What you are doing is all the right things to find the right combination
to drive the IRF510's. Get a hold of and really read the original article by WA2EBY
for all the things he discovered along the way. The primary thing is taking care of heat.
You seem to have accomplished that. The second is coming up with the proper layout to eliminate
the propensity of the finals to oscillate -- that is probably what you are experiencing.
"I'm working on my uBitx which is completely homybuilt, so not the currently sold PCB version. By building it from scratch you encounter a lot of interesting things and learn a lot about it. Receiving of the rig is very good now but I am struggling with the TX. When aligned for output power of 8 watts, using a sine wave generator and full modulation level, the rig draws about 3,5 amps of current on 40M. At 80M, power output increases to about 12 watts, but current jumps up to around 4 amps. Sometimes my IRF's get killed at that level. There's now way so far I've got it set to say 8 watts with less than 2 amps of current. Things I've done so far are:"
Are you sure you are using SSB? And the amplifier class is really AB? Remember, SSB output is Pk-Pk, while AM is "100% modulation". I would be interested in
seeing if there is any carrier output. This sudden current jump is suspicious of extra oscillation.
"Play around with the HF power gain pot RV1 in combination with more audio drive level"
Don't go above 3v or so on the IRF gate. Maybe even 2.5v. These are very touchy devices and hit a full on current condition with no warning...as you have
already found out.
"Play around with the IRF's bias adjustment. Setting it too high will give a current thermal runaway and a fried IRF within a second. Careful with that...
Output toroids of the IRF balance, changed the windings bifilair, trifilair en more in both directions, so in fact play around with the output impedance with the IRF's see. The trifilair original option seems to be the best bet, but with these high current values. Toroid is a 37-10 with 7 windings trifilair.
Changed the emitter resistances on the 2N2219 to decrease the input power."
IRF MOSFETS are notorious for being extremely variable. You are right that a different batch will give entirely different results. The use of #10 material for the toroid is very good, but I would increase the size to at least a T50, and probably would go to a T68 or even a T100 for high power levels. It doesn't take much to saturate a T37, and then the transformer is an entirely different one. Also, it might be prudent to look for a different form factor for the output transformer, one with a bit more shielding than a toroid. I know, they have a good
reputation, but they are not perfect in a strong RF field.
"I've measured the LPF and it has a cutoff at 7300 KHz so that should be fine.Testing done at 12volts, when at 13,8 volts, the output power and IRF amp values increase dramatically (got over 16 watts of RF) but this often kills one of the IRF's. No way I would keep the IRF's alive on 24 volts.
I don't have the LPF's for the higher bands in place yet, but it seems that at 20m and higher the overall gain of the TX chain decreases. When measured WITHOUT LPF (yes on dummy load, for sure) output power decreases to less than a single watt... This could also be an effect of the bidirectional part of the TX chain where the gain decreases with higher frequencies. For now I'm concentrating on the difference between the behaviour at 40 and 80 meters."
It's unlikely that the bidirectional amps have much to do with radiated power...provided the throughput is adequate. In the case of the BITX, that is roughly 10 dB.
Know also that the IRF510 is not a very good device above 40m. Yes, they can be made to work well, but it takes a good impedance match and heat control to
do so. They also work better at 24v than at 12v. So it is likely that you are seeing VHF oscillation drawing the extra current. That will SEEM like the output to the
antenna is better, but it really is not. Offhand, I would say that you have a layout problem and are using a little too small toroids to handle the load. You might consider
divorcing the final amplifier from the rest of the rig, and working out the rest. Farhan designed the uBITX for a wide range, but rather low power output -- certainly less
than the IRF510 is capable of handling. Any experimental work is good, but it is really in a different league here. I like what you are doing...it is pushing the design into new territory.
"Could it be I am just using a wrong brand of IRF's? I have a load coming in from a second supplier to check.
Any other ideas? I have got rechargeable batteries built into my rig and they would last much longer if I could keep the total peak current below 2 amps.
Oh by the way, I've got a very large cooling heat sink and never got it really hot, just 'warm'
Lovely experimental rig!