Valve amp volume control implementation.

[Diabolical Artificer]

Attached is a circuit for a common cathode triode gain stage, which is pretty bog standard in most valve amplifiers, apart from the volume control that is. The problem is with the volume pot set at low/zero, , the grid resistor is "nulled" out by the bottom half of the pot set to around zero ohms,with the control grid sitting pretty much at ground, which AFAIK, is bad practice.

I've seen some guitar valve amps with this sort of setup, apart from the addition of a grid leak resistor or two, but better amp's and some pre-amps use one triode as a fixed low gain
"buffer" prior to the volume control.

How would one implement a volume control without the use of another triode "buffer"? Is there a configuration/circuit that allows the use of a pot before the gain stage but doesn't have the problem as above?

Cheers, TFL, Andy,

[ppppenguin]

Unless the valve is drawing grid current the resistance from grid to 0V is pretty much irrelevant. Whether it's zero or 100K won't make much difference. The grid will be at 0V as intended in any such amplifier stage. The bias is set by the cathode resistor which ensures that the cathode is +ve wrt the grid. In other words the grid is -ve wrt the cathode which is what you want.

If you really don't want the resistance of the pst in parallel with the grid resistor then just AC couple from the slider to grid.

[Murphyv310]

I've added a cathode follower on some of my amp projects and used high quality pots of 20k between cathode and chassis, the slider AC coupled to the following grid with a low value of grid resistor of 100k the coupling cap was .22uf. It works well with no pick up but can cause some DC shift on the next stages when adjusting the volume due to charging and discharging the coupling cap.
I've also tried DC coupling stages, a little like the 3-3 but with an ultra linear PP output stage, it works well enough but any poor components, valves or supply that is not stable gives DC shifts, seen by the speaker cone being less than stable with no signal, adjusting controls can also cause problems, when it right the sound quality is outstanding though.

[Diabolical Artificer]

Now you mention it Jeffrey Vgk is set by the cathode resistor, therefore it isn't really a problem.

That's an idea Trevor, I'll look into that.

A bloke mentions the "issue" here - https://www.youtube.com/watch?v=fD83Oi6gCUs at about 33mins. I'd read it elsewhere, but thinking about it isn't really an issue, so now I'm uncertain as to what he's on about.

Andy.

[ppppenguin]

In theory the grid leak resistor is not needed if the wiper of the pot directly feeds the grid. In practice it's worth having so that if there's a problem with the pot then the valve's grid won't be left floating. If you have a pot whose wiper is making intermittent contact the resulting noises won't be pleasant. Just make the grid leak resistor at least 10x the value of the pot.

Actually you can use this arrangement to simulate a log pot with a linear one. Log pots are generally pretty crude approximations. If you use a linear pot and the grid leak is smaller than the pot value you can get a good approximation to log. You can do the sums but somewhere between a half and 20% of the pot value is about right. you can just try it and see if the control action feels nice. The downside is that the loading on the previous stage will vary with volume control setting. Whether this matters depends on the design of that stage.

I don't like Trevor's idea of putting a pot between cathode and ground. DC flowing in a pot's track can often make the action sound rather scratchy. Also, as Trevor points out, if you turn the pot rapidly you're putting a low frequency thump into the next stage as the coupling cap charges to the new voltage.

[Murphyv310]

Hi.
Yes I agree with Jeffrey.
On one amp I was playing about with a few years back I used a linear pot on the cathode of the follower, at the time I searched for a quality pot that would handle the little bit of cathode current and not sound scratchy, a hard task indeed and I was just about to give up, I then saw some pots on a Ukrainian site on E-bay and ordered a few of them. These were the biz, no discernable noise when adjusting the pot.
If you notice the value of the grid leak at 100k and the pot at 20k, the circuit can and does work with a much lower grid resistor and the effects of charging reduces but of course the bass response falls off too so although the circuit does work it can be a compromise, DC coupling is by far the best as phase shifts due to cap coupling are obviated but to get a good design to work all the way through a valve amp is challenging to say the least.

[Diabolical Artificer]

Thanks both. A KF as a buffer sounds a good idea, but a low gain stage with local FB might be better, see this circuit - http://www.angelfire.com/electronic/funwithtubes/Amp- Gain_Block.html

I'll have a play with both your idea's at a later date.

Andy.

[Jez]

I use the log faking method in the diy valve pre I'm listening to right now. Works very well and I've got it so although the pre has 10x gain, volume is nicely adjustable even with my Leak St20 with 100mV sensitivity. As it does have the x10 gain then if you happen to have a power amp of 3V sensitivity then just keep turning the volume up and you'll get there.

[Kalee20]

I can't see any problem with the circuit in the opening post of this thread! Why is it bad practice?

[Mike Watterson]

The grid resistor is so the grid is still at 0V if the pot momentarily goes open circuit as it's moved, or to change the approximate log law (depending on value). The resistance at the grid can be 0 ohms to maybe 10 M Ohms with no change to the bias.

There is no problem with the circuit.