Distortion/Oscillation

[Leevi]

I have a Deluxe reverb type amp where I have probably an oscillation issue?
I get like a "farting" sound on low notes and bigger preamp volumes.
It comes even if the reverb is off. If I disconnect the 500pF cap on reverb driver grid or the 1M grid leak resistor the problem disappears.
The reverb driver is 1/2 12AX7 (not 12AT7). I have moved the reverb transformer and other components around the reverb tube without success.
Grid stoppers do not help. The tube has also been changed. I could understand the problem if it comes when reverb is on but...
Could it be related to voltage supply? The reverb transformer is now powered from "B" and the triode after tank from "D" like the other preamp tubes.

https://el34world.com/charts/Schematics/files/Fender/Fender_deluxe_reverb_ab763.pdf

/Leevi

[shooter]

If I disconnect the 500pF cap

I'm not a reverb guy so.....
what are you using the "other" 1/2 of the tube for?
have you tried moving B+ tap?

the circuit has been bullet proofed many times, but you changed it so it might be an un-intended consequence thing 

[PRR]

> a "farting" sound on low notes

Just to look another way: have you tried a different speaker? Some speaker overloads or faults "fart".

[SILVERGUN]

Is there any chance that you are sharing a cathode R/C on two sequential gain stages?


Did this just appear in an existing build or is it a problem in a new build?

[Leevi]

Some new information regarding the case.


The schematic is visible in the blueprint attached.


I don't think it is a speaker issue since I have tried with several different speakers
and it works if I remove the reverb tube.


It's a new build.


So once again:
- by disconnecting the 500p cap or 1 M resistor the problem disappears.
- the reverb is all the time off even shorted by foot switch


/Leevi

[PRR]

Sorry for the side-track.

> The reverb driver is 1/2 12AX7 (not 12AT7).

2 things smell wrong.

* 12AT7 will bias-up at 8V; 12AX7 nearer 2V. So the grid of the 12AX7 will start distorting stuff 4 times sooner than 12AT7.

* The 220k are for mixing the two channels. If you are switching instead, the 220k should be shorts.

It may be *both* items. The 12AX7 grid distorts sooner, but it might not matter so much if fed directly from a preceding stage; but the added 220k allows 12AX7 grid distortion to screw-up the main-path signal more.

I am wondering why 12AX7 for what is really a heavier load than an AX should handle. Can you put a 12AT7 in there?

If you insist on one tube-type throughout, try one of the 220k you took out directly at the grid of the 12AX7 reverb driver (between 500p+1Meg and the AX grid) so when it overloads its input goes to 221k not 1k. Without the now-pointless 220k mix-resistors, so the signal comes from the 39k of V1 or V2, an added 221k load is small flavor not fart.

A more complicated option is to change the 1Meg at driver grid to a 1Meg "Drive" pot, wiper to grid. Logic suggests 12AX7 needs 4 times -less- drive than 12AT7, so you turn-down considerably. The 12AX7 grid overloads later, and the top-of-pot resistance isolates this distortion from the main path. This might not be a permanent panel control-- you might find that setting it at 30% from the bottom works great, and replace it with a 680k:330k divider.

[glass54]

This may be a silly question  .
Is the switch indicated an actual switch or relay and is there some length of wire, shielded cable, etc. (maybe a photo?)
Regards
Mirek

[Leevi]

* 12AT7 will bias-up at 8V; 12AX7 nearer 2V. So the grid of the 12AX7 will start distorting stuff 4 times sooner than 12AT7.   

That does make sense but how the distortion goes to the signal way if the reverb is off? Via the the plate?

* The 220k are for mixing the two channels. If you are switching instead, the 220k should be shorts.

Sorry the picture was not complete, see the updated picture

I am wondering why 12AX7 for what is really a heavier load than an AX should handle. Can you put a 12AT7 in there?

I can try 12AT7 but parallel wiring is not possible

If you insist on one tube-type throughout, try one of the 220k you took out directly at the grid of the 12AX7 reverb driver (between 500p+1Meg and the AX grid) so when it overloads its input goes to 221k not 1k. Without the now-pointless 220k mix-resistors, so the signal comes from the 39k of V1 or V2, an added 221k load is small flavor not fart.

See the updated picture

A more complicated option is to change the 1Meg at driver grid to a 1Meg "Drive" pot, wiper to grid. Logic suggests 12AX7 needs 4 times -less- drive than 12AT7, so you turn-down considerably. The 12AX7 grid overloads later, and the top-of-pot resistance isolates this distortion from the main path. This might not be a permanent panel control-- you might find that setting it at 30% from the bottom works great, and replace it with a 680k:330k divider.

Thanks, I'll try that

Is the switch indicated an actual switch or relay and is there some length of wire, shielded cable, etc. (maybe a photo?)

There are two relays one for channel switching and one for mixing the channels, cable from relay is shielded.


AUDIO CLIP:


http://rikstone.fi/DistortionReverb.mp3

/Leevi

[sluckey]

I'm guessing pins 1,2,3 are the driver and pins 6,7,8 are the recovery? If so, put a 12DW7 in the socket to see if that cures the problem.

[Leevi]

I'm guessing pins 1,2,3 are the driver and pins 6,7,8 are the recovery? If so, put a 12DW7 in the socket to see if that cures the problem.

Good idea but unfortunately the wiring is in different order.


I tried the voltage divider which solves the problem. I used 560K/330K which is enough to kill the distortion and keeps the reverb still on acceptable level.


I also have added Sluckey's tremolator after the reverb with 50K RA-pot, which works excellent there.


Thanks
/Leevi

[sluckey]

Good idea but unfortunately the wiring is in different order.

In that case, the ECC823 is a mirror image of the 12DW7.

     https://www.tubesandmore.com/products/vacuum-tube-ecc823-jj-electronics-dual-triode

A few years back another member experienced the same weird symptoms you are having. I think his solution was changing driver tube type, but I'm not certain. The solution made no sense to me so I just accepted it and soon forgot the details. You can probably find the thread if you are willing to search long and hard enough. Sorry I can't give you more clues. Maybe some of the other members remember more.

[PRR]

> how the distortion goes to the signal way if the reverb is off?

Off where? The recovery stage? Then you are still driving the driver stage grid?

Grid-cathode is a *diode*. If you pull it positive, it conducts well; negative no-conducts. When conducting it is about like a 1k resistor; no-conduct is >100Meg in the grid (1Meg with the resistor).

You have equvalent 150k-300k coming from your preamps and switching.

A diode on a 100+k load will clip the positive tips of loud signals.

Here it is more complicated. There is a capcitor in series with the diode so more effect on higher frequencies. The diode is standing on a nominal 2V made by a cathode bias network with R and C. First positive peaks charge-down the 500pFd cap and it tends to stop clipping. But this also turns off the tube so the 2V goes away, at a rate determined by the 22u cap. Clipping tends to stop for a while. We could study this for steady tone. For guitar transients the attack and decay times of guitar and the attack and decay times of the two caps may interact in farty ways.

[PRR]

Here is an Idiot Assistant's opinion. This 12AX7 model models grid current but not real exactly-- shows trend not exact values.

The Grid (green) line is clearly clipping. I was wrong about the cathode: for these values at this frequency the cathode voltage rises.

The "main signal" voltage is "clean enough for guitar", for this condition. If you zoom in there is a significant glitch where the reverb drive grid starts to conduct, but I have seen far worse in amplifiers driven hard. Maybe it is worse at lower frequency. (I won't putz with it all day, my shoulder hurts.)

In any case, a divider (470k:470k or 680k:330k) in place of the 1Meg will both cut-down the excess drive to grid and reduce the loading when it does clip.

[Leevi]

Thank you PRR for those lessons. I have not even thought the issue from that point of view.
Maybe thar's why Fender uses parallel 12AT7 in their amps?

Off where? The recovery stage? Then you are still driving the driver stage grid?

The recovery stage

In that case, the ECC823 is a mirror image of the 12DW7.

Thanks for the info Sluckey

/Leevi

[pdf64]

A simple way to get rid of reverb driver grid clipping contaminating the dry signal path, without reducing the reverb drive, is to use a grid stopper on the reverb drive tube, eg 10k-47k though anything is better than nothing.

[Leevi]

A simple way to get rid of reverb driver grid clipping contaminating the dry signal path, without reducing the reverb drive, is to use a grid stopper on the reverb drive tube, eg 10k-47k though anything is better than nothing.

The voltage diver 560K/330K I installed works as a grid stopper or do you mean that I should replace the 560K with 10-47K?


Update:


Tested with 22K, is not enough to kill the distortion.

/Leevi

[pdf64]

A simple way to get rid of reverb driver grid clipping contaminating the dry signal path, without reducing the reverb drive, is to use a grid stopper on the reverb drive tube, eg 10k-47k though anything is better than nothing.

The voltage diver 560K/330K I installed works as a grid stopper or do you mean that I should replace the 560K with 10-47K?


Update:


Tested with 22K, is not enough to kill the distortion.

/Leevi

A grid stopper is a resistor that is (ideally) attached directly and immediately adjacent to the grid terminal, eg http://3.bp.blogspot.com/-mf1Em7fe8l0/Tzw9iAliXQI/AAAAAAAAAWw/5dLBv8GarAw/s400/Grid+Stoppers.jpg
My idea is to use the regular 1M grid leak resistor and the 500pF cap connection to it, but then insert a grid stopper between that node and the grid itself.
22k probably didn't work well for you because the source impedance feeding the circuit was much higher than with a regular Fender, due to the use of 220k mixer resistors after the 2nd stage plate. And the much lower clipping threshold of the 12AX7 won't help.
Another option that may help is to remove the cathode bypass from the reverb driver; that will maybe double the clipping threshold and may also make it smoother (hence less noticeable).
Sorry, I didn't have a good look at your particular circuit variation before making the previous suggestion.

[Leevi]

Thank you for the tips. The circuit I'm using is a bit challenging having Fender and Marshall type preamps and different power amp options and lot of relay controled features.


/Leevi