Fender The Twin Problems
Moderators: pompeiisneaks, Colossal
Fender The Twin Problems
I have a Fender The Twin red knob that’s giving me issues. The guy I got it from says it was worked on by a tech and it came back to him like it is. After warmup it has a hum in both channels with volume all the way down. When playing, the output is dirty with a weird octave ghost note on single notes. When strummed, sometimes it has an almost chorus effect on it. On the scope, the signal on the output is pretty nasty at idle with volume down and no instrument input. The caps all look good and don’t seem to be leaking. Unless I’m missing something, voltages seem to be ok. Attaching videos and pic. Any help with this is very greatly appreciated.
You do not have the required permissions to view the files attached to this post.
Last edited by wattsup on Thu Apr 10, 2025 5:59 pm, edited 1 time in total.
Re: Fender The Twin Problems
Evil Twin has FX loop? Divide and conquer.
I build and repair tube amps. http://amps.monkeymatic.com
Re: Fender The Twin Problems
Even if the caps look fine, they might still be dried out or failing. Same goes for resistors, especially plate or screen resistors. Swap out your tubes if you haven’t already, especially the power and phase inverter ones
Re: Fender The Twin Problems
Changed out power and preamp tubes. No change. Tried pulling preamp tubes one at a time also with no change.
-
solderhead
- Posts: 151
- Joined: Thu Jan 16, 2025 5:42 pm
1 others liked this
Re: Fender The Twin Problems
Thanks for posting that 'scope trace -- if my hunch is correct then it provides all of the information necessary to fix your problem.
TLDR: you need to add a 50pF cap across pins 1 and 6 of V4.
Long Version:
This amp would be fun to have on the bench. The 'scope trace clearly indicates that your amp is oscillating within the NFB circuit. For whatever reason, the amp came back from the previous tech in a marginally stable state. There's no way for me to determine (based on the limited information you've provided) what happened when the other tech had it to render it marginally stable, but it could have been as simple as a tube change or a problem with lead dress that caused a phase shift in the negative feedback circuit to approach 180-degrees.
What is your bias current? Please check it. Try re-biasing the amp using A) the crossover notch method and B) blindly biasing the amp to 40mA, then report back on if the amp's behavior is improved with either method. The Twin has test points 201-203 for checking bias current by measuring the voltages across R215/216.
Looking at your scope trace, it appears that the dominant pole of the amp's NFB circuit has not been compensated correctly which has rendered the amp marginally stable and prone to oscillation when excited.
It appears that oscillation is being triggered by notch distortion as the sine wave crosses the origin. The square edge of the bias notch contains enough HF content to provide the octave-up harmonics/chorusing that you're hearing by creating an impulse edge that excites oscillation in a marginally stable amp that is otherwise effectively damped The solution to the problem is to damp the dominant pole frequency in the NFB loop to exclude the excited frequency. To do this requires that you determine the dominant pole in the amp's NFB loop and the secondary pole in the NFB loop, and slug the dominant pole with sufficient capacitance to form an RC low pass filter at a frequency that will limit the bandwidth of the amp so that it will not have unity gain by the time that the signal frequency reaches the secondary pole.
I could post the math here for how to calculate all of the frequency poles for the NFB loop but I swear that nobody is going to want to read it.
The quick and dirty method to solving this problem is to limit the bandwidth of the amp so that it isn't prone to HF oscillation. To do this you want to create a -6dB/octave RC filter at a frequency that is high enough not to interfere with the necessary BW of the amp, but low enough that given the -6dB/octave slope, the frequency response of the amp will be greatly diminished by the time that the signal reaches the secondary pole. The reason for this is that each RC pole induces 90-degrees of phase shift, and at the second pole you've aggregated 180-degrees of phase shift. In an NFB circuit 180-degrees of phase shift makes the feedback positive and oscillation ensues at that frequency. By diminishing the frequency response of the amp at the second pole you'll shrink the size of the oscillating waveform to insignificance.
You could use your scope to zoom in on the hash that is prevalent on both sides of your test signal, to determine the frequency of oscillation. And then do the math. Or you could take a simpler approach. Say that you're willing to arbitrarily constrain the BW of your amp to be 8kHz at the top end. You want to set the knee of your slugging filter to be 2 octaves higher, or 32 kHz, so there will be no audible effects. So you want to create a 32kHz LP filter. You've got ~100k of effective impedance at play during each half of the P-P cycle and you want to solve for C in the equation f=1/(2PiRC). The math says that C = 50pF. Add 50pF of capacitance to the amp across pins 1 and 6 of V4 and let us know how it goes.
That's the best that I can do for a scientific wild-assed guess based on the limited information that's been provided.
TLDR: you need to add a 50pF cap across pins 1 and 6 of V4.
Long Version:
This amp would be fun to have on the bench. The 'scope trace clearly indicates that your amp is oscillating within the NFB circuit. For whatever reason, the amp came back from the previous tech in a marginally stable state. There's no way for me to determine (based on the limited information you've provided) what happened when the other tech had it to render it marginally stable, but it could have been as simple as a tube change or a problem with lead dress that caused a phase shift in the negative feedback circuit to approach 180-degrees.
What is your bias current? Please check it. Try re-biasing the amp using A) the crossover notch method and B) blindly biasing the amp to 40mA, then report back on if the amp's behavior is improved with either method. The Twin has test points 201-203 for checking bias current by measuring the voltages across R215/216.
Looking at your scope trace, it appears that the dominant pole of the amp's NFB circuit has not been compensated correctly which has rendered the amp marginally stable and prone to oscillation when excited.
It appears that oscillation is being triggered by notch distortion as the sine wave crosses the origin. The square edge of the bias notch contains enough HF content to provide the octave-up harmonics/chorusing that you're hearing by creating an impulse edge that excites oscillation in a marginally stable amp that is otherwise effectively damped The solution to the problem is to damp the dominant pole frequency in the NFB loop to exclude the excited frequency. To do this requires that you determine the dominant pole in the amp's NFB loop and the secondary pole in the NFB loop, and slug the dominant pole with sufficient capacitance to form an RC low pass filter at a frequency that will limit the bandwidth of the amp so that it will not have unity gain by the time that the signal frequency reaches the secondary pole.
I could post the math here for how to calculate all of the frequency poles for the NFB loop but I swear that nobody is going to want to read it.
The quick and dirty method to solving this problem is to limit the bandwidth of the amp so that it isn't prone to HF oscillation. To do this you want to create a -6dB/octave RC filter at a frequency that is high enough not to interfere with the necessary BW of the amp, but low enough that given the -6dB/octave slope, the frequency response of the amp will be greatly diminished by the time that the signal reaches the secondary pole. The reason for this is that each RC pole induces 90-degrees of phase shift, and at the second pole you've aggregated 180-degrees of phase shift. In an NFB circuit 180-degrees of phase shift makes the feedback positive and oscillation ensues at that frequency. By diminishing the frequency response of the amp at the second pole you'll shrink the size of the oscillating waveform to insignificance.
You could use your scope to zoom in on the hash that is prevalent on both sides of your test signal, to determine the frequency of oscillation. And then do the math. Or you could take a simpler approach. Say that you're willing to arbitrarily constrain the BW of your amp to be 8kHz at the top end. You want to set the knee of your slugging filter to be 2 octaves higher, or 32 kHz, so there will be no audible effects. So you want to create a 32kHz LP filter. You've got ~100k of effective impedance at play during each half of the P-P cycle and you want to solve for C in the equation f=1/(2PiRC). The math says that C = 50pF. Add 50pF of capacitance to the amp across pins 1 and 6 of V4 and let us know how it goes.
That's the best that I can do for a scientific wild-assed guess based on the limited information that's been provided.
Better tone through mathematics.
-
WiderGates
- Posts: 12
- Joined: Tue Sep 04, 2007 3:31 pm
- Location: Germany
-
martin manning
- Posts: 14058
- Joined: Sun Jul 06, 2008 12:43 am
- Location: 39°06' N 84°30' W
Re: Fender The Twin Problems
I would!solderhead wrote: ↑Fri Apr 11, 2025 8:06 am I could post the math here for how to calculate all of the frequency poles for the NFB loop but I swear that nobody is going to want to read it.
-
solderhead
- Posts: 151
- Joined: Thu Jan 16, 2025 5:42 pm
Re: Fender The Twin Problems
Uh-oh. Am I looking at the right schematic? I was looking at this one:
You do not have the required permissions to view the files attached to this post.
Better tone through mathematics.
Re: Fender The Twin Problems
Please do!solderhead wrote: ↑Fri Apr 11, 2025 8:06 am I could post the math here for how to calculate all of the frequency poles for the NFB loop but I swear that nobody is going to want to read it.
Re: Fender The Twin Problems
That would be the one!solderhead wrote: ↑Fri Apr 11, 2025 11:35 amUh-oh. Am I looking at the right schematic? I was looking at this one:
TheTwin.pdf
Re: Fender The Twin Problems
No, you're missing the point, which is, using the FX loop you can test the preamp for the noise, and separately test the power amp for the noise--this will help you eliminate half the amp as the culprit.
I build and repair tube amps. http://amps.monkeymatic.com
-
solderhead
- Posts: 151
- Joined: Thu Jan 16, 2025 5:42 pm
Re: Fender The Twin Problems
Divide and conquer is a smart way to approach the problem. But that trace looks like ringing to me. If my S.W.A.G. is on-target then this is a NFB problem and we should focus our attention on the NFB loop.
I should have mentioned a couple of more things in my first post. First, it should be obvious that the easiest diagnostic method to confirm the NFB loop as the source of your problem would be to disconnect the NFB loop and see what happens. I thought about that but in my long-winded rant I forgot to type it. If doing that solves the problem then we know we're on the right track and can proceed with the steps that I mentioned to: a) suppress the triggering events, and b) increase the amp's margin of stability.
Another thing that I should have mentioned is that I determined that the OP's amp was 'marginally stable' because it was ringing rather than going into sustained oscillation when it was triggered. In the 'scope trace we can see that the amp is triggered every time the signal hits a crossover notch, and it rings, but the amp is sufficiently damped so that the ringing quickly subsides before the next crossover occurs. This tells me that the rate of closure between the open-loop gain and the closed-loop gain is more than 6dB/octave but the loop gain is already less than unity. (If the loop gain were more than unity then the amp would likely undergo sustained oscillation when triggered.)
Before we consider going into any math, I'd like to ask the OP to please perform all of the tests that I asked him to perform and post the results here. I would be particularly interested in seeing him dial-in the frequency of the ringing on his 'scope so that we can calculate the proper amount of capacitance to solve the problem. It would be cool if he should shoot video while he does this, to compare before/after traces when the tests are performed. The nerd in me finds that watching oscillation on an oscilloscope, and seeing it disappear when you add a cap of precisely the right calculated value amounts to great fun.
The solution that I recommended (50 pF cap across the PI plates [or even up to 220pF if needed]) is but one approach to solving the problem. Another approach is to add ~150pF in parallel to the plate resistors (which is equivalent of anode-to-ground), or to add 220pF caps from grid-to-cathode on the power tubes. All of these methods follow the same frequency calculations cited in the first post and have been used in various amps from Fender to Mesa to Marshall using whatever values were needed to solve the problems in their amps. Exactly how to approach the problem would depend on the results of the OP's experiments.
The method that I recommended for taking a blindfolded/dumbass approach to slugging the dominant pole of the amp, using a simplified method of bandwidth limiting, is called "pole-splitting." There's a Wikipedia article cited below. The traditional method for solving this problem is mathematically cumbersome. Pole-splitting methods derived by Cherry use graphical methods to avoid getting into any complicated math and provide a more simple approach that most people prefer to drawing Nyquist or Nichols plots, hand-drawing Bode approximations or performing Laplace transformations. The pole-splitting article is here:
https://en.wikipedia.org/wiki/Pole_splitting
Cherry's paper that gives a generalized example of how to do the quick and dirty approximations is here: Cherry provides the simplest, quick and dirty method that can be used to avoid hours of math derivations that would waste an entire evening. Truth be told, most people who work in production probably don't bother with any math -- they just blindly hang caps by trial and error and pronounce the amp as fixed when it stops oscillating and sounds OK.
By posting Cherry's work I've put the cart in front of the horse, just because multiple people asked for math. For this stuff to be relevant the OP needs to confirm that the ringing is occurring within the NFB loop, otherwise I'm wasting everyone's time talking about this.
I should have mentioned a couple of more things in my first post. First, it should be obvious that the easiest diagnostic method to confirm the NFB loop as the source of your problem would be to disconnect the NFB loop and see what happens. I thought about that but in my long-winded rant I forgot to type it. If doing that solves the problem then we know we're on the right track and can proceed with the steps that I mentioned to: a) suppress the triggering events, and b) increase the amp's margin of stability.
Another thing that I should have mentioned is that I determined that the OP's amp was 'marginally stable' because it was ringing rather than going into sustained oscillation when it was triggered. In the 'scope trace we can see that the amp is triggered every time the signal hits a crossover notch, and it rings, but the amp is sufficiently damped so that the ringing quickly subsides before the next crossover occurs. This tells me that the rate of closure between the open-loop gain and the closed-loop gain is more than 6dB/octave but the loop gain is already less than unity. (If the loop gain were more than unity then the amp would likely undergo sustained oscillation when triggered.)
Before we consider going into any math, I'd like to ask the OP to please perform all of the tests that I asked him to perform and post the results here. I would be particularly interested in seeing him dial-in the frequency of the ringing on his 'scope so that we can calculate the proper amount of capacitance to solve the problem. It would be cool if he should shoot video while he does this, to compare before/after traces when the tests are performed. The nerd in me finds that watching oscillation on an oscilloscope, and seeing it disappear when you add a cap of precisely the right calculated value amounts to great fun.
The solution that I recommended (50 pF cap across the PI plates [or even up to 220pF if needed]) is but one approach to solving the problem. Another approach is to add ~150pF in parallel to the plate resistors (which is equivalent of anode-to-ground), or to add 220pF caps from grid-to-cathode on the power tubes. All of these methods follow the same frequency calculations cited in the first post and have been used in various amps from Fender to Mesa to Marshall using whatever values were needed to solve the problems in their amps. Exactly how to approach the problem would depend on the results of the OP's experiments.
The method that I recommended for taking a blindfolded/dumbass approach to slugging the dominant pole of the amp, using a simplified method of bandwidth limiting, is called "pole-splitting." There's a Wikipedia article cited below. The traditional method for solving this problem is mathematically cumbersome. Pole-splitting methods derived by Cherry use graphical methods to avoid getting into any complicated math and provide a more simple approach that most people prefer to drawing Nyquist or Nichols plots, hand-drawing Bode approximations or performing Laplace transformations. The pole-splitting article is here:
https://en.wikipedia.org/wiki/Pole_splitting
Cherry's paper that gives a generalized example of how to do the quick and dirty approximations is here: Cherry provides the simplest, quick and dirty method that can be used to avoid hours of math derivations that would waste an entire evening. Truth be told, most people who work in production probably don't bother with any math -- they just blindly hang caps by trial and error and pronounce the amp as fixed when it stops oscillating and sounds OK.
By posting Cherry's work I've put the cart in front of the horse, just because multiple people asked for math. For this stuff to be relevant the OP needs to confirm that the ringing is occurring within the NFB loop, otherwise I'm wasting everyone's time talking about this.
You do not have the required permissions to view the files attached to this post.
Better tone through mathematics.
-
Stratscratcher
- Posts: 10
- Joined: Tue Dec 05, 2017 2:53 pm
Re: Fender The Twin Problems
I had "The Twin" that had a severe hum issue that I fought for while. Finally one day I noticed that the Hum would change as I pushed the buttons on the foot switch. It turned out to be grounds on the Power board where IC101 is located. The foot switch/switching uses 38VAC and IC101 is a comparator that "reads" the switched AC, and the hum varied with switch operation... The grounds on the that board are very thin wire, and has a lot of stuff going to ground on that board. I ran new ground wires and an additional ground from the middle of the board where IC101 is located to fix the hum issue