I agree, it will make everything easier to be seen, then you can apply fbk to iron remaining harmonics.
The SS tail Mr Steward has another advantage. I like DC coupling of the first stage to the LPT as it makes the LF stability better. However the ef86 plate operating voltage is very variable form device to device. Adding a CCS in the second stage prevents the LPT plate voltages being a function of the ef86 plate voltage.
Thru previous work I found that there is ~10 db stabilization of the Q point of a small signal pentode by the screen resistance. The difference between the bypassed & unbypassed stage gain measured on the bench is the indicator. Further stabilization of the Q point is provided by the cathode resister.
Thru previous work I found that there is ~10 db stabilization of the Q point of a small signal pentode by the screen resistance. The difference between the bypassed & unbypassed stage gain measured on the bench is the indicator. Further stabilization of the Q point is provided by the cathode resister.
No worries Jan, you're here, that's all that matters! And yes, I didn't expect this either, although I need to revisit as I've obviously made changes to the prototype since then, so this may no longer be the case with the current iteration of the schematic.
Yes, that part I understand, I'm just wondering what the best approach would be to measure the various sections in isolation?
More on this, I'm trying out different ways of hooking up the cathode resistors on the KT88s. Have a look at the two configurations I just tried, and the resulting measurements. There's a clear difference, however with the bypass capacitors the distortion plot doesn't have the dip around 30W and is flattened out. Interesting, just not sure what sort of conclusions I can draw from this?
Attachments
Alright, this is where I'm at at the moment, and I am running out of ideas to further optimize things. I've reduced the parts count and simplified the circuit around the LTP without any adverse effect on performance, or at least as far as I can see/measure. I believe it was Albert Einstein who said: 'Everything should be made as simple as possible, but not simpler' and I couldn't agree more. Thanks guys for all the helpful suggestions thusfar!
Attachments
Actual anode voltages on the 6922 are lower, around 300V, I forgot to add these to the schematic, please see an updated schematic below. The voltages noted with a * are the measured voltages. So there's no problem, Also many manufacturers (JJ and Tung-Sol for example) specify a 400V Ua-max for the 6922/E88CC so we're nowhere near the max, with about 100V of headroom.
Attachments
I don’t see the problem, the 6922 is routinely used, also in commercial designs, running at much larger current and anode voltages than I’m using here. If you worry about the longevity of the tube then I suggest you stay away from tube amplifiers as they all need their tubes replaced at some point, some more often than others.
The 6922 is a long-life tube for 10.000h, in your amplifier a lot less with those voltages, it’s your choice for less reliabilty.
More on this, I'm trying out different ways of hooking up the cathode resistors on the KT88s. Have a look at the two configurations I just tried, and the resulting measurements. There's a clear difference, however with the bypass capacitors the distortion plot doesn't have the dip around 30W and is flattened out. Interesting, just not sure what sort of conclusions I can draw from this?
Here is an interesting read on that subject.
Here is an interesting read on that subject.
Attachments
About the cathode capacitors for output tubes, there’s also VacuumState trick: use separate resistors for each output tube cathode, then use one capacitor for each output tube cathode, join the other side of the two capacitors together, then add a resistor with a value of 16% of Rk to ground. That proved to dramatically reduce IMD.
Thanks! I’ll whip that up tomorrow and see whether I see an improvement on the prototype.
P.s. any suggestions for a gyrator load for the EF86 you mentioned earlier? I mean I can cobble something together, but if you have a proven design you’d like to share I’d rather hit the ground running.