Power tube grid stop resistor and pF cap in PI.

[Mike_J]

Looked at a schematic showing the 47pF cap in the PI and the 1.5K grid stopper resistor as being in the signal path. Looked on the Fender Bassman reissue amp that has 47 ohm grid stoppers and the 47pF in the PI. Have heard a comment the grid stopper resistor at 1.5K doesn’t impact audible frequencies. Just helps with high frequency oscillation issues. Have also seen amps where the 47pF cap is eliminated. Would anyone be willing to give their two cents on these issues? For example is there anything magical about the 47pF cap? Might a 10pF cap be better? Same with a grid stopper feeding the grids of th power tubes.

[thetragichero]

47pf corner frequency is above the audible range (I'm getting about 33khz with 100k plate resistor). next most common value (at least in my parts bin) is 100pf which is it slightly above 15khz. may slightly dull the upper upper frequencies but probably okay. that's my guess why 47pf is so common

[pdf64]

47pf corner frequency is above the audible range (I'm getting about 33khz with 100k plate resistor)...

Bear in mind that the corner freq will differ according to the valve type etc.

[tubeswell]

Looked at a schematic showing the 47pF cap in the PI and the 1.5K grid stopper resistor as being in the signal path. Looked on the Fender Bassman reissue amp that has 47 ohm grid stoppers and the 47pF in the PI. Have heard a comment the grid stopper resistor at 1.5K doesn’t impact audible frequencies. Just helps with high frequency oscillation issues. Have also seen amps where the 47pF cap is eliminated. Would anyone be willing to give their two sense on these issues? For example is there anything magical about the 47pF cap? Might a 10pF cap be better? Same with a grid stopper feeding the grids of th power tubes.

There are different ways to skin the HF oscillation cat. Simple answer is build the amp, and if it has HF oscillation, then try one of those ways.


100pF 3kV between the PI plates is pretty common. (47pF doesn't do much IMO).


220pF to 2000pF grid-to-ground on the output tubes is another way (if you have good caps that don't leak).


1k5 grid stoppers on output tubes is pretty ineffectual. You can easily go up to 47k on pentode and tetrode grids without any audible HF rolloff (because of the low Miller capacitance enabled by the screens).


YMMV

[PRR]

> You can easily go up to 47k

Agree insignificant effect on frequency response.

If you do that, you change the overload action. Better, worse, depending on tubes, how you play, what you call good/bad.

Also 50k may be a large fraction of the allowable grid resistance.

I have a suspicion that 5k for guitar and 50k for hi-fi are fine starting points, but experiment.

[Mike_J]

47pf corner frequency is above the audible range (I'm getting about 33khz with 100k plate resistor). next most common value (at least in my parts bin) is 100pf which is it slightly above 15khz. may slightly dull the upper upper frequencies but probably okay. that's my guess why 47pf is so common

Knowing this information is very useful. Might try a 10pF just to see if it does anything. Guess is only negative things to the stability of the output stage. Thanks much for the information.

[Mike_J]

47pf corner frequency is above the audible range (I'm getting about 33khz with 100k plate resistor)...

Bear in mind that the corner freq will differ according to the valve type etc.

Tubes I am likely to be using are 5881s and 6L6GCs. Thanks for the help.

[pdf64]

Sorry, I meant the LTP valve type.

[Mike_J]

Sorry, I meant the LTP valve type.

That would most likely be a 12AX7 or perhaps a 12AT7 in that position.

[Mike_J]

Looked at a schematic showing the 47pF cap in the PI and the 1.5K grid stopper resistor as being in the signal path. Looked on the Fender Bassman reissue amp that has 47 ohm grid stoppers and the 47pF in the PI. Have heard a comment the grid stopper resistor at 1.5K doesn’t impact audible frequencies. Just helps with high frequency oscillation issues. Have also seen amps where the 47pF cap is eliminated. Would anyone be willing to give their two sense on these issues? For example is there anything magical about the 47pF cap? Might a 10pF cap be better? Same with a grid stopper feeding the grids of th power tubes.

There are different ways to skin the HF oscillation cat. Simple answer is build the amp, and if it has HF oscillation, then try one of those ways.


100pF 3kV between the PI plates is pretty common. (47pF doesn't do much IMO).


220pF to 2000pF grid-to-ground on the output tubes is another way (if you have good caps that don't leak).


1k5 grid stoppers on output tubes is pretty ineffectual. You can easily go up to 47k on pentode and tetrode grids without any audible HF rolloff (because of the low Miller capacitance enabled by the screens).


YMMV

Thank you and the other people who responded for their valuable responses.

[pdf64]

Sorry, I meant the LTP valve type.

That would most likely be a 12AX7 or perhaps a 12AT7 in that position.

They have rather different (internal) anode resistances, so different cap values would be needed to achieve the same corner freq.

[AmberB]

Definitely useful information here!

[pdf64]

The 50s/60s Fender development was that for the late Tweed and early brownface eras with a 12AX7 LTP phase splitter, an anode to anode (a-a) 47pF stability cap seemed to be used, without power valve grid stoppers.
With the BF era move to 12AT7 LTP phase splitter, the 47pF a-a stability cap was dropped, and 1k5 grid stoppers fitted to the power valve control grid socket terminals.
The somewhat lower anode resistance of the 12AT7 would make a 47pF a-a stability cap less effective, as its corner freq would be higher.

Having discovered the weird common mode signal output from an LTP that occurs when the anode to anode load impedance becomes very low (eg <10k with a typical 12AX7 LTP), I've come to appreciate that the SF era arrangement, of fitting stability caps between the power valve g1 and k terminals, as being a technically superior arrangement.

[Mike_J]

The 50s/60s Fender development was that for the late Tweed and early brownface eras with a 12AX7 LTP phase splitter, an anode to anode (a-a) 47pF stability cap seemed to be used, without power valve grid stoppers.
With the BF era move to 12AT7 LTP phase splitter, the 47pF a-a stability cap was dropped, and 1k5 grid stoppers fitted to the power valve control grid socket terminals.
The somewhat lower anode resistance of the 12AT7 would make a 47pF a-a stability cap less effective, as its corner freq would be higher.

Having discovered the weird common mode signal output from an LTP that occurs when the anode to anode load impedance becomes very low (eg <10k with a typical 12AX7 LTP), I've come to appreciate that the SF era arrangement, of fitting stability caps between the power valve g1 and k terminals, as being a technically superior arrangement.

Used a 12AX7 or a comparable 7025 for the PI. Had 47pF across the plates in the pi but no no grid stoppers to the power tubes. Had an oscillation that was cured by 1.5K grid stoppers. One of the reasons I posted this was the reissue amp schematic from the’90s used 47 ohm resistors for grid stoppers. Wondered if it made a difference from an audible range point of view. Sounds like probably not so much.


I am also planning on changing the pi tail resistor from the 10K to a 6.8K. The 6.8K is the value on a vintage ‘59 some guy owned and liked as well as the 90s reissue amp per the schematic. The 90s reissue amp is supposedly taken from a good sounding’59.


The reason I mention this is I would imagine the lower tail resistor value will drive a higher signal out of the pi. Don’t know how much but will probably find out.

[thetragichero]

https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/calculator/

[Mike_J]

https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/calculator/

Getting back to what we can do something about which is building amps. Can you explain what inverted and non-inverted voltage gain is and what is an ideal relationship?

[shooter]

Can you explain

surf up something like "OP AMP basics"
the real basics;
the inverting "side" is simply 180 degrees out of phase with the NON side,
you can get real deep in the weeds fast, i did my study in the '80s, then forgot the complex n just learned what wasn't normal and fixed it 

[pdf64]

https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/calculator/

Getting back to what we can do something about which is building amps. Can you explain what inverted and non-inverted voltage gain is and what is an ideal relationship?

With respect to the signal voltage at ‘in’, the inverting gain is the dB increase in signal voltage at ‘out1’, the non inverting gain is that at ‘out2’.
Ideally, ie for best linearity, they should be the same.
To calculate, voltage gain in dB = 20x log(VOUT / VIN)
https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/

[Mike_J]

Can you explain

surf up something like "OP AMP basics"
the real basics;
the inverting "side" is simply 180 degrees out of phase with the NON side,
you can get real deep in the weeds fast, i did my study in the '80s, then forgot the complex n just learned what wasn't normal and fixed it 

Everything I was able to find on the subject dealt with opamps as well. However, the calculator is for a long tail pi as I recall. My issue is with the tail resistor for the long tail pi. The schematic shows a 10K tail resistor but the actual 5f6a Bassman as built actually used 6.8K from what I have been able to discern. The tail resistor changes both the inverted voltage gain and non-inverted voltage gain per the calculator.

[Mike_J]

I looked for the tail resistor that had approximately the same inverted voltage gain and non-inverted voltage gain which happened at 8K, 25.31 and 25.25. The 6.8K resistor reflects 25.57 and 24.97 whereas the 10K reflects 24.99 and 25.59 respectively. Don’t know exactly what this means but I will be trying an 8K tail resistor as well as a 6.8K.

[Mike_J]

https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/calculator/

Getting back to what we can do something about which is building amps. Can you explain what inverted and non-inverted voltage gain is and what is an ideal relationship?

With respect to the signal voltage at ‘in’, the inverting gain is the dB increase in signal voltage at ‘out1’, the non inverting gain is that at ‘out2’.
Ideally, ie for best linearity, they should be the same.
To calculate, voltage gain in dB = 20x log(VOUT / VIN)
https://www.ampbooks.com/mobile/amplifier-calculators/long-tailed-pair/

If you look at reply 28 an 8K would provide the best linearity based on calculation of different values. Although I am still not sure I know what that means I appreciate you giving me the tool to calculate it.


Thanks
Mike

[pdf64]

For any reasonable given values of cathode and tail resistors, the most effective way to balance up the outputs is to tweak the anode load resistors. 
eg if the inverted gain is lower than the non inverted gain, then increase the value of RL1.
Note that use of trimmers is feasible, as per Dumbles.
But in reality, it’s probably not worth getting too concerned about, as valve guitar amps get overdriven. Such calculators don’t tell you anything about how circuits respond to their inputs being overdriven / their outputs subjected to halfwave clipping.
So it’s an iterative design / tweak process, using them in conjunction with sig gen-scope and guitar-speaker-ears.

[Mike_J]

To arrive at 8K for the tail resistor set RL1 to 82K, RL2 to 100K, RK to 470, RT to 8 and RG to 220K. All Fender pi values except the tail resistor which was calculated at 8K to balance the inverted voltage gain with non-inverted voltage gain.


I have made three Dumble clones and used 10K 500VDC trimmers for the reasons you mentioned although I didn’t know why I was doing it then. Played with RL1 value of 110K and RL2 value of 91K in the calculator with an RT value of 8K and the voltage gain numbers were off by quite a bit. Have decided for now my first experiment is going to be with resistors measured to as close to Fender values as I have in my parts inventory with with the exception of RT being 8K. Changing the tail resistor to balance the pi seems more sound than using a trimmer to accomplish the same goal. Especially since I have anecdotal evidence Fender used 6.8K tail resistors and I know for a fact that Allen Bradley carbon comps were not precision resistors. The 6.8K resistor probably worked best with the resistors in the pi for their test amp but you put resistors in that deviate from the schematic because of the lack of precision with the CCs they were using then maybe not so much.


I am very thankful for the calculator. As you mention over driving the amp may change things but I think this may be putting me on the right track. Want a growler. Not an amp that purrs.


Thanks
Mike

[Mike_J]

Played around a little more with the calculator and changing the 82K resistor to an 81.8K resistor gets me to the point where the dB values are the same for both the inverted and non-Inverted gains. I am sure it is nutty over kill. But that is where I live.


Thanks
Mike

[Mike_J]

For any reasonable given values of cathode and tail resistors, the most effective way to balance up the outputs is to tweak the anode load resistors. 
eg if the inverted gain is lower than the non inverted gain, then increase the value of RL1.
Note that use of trimmers is feasible, as per Dumbles.
But in reality, it’s probably not worth getting too concerned about, as valve guitar amps get overdriven. Such calculators don’t tell you anything about how circuits respond to their inputs being overdriven / their outputs subjected to halfwave clipping.
So it’s an iterative design / tweak process, using them in conjunction with sig gen-scope and guitar-speaker-ears.

You need to start somewhere and it makes sense to me to start with a balanced pi. Have to agree with you the plate load resistors are an effective way to get to a balanced pi. The realization Fender changed their tail resistor value without changing their schematic is what got me thinking about this. If you start with changing the tail resistor then you can go with Fender values for the other resistors. Tweaking RL1 can then get you to perfection if the exact value resistor you need is in your inventory. As you mentioned it might not mean anything but Dumble put some effort into those trimmers to accomplish the same thing. My guess is that it will be a fruitful exercise.

[pdf64]

Bear in mind that with Fenders etc, there’s (almost) always a further tail resistor and a global NFB loop, both of which act to improve the balance.
And that the Dumble balance trimmer also acts to bring any discrepancies in the whole balanced signal system into line, eg inc the power valves and OT, so not just the LTP. 
And that you may get better ‘growl’ with the harmonics that some degree of imbalance will introduce.

[Mike_J]

Bear in mind that with Fenders etc, there’s (almost) always a further tail resistor and a global NFB loop, both of which act to improve the balance.
And that the Dumble balance trimmer also acts to bring any discrepancies in the whole balanced signal system into line, eg inc the power valves and OT, so not just the LTP. 
And that you may get better ‘growl’ with the harmonics that some degree of imbalance will introduce.

Come on man. As a demented clown might say. Just as I think I have it all figured out you throw a wrench into my plans. Know where I am going to start now though. Thanks much for your help.

[Mike_J]

Bear in mind that with Fenders etc, there’s (almost) always a further tail resistor and a global NFB loop, both of which act to improve the balance.
And that the Dumble balance trimmer also acts to bring any discrepancies in the whole balanced signal system into line, eg inc the power valves and OT, so not just the LTP. 
And that you may get better ‘growl’ with the harmonics that some degree of imbalance will introduce.

How right you are. Spend $3.03 on a precision 8K resistor for the tail to balance it and now I am finding out the value across the 5K presence pot needs to be added to my 8K resistor which makes my balanced pi not  so much. Fender’s 6.8K is closer but to get it truly balanced I would need to take 8K less the value across the pot in my amp which will be +/~ 3K for the first tail resistor. Pretty big jump from 10K but it is starting to make sense why Fender liked the lower 6.8K value. Wonder if anyone has traveled down to the 3K area. Know someone plans on doing it but not until the amp is up and stable. Will probably throw in a 6.8K since it has been used successfully.


A question I have is why use a 25K resistor with a 4.7K pot across it. Have seen it on a vintage ‘59 as well as reissue schematic. Wrote off as out of 5K pots so they made one kind of.


Was trying to research it last night and someone said the pi tail is extended by the 4.7K resistor and for some reason the 25K was out of it. Only way that would make sense is if the .1 cap from  the wiper to CCW makes the 4.7K resistor a more attractive path to ground. If that is the case it would make sense to use a 10K pot and a lower resistor value for the presence pot.


Could probably search around and find a pot that measures low enough that the resistor and a 6.8K cap could get me to the 8K I was shooting for.

[2deaf]

47pf corner frequency is above the audible range (I'm getting about 33khz with 100k plate resistor). next most common value (at least in my parts bin) is 100pf which is it slightly above 15khz. may slightly dull the upper upper frequencies but probably okay. that's my guess why 47pf is so common

The output impedance that you would use to determine the corner frequency is the external plate resistor in parallel with the internal plate resistance.  The internal plate resistance for the inverting stage of the 12AX7 in a 5F6-A LTPI is 54K, so the effective output impedance is 33K.

The 47pF capacitor doesn't go to ground.  It goes between opposite phases.  When you do that, the effective impedance seen by each end is half the nominal impedance.  Half the nominal impedance would be twice the nominal capacitance, so the effective capacitance is 94pF.

Using 33K and 94pF, the corner frequency would be 51KHz.

[jjasilli]

For grid stoppers, see:
https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used

https://www.aikenamps.com/index.php/what-is-blocking-distortion

http://www.valvewizard.co.uk/gridstopper.html

For above & for snubber (fizz cap), see:
https://robrobinette.com/Voicing_an_Amp.htm#Phase_Inverter_Plate-to-Plate_Snubber   and scroll around the rest of that page.




Moderators: I think this Topic belongs in the Amp Section

[Mike_J]

For grid stoppers, see:
https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used

https://www.aikenamps.com/index.php/what-is-blocking-distortion

http://www.valvewizard.co.uk/gridstopper.html

For above & for snubber (fizz cap), see:
https://robrobinette.com/Voicing_an_Amp.htm#Phase_Inverter_Plate-to-Plate_Snubber   and scroll around the rest of that page.




Moderators: I think this Topic belongs in the Amp Section

Thank you so much for referring me to this reading material. The referrals to Aiken and Merlin were extremely valuable in my understanding grounding techniques. Never would have created the quasi-bus bar method I am using had I not read Merlin. Hoping it will make for a very quiet amp. Once it is sound then I can use the information above and I am sure some other things to tweak the amp to get it exactly where I want it.

[Mike_J]

For grid stoppers, see:
https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used

https://www.aikenamps.com/index.php/what-is-blocking-distortion

http://www.valvewizard.co.uk/gridstopper.html

For above & for snubber (fizz cap), see:
https://robrobinette.com/Voicing_an_Amp.htm#Phase_Inverter_Plate-to-Plate_Snubber   and scroll around the rest of that page.




Moderators: I think this Topic belongs in the Amp Section

Read Rob Robinette's article about ice pick highs and fizzy overdrive brought back memories about this amp. It was very ice picky which I attributed to speakers that needed some break in time. Also the overdrive was fizzy. I think when I say I want an amp that growls I mean one that can be pushed into overdrive without getting fizzy and doesn't have the ice pick highs. Never did put enough time on the speakers to break them in because of it. I have 68pF, 82pF and 100pF caps to try which may tame what I didn't like at first. Later the amp was played with my LTD and they sounded pretty much the same but they were probably both played clean and away from the treble range. Don't remember but am looking forward to trying different values to see. Thanks again for this information. I am guessing it will be useful to get me where I want to go.


Thanks
Mike

[Mike_J]

Bear in mind that with Fenders etc, there’s (almost) always a further tail resistor and a global NFB loop, both of which act to improve the balance.
And that the Dumble balance trimmer also acts to bring any discrepancies in the whole balanced signal system into line, eg inc the power valves and OT, so not just the LTP. 
And that you may get better ‘growl’ with the harmonics that some degree of imbalance will introduce.

How right you were. Fender changed the tail resistor from 10K to 6.8K and the presence pot from 5K to somewhere near 3.4K sometime in 1960 and on their reissue amps. Fender didn't change the 27K feedback resistor when they changed the pot value. Supposedly the later amps sound better and that is what their reissue amp schematic shows which would seem to support that.



I was speaking with Randall Aiken a few days ago about his white paper on the LTP and I think he said the relationship between the 27K feedback resistor and the presence pot is as a voltage divider so if you changed the presence pot value the feedback resistor would need to be changed to get the same voltage. Some lady was calling me over and over again about doing her tax return while I was talking to him so I was getting the constant beeping you get while you are on the phone and someone else is trying to reach you. It is therefore possible I misunderstood him. He is a very nice man. Enjoyed speaking with him except for the beeping caused by the woman who wouldn't give up on reaching me.


Thanks
Mike

[pdf64]

How do you get 3.4k?

[Mike_J]

How do you get 3.4k?

Measured from cw to ccw with the 4.7K resistor across those points. The pot is supposedly a 25K. Don't remember what the value actually was without the resistor across it or if I even measured it.

[pdf64]

The 0.1uF presence cap should be in series with the presence pot.
The pot is not wired directly in parallel with the 4k7 tail / NFB shunt resistor.
At least by measuring it, you've got the 27k NFB series resistor in parallel with it though, which I was going to remind you about needing to include 

[Mike_J]

The 0.1uF presence cap should be in series with the presence pot.
The pot is not wired directly in parallel with the 4k7 tail / NFB shunt resistor.
At least by measuring it, you've got the 27k NFB series resistor in parallel with it though, which I was going to remind you about needing to include 

The latest will be a 2.632K pot with nothing wired across it. The presence cap may or may not be .1uF. Depends on how different values sound using my sour cream container cap tool, wrapped in aluminum tape which may or may not help with EM interference. The NFB series resistor will be of some value using another tool with a pot connected to a thin strip of G10 to test different values for it. The NFB series resistor may be calculated as well. The tail resistor at least for the moment will be 5.36K. Could you explain how the NFB series resistor impacts my calculations for balancing the PI? It is not shown at all on the LTP calculator. I am so confused. All I remember Randall Aiken saying is the presence pot is connected to the NFB series resistor as a voltage divider and if the presence pot value is changed the NFB series resistor must be changed to keep the relationship consistent. The original 5f6a schematic had a 5K presence pot with nothing across it and the 27K NFB series resistor. Not entirely certain which would be considered R1 and which R2 so I don't know the component placement for calculating a voltage divider so a new value for the NFB series resistor can be determined. I am currently a muddled mess and could use your help getting unmuddled.


Thanks
Mike

[Mike_J]

If you read the article on the PI referenced below I think it explains it.


https://www.aikenamps.com/index.php/the-long-tail-pair


Aiken says, "the feedback divider ratio is the ratio between the series feedback resistor (R7) and the shunt feedback resistor (VR1).  The amount of feedback also controls the effective bootstrapped input impedance." If we are talking ratios here then before it was a 5K pot (VR1) and a 27K series feedback resistor. If I put a 2.632K pot in for VR1 then the series feedback resistor would need to be 14.2K to maintain the ratio.


I am not entirely certain when Fender changed the pot from a 5K to a 25K with a 4.7K resistor across it that anecdotally sounds better. If changing this ratio made the amp sound better using 14.2K may not be the ideal value for best tone. Not a problem, I have my thin strip of G10 with a pot on it that can help determine the best value. This would at least provide a range though. Only needed because I hope to balance the out-of-phase and in-phase outputs. Still have no idea what the series feedback resistor has to do with that. Only think the shunt feedback resistor is part of the tail resistor for purposes of the tail resistor value to use as input into the PI calculator.


VR1 is also known as the presence pot from what I can tell.


Thanks
Mike

[pdf64]

Just to note that the 27k series feedback resistor is connected to the 5k pot (or 4k7 resistor) shunt feedback resistance at one end, and the OT secondary at the other. Being an AC voltage source, the OT secondary has a low(ish) impedance path to 0V common. So the 27k series feedback resistor and the 5k (or 4k7) shunt feedback resistors are effectively in parallel.

I think Fender may have moved the 5F6A's shunt resistance arrangement from the 5k presence pot with 0.1uF presence cap, to a 4k7 shunt resistor in parallel with the presence cap and pot series pair, because it removes DC from the pot. The old way causes a rustle when the presence control is adjusted, the later way doesn't.

As you've halving the shunt resistor value, the series resistor value should also be halved.

The feedback ratio (beta B) = shunt / (shunt + series)

Regarding terminology, I try to avoid words that have more than one meaning, as that can't help but cause confusion. Hence I only use 'ground' when referring to an electrical connection to planet earth. Because I've found that people tend to conflate an amp's grounding system with its safety ground connection. So I refer to 0V / circuit common etc. Plus my electronics education was in the 80s, and the terminology I learnt was generally as I describe.

Also I try to avoid the conflation of a signal's 'polarity' with its 'phase'. So having to use the term 'phase inverter' pains me, and I'll try to describe it as cathodyne / LTP etc instead

[2deaf]

What output transformer are you using?  What power tubes?

[Mike_J]

What output transformer are you using?  What power tubes?

Using 5881s. Believe they are NOS RCAs. Have a considerable number of NOS Tungsols as well. The OT was made by Classictone and was I think a good attempt at remaking the original. It has only a 2 ohm secondary. Think I may have one somewhere that has 2, 4 and 8 ohm secondary taps.


Thanks
Mike