New Build, a mutt.

[mountainhick]

I lifted the connection to tone cap and pot at V3A, the PI plate... silence. Now to put the preamp and PI tubes back in.

All working, I'll add the bright switch on the 12AY7 channel. The 5879 is FAT!

Now voltage chart and bias...

 

[mountainhick]

So any ideas why the feedback/oscillation from the tone circuit? This is the exact circuit I had. The oscillation stopped when I detached the tone cap from upper PI plate:

 

[mountainhick]

Voltages with 5Y3... Might try to bump it with a 5V4GA. Room to move on bias, at 80%

Wall 122.4VAC
B+ 296

V1A P 116.1 K 1.79
V2A P 109, K shared

V2 P 159.5 SC 65.2 K 2.64

V3A P 138.7 K 1.06
V3B P 133.2 K shared

V4 P 292 Sc 293 K 18.4
V5 P 294 SC 294 K shared

Heaters 6.9V, not bad for a 1940's PT on modern voltage... I'll have to monitor its temp, but it ran more heater currentthan that in the radio it was pulled from.

 

[NickK_chugchug]

Is the ground from the LTP and to the tone stack low impedance? If not that could send a signal through the NFB.

Hmm but you've said silence when the cap-to-pot is broken which would indicate that the ground is working.

I was thinking that the power tube cathode signal will be through one cap =+90deg, then another cap to the plate load =+90deg which gives a +180 but it depends on the bandwidth of the pot and the cap. It may be a high frequency going through as a resistive load on the caps, which would not be +180 but instead +0.. which would then be a constructive interference.
You could try a grid stopper on the power tube to block out some of the high frequency.

The capacitance figures of the tube should give an indication if it's capable of amplifying HF noise above audio.

Lastly it could be that the ground is causing a ground loop that feeds back through the pot and the treble cap.

Just theorising.

 

[mountainhick]

Is the ground from the LTP and to the tone stack low impedance? If not that could send a signal through the NFB.

It is paraphase. There is no NFB from the speaker output. Did you look at the schematic in post #22?

 

[NickK_chugchug]

It is paraphase. There is no NFB from the speaker output. Did you look at the schematic in post #22?

Yes, not from the speaker but if the ground is high impedance at a high frequency then the connection you have in your schematic from the cathode of the power tube then becomes an AC loop.

Also I see a difference between your schematic and the post 1 schematic. The tone 1M is fed from the other tube (ie 180deg out of phase) and not the cathode (which is in phase).

Let me make a picture.

 

[mountainhick]

Is the ground from the LTP and to the tone stack low impedance? If not that could send a signal through the NFB.

Hmm but you've said silence when the cap-to-pot is broken which would indicate that the ground is working.

I was thinking that the power tube cathode signal will be through one cap =+90deg, then another cap to the plate load =+90deg which gives a +180 but it depends on the bandwidth of the pot and the cap. It may be a high frequency going through as a resistive load on the caps, which would not be +180 but instead +0.. which would then be a constructive interference.
You could try a grid stopper on the power tube to block out some of the high frequency.

The capacitance figures of the tube should give an indication if it's capable of amplifying HF noise above audio.

Lastly it could be that the ground is causing a ground loop that feeds back through the pot and the treble cap.

Just theorising.

Thank you for updating that post with more info. I get the feedback loop ideas.

My bad, sorry, I do have 1.5K grid stoppers on the power tubes. I updated post #22 schematic:

 

[mountainhick]

Yes, not from the speaker but if the ground is high impedance at a high frequency then the connection you have in your schematic from the cathode of the power tube then becomes an AC loop.

Also I see a difference between your schematic and the post 1 schematic. The tone 1M is fed from the other tube (ie 180deg out of phase) and not the cathode (which is in phase).

Let me make a picture.

I am a visual learner, a diagram would be helpful, Thanks.

The schematic in post #22 is a later version of the GA-40 schematic. It is the one I decided to use during construction because the tone pot is not connected to plate HV on the wiper like the older version.

The ground wire from the pot is run to a star like point on the ground buss. There may be a possibility that there is a ground loop not being directly connected to power tube cathode resistor. I'll look at the wiring closer. You can see it in the pic above.

 

[NickK_chugchug]

Cool... this was the observation - the original tone cap is simply using a resistor to use phases to cancel. By adding a ground point to the cathodes that could cause a problem IF the ground seems high resistance to a high frequency. But it sounds (sensibly) you have grid stoppers in etc.

I agree a non-HV carrying pot is a good safety feature!

You could power down and bleed, then unplug and use a small signal generator on the tone cap to see measure the impedance of the ground to that high frequency say at the decoupling cap.

The idea is you're looking for where the HF wave is the largest and the smallest on the ground (signals will follow back to the decoupling cap). That way you can find the highest impedance (large drop) and the lowest impedance (lowest drop). If that ground note you have next to the Tone 1M is high impedance.. then the HF will prefer to go a different route.. perhaps back through the tone pot.

The key here is phase. If the phase of the signal is rotated as it passes through components, if the phases result in a signal that rotates either by:
a) 0 or 360deg on the positive input of the amplifier
b) 180deg rotation to the negative input of a differential amp stage.

Then you'd get a signal that reinforces.. and the amp becomes unstable.

Oscilations are a pain to track down.. but it's typically a loop. A ground loop or something else that's causing a phase change in the loop to make it unstable.

 

[mountainhick]

Dropped bias resistor to 200R which pulls B+ down to 288V, and 88.3% dissappation.

Also tried a 5V4GA, and started getting strong 60hz hum. B+ up to 320V, power tube cathode voltage started climbing as hum got louder, then a little redness on the 6AQ5 plates.

Going back to 5Y3 and calling it good for now. I'll play around a little more with 250R vs 200R, if 200 is preferable may even try 180.

Would definitely benefit from bright caps on both channels.

I will just move the tone circuit over to pre PI, pre FX loop. Might also add a bass control. Both channels are fat. I was not expecting the 12AY7 to be that thick.

Tweak time!

 

[mountainhick]

Cool... this was the observation - the original tone cap is simply using a resistor to use phases to cancel. By adding a ground point to the cathodes that could cause a problem IF the ground seems high resistance to a high frequency. But it sounds (sensibly) you have grid stoppers in etc.
View attachment 1110481View attachment 1110482

I agree a non-HV carrying pot is a good safety feature!

You could power down and bleed, then unplug and use a small signal generator on the tone cap to see measure the impedance of the ground to that high frequency say at the decoupling cap.

The idea is you're looking for where the HF wave is the largest and the smallest on the ground (signals will follow back to the decoupling cap). That way you can find the highest impedance (large drop) and the lowest impedance (lowest drop). If that ground note you have next to the Tone 1M is high impedance.. then the HF will prefer to go a different route.. perhaps back through the tone pot.

The key here is phase. If the phase of the signal is rotated as it passes through components, if the phases result in a signal that rotates either by:
a) 0 or 360deg on the positive input of the amplifier
b) 180deg rotation to the negative input of a differential amp stage.

Then you'd get a signal that reinforces.. and the amp becomes unstable.

Oscilations are a pain to track down.. but it's typically a loop. A ground loop or something else that's causing a phase change in the loop to make it unstable.

I appreciate and understand your explanation. Thanks

 

[NTC]

One thing to note: 6AQ5's share a lot of ratings with 6V6'S, but they really are 12W tubes and they can't stand voltages much higher than their ratings, unlike 6V6's. Keep the plate voltage lower on these. What you are getting with the 5Y3 is where they want to be.

 

[printer2]

I see blue.. are you using metal foil resistors? (They can pass higher frequencies than the old carbon comp).

Wonder what kind of resistors they used in radio and TV circuits before the advent of metal foil resistors?

Is the ground from the LTP and to the tone stack low impedance? If not that could send a signal through the NFB.

The capacitance figures of the tube should give an indication if it's capable of amplifying HF noise above audio.

Already has been a mention it is a Paraphase PI not a LTP. In case you want to read up on it.

Page Title

The Tube CAD Journal for tube audio circuit design. From GlassWare the publishers of tube audio design software.

www.tubecad.com

Metaphrase Phase Splitter

www.tubecad.com

Cynosure Variations

www.tubecad.com

So how much capacitance would show that the power tube will pass frequencies above audio?

Cool... this was the observation - the original tone cap is simply using a resistor to use phases to cancel. By adding a ground point to the cathodes that could cause a problem IF the ground seems high resistance to a high frequency. But it sounds (sensibly) you have grid stoppers in etc.

You could power down and bleed, then unplug and use a small signal generator on the tone cap to see measure the impedance of the ground to that high frequency say at the decoupling cap.

The idea is you're looking for where the HF wave is the largest and the smallest on the ground (signals will follow back to the decoupling cap). That way you can find the highest impedance (large drop) and the lowest impedance (lowest drop). If that ground note you have next to the Tone 1M is high impedance.. then the HF will prefer to go a different route.. perhaps back through the tone pot.

The key here is phase. If the phase of the signal is rotated as it passes through components, if the phases result in a signal that rotates either by:
a) 0 or 360deg on the positive input of the amplifier
b) 180deg rotation to the negative input of a differential amp stage.

Then you'd get a signal that reinforces.. and the amp becomes unstable.

Oscilations are a pain to track down.. but it's typically a loop. A ground loop or something else that's causing a phase change in the loop to make it unstable.

What would be a high resistive ground when the tone pot is 1M?

Mountainhick, I have not looked at your layout much, just trying to comprehend the posts, covid kicking my butt at the moment. Back to bed, hope to look in again later.

 

[NickK_chugchug]

The Valve Wizard -Paraphase

So that makes sense. It’s using the phase output to drive the phase splitting. As printer has eluded to - the 1M is not a tone pot. (Mental note to self don't try to be helpful after a week of sleep deprived travel).
The paraphase normally creates a phase inversion using the grid signal to one of the power tubes. Todo that it is normally scaled and fed into the grid of the other half of the phase inverter. They have made the ability to bleed some of the AC to ground (the cap you have labeled “tone” is essentially a adjustable snubber as the AC signal will see past the cap to ground via the 1M resistor it allows you to attenuate the signal).

 

[mountainhick]

Hmm,

Wonder what kind of resistors they used in radio and TV circuits before the advent of metal foil resistors?

Already has been a mention it is a Paraphase PI not a LTP. In case you want to read up on it.

Page Title

The Tube CAD Journal for tube audio circuit design. From GlassWare the publishers of tube audio design software.

www.tubecad.com

Metaphrase Phase Splitter

www.tubecad.com

Cynosure Variations

www.tubecad.com

So how much capacitance would show that the power tube will pass frequencies above audio?


What would be a high resistive ground when the tone pot is 1M?

Mountainhick, I have not looked at your layout much, just trying to comprehend the posts, covid kicking my butt at the moment. Back to bed, hope to look in again later.

Get better soon!. I just had COVID, no fun. Tested negative for the first time in a month yesterday... then today went for my overdue appointment for blood letting related to the leukemia etc... I am feeling a little drained. LOL!

 

[printer2]

The Valve Wizard -Paraphase

So that makes sense. It’s using the phase output to drive the phase splitting. As printer has eluded to - the 1M is not a tone pot. (Mental note to self don't try to be helpful after a week of sleep deprived travel).
The paraphase normally creates a phase inversion using the grid signal to one of the power tubes. Todo that it is normally scaled and fed into the grid of the other half of the phase inverter. They have made the ability to bleed some of the AC to ground (the cap you have labeled “tone” is essentially a adjustable snubber as the AC signal will see past the cap to ground via the 1M resistor it allows you to attenuate the signal).

No, the pot is a tone pot. Forget the paraphase part of the amp, think of the top half of the amp as a SE amp. The pot and cap is the same as any tone cut knob, say the AC-4.

It just before the 6AQ5. The paraphase inverter and the bottom output tube are just add-ons to a SE amp, just happends to need less compromise with the output transformer due to the magnetic field being set up. In the early days of radio the output section was always in Class A, the paraphase circuit just a first step on the way to modern hifi.

 

[printer2]

Been scratching my head for a little while, finally figured out your layout. Nothing I can see other than the tone control lines to the pot. I would keep the wires close to each other if not twisted together. I look at layouts with the idea that each wire can be a transmitting antena or receiving (did I say this before?) and without getting too in depth into antennas (not my field) I see longer wires as better antennas than shorter ones, also the loop area of the wire. An equal length of wire, one going north and then south again along the same line will not pick up as much noise as a wire (same length as the previous one) going in a big loop. Think of it as a minnow net, much easier to catch them with the net wide open than if you turned it 90 degrees and have little area to scoop fish. On runs back and forth like the tone circuit I also give the wires a bit of a twist which helps cancel out any induced voltage. Can not see anything really wrong. Try chop-sticking the wires maybe?

 

[mountainhick]

Been scratching my head for a little while, finally figured out your layout. Nothing I can see other than the tone control lines to the pot. I would keep the wires close to each other if not twisted together. I look at layouts with the idea that each wire can be a transmitting antena or receiving (did I say this before?) and without getting too in depth into antennas (not my field) I see longer wires as better antennas than shorter ones, also the loop area of the wire. An equal length of wire, one going north and then south again along the same line will not pick up as much noise as a wire (same length as the previous one) going in a big loop. Think of it as a minnow net, much easier to catch them with the net wide open than if you turned it 90 degrees and have little area to scoop fish. On runs back and forth like the tone circuit I also give the wires a bit of a twist which helps cancel out any induced voltage. Can not see anything really wrong. Try chop-sticking the wires maybe?

Thanks for taking the time. Hope you are feeling better.

How about the power tube grid stopper wiring? Maybe better to switch the lead wires with the resistors? i.e. place resistors from board eyelets to tube pins?

 

[NickK_chugchug]

Grid stoppers are best at the pin. That way there is less lead (antenna) before amplification.

You can measure the phase & frequency for the feedback. That would identify what frequency range is causing the issue.
Todo that you need a frequency generator to inject a low amplitude signal into the NFB lne, then measure the signal of the EF80 grid.
It’s essentially doing a bode plot on the feedback rather than the amp itself. If the frequency phase shift is too high then the circuit will be unstable (nyquist diagram will show a crossing >1) and the circuit will oscillate.
In the original 12ax7 with EL34s, the grid capacitance of the el34 was ~8pF so that would make the el34 ignore RF noise. The ef80 had a grid cap of ~1.6pF IIRC so it will be happy to amplify RF thus possibly causing this issue.
An option is to put a tiny cap to ground directly - this shorts RF but not the audio frequencies. The value of cap and viability (ie if it’s actually rf causing this rather than <100khz) is really why the oscillation needs characterisation.
However lead dressing and also checking the ground path impedance would be useful. The length if the lead also increases inductance, so distance to the decoupling cap (ie filter caps) may also be playing a part. The use of EF80 may need the caps a little closer.
RF is funny in that it will follow the path of least impedance for the return current which may be a path to ground, DC or audio signal.

If you’ve not tried, the shield on the ef80 will increase the capacitance a little but i suspect not by enough.

If you can put a scope on with an FFT it should give better visibility (although the value of the cap should give a rough frequency range).

 

[mountainhick]

Grid stoppers are best at the pin. That way there is less lead (antenna) before amplification.

You can measure the phase & frequency for the feedback. That would identify what frequency range is causing the issue.
Todo that you need a frequency generator to inject a low amplitude signal into the NFB lne, then measure the signal of the EF80 grid.
It’s essentially doing a bode plot on the feedback rather than the amp itself. If the frequency phase shift is too high then the circuit will be unstable (nyquist diagram will show a crossing >1) and the circuit will oscillate.
In the original 12ax7 with EL34s, the grid capacitance of the el34 was ~8pF so that would make the el34 ignore RF noise. The ef80 had a grid cap of ~1.6pF IIRC so it will be happy to amplify RF thus possibly causing this issue.
An option is to put a tiny cap to ground directly - this shorts RF but not the audio frequencies. The value of cap and viability (ie if it’s actually rf causing this rather than <100khz) is really why the oscillation needs characterisation.
However lead dressing and also checking the ground path impedance would be useful. The length if the lead also increases inductance, so distance to the decoupling cap (ie filter caps) may also be playing a part. The use of EF80 may need the caps a little closer.
RF is funny in that it will follow the path of least impedance for the return current which may be a path to ground, DC or audio signal.

If you’ve not tried, the shield on the ef80 will increase the capacitance a little but i suspect not by enough.

If you can put a scope on with an FFT it should give better visibility (although the value of the cap should give a rough frequency range).

Thanks, yes I already got the grid stopper location part of that which is why I was asking @printer2 specifically about the location of them in this build since he took the time to understand the layout.

May I ask, have you actually reviewed the latest schematic on this discussion, the layout and the photo of the guts to actually see what I have done here? That you are talking about EF80 and EL34, seems you are confusing this project with something else.

Principles are still the same, and I understand and appreciate the input as general information to consider, but it is confusing when you seem to be talking about a different amp.

I do not have a scope of frequency generator, but am seriously considering the investment.

Thanks again