Parallel ECC83 gain stage - how to mix the signal

[StingRay85]

Basically I'm going for a single input Super Bass where both channels are blended in the most classic way. Hi normal input with a bridged Lo normal - Hi bright. Got this first part sorted out already. But how to deal with the plates and only use a single vol pot. Basically there are two options.

This seems like the most sensible way to do it:



But when looking at the 2204 schematic, and the gain stage where the preamp volume is positioned, it seems the position of the resistor is reversed in the signal chain. But with preamp volume only slightly on 1-2, the grid leak resistance value would be quite low.



Anyone who can advise me on which option to go for, and provide some more insights on this? Also, is it the right choice to go for a combined 50K anode resistor? Cheers

PS: Obviously you can still argue, why not use a grid leak of 1M instead of 532K, and also a 68K for the 2nd grid stopper instead of 102K, but I'm not too much bothered with this right now

 

[Chris-in-LA]

Yeah, but the classic way uses 2 pots so that the blending can be controlled or mixed to your liking. Why not go with the 2204 schematic so that you can control volume and gain independently?

 

[StingRay85]

That I will also do, but the only adaption needed for that is only putting the MV pot behind the treble pot. All the rest is now identical, unbypassed 820 ohm gain stage DC coupled to the cathode follower. I'm trying to make two amps into one. This single input would be the "low" input in a 2204 style preamp, the hi input will also have two additional gain stages, but with lots of intermediate attenuating resistors until it sounds good

 

[TheKman76]

Scheme 1 will work but will alter the tone as you move the pot. Grid leak is not a concern, it's *no* grid leak which you need to be careful with.

Scheme 2 is the traditional approach with a divider and bright cap and will give you the more consistent tonal variation to the pot.

Parallel triodes are all a bit redundant here though. You don't need the lower output impedance, what's the point?

 

[Pete Farrington]

This seems like the most sensible way to do it:

I don’t see why you’ve put the 470k // 500pF between pot wiper and grid. What’s your intention with that?

the grid leak resistance value would be quite low

Why might that be a problem? It may be helpful to explain your thinking.

This single input would be the "low" input in a 2204 style preamp

Just to note that the low input stage has much lower than normal gain, due to its 10k unbypassed cathode.

 

[StingRay85]

Scheme 1 will work but will alter the tone as you move the pot. Grid leak is not a concern, it's *no* grid leak which you need to be careful with.

Scheme 2 is the traditional approach with a divider and bright cap and will give you the more consistent tonal variation to the pot.

Parallel triodes are all a bit redundant here though. You don't need the lower output impedance, what's the point?

Well, in the Super Bass schematic, the 470K resistors are there and used to mix the signal of the two channels. So when trying to work around having to use a single vol pot, but preserve the idea of a two channels amps, the resistor is unlikely to be obsolete, so the original question was where to put it then, before or after the pot. But you basically answered the question, I need to use the second schematic.

It was not entirely clear to me what is the real function of the grid leak resistor, and how low/high it really needs to be. But since you mention it is of no concern, then I guess it's alright with me, and I can proceed

 

[StingRay85]

Just to note that the low input stage has much lower than normal gain, due to its 10k unbypassed cathode.

The goal is to use the low input for a clean-to-overdriven bridged Super Bass sound, with a pre-phase inverter MV behind the treble pot. The high input will be a double gain stage that cascades into the low input, just like a 2204. There I will have to experiment with sufficient attenuating resistors, and probably look a bit into the SLO100 values. But I will not use a cold unbypassed gain stage

 

[ELS]

If you go for the top setup, you will reduce blocking distortion a bit without changing the tone.
If you go for the bottom one, you will reduce gain and change the tone because of the cap bypassing the resistor.

You don't need 2 seperate grid stoppers for the 1st stage, just connect them together and use 1/2 the grid stopper value (~33k usually)

 

[TheKman76]

If you go for the top setup, you will reduce blocking distortion a bit without changing the tone.

I should think blocking distortion is extremely unlikely here. In either scheme there is a 470K grid stopper (at least) bypassed by a small cap. Providing there's no direct current path to the coupling cap it's golden.

And you're right, in scheme one the bright cap will do mothing at all. Just to correct myself.

 

[ELS]

I should think blocking distortion is extremely unlikely here. In either scheme there is a 470K grid stopper (at least) bypassed by a small cap. Providing there's no direct current path to the coupling cap it's golden.

And you're right, in scheme one the bright cap will do mothing at all. Just to correct myself.

Wrong, in the top schematic the bright cap will increase top end because the grid stopper resistor is so high that you'll loose high end above like 8khz probably, the grid has capacitance to the plate.
I don't think you know how blocking distortion works... (It wont be more top-end than without the grid stopper. With the cap you wont loose the top end you would with just the grid stopper)

When you sent a ton of signal trough the grid of a tube stage, when the signal goes positive the grid starts to conduct current, which then charges the coupling capacitor negatively on the grid side, which biases the tube stage cooler and creates blocking distortion - when most of your wave is lost because the tube goes into cutoff.
The grid stopper resistor greatly reduces blocking distortion because it limits the current that can be drawn from the grid, that in conjunction with the fact that the volume control will discharge the coupling cap to ground all the time means that you can effectively reduce blocking distortion just by adding a grid stopper, Then add a capacitor across the grid stopper (don't make it too big as you'll ruin the point of the grid stopper in the first place) but ~0.001 works fine, and you don't loose top end, you don't loose gain, but you reduce blocking distortion.

The blocking distortion wont be a problem in these 2 stages but when it gets further down to a 3rd or even 4th stage, it becomes very apparent. Since OP mentioned a 2204 design, with that you easily could run into problems with that
Don't doubt me, I did the exact thing like 5 years ago when making my amp - I wanted the gnarliest amp ever and go it https://www.marshallforum.com/attachments/brassman-2a4-schematic-png.126031/
3 hot gain stages, high value coupling caps, grid leak biased 1st stage, but sounds great. to fix the blocking distortion I just added a 220k grid stopper bypassed with a 1nF cap on the 2nd stage. I intentionally left a bit of blocking distortion by choosing a lower value, a 470k should pretty much remove all blocking distortion!
And since I left the high value coupling caps, it didn't lack bass when playing clean, AT ALL!

 

[TheKman76]

Wrong, in the top schematic the bright cap will increase top end because the grid stopper resistor is so high that you'll loose high end above like 8khz probably, the grid has capacitance to the plate.

I understand all that. In practice a clean signal after just one amplification stage has so little content in this frequency range as to be meaningless. After clipping has begun it's a completely different story, but in this example I'd expect no effect except maybe extra noise.

That said, I'll test my thinking and model the stage in LTSpice.

I don't think you know how blocking distortion works...

Here's my thinking. A 500p cap coupled directly to the grid with a leak path of 1M (at maximum) in either scheme simply isn't big enough to alter the grid bias. The coupling cap may well be big enough to alter bias, however, in either scheme there is a 470k grid stopper, at minimum. For clarity, whether it's inline with the grid directly, or simply between the coupling cap and the grid at another location matters not. In both schemes there is a grid leak path and coupling discharge path.

So, even with the second stage driven to heavy clipping I don't see blocking distortion happening.

to fix the blocking distortion I just added a 220k grid stopper bypassed with a 1nF cap on the 2nd stage. I intentionally left a bit of blocking distortion by choosing a lower value, a 470k should pretty much remove all blocking distortion!

I think you have that backwards. A larger resistor will make it worse. Infinite resistance means no grid leak, short circuit would mean no capacitance.

 

[ELS]

I understand all that. In practice a clean signal after just one amplification stage has so little content in this frequency range as to be meaningless. After clipping has begun it's a completely different story, but in this example I'd expect no effect except maybe extra noise.

That said, I'll test my thinking and model the stage in LTSpice.


Here's my thinking. A 500p cap coupled directly to the grid with a leak path of 1M (at maximum) in either scheme simply isn't big enough to alter the grid bias. The coupling cap may well be big enough to alter bias, however, in either scheme there is a 470k grid stopper, at minimum. For clarity, whether it's inline with the grid directly, or simply between the coupling cap and the grid at another location matters not. In both schemes there is a grid leak path and coupling discharge path.

So, even with the second stage driven to heavy clipping I don't see blocking distortion happening.


I think you have that backwards. A larger resistor will make it worse. Infinite resistance means no grid leak, short circuit would mean no capacitance.

Then walk your own road genius.

 

[StingRay85]

Merlins preamp book covers blocking distortion quite nicely.

 

[StingRay85]

Interestingly, the Soldano SLO100 simply does both. They have a 470K before and after the preamp volume pot.

 

[TheKman76]

@ELS I stand corrected on the bright cap in scheme 1. This has an effect down as low as 2kHz, would be very noticeable.

 

[ELS]

@ELS I stand corrected on the bright cap in scheme 1. This has an effect down as low as 2kHz, would be very noticeable.

The signal from a guitar is far from a perfect sinewave, it has harmonics even in ultrasonic frequencies. That's all I'm gonna say...

 

[StingRay85]

This is probably what I'll aim for. It's still work in progress though.

 

[Pete Farrington]

I think you have that backwards. A larger resistor will make it worse. Infinite resistance means no grid leak

Clearly a grid stopper of infinite resistance would be a bad idea.
Aiken suggests something in the range 100k - 470k for preamp clipping stages

What Is "Blocking" Distortion?

Guitar amplifiers

aikenamps.com

Look for the problem in the preamp stages. Blocking can (and does) occur just as bad on RC-coupled preamp stages that are overdriven. The solution is similar. Add large value series grid resistors (100K - 470K), reduce coupling capacitor values to the minimum required for the desired low frequency response, add interstage attenuators to limit the amount of grid drive to the next tube, reduce the size of, or eliminate, the cathode bypass capacitors, as they also contribute to the blocking due to the long time constant associated with the recovery from a transient signal, or add a diode clipper bounding circuit to prevent the grid from being driven too far.

 

[TheKman76]

The signal from a guitar is far from a perfect sinewave, it has harmonics even in ultrasonic frequencies.

I agree wholeheartedly. When I've used EQ filters to clean up the input to an amp I can get away with as low as a 5kHz knee with -12dB/oct for clean guitar signals, 8-10kHz for distorted signals. I *was* thinking scheme 1 would have a knee higher than this and proved myself wrong with the LTSpice model. According to the model there would have been -3dB by around 2kHz for a generic 12AX7 without the bright cap, clearly not great.

@StingRay85 This looks like you're using two valve to do the job of one. What's your thinking here? Also, no attenuator before the third stage plus the unusual bright cap arrangement might be an unexpected result.

 

[mickeydg5]

Man this is getting out of hand.