling of trace widths. Yet another thing i really don’t have much of a clue about. So i need to unroute the whole board almost and kind of start again. But for now, its off to find some credible source on trace widths and rf power levels.
Lots has changed since i posted this, this morning. Pretty much changed everything in the board layout. Here is a 3d render of where its at. Yes, i changed back to through hole parts because it would allow me to jump a trace using resistors and keep the bottom copper layer almost completely unbroken. So its a perfect ground plane to stick the broken top copper to.
No its not an April fools joke. I actually am going to build an rf amplifier capable of over 100w. Sacrilege i know for all the qrp’s out there. But there comes a time when everyone has to have a go at making some power. To date the most power I have made is a little over 10w and typically i would be happy with this kind of power. But for some unknown reason, the other day I decided while waiting for receiver bits, to design and layout a pcb capable of doing a few hundred watts or a little more. Actually, the board should handle 400w comfortably. The limiting factor being the T106 toroids wont handle much more than that on a good day.
The schematic can be viewed here: vk4hat_amp
Now, i do not claim credit for the design. A lot of it was directly copied from a 6m amp design by OZ1PIF http://frenning.dk/OZ1PIF_HOMEPAGE/50MHz_IRF510.htm all i did really was add my own spin on it. Grounding the unused side of the Omron G2RL relays for better isolation and the relay itself was recommended by Mark VK5EME from https://www.minikits.com.au/ and other bits of inspiration were taken from people like LZ2JR http://lz2jr.com/blog/index.php/mosfet-linear-power-amplifier-50-watts/ from whom i learned about the FQP13N10 mosfet that lead me to a lot of Cb’rs who use that same transistor in a lot of their illegal amps. From what I have read, its a superior transistor to the IRF510 and should give better output. I will not be disappointed if i do not make a lot of power out of this, but somewhere between 100 and 300w should definitely be on the cards with a supply voltage over 24v.
They say short leads and short traces make everything better. You can see in the 3d view from my ecad how I plan on mounting the transistors. The tab on both the IRF510 and the FQP13N10 is connected to the drain, so i will, with the aid of tab under the mounting screws, soldered directly to the pcb. Other than going bonkers with the via’s its a pretty standard affair with a single 7th order lowpass filter. Yes, I am making this for one band only. I will likley made 2 of them, one for 40 and one for 20m, but making this multi band is just way to spendy. For each band you are adding $50 in relays, toroids and capacitors to the build cost. There is just no way to make power cheap. If the board turns out ok and the amp is a good thing and works. I will probably release the gerbers and let everyone have at it and build their own.
To add to the PA madness that is going on here at the moment, I decided today to build one of the 2 QRP Labs 10w PA’s kits i have here. At the risk of inflaming all the QRP Labs fan boys out there, i both love and hate their kits.
I love that they are well thought out, designed and tested and bring a level of professionalism to the whole kit process, but, and this is the clincher for me, they are mostly a total pain in the arse to build and fix when something goes wrong because they are so densely packed.
Now, some of this has a lot to do with my own personal situation, I have arthritis in my fingers and only have about 75% use of some of my fingers and these boards are backed dense. Might be ok if you are perfect dexterity in your hands, but if not, well the build is not impossible, just frustrating. It would be nice to have say 30% more board space on all their kits. It would really make things much easier to work on, for me at least.
Anyway, it works, with 12v, 250ma bias and 15dbm drive I am seeing about 5W out, not 10. To get 10W out i need to drive it into hard clip with 1W input and turn the output into a square wave. Obviously not good. I figure it could probably do with some more bias current. Gain is pretty flat which is a nice thing. I will have to come back to this and play further at some point.
What i might end up doing is copying the design and laying out my own board and having some boards made up using larger ferrites, even if I lose some gain at 28mhz, it would be worth it to me as I dont use 10m anyway. If i do this, i will not be releasing my gerbers, because you know that someone will take them and start selling boards and that would not be good.
So its been a busy day here in the workshop. Caps arrived today and we set about to making low pass filters for the transmitter. I started by sticking in the caps and then in turn winding the inductors for each of the 3 filters, 80, 40 and 20m bands and giving them a test by feeding in a square wave from my it makes nice distortion signal generator.
80m Low Pass Filter Under Test
The output was fed into the spectrum analyzer on the Red Potato. The spurs either side of the fundamental are crud from my signal gen, it does that when you start to turn up the output. This was a 10 volt peak 3.6mhz square wave being fed into filter and I through up a few cursors to get some levels, first harmonic is -60dBm down on the fundamental. That should be more than good enough.
I did the same for the 40m and 20m filter, again -60db down on the first harmonic, the above image being for the 40m band.
Something I was worried about was isolation on the relays. These are no uber doober premium rf relays, no no no, they are 10 for $1 Chinesium Grade floor sweepings and seconds off Ebay. So we stick the signal in the in port and connected the 100meg Hantek scope up to the other end and turned the volts per division down to the noise floor and looked for signal leakage. Nothing, maybe a couple of microvolts but that might have been an aberration. Either way, good enough for the kinds of girls i go out with.
Final board all together and soldered up.
A bit of an idea of how its all going to go together as a stack. I think things might actually work ok after all. Touch wood. Tomorrows job is to program an ATMEGA IC to work as an iambic keyer, add in some switching and buffering to get the transmitter up and working. Not sure if i am going to have an external PA yet, or just use a couple of BD139’s and get 3 or 4 watts out. Will see how we go.
So when i designed the filter boards one thing i took into consideration was that often you need to use multiple caps to get close to the value required. Now when i first started home brewing, i used to obsess about values, oh i dont have this value, oh i dont have that value, but now i really do not give a crap, as long as its in the ball park she’ll be right mate, mostly the values are not super critical as long as they are ball park.
I mean does it really matter if the corner frequency of a LPF is shifted up or down a few 10’s of Hz, or Khz for that matter because your cap values are 5% off ideal? I am using 10% caps as it is, so there can be a bloody lot of variation here. But I did manage to get pretty close to the numbers using the one range of MLCC caps that minikits.com.au most of which are also NP0 types also. So below are the schematics for the filters.