Dumble SSS/ODS issue ! Any Advice ?

[Taitoa]

Hi everybody,

I currently doing preamplifier tests before building the power amp and I meet some issues with the dumble channel.

The pre-amplifier have severals channels with different mod on them, each of them turn on/off some relays and everything is controlled by an μC.

So, each mods of the channel 1 are a dumble amp :
Mod A = ODS
Mod B = SSS
Mod C = TwoRock JM

Tubes : EHX 7025

I used schematic files available on this forum for this channel design. For my tests I pick the signal out of the loop and go to my audio interface.

This is for the fast presentation.

Now, I meet some issue with the preamplifier sound and use :

1st case : If I plug my guitar (with single coils) direct into the amp, there is some kind of 'over transient" on strongs attacks. It's hard to explain but output level on theses transient is insane comparing to regular signal. This tend to disappear if I reduce my pick-up volume on the guitar or if I use a pedal as buffer.

2nd case : Guitar still plug into the amp, my "dumble" channel tend to saturate fast (really fast) when I turn the gain up, maybe this is normal but if I go over 2 o'clock, preamp start to oscillate at something like 2kHz. If I unplug the guitar, the oscillation is constant regardless of the gain. If I use a buffer pedal, the oscillation tend to disappear or need an higher gain to came (almost 100%).

I've got to say that I have this oscillation problem on every channels. For exemple, on the distortion channel (mesa boogie rectifier) the sound is a bit fuzzy/farty and unpleasant, its looks like I've got too much gain on the 1st or 2nd preamp stage. I don't have a LFG/oscilloscope and I can't measure voltage gain of each stage. But all my dc voltage, component value and routing seems to be ok.

Has anyone ever met this kind of issues with tube amp ? If you have any leads it could be really useful for me.

One last thing a little bit out of topics, I would decrease the famous fizzy sound of my distortion channel, like a voodoo moded rectifier, what parts of the schematic should I focus to make some test ? Earlier or late stages ? Any leads ?

Thanks you in advance

[erwin_ve]

Hi everybody,

I currently doing preamplifier tests before building the power amp and I meet some issues with the dumble channel.

The pre-amplifier have severals channels with different mod on them, each of them turn on/off some relays and everything is controlled by an μC.

So, each mods of the channel 1 are a dumble amp :
Mod A = ODS
Mod B = SSS
Mod C = TwoRock JM

Tubes : EHX 7025

I used schematic files available on this forum for this channel design. For my tests I pick the signal out of the loop and go to my audio interface.

This is for the fast presentation.

Now, I meet some issue with the preamplifier sound and use :

1st case : If I plug my guitar (with single coils) direct into the amp, there is some kind of 'over transient" on strongs attacks. It's hard to explain but output level on theses transient is insane comparing to regular signal. This tend to disappear if I reduce my pick-up volume on the guitar or if I use a pedal as buffer.

2nd case : Guitar still plug into the amp, my "dumble" channel tend to saturate fast (really fast) when I turn the gain up, maybe this is normal but if I go over 2 o'clock, preamp start to oscillate at something like 2kHz. If I unplug the guitar, the oscillation is constant regardless of the gain. If I use a buffer pedal, the oscillation tend to disappear or need an higher gain to came (almost 100%).

I've got to say that I have this oscillation problem on every channels. For exemple, on the distortion channel (mesa boogie rectifier) the sound is a bit fuzzy/farty and unpleasant, its looks like I've got too much gain on the 1st or 2nd preamp stage. I don't have a LFG/oscilloscope and I can't measure voltage gain of each stage. But all my dc voltage, component value and routing seems to be ok.

Has anyone ever met this kind of issues with tube amp ? If you have any leads it could be really useful for me.

One last thing a little bit out of topics, I would decrease the famous fizzy sound of my distortion channel, like a voodoo moded rectifier, what parts of the schematic should I focus to make some test ? Earlier or late stages ? Any leads ?

Thanks you in advance

Hi Taitoa,

Welcome to the forum!
That's quite a undertaking you have there.
I don't know for other members here but I find it difficult/impossible to give a appropriate answer to your questions without a full schematic on your build and detailed pics.

Erwin

[Taitoa]

Thanks erwin_ve,

The full schematic of the current build would be complicated to understand because of the microprocessor, i had design it for me and there is probably a lot of missing things about the switching matrix. I will make something more understable and post it here with pics an audio.

[Richard1001]

Is this 'microprocessor' just to control the relay switching by midi? There is a good chance this causes some of your problems due to emi and interference (wiring/ground) injecting high frequency switching noise of the processor in your audio circuit (harsh sound) and detection of the clock frequency (whistle)
Did you buy an of the shelf kit like the 8 channel dc switcher from 64pix?

[Taitoa]

Is this 'microprocessor' just to control the relay switching by midi? There is a good chance this causes some of your problems due to emi and interference (wiring/ground) injecting high frequency switching noise of the processor in your audio circuit (harsh sound) and detection of the clock frequency (whistle)
Did you buy an of the shelf kit like the 8 channel dc switcher from 64pix?

Thanks for your response Richard1001,

The microprocessor is a teensy, outputs control mosfets and theses mosfets control relays switching and led's. There is no direct signal from the uC through the audio circuit but only a ground or a clean dc voltage (for states 0(OFF) or 1(ON)). Moreover the uC is not physically near of the audio circuit and have is own power line/ground circuit. For me, there is no reason my problems come from this uC, I didn't have switching noise or other interference. I didn't use DC/DC converter or other switching power supply and I didn't use the 64pix kit.

I would like to add that my 3rd clean channel almost work, it based on mesa boogie mark I and it sound great, there is just too much gain too fast (like any channels).

Thanks erwin_ve,

The full schematic of the current build would be complicated to understand because of the microprocessor, i had design it for me and there is probably a lot of missing things about the switching matrix. I will make something more understable and post it here with pics an audio.

Here is the schematic for one of the dumble as used in the amp :

CH1-ModC.pdf

Please not that I remove all relay/switching, and keep only one configuration. I think it will be more understandable. the schematic goes from the input to the fx send output (which corresponds of signal chain I use for tests).

Here is the sound i'v got at low gain (the one with the strange attack) :

CH1 Issue.wav

And here the oscillation at higher gain :

CH1 Osc Issue.wav

[erwin_ve]

1. Schematic: What are those zeners doing in front of the cathode follower?

2. CH1 issue: sounds like a bad (solder) connection or a bad tube.

3. CH1 Osc issue: 6 kHz parasitic oscillation. I think you got positive feedback somewhere in the circuit, the proximity between components/wires/pcb traces are causing signal induction ; if your layout is identical for the remaining channels this could explain why it is happening on all channels.
This is just a wild guess; pics couyld help.

[luix]

Can you check the circuit voltages? The first sample sounds like a misbiased tube (or a broken one).
It could also be a blocking resistor value mistake (the resistors that are in series with the grid).

With tubes hi-gain design is really difficult to not get oscillation if all catode of the first tube are bypassed (triode max gain) you could see that in Marshall and Soldano, the first tube has one katode bypassed and the other not bypassed, the same for V2... I've learned this at my expence when I tried to replicate the Soldano lead channel on a board and I had V1 katodes both bypassed, the solution was:
1) adding some capacitors of small value between anode and grid or between anode and katode and also in parallel with anode resisstor.
2) a grounded metal screen between V1 and V2.

Oscillation also depends on layout and wiring but start trying adding some AG or AK capacitors.

[Taitoa]

1. Schematic: What are those zeners doing in front of the cathode follower?

2. CH1 issue: sounds like a bad (solder) connection or a bad tube.

3. CH1 Osc issue: 6 kHz parasitic oscillation. I think you got positive feedback somewhere in the circuit, the proximity between components/wires/pcb traces are causing signal induction ; if your layout is identical for the remaining channels this could explain why it is happening on all channels.
This is just a wild guess; pics couyld help.

1. These are clipping diodes for the lead channel (like mesa boogie amp). (I will try without, it could be a good point about why my clean channels saturate so fast).

2. Yes it's sounds exactly like a bad tube, but the tube works and I don't get this trouble with another channel. (I try différents tube). However a bad solder is not impossible.

3. Good point, V1 is the input tube for every channels and the only one which is present on all of them. V1B looks like the perfect suspect (maybe because of the wiring). I would add that the oscillation frequency move with the treble pot position. And why oscillation disappear when I use a pedal in front of the amp ?
Here's a picture of the inside :

IMG_2860.jpg

Can you check the circuit voltages? The first sample sounds like a misbiased tube (or a broken one).
It could also be a blocking resistor value mistake (the resistors that are in series with the grid).

I check voltages each time I power the amp, voltages are all ok and according (more or less) with calculs/simulations. I don't think a grid stopper could cause this kind of results ?

With tubes hi-gain design is really difficult to not get oscillation if all catode of the first tube are bypassed (triode max gain) you could see that in Marshall and Soldano, the first tube has one katode bypassed and the other not bypassed, the same for V2... I've learned this at my expence when I tried to replicate the Soldano lead channel on a board and I had V1 katodes both bypassed, the solution was:
1) adding some capacitors of small value between anode and grid or between anode and katode and also in parallel with anode resisstor.
2) a grounded metal screen between V1 and V2.

Oscillation also depends on layout and wiring but start trying adding some AG or AK capacitors.

1 - AG and AK will works as low pass filter in high frequency that's right ? On the high gain channels the 2nd stage already have a small caps in parallel with the anode resistor (which work as lowpass too) like this :

Screenshot 2021-01-10 at 19.33.10.png

However it's a really good idea to try LPF on each stages and see if the oscillation come from one or more stages.

2- Each preamp tube have a grounded metal screen socket

Thanks for your reply guys, this really allows me to take a step back.

[Goldguitarguy]

You might be getting your oscillation from your lead dress. Chop stick the wires while the amp oscillates and listen for changes.

[erwin_ve]

Thats a hefty build!
Impressive that you could pull it off to put it in one chassis.
But frankly that wiring is a complete mess. Lots of wires, from different tube stages, touching/ running parallel is a receipe for oscillations.
No shielded wire?
If I had to start, look at pictures of a real Dumble. Look how carefully wires are apart from each other and which one touch each other.
Tonestack wires from the same stage can touch each other. Wires for volume pots from different stages touching each other is a big nono.

Zener clippers in front of the cathode follower. I know this works well for some high gain amps like a Marshall, not for Dumble IMO. Im certain you have a lot more headroom and dynamics if you remove them.

Erwin

[Taitoa]

You might be getting your oscillation from your lead dress. Chop stick the wires while the amp oscillates and listen for changes.

It's a great idea you had here ! I did it and oscillation frequency change and tend to disappear in a certain position for V1 wiring. Not sure, but V1 anodes wires seems doing that mess. I will try with shielded wires.

Thats a hefty build!
Impressive that you could pull it off to put it in one chassis.
But frankly that wiring is a complete mess. Lots of wires, from different tube stages, touching/ running parallel is a receipe for oscillations.
No shielded wire?
If I had to start, look at pictures of a real Dumble. Look how carefully wires are apart from each other and which one touch each other.
Tonestack wires from the same stage can touch each other. Wires for volume pots from different stages touching each other is a big nono.

Zener clippers in front of the cathode follower. I know this works well for some high gain amps like a Marshall, not for Dumble IMO. Im certain you have a lot more headroom and dynamics if you remove them.

Erwin

Yes wiring is really ugly, but it's a kind of prototype and final wiring are not done yet.
Only the input jack wire and the distortion channel gain are shielded, but as I say just before I will shielded V1 wiring.

I had seen some dumble pictures, wiring is so nice and well organised. Honestly it's would be a way more difficult to optimise wiring like this in my amp because of density, but I'll took in consideration for the next version.

I will make a bunch of test tomorrow and try with and without clipping diodes.

[Richard1001]

A buffer or pedal will lower the inpedance of the input signal making the input signal wiring inside the amp less sensitive for capacitve coupling.

I think gridblocking may be an issue on your clean channel. This happens when the input impedance of a following stage is to high and the input capacitor to big.
Suspect is your buffer i think. It has got a very high input impedance and the grid has no reference to the cathode. Your signal has to come in through a 47nF cap. In order for any cap to pass ac, is has to be able to load an unload fast enough. When the cap has no time to unload( RC time to high), it can not pass ac. You also don't need such high input impedance at this point.