Help me pick an amp design: P-P EL84 with 12ax7 driver

If I were to experiment I suppose I'd need to know what oscillation looks like on the scope (or more accurately, the onset of oscillation) and adjust the values of the FB loop accordingly. Or something like that.
By the way, one easy way to see what oscillation looks like is after you get your amp built, hook up a 5K pot or something in place of the feedback resistor. Hook a resistor dummy load across the output speaker terminals, then put your scope probe across the dummy load resistor. Then turn down the feedback pot. Start at 5K and slowly turn it down to less and less resistance. Eventually you will get to a point where the flat line on the scope will just go wild. That's the point where it breaks out into sustained high frequency oscillation.

If you run a 1KHz sine wave through the amp at 1 watt output, and do the same test, many times you will see "nodgules" start to appear on the sides of the sine wave, or other oddities that look like puss packs or something attaching themselves to the wave form. That's the onset of high frequency oscillation also.

Other times everything will be fine at high frequency, but when you pulse any random low level signal across the inputs, the output will exhibit a low frequency oscillation of maybe 1 to 10 Hertz. What that looks like on a scope (on the DC probe setting) is a flat line that is up then down then up then down, etc. This is low frequency oscillation.
 
Good info, thanks.

I did spend some time this week getting to know my 'scope and signal generator. What I thought was an issue with them was operator error :)
 
I'm working on it, but slowly.

Question: when routing wires, is it best practice to route signal carry wires in such a way that they don't cross the wires carrying B+, even if it means taking the long way around the chassis? Or does it matter?
 
B+ you're mostly OK with, its the AC heater wiring where you're liable to pick up hum. Personally I try to keep signal as far away from power supply as I can. Sometimes it just can't be helped though. If needed, you can always use shielded wiring for the signal to keep things quiet.
 
Try to keep signal carrying wires as short as possible and as close to chassis as possible. Keep them away from AC lines. If you need to cross filament AC lines, do so at 90 degree angles. Should never need signal carrying wires to cross 120V line. I always use shielded input lines from input jacks to first gain stage, but I prefer input jacks on back of amp, and that usually means longer runs to get to the first gain stage.
 
I also like inputs at the back, but not having to think about the wiring from input jack to input tube is an argument for putting the jacks at the front.
 
OK! Shielded wire it is. I have some nice Mogami mic cable pieces.

Noise is a concern: In its original form this amp had lots of hum/noise, even with new caps. I decided to first try the build with the power transformer in its original location. Depending on the outcome I may opt to remote the PT, as Dave did on his Eico HF-81 thread. The Eico and Knight transformers look very similar.
 
Last edited:
After triple checking the wiring, started bringing it up on the varaic (with dummy loads) and the 430 ohm 2w resistor connected to the MOSFET (see post 84) began smoking at about 40v input. I used a 470 because that's what I had on hand, still should be within tolerance I'd think.

Maybe I should bypass the EFB part first and see if the basic circuit is OK?
 
Ouch. Something's not right. May need to replace the mosfet or the zener if either were damaged. If in doubt, replace them. Careful handling those mosfets (or any transistor, really). Touch ground each instance before handling them or soldering the pins. Go light touch on the soldering iron. Iron should be just hot enough to melt solder. For the solder I use, that's 713 degrees.

Make sure mosfet and zener are correctly oriented and wired correctly. For next test, try powering it up without output tubes installed. That will put zero load on the mosfet, then check voltage between source pin (mosfet output) and ground. Hook up meter with grabbers on meter probes for hand free measurement, so you can be ready to power down if you see something start to smoke. Top resistor in the gate voltage divider should be 82K and lower 270K, just wanted to check that's what you're actually using.

Overall not to worry though--all part of learning. I bet I fried 30 to 40 mosfets, BJTs, zeners, 3 terminal regulators, etc. before I got it all right on one of my builds from last year.
 
Last edited:
OK thanks I'll replace the 470 resistor, MOSFET and Zener and test without tubes.
Good thing I ordered 5 of the MOSFETs.

Also 'yes' to the 82K and 270K resistor question

Edit: DG has pointed out to me that I need to isolate the case on the MOSFET (it's currently grounded) Easy fix.

I'll redo and report back.
 
Last edited:
ohh yes. case grounded...not good. (you must not be using the type completely encased in plastic)
 
A trip to Radio Shack's Final Clearance sale (kind of sad, I'll miss them) yielded a mica insulation kit for the MOSFET. After replacing the toasted components and installing the insulator kit, I powered it up without output tubes installed. No smoke this time: progress! Voltage between source pin (mosfet output) and ground is 238; the goal is 260. It also built up quite slowly to this voltage, maybe 5 minutes or more. I did not replace the zener diode since Mr. Fluke said the one in there now is OK.

Do I need to bump that up to 260 or is the output tube-less test altering the voltages?
Or is my 470 ohm vs the 430 Dave specified dropping too much voltage before it hits the drain?

Edit- I clipped a 5k resistor in parallel with the 470, bringing it down to 430-ish ohms. Source to ground voltage increased slightly to 241VDC.
 
Last edited:
The output voltage at the mosfet source pin should be about 77% of the voltage on the high side of the dropping resistor. Report all four voltages (voltage at high side of dropping resistor, voltage at Drain, Gate, and Source). Under no load (output tubes removed) shouldn't matter what the dropping resistor value is, unless there is still a large current draw through the dropping resistor.
 
Last edited:
The output voltage at the mosfet source pin should be about 77% of the voltage on the high side of the dropping resistor. Report all four voltages (voltage at high side of dropping resistor, voltage at Drain, Gate, and Source). Under no load (output tubes removed) shouldn't matter what the dropping resistor value is, unless there is still a large current draw through the dropping resistor.

Before dropping resistor: 408vdc
@Drain: 407vdc
@Source: 240vdc
@Gate: 232 vdc

note: line voltage is running a bit higher than yesterday, more like 120 vs 117
 
The resistors in your gate voltage divider are not as advertised, or you have extra current draw through the voltage divider itself. For example if the 82k resistor has extra unexpected current running through it, that will drop more voltage and would explain the results you are seeing.

If everything was working as expected, and under the voltage conditions stated and no output load present, you should be getting about 308V out of the MOSFET.
 
Last edited:
I had the source and drain reversed. Another new MOSFET and zener just in case, now voltages are:

Before dropping resistor: 408vdc
@Drain: 407vdc
@Source: 295vdc
@Gate: 294 vdc
 
That's better. Now plug the output tubes in and try again. Should be getting close to the desired voltages now.
 
With output tubes (not matched)

Source varies from 270v-292v depending on position of 5k pot in EFB circuit
Gate varies from 271 - 294v as above
Drain varies from 360 - 393v as above
 
Back
Top Bottom