Your 50K FB VR is wired in so that it has the potential to be set to 100% FB. You should have a ''minimum value'' resistor ahead or behind on the wiper to set a maximum FB to the VA. With the VR that's in place now, the VA cathode biasing is going to be affected as you adjust it because it is a direct line back to the OPT winding to ground. Assuming you have plans to set it as fixed once you get the best performance decided. You can then try some phase caps across it to get the overshoot smaller if you want. Just a tweak at that point, but because you have another stage coupling cap now, it could be beneficial.
What does the 5pF FB provide?
What does the 5pF FB provide?
Last edited:
I had forgotten to put it on the schematic.
I have a 3k resistor, but failed to add it to the schematic.
The 5pF provides some compensation.
I tried a cap across the feedback resistor first, however that just made the amp want to go unstable.
Here's the corrected schematic.
Doing a basic test the amp is flat from 20Hz-20kHz and there's zero phase shift at 20Hz with only a small bit of phase shift at 20kHz. In fact if not for needing to reduce the output level at 20kHz with the 500k pot and 47pF cap there would be even less phase shift at 20kHz.
Given the idle 6V6 cathode currents are only 1mA off from being matched do I really need to get a matched quad of 6V6 tubes?
At the moment I'm just happy I got that Edcor transformer to play so nicely with the amp given someone on another forum had to do a lot of work to get the Edcor they used to work good in a Magnavox 9300 amp.
Last edited:
What compensation did you calculate with the 5pF cap? My calculator puts its affective rolloff point at 45MHz. But there could be something I'm missing. But I don't think it's far off. The filter is based on the 820R cathode resistor and parallel R of the FB line components estimated to take the K resistance to 750R.
Didn't calculate. Just tried various caps I had on hand which were 390pF, 200pF, 47pF and 5pF.
5pF was the value that provided the best looking 10kHz square wave, although without the 47pF cap and 500k pot with a clip lead connected I think to the junction of the .47uF cap and 1 meg resistor the 10kHz square wave looked even better so maybe the amp wants slightly more than 5pF. I do have a 1.5-20pF cap, however that would put B+ on one part of the variable cap.
The square wave looked a little better until I added the 47pF cap and 500k pot to flatten the amp's response at 20kHz as without it the output increased slightly starting at 10kHz.
5pF was the value that provided the best looking 10kHz square wave, although without the 47pF cap and 500k pot with a clip lead connected I think to the junction of the .47uF cap and 1 meg resistor the 10kHz square wave looked even better so maybe the amp wants slightly more than 5pF. I do have a 1.5-20pF cap, however that would put B+ on one part of the variable cap.
The square wave looked a little better until I added the 47pF cap and 500k pot to flatten the amp's response at 20kHz as without it the output increased slightly starting at 10kHz.
Last edited:
OK, so that says that any amount of cap, on their own, without the correct amount of a series resistor too, coming from the plate, is a bad FB circuit to place into the cathode FB circuit. The 5pF cap is blocking all of the signal going to it, no audio frequencies are passing it. So in affect it's an open circuit, as is. Might as well just disconnect it.
Without the 5pF cap the 10kHz square wave was not nearly as good.
I'm thinking that the 5pF cap will pas some signal although at a greatly reduced level, but at the plate it's got over 100V of signal.
Odds are I could replace the 5pF cap with a larger cap and a trimpot and get the same or better result.
I'm thinking that the 5pF cap will pas some signal although at a greatly reduced level, but at the plate it's got over 100V of signal.
Odds are I could replace the 5pF cap with a larger cap and a trimpot and get the same or better result.
''Odds''.... is not a plan.
Make sure you have a high enough voltage rating on that cap in the plate circuit. It could be subject to 2X the plate voltage under max signal conditions.
Make sure you have a high enough voltage rating on that cap in the plate circuit. It could be subject to 2X the plate voltage under max signal conditions.
Not sure what it's rated at, but it came from a defunct vacuum tube tuner so it's likely rated at a higher voltage.
Yep it was a vacuum tube FM tuner that was beyond repair.
Now should I try a larger value cap and a variable resistor in place of the 5pF cap or is it best to just leave well enough alone?
I might temporarily try the 1.5-20pF variable cap just to see if the square wave can be improved any.
Now should I try a larger value cap and a variable resistor in place of the 5pF cap or is it best to just leave well enough alone?
I might temporarily try the 1.5-20pF variable cap just to see if the square wave can be improved any.
Honestly, I just don't see that 5pF cap doing anything. It makes sense it's out of an FM tuner because it would be needed forthe MHz frequencies.
The Dynaco amp uses 390pF coming from the screen tap but I believe its purpose is to send only ultrasonic frequencies back for stability enhancement. Most amps don't even do that. Not needed. That's tailored to the Dynaco amp. And that is from the screen tap and not the plate. Maybe little difference, but it's something to consider.I have to say that from my experience the right amount of GNFB and sometimes a parallel phase cap, sometimes not, get the job done.
The Dynaco amp uses 390pF coming from the screen tap but I believe its purpose is to send only ultrasonic frequencies back for stability enhancement. Most amps don't even do that. Not needed. That's tailored to the Dynaco amp. And that is from the screen tap and not the plate. Maybe little difference, but it's something to consider.I have to say that from my experience the right amount of GNFB and sometimes a parallel phase cap, sometimes not, get the job done.
Maybe also a simple Zobel network on the output/FB tie in, too, could help. 50R and .1uF cap to ground. Easy to try. Used a lot by DCG.
The idea for the cap came from here.
https://audiokarma.org/forums/index...-magnavox-the-9300-series.687735/post-9213408
https://audiokarma.org/forums/index...-magnavox-the-9300-series.687735/post-9213408
I'm hoping someone can explain the purpose of the 1.68uF capacitor connected in parallel with the 1H filter choke in the B+ supply. As below:
As for negative feedback compensation...
If the problem is an ultrasonic resonance or phase shift caused by the OPT, you can limit the bandwidth of the driver stage at higher frequencies to attenuate those a bit. The traditional way to do that would be in two steps:
1. You could put a small value capacitor from the plate of the 1st stage voltage amplifier to ground, or put a resistor in series with that capacitor to make it a step network. This step network is highlighted in green in the schematic below:
Note that a small value capacitor in parallel with the plate load resistor would do the same thing, as signal ground can be either the 0 (zero) volts ground bus, or the filtered B+ plate supply. Both are at ground as far as AC signals are concerned.
2. You can instead—or also—put a small value capacitor in parallel with the negative feedback resistor, as shown in the schematic below (Dynaco ST35):
You can do both, or just one, depending on what works for your particular circuit with your particular OPT. People usually have to experiment a bit to get the best looking 10kHz square wave.
--
EDIT TO ADD:
1) I just noticed that both of these classic circuits don't have a decoupling network between the 1st stage voltage amp and 2nd stage phase inverter. FWIW...
2) In the Dyna ST35, the extra small value compensation capacitor connects the cathode of the 1st voltage amplifier to the screen grid of one of the output tubes. In your circuit, are you sure the 47pF capacitor is in the right location to provide high frequency compensation?
If the problem is an ultrasonic resonance or phase shift caused by the OPT, you can limit the bandwidth of the driver stage at higher frequencies to attenuate those a bit. The traditional way to do that would be in two steps:
1. You could put a small value capacitor from the plate of the 1st stage voltage amplifier to ground, or put a resistor in series with that capacitor to make it a step network. This step network is highlighted in green in the schematic below:
Note that a small value capacitor in parallel with the plate load resistor would do the same thing, as signal ground can be either the 0 (zero) volts ground bus, or the filtered B+ plate supply. Both are at ground as far as AC signals are concerned.
2. You can instead—or also—put a small value capacitor in parallel with the negative feedback resistor, as shown in the schematic below (Dynaco ST35):
You can do both, or just one, depending on what works for your particular circuit with your particular OPT. People usually have to experiment a bit to get the best looking 10kHz square wave.
--
EDIT TO ADD:
1) I just noticed that both of these classic circuits don't have a decoupling network between the 1st stage voltage amp and 2nd stage phase inverter. FWIW...
2) In the Dyna ST35, the extra small value compensation capacitor connects the cathode of the 1st voltage amplifier to the screen grid of one of the output tubes. In your circuit, are you sure the 47pF capacitor is in the right location to provide high frequency compensation?
Last edited:
That forms a LC resonant circuit with the choke at 120Hz. Reduces ripple voltage.
In this case the ripple voltage is 5.4V before the choke and 35mV after the choke. It was higher after the choke before adding that cap and I had a slight hum in the speaker.
In another amp the ripple voltage was 200mV at the choke output and the cap across the choke dropped it to 12mV.
That said the choke value changes with B+ load so as the amp puts out more power the effect of the cap is less, however as the ripple voltage increases some it won't be heard over the music.
That's the purpose of the 47pF cap and 500k pot.
Tried that initially and the amp didn't like that given it looked like the amp was more unstable based on how the 10kHz square wave looked.
In this amp I had to connect that cap to the plate given I am running in pentode mode.
It connects where the step network in green connects.