Grid Leak Resistor Sizing

[simon]

I'm sure this has been discussed before but I've searched to no avail. MJ doesn't seem to cover the considerations when sizing a grid leak resistor (unless it's hidden away somewhere) but what should we consider when choosing a grid leak resistor?

For simplicity let's assume self biased DHTs. The WE 300B datasheet specifies a maximum of 250k in self bias (small signal IDHTs often specify a max. of 1M). But the 26 and 10 datasheets don't mention any maximum.

Now the combination of the grid leak resistor and coupling cap set the bass rolloff, but I seem to recall there's more to it than this and that biggest isn't always best. Is it as simple as limiting how much current might flow through it, or is there more to it?

[Paul Barker]

every part which the grid leak relates to has a different demand of it.

the former valves asks it to be as large impedance as possible, the valve it's connected to wants it as small dc resistance as possible so those two valves are pulling in opposite directions.

As you rightly say the coupling cap it interacts with pairs up to gang up on your bandwidth compromise, and the cap introduces blocking distortion and possibilities of dc voltage leakage before you begin to consider sound of the capacitor itself.

IT's and grid chokes relieve one part of the compromise grid leak resistors present. That is they show a low dc resistance to the grid, and if correctly chosen they offer good ac impedance (and anode load in case of IT) to the former valve.

Direct coupling is a great solution to all the above whilst introducing it's own problems.

LCL or CCS CL or RCL coupling are solutions.

sometimes the grid choke is bearly an improvement and possibly at times when I have AB compared I may have only plumeted for what my psycho acoustic hearing preferred. But if I am not direct coupling (which is as rare as hens growing teeth) I use a grid choke.

[IslandPink]

I've never had too much trouble with using a pretty low time constant like 2Hz , so hence something like 0.47uF/220k . People say the bigger caps sound sluggish, or there is some problem charging them up/down, which seems a bit odd.
I think the main thing is that bigger caps like 0.47u or 1u are physically big and need a decent driver current . I just don't think you can drive big caps with a weedy driver valve ; but then again you can't drive decent-sized output valves with a weedy driver, either.

If you go for the low ( < 3Hz ) time constant then you get extended low bass and good timing at the bottom end. That's the requirement . Once you go above 5Hz, it starts to sound bass-light and a bit slow , to my ears .

I've used 0.47u/220k on the 300B amp for years . I had 0.1uF and 0.22uF , but the bass got better with the bigger cap . On the other hand I'm always using at least 17mA in the driver .
I wouldn't go to the upper limit of the grid-leak value on most valves though - 220k is pushing it a bit on a 300B , might be better with 150k or 100k to be honest . Respect the data sheet !

I asume you know how to calc the time constant ?
Hz = 1/( 2xPIxRxC ) ( ps 0.47u = 0.47x10e-6 )

[simon]

Yes, thanks chaps. I'm reworking my way through the design of 26-10Y-300B(XLS), hence the query over 26 and 10(Y). The 10Y is running at 20mA but it's IT coupled to the 300B, so I could just use 100k or 150k here.

The 26 is running at 5.5mA via 300H anode choke and a AN 0.47uF copper coupling cap to the 10Y. So as you say Mark I was planning on using something like 200k grid leak with the 10Y. I do wonder about using a physically large coupling cap here too, though I've nothing to back up my concern.

Which leaves the 26. It's the first valve so 100k might be appropriate, if it's big enough?

[IslandPink]

The 10Y is running at 20mA but it's IT coupled to the 300B, so I could just use 100k or 150k here.

Not sure what the 100k or 150k does here - if you are IT-coupled, isn't the 300B grid coupled to ground via the low impedance of the transformer secondary ?

Which leaves the 26. It's the first valve so 100k might be appropriate, if it's big enough?

Sounds fine. I'm sure 220k would be fine too. It's a relatively small low-current valve.

[simon]

Not sure to be honest Mark. I have thought about it but I've noticed other's circuits that have them so was swayed. But maybe I'm just perpetuating another internet myth?

[cressy]

On my kt88 im using a 2.2uf coupling cap and a 220k grid leak. I never bothered calculating the correct cap, it was what I had in the box of bits. Tbh ive almost always used a .47 and 220k. I might have some .47 russian pio caps kicking about I could sub in to see what the effect is

[Nick]

Not sure to be honest Mark. I have thought about it but I've noticed other's circuits that have them so was swayed. But maybe I'm just perpetuating another internet myth?

It does provide a defined load for the transformer and hence for the valve that is driving it, maybe better than letting the transformer do its thing.

[simon]

That makes sense Nick. So if I use, say, a 150k grid leak the 10Y will see 150k as the IT is 1:1. Which should be good for the 10Y and the 300B.

[simon]

Ant, 220k and 2.2uF gives 0.32Hz, 220k and 0.47uF gives 1.54Hz. I suspect you won't hear any difference in bass, but it might be better for using a smaller cap. Or not.

[Paul Barker]

That makes sense Nick. So if I use, say, a 150k grid leak the 10Y will see 150k as the IT is 1:1. Which should be good for the 10Y and the 300B.

It depends what your transformer requires to make the square wave look better, transformer specific, either already worked out for you or you work it out. then you have to make sure it doesn't just sound better without a resistor load at all.

[simon]

Okay, thanks Paul. The ITs are 11k 1:1 specials from Bud Purvine. So what I should do is wire the amp up, then input a square wave (1kHz ?) Then scope the grid of the following 300B and AOT the 300B grid resistor for best square wave?

Provided the value isn't too big or too small for the 300B job done?

[Paul Barker]

It can't be too big because the 300b is only conscerned with dc resistance and the limiting factor of that is buds secondary dc resistance which may be 200 ohm or so. Bud himself may have a suggestion on the matter ref the damping.

Bud's IT's may not suffer from any peaks. If SW looks square, don't use a resistor is my hunch.

[Nick]

It depends what your transformer requires to make the square wave look better, transformer specific, either already worked out for you or you work it out. then you have to make sure it doesn't just sound better without a resistor load at all.

There is more to it that that. By choosing a actual slope for the load line the driving valve is seeing you have control over the distortion that you may want to cancel with the driven valve. If you leave it up to the transformer, it will vary with frequency from a flat line to something less flat.

[simon]

Mmm, good food for thought, thanks.