A strange signal artefact with a 1970 Fender Vibrolux Reverb.

[Helmholtz]

While servicing the amp I noticed a strange kind of signal distortion at large output in the Vibrato channel.
Especially with a square wave signal around 1 kHz signal it looks like ringing.
It only shows on one half-cycle and the amplitude of the ringing is relatively small.

But any sign of ringing makes me suspect a stability problem, so I was curious to find the root cause.
The effect disappears when I pull the reverb driver tube.

I did some tests to exclude a grounding problem, crosstalk between wires, bad filter caps or a conductive board.

I could trace the ringing artefact back to the grid of the reverb driver.
But there are no signs of grid conduction/clipping. Adding a 10k grid stopper doesn't change anything.

The ringing also disappears when I disconnect the tank input, i.e. the transducer load.

Scoping the driver plate signal shows a weird kind of distortion which disappears when the transducer is disconnected.

Using a square wave signal revealed that the distortion is caused by (asymmetrical) transducer core saturation:
At one side of the square wave the voltage collapses after a short peak, at the other side not.
Collapsing voltage is a sign of core saturation.

The asymmetry indicates a magnetized core.

Tried 2 other tanks and all showed the same effect.


So my conclusion is that the „ringing“ originates from reverb transducer core saturation and higher harmonics of the distorted plate signal get to the grid via the grid-to-plate capacitance of some pF and find their way back to the grid of the reverb/dry mixer tube V4A.
It is most likely a normal effect with most Fender amps and avoiding it would require to significantly reduce the reverb drive.


The effect might be common with other amps having a similar architecture.

[Chuck H]

Ah... So what if you used a "conjunctive filter" on the reverb transformer.? Not my term, but the most commonly used. Some just call it a Zobel network but that typically applies to a transformer secondary and this would be on the primary. Anyway, I hope you know the circuit I'm talking about. If the resistor and capacitor values were carefully chosen (or experimented with) it would probably be only barely audible. On amps that have the 220p cap load at the reverb recovery grid you might even be able to reduce that value and get back what was lost.

[Helmholtz]

Ah... So what if you used a "conjunctive filter" on the reverb transformer.? Not my term, but the most commonly used. Some just call it a Zobel network but that typically applies to a transformer secondary and this would be on the primary. Anyway, I hope you know the circuit I'm talking about. If the resistor and capacitor values were carefully chosen (or experimented with) it would probably be only barely audible. On amps that have the 220p cap load at the reverb recovery grid you might even be able to reduce that value and get back what was lost.

Yes, I know what you mean. A conjunctive filter/Zobel network is typically used at the primary of the OT with tube amps or directly across the output of SS amps.
It's main purpose is to compensate the increase of load impedance caused by the speaker inductance, which in turn causes an increase of open loop gain at HF.
The latter could cause a stability problem with NFB, so the Zobel improves stability.
In any case the Zobel is a low pass filter.

I tend to think that the effect I described is normal with Fender amps ( I recall having seen it with other Fender amps before) and attenuating the drive harmonics might change the signature reverb sound. And I love that Fender reverb sound.
As there's no stability issue, I don't see a reason for modification.

Not sure what you mean with the 220pF cap. Did you mean the 2nF cap? That cap actually boosts (reverb recovery) treble by a resonance effect.

[Pedro Vecino]

Not sure what you mean with the 220pF cap. Did you mean the 2nF cap? That cap actually boosts (reverb recovery) treble by a resonance effect.

Modern units (Deluxe / Twin Rev, Princeton, etc.) they usually have a 220p capacitor in parallel on the input and another of the same value on the reverb output. It is not used by older models like yours.
I also remember that the Super Twin reverb used in the pentode part of the 6CX8 (the tube that drives the tank) a 560pf / 3000V capacitor between the plate and the cathode.
I don't know if it is directly related to the effect you mention.

[pdf64]

How about if the V3 driver stage has its cathode bypass cap lifted? Or a resistor inserted in series with it, in the range 470ohms to 1k?
I find that helps to sort out quirks there.

[Chuck H]

Not really trying to find a solution for a problem that doesn't exist. And I know your observation of the "ringing" was just an observation. But since it was noticed as an artifact that occurs with clipping it occurred to me that the built in reverbs can usually use any improvement they can get when it comes to clipped tones and clarity. It's probable that eliminating the ringing wouldn't even make an audible difference but we don't know if we don't try. I don't have any amps with built in reverbs or I might recreate your test and see for myself.

[Mick Bailey]

Cascade folded line trays (Manufactured by beautiful girls in Milton) had a cap wired across the input transducer. I often wondered why, but I experimented with switchable caps across different trays and the results were quite interesting, to the extent that I thought an inclusion in my next standalone reverb build would be a rotary switch with a selection of capacitors. The effect is to give a very marked peak at a certain frequency and depending on the cap size can produce ringing - it sounds like a Minimoog filter where the resonance is turned right up. Just the right cap value increases the drive to the transducer over a broader range.

[Helmholtz]

Modern units (Deluxe / Twin Rev, Princeton, etc.) they usually have a 220p capacitor in parallel on the input and another of the same value on the reverb output. It is not used by older models like yours.

That 220pF cap across the rev. transformer secondary shouldn't have an effect at all.
Firstly because it's across a very low impedance (8 Ohm).
Secondly because the rev. transformer's self-capacitance at the secondary is around 100nF (!) and the 220pF will just add to that.

The 220pF cap at the recovery side adds to the cable capacitance (measured values around 200pF) and will produce a resonant boost around 10kHz.

[Helmholtz]

How about if the V3 driver stage has its cathode bypass cap lifted? Or a resistor inserted in series with it, in the range 470ohms to 1k?
I find that helps to sort out quirks there.

I'm sure lowering the gain would mitigate the effect, which uses the "Miller" feedback path from plate to grid. And the Miller effect depends on gain.
(BTW, the signal at the driver plate(s) is huge, around 480Vpp.)
But I don't want to change the reverb sound.
I always thought that there's some distortion (i.e. added harmonics) involved with the reverb drive of old Fender amps.
And I prefer the fully bypassed version.

I think the effect is normal with these amps.

But what fascinated me were 3 things (well, I'm an experimental physicist ):
1) that part of an output signal finds its way back to the input without any obvious NFB loop,
2) that I found heavy rev. transducer distortion,
3) that the distortion is asymmetrical, meaning a magnetized core (with all 3 tanks).

It all started because I wanted to see why the Normal channel puts out about 3W more than the Vibrato channel before any clipping.
The output with the Vibrato channel is limited by single side clipping.
Reason seems to be a slight asymmetry in the plate signal of V4A.

No wonder Fender specifies output power and distortion figures with the Normal channel only.

Schematic: https://docs.google.com/viewer?a=v&p...ZmMjg4ZWExYjE3

[Helmholtz]

But since it was noticed as an artifact that occurs with clipping it occurred to me that the built in reverbs can usually use any improvement they can get when it comes to clipped tones and clarity.

No clipping involved.
Core saturation doesn't produce voltage clipping.

[Chuck H]

Especially with a square wave signal...

I guess I misinterpreted this ^^^^^^^^^^^^^^

I know the square wave input signal was (and is typically) used to reveal amplifier performance. I took this as an indication of how the circuit might also manage actual amplifier clipping.

Considering it now I realize that (short of using a boost pedal) the input to the reverb driver never reaches clipping levels in these amps.

[pdf64]

… the input to the reverb driver never reaches clipping levels in these amps.

I dunno about that, at typical settings and bias, as signal level rises from 0, the reverb driver will clip the top of higher frequencies on that channel at levels not that much higher than that at which the output valves reach grid clipping, ie it’s the earliest clipping point on the reverb channel.
My hypothesis is that the horrid distortion that causes is why when master volumes were added, the cathode bypass on the reverb driver was removed. That moves the earliest preamp clipping point to the 3rd stage grid.

[Helmholtz]

I dunno about that, at typical settings and bias, as signal level rises from 0, the reverb driver will clip the top of higher frequencies on that channel at levels not that much higher than that at which the output valves reach grid clipping, ie it’s the earliest clipping point on the reverb channel.

I could not detect any sign of reverb driver grid (and plate) clipping even with heavy power amp clipping.
Rev. driver grid conduction will not start before 17Vpp. But grid signal is only around 11Vpp at full output.
So I don't agree with the above.

But what can be seen at the grid signal is the effect I described earlier.
It's a distortion at the negative peaks only, so it's easy to tell from grid clipping.
It gets noticeable at a grid signal above 9Vpp.

So I think the early high frequency distortion heard in the Vibrato channel is essentially due to the "effect".

[pdf64]

…
Rev. driver grid conduction will not start before 17Vpp. But grid signal is only around 11Vpp at full output.
…

I suppose that the ratio between the output grids clipping and the reverb driver’s grid clipping will be affected by the particular bias setting, and the model and its degree of NFB?
Even so, it might be that when set it a sweet spot ‘just slightly overdriven’, the note attack clips both the output valve and the reverb driver grids?
I’m not suggesting that with the stock circuit, the reverb grid clipping is audible, rather it only becomes noticeable when a Type 4 master vol is brought into play.

[Helmholtz]

I suppose that the ratio between the output grids clipping and the reverb driver’s grid clipping will be affected by the particular bias setting, and the model and its degree of NFB?
Even so, it might be that when set it a sweet spot ‘just slightly overdriven’, the note attack clips both the output valve and the reverb driver grids?
I’m not suggesting that with the stock circuit, the reverb grid clipping is audible, rather it only becomes noticeable when a Type 4 master vol is brought into play.

All I can say it that the reverb driver grid seems to have a lot of headroom before grid conduction.
In fact I couldn't see grid conduction before a level of 28Vpp with a continuous signal.
Reason is that the cathode voltage strongly increases with drive level, probably because of the asymmetric plate load.
(Cathode voltage does not increase with signal when the tank is disconnected.)
With short pulses the cathode voltage should be constant at 8.5V though.
Still the difference between 17Vpp and 11Vpp is significant and bias setting shouldn't change much (say less than 10%).

What makes you suspect reverb driver clipping?

If you're referring to the horrid distortion of Fender amps with the push/pull switch, the explanation is simple.
Signal is taken from the secondary of the reverb transformer.
That signal is heavily distorted by core saturation, the latter being known to sound ugly.
And core saturation distortion does not cause desirable compression, typically associated with "normal" distortion.