Why is this Princeton style amp not loud?

[Sonny ReVerb]

I built this amp with blackface Princeton and brownface Vibrolux channels and a 7591 output section. This is the second version of this amp I've built. The first one is a great amp. This one has a bigger power transformer and a 2x10 cab. It's supposed to be louder than the first one, but it's not. Both channels are relatively low volume until you get to about 7-8 on the volume knob. It starts to get loud at that point, but it's no longer clean. This amp should be 35-40W. I have a Vibrolux Reverb that is very loud at 3-4 on the volume. I know a LTP PI would provide more drive, but the Princeton PI should provide plenty of drive for 7591's. I need another set of eyes to check my work.

[Sonny ReVerb]

Some pics

[sluckey]

The pic that shows the actual PI circuitry does not match the schematic PI circuitry. Do you have a layout drawing?

[Sonny ReVerb]

Thank you, Sir. I tried this idea I found on TDPRI for a fixed bias cathodyne phase inverter and forgot to document it in the schematic. Could be part of the problem...

https://www.tdpri.com/threads/fender-cathodyne-phase-inverters.755580/post-7796034

[tubeswell]

Thank you, Sir. I tried this idea I found on TDPRI for a fixed bias cathodyne phase inverter and forgot to document it in the schematic. Could be part of the problem...

https://www.tdpri.com/threads/fender-cathodyne-phase-inverters.755580/post-7796034

^that^ cathodyne grid is set at 42.5V, which will significantly reduce clean output voltage and cause early clipping (the cathode voltage will not stay at 56V - it will automatically drop to follow the grid and this will cause clipping in this stage as the amp is cranked) because it will muck up the B load-line.

[wsscott]

Tubeswell-how did you calculate that number? I’d like to learn something. Thanks

[sluckey]

Tubeswell-how did you calculate that number? I’d like to learn something. Thanks

It's a simple voltage divider that can be solved with Ohm's Law. You have 250V applied across a 3.3M in series with a 680K. Solve for the voltage dropped across the 680K and you have the grid voltage.

R_t = 3,300,000 + 680K = 3,980,000Ω
I_t = 250/3,980,000 = .000063A
Voltage across the 680K is .000063 X 680,000 = 42.8V

There is no grid current so no voltage is dropped across the 1M or 470K so that 42.8V will also appear on the grid.

[wsscott]

Sluckey-Thanks.  I had thought that the grids on the preamp tubes were supposed to be at or near 0 VDC, and maybe similarly for the cathodes.  So seeing these values of 42.8VDC on the grid and 56VDC on the cathode has gotten me confused.

I know I'm missing a lot about operation and theory, but is it different for the grids and cathode on the phase inverter part of the circuit?  Or is it something else that's going on?

So what's the fix for his problem that you describe?

[sluckey]

The actual voltage on the grid is not that important. It's the voltage ***BETWEEN*** the grid and cathode that's important. Go read some Merlin pages.

So what's the fix for his problem that you describe?

The fix is to revert back to the original bootstrap bias circuit used in the original PR.

[wsscott]

Thanks.  I'll check out his website.

[Sonny ReVerb]

Well, I dug into the amp today. I had already removed that fixed bias circuit for the PI, so that wasn't the problem. I retraced the circuit. I didn't find any wiring problems or incorrect components. I used a 100 ohm tail resistor and a 3300 ohm NFB resistor from the 4 ohm speaker tap for the NFB circuit. I tried disconnecting that. No significant change. I'm not sure there's a real problem, because everything works. It just doesn't seem as loud as it should be. I ordered some CTS "J taper" 1 meg pots. I'm going to replace the Alpha volume pots. Maybe that will make me happy...?

[Sonny ReVerb]

Updated pics

[Sonny ReVerb]

More pics

[Sonny ReVerb]

More pics

[wsscott]

I know exactly what you mean!  Been there--Am there!  Good luck.

[HotBluePlates]

... everything works. It just doesn't seem as loud as it should be. ...

What speaker(s) are you using?

... This amp should be 35-40W. I have a Vibrolux Reverb that is very loud at 3-4 on the volume. ...

What speakers are you using in the Vibrolux Reverb?

How about plugging the 7591 Princeton into the Vibrolux' speakers to verify it's not just a case of "low-sensitivity speaker"?

[Sonny ReVerb]

That's a good suggestion! I have a pair of the Jupiter 10LC speakers in the new amp. The Vibrolux Reverb has the original pair of Oxfords.

[acheld]

Those Jupiters should be sensitive enough to give you the juice.  Once you check that your low output follows the amp (ie, your speaker choice is not the problem -- then measure your actual power output across a dummy load.

One thing that comes to mind when you mention behavior of your other amp's behavior "loud at 3-4" is that your volume pot may be linear, not audio.   Or the pot could actually be bad. 

Nice looking build!

[Sonny ReVerb]

Update - I made some changes to the amp today. I swapped in the CTS J taper volume pots. I changed the reverb mix resistor to a 2M2 instead of the 3M3. I reconnected the NFB. Lastly, I changed the PI bias resistor from the 1k to a 1k5 as in the Pro 5E5-A.

There's definitely an improvement in the overall volume. You still have to get up to '7' or so for a usable stage volume. The CTS pots are nice, but there is less turning resistance than the Alphas. No biggie, but it is noticeable. The screen voltage is about 10 volts over the maximum for 7591's. If they melt, they melt. I hope to Leo, they don't :)

I'm going to take it to a local pub on Tuesday to see if a friend of mine can give it a test run in a band setting.

[HotBluePlates]

I built this amp with ... a 7591 output section. ... Both channels are relatively low volume until you get to about 7-8 on the volume knob. ...

Update - I made some changes to the amp today. I swapped in the CTS J taper volume pots. ...

... You still have to get up to '7' or so for a usable stage volume. ...

Which-brand "7591" are you using?  Any chance you can get some old Sylvania 7591s?

On another forum a question was asked about how modern tubes marked "7591" compare to originals.  Someone noted one brand were simply 6L6s with a different pin-out.

   - Old 6591 tubes have a much higher Transconductance than 6L6s, but have a plate dissipation rating like first-version 6L6 tubes.

   - High Transconductance (compared to a 6L6) means the 7591 delivers the same-output with less drive-signal.  Or said another way, it delivers a bigger output with the same drive-signal.


I suggested on that other forum that folks with tube testers should measure the Transconductance of their supposed-7591 and see whether it shows figures consistent with the 7591 data sheet.  An alternative approach would be to see whether the "7591" has a fairly small negative bias-voltage to control its plate current.

   - Your schematic shows -26.5v of bias yields about 28mA of cathode current with ~430v on the screen.

   - The lower graph on Page 5 of the 7591 data sheet suggests that with this screen voltage, -25v of bias should pretty much cut off plate current.  That suggests these tubes do not have all the Transconductance they should have for healthy 7591s.

   - However, curves for a 6L6 (Page 8 of this 6L6 data sheet) suggest plate current would be over 100mA for the same screen & bias voltages.

   - If these are old-production 7591 tubes, are they simply worn out?  That could explain "7591-ish bias" but also explain why the bias voltage is more-negative than it should be for this idle current, and also the low output power experienced in-circuit.

[Sonny ReVerb]

These 7591's are new version Tung Sol.

[HotBluePlates]

These 7591's are new version Tung Sol.

The other guy's tubes were Electro-Harmonix 7591s.

They tried to melt-down when plugged into a cathode-biased 7591 socket, so he had to modify the amp to cool off the idle current.  That was likely because the tubes had 6L6-like Transconductance, rather than 7591-like Transconductance.

Your schematic figures suggest a less-extreme situation, but still probably low-Gm.

[Sonny ReVerb]

Thanks for the info. I will test these Tung Sol's when I get a chance. I also have a pair of JJ's and I ordered a pair of Philco's (Sylvania?). I'll see how they all compare.

[Helmholtz]

Your PI supply seems to be around 100V lower than the Prince schematic value.
What is the main B+ (schematic point A)?