Eliminating channel bleed through in dual channel preamp

[BiBi]

I've been thinking about your analogy and would like to pose a question. Taking the fire hydrant-- if the flow of water at maximum actually overwhelmed the valve, causing water to spray all over and possibly damaging the valve then this would be inefficient. As well, if the pressure from the hydrant were actually low, one would like to adjust the valve to this pressure.
To go back to the original circuit: if the input level is fixed, you are using a relatively high level as your benchmark. This means putting a larger series resistor to create a voltage divider. Problem being when you have a small signal, that series resistor will be a source of noise. The gear related noise on the output should actually be louder as the output level will have to be adjusted higher with lower input signal levels.

[mooreamps]

First of all, I think there's a misunderstanding in this thread. Mooreamps says the cause of channel bleed is that one of the tubes is configured as a mixer. That's true, but it's not mixing the dirty channel and the clean channel. It's mixing the reverb return with the dry signal, all in the context of a single channel. The bleed issue is separate.

You're right... I was looking at the Reverb. As far as channel bleed in the pre-amp, I am not seeing anything too obvious in the print. Perhaps it is issues with the wiring and/or the layout. Shielded wire for all the signal paths as Steve pointed out would be the first thing I would try...

-g

[Wilder Amplification]

I've been thinking about your analogy and would like to pose a question. Taking the fire hydrant-- if the flow of water at maximum actually overwhelmed the valve, causing water to spray all over and possibly damaging the valve then this would be inefficient. As well, if the pressure from the hydrant were actually low, one would like to adjust the valve to this pressure.

It's funny every time I explain about "maximizing clean gain" everyone misses the "clean" part and assumes I'm telling them to turn the input all the way up regardless of input clipping.

This is what I mean by "maximizing CLEAN gain". You don't want it so hot that you clip the input...but you want as much possible clean input signal as you can possibly get without clipping the input.

To go back to the original circuit: if the input level is fixed, you are using a relatively high level as your benchmark. This means putting a larger series resistor to create a voltage divider. Problem being when you have a small signal, that series resistor will be a source of noise. The gear related noise on the output should actually be louder as the output level will have to be adjusted higher with lower input signal levels.

Not sure I follow you on this statement...could you clarify a bit?

[R.G.]

Taking the fire hydrant-- if the flow of water at maximum actually overwhelmed the valve, causing water to spray all over and possibly damaging the valve then this would be inefficient.

It would be inefficient - but it's also impossible. The flow of water can't overwhelm the valve, by the very nature of valves. You need to make the distinction between water pressure (analogous to voltage or electron pressure) and water flow rate (analogous to current or electron flow rate).

A static pressure that's too high may burst a valve. However, it's the pressure from one side of the valve to the other (outside) that bursts the valve. This pressure is greatest when the valve is not open at all. If you open the valve, the water flow (i.e. current) reduces the pressure across the valve by allowing the flow to start equalizing the pressure from the water main side to the atmospheric side. In fact, the pressure at the output opening of the valve remains somewhat high. If it were the same as atmospheric pressure, the water could not move through the air. The pressure from one side of the valve to the other *drops* when you open the valve and increase the flow. The valve's resistance is by definition the pressure across the valve divided by the flow rate through it. A closed valve has a zero flow rate, and the fluid resistance is literally infinite. Opening the valve lets some flow happen, and the valve fluid resistance becomes finite, and the pressure across the valve drops.

Valves do not burst from too much flow. In fact, it's impossible to damage a valve by too much flow. It's always the pressure across the valve that breaks it. In this way, valves are different from electrical resistors, as the fluid flowing through the valve very efficiently carries away the heating from frictional losses in the valve, in a way that electrons don't do.

So what you propose is actually impossible.

As well, if the pressure from the hydrant were actually low, one would like to adjust the valve to this pressure.

Actually, no, this does not follow. If the pressure from the water main behind the valve were low, the firemen would again open it all the way. Adjusting the valve to less than fully open does nothing but cause a pressure drop proportional to the flow. The valve position adjusts the valve's fluid resistance. Closing the valve down because of low water pressure behind it would lower the pressure at the end of the hose even more as the flowing water lost pressure going through the partly closed valve. This would make the nozzle situation even worse, and would not happen.

I also cannot think of a situation in the electrical analogy where you'd close a resistor/valve down because the voltage was too low for what you needed.

To go back to the original circuit: if the input level is fixed, you are using a relatively high level as your benchmark. This means putting a larger series resistor to create a voltage divider. Problem being when you have a small signal, that series resistor will be a source of noise.

A series resistor is *always* a source of noise, proportional to the resistance and absolute temperature. The input voltage changes this not at all.

And no, I don't think that's what he had in mind.

The gear related noise on the output should actually be louder as the output level will have to be adjusted higher with lower input signal levels.

I think what you're saying is that a user would increase the amp *gain* to make up for a lower signal. That's true as far as it goes, but one would never intentionally reduce input signal with a divider if noise were an issue.

The bottom line on low noise design is to
(a) know your signal's level, bandwidth, and impedance
(b) get as much of the necessary gain in the input/first stage as you can, with proper attention to low noise components, restricting the bandwidth, and suiting the input biasing and impedance levels to the signal source impedance.

The first stage of any amplifier is where substantially all of the whole device's noise performance is determined. Noise from the first stage is amplified by all the rest of the gain in the amp, so the biggest *clean* and *noise free* gain possible from the first stage makes the noise performance of the rest of the amp the best possible.

Padding down an input is only acceptable when the signal is so big that noise is not an issue. This is why guitar amps never have the volume pot as the first component. It is always placed after the first gain stage so the signal is amplified as much above the noise as possible.

[BiBi]

Ok the firehose analogy wasn't mine so maybe I bogged it up! But take this situation: when a hose has low pressure, you can squeeze the hose to increase the velocity of the output. This is what I was referring to with the valve. As well, try whistling with an open mouth!

In regards to the gain structure of the original schematic-- the first gain stage is followed by a pot to set the level into the effect. I'm still uncertain why this isn't the most efficient method as it seems to follow the guidelines you've posted on low noise design.
Yes, the series resistor is always a source of noise, but if your signal isn't large enough to overwhelm it then it becomes a problem, right?

I'm happy to get the input from everyone, my hope is to get a better preamp out of this and of course, a little better understanding!

[cbarrow7625]

I think if you follow Wilder's advice you will use the pot to set the level into the effect to the maximum level possible without clipping the input of the effect. For the best noise performance you would use that pot as a level trim pot (set and forget) rather than a "dwell" control. It would also work as a dwell control if you turn it down from maximum clean input but the trade off would mean paying a penalty of increased noise.

[BiBi]

Ah! Now I can see what you all are getting at! I never even considered using the pot prior to the effect as anything other than setting the maximum clean volume. I assumed that is what was meant by dwell-- but after a little searching I see it's used as a depth control. I have a bit of egg on my face on this one then.
Problem is I always looked at like a mixing desk where setting the gain is of the utmost importance while faders are used for mixing. I could always get by with the faders set at 0dB but the difference in signal levels in say a voice and kick drum necessitates a gain control.
Thanks for that cbarrow!

[cbarrow7625]

Right. In reverb lingo, "dwell" typically equates to the length of reverberation time. With a spring reverb, if you feed it less signal at the input the length of time it can sustain reverberation is proportionally reduced. So a long reverb tank can be made to sound (somewhat) like a shorter one. Dick Dale could get days of reverb out of an old Fender reverb unit by turning up the dwell control but the same unit could be used for subtle "pretty" reverb as well if the dwell control was turned down for a shorter reverb time.