Request for design guidance with Fixed Bias Output Tube

[RhythMick]

Hi all,

I'm after having a play with fixed bias, partly to satisfy my own curiosity about the sound and partly to learn some more about this wonderful hobby. I may or may not like it, the object is to be able to try it and let my ears be the judge.

As a starter for ten I have the attached rough sketch and some questions. I'd be really grateful for any insights and inputs from the learned audience. Essentially cathode at 0v and grid bias adjusted within an allowable range as I see fit.

• (not shown) signal through coupling cap from driver stage direct to the grid joining the bias
• -ve rail which I plan to regulate at -100v.
• 10mA across the 5K 2W pot should give me adjustment from -50v to -100v for the 300B grid - 10mA is maybe higher than others would opt for, but the regulator will have a minimum current level and I figure this will draw enough to satisfy that
• 1M failsafe to -ve rail is intended to shut the valve down if the pot fails open wiper - any comments ?
• C1 - I've seen this in other designs but what is it's purpose ? Wouldn't C1 attenuate the signal, which would see a capacitative reactance to ground ?
• I assume that to calculate the value of C1 we take the parallel resistance through the wiper to 0v and to -100v (about 1875R at the midway point, 2k5||7k5) and use that in 1/2pi*fR, aiming for a 10Hz corner ? About 10uF ?
• bias rail must be up before valve B+ so I'm thinking of options for ensuring that - bias supply will use bridge rectifier while valve B+ uses tube rectifier - optocoupler on bias rail, engaging relay on the B+ rectifier heaters maybe? that would also cut B+ if the bias supply fails
• I inherently don't like the fuse/shutdown resistor arrangement in the cathode for overcurrent protection, fearful of the sound of that fuse right there in the cathode - what do others recommend ?

All comments welcome and thanks in advance.

[Nick]

-ve rail which I plan to regulate at -100v.

If the B+ is regulated then fine, if not I would go with filtered not regulated. That way the bias and the B= will track the incoming mains supply voltage and give more stable operating point. Also, if its a regulated supply, why not use a adjustable regulated supply ti set the bias voltage instead of the potential div.

10mA is maybe higher than others would opt for

Much more, you will end up with noisy bias adjust pots. I would restrict the current to at most 1ma, then use more resistors eitehr side of the pot to give you a smaller adjustment range but finer control and less damage if the pot fails.

I've seen this in other designs but what is it's purpose

Well, maybe a bit of smoothing, and will give something for the valve to charge up if it goes in to A2. As you say it will remove any signal. If its used, you need a bigish resistor between it and the grid, signal injection point.

bias rail must be up before valve B+

Well, up before the valve in conducting anyway.

optocoupler on bias rail, engaging relay on the B+ rectifier heaters maybe? that would also cut B+ if the bias supply fails

By overcomplicating things you are creating more things that could fail. I would just use a fuse in the transformer secondary. Valves are sturdy things, they can take a lot of misuse. IMHO, the protection circuit will fail and do more damage than what its trying to protect.

what do others recommend

See above. Or at least put a cap across it.

[izzy wizzy]

There is a bit missing off your circuit; the feed to the grid so either the grid resistor or the secondary of a transformer. The fixed bias sits under either one of these.

The 1M will not be a fail safe in case of wiper opening. If you have the adjustment as a variable resistor from the grid tapping point in the leg going to 0V and have a fixed resistor from the grid tapping point to the negative bias supply, then that will be fail safe as on wiper opening, the grid will then only see the most negative point on the bias supply.

As Nick says, you can only regulate fixed bias if the HT if regulated OR if you take a portion of HT as a reference for the bias regulator.

The grid resistance for a fixed bias circuit is usually quite a bit less than for self bias to ensure the valve doesn't run away. The data sheet normally gives two values for max grid resistance for self and fixed bias. Sometimes it can be quite low and so place extra demands on the driver.

Make the fixed bias supply as good as you can. Soooo many use some half wave rectified, barely smoothed arrangement and then stick it up the grid. A better supply does sound better. Something I haven't got to the bottom of in mine but I've found same on other amplifiers.

I've recently seen a fixed bias scheme fed into the cathode but that was for PP so not sure of that applies here.

Edit: I just looked up a data sheet and it says 250k max self bias, 50k max fixed. So that means the total resistance in the grid to ground must be 50k max.

[JamesD]

In theory the 1M failsafe isn't...

If the 5K wiper goes open circuit then you are in -100V fixed bias mode and WE recommends fixed bias operation has a maximum Rg of 50K (from memory - couldn't find the datasheet to check it could be 150K but a it will be a lot less than 1M).

The reason for this is to stop grid current developing positive bias in normal operation so uless you are driving the beastie with, say, more than an 90V peak signal you should be fine with the 1M.

Otherwise comments much as Nick has said - he probably has the most experience with this and driving the beastie into mild A2 too.

Looks like an interesting project and it will be interesting to hear what you think of fixed bias via cathode bias.

James

[izzy wizzy]

Also Rod Coleman has a fixed bias gizmo that might be of interest.

[RhythMick]

As always, useful input thanks Nick.

I already have fused on both transformer secondaries, so yes you're quite right that protection is already there. Thanks.

Adjustable regulated supply - could you point at an example ?

[RhythMick]

As always, useful input thanks Nick.

I already have fused on both transformer secondaries, so yes you're quite right that protection is already there. Thanks.

Adjustable regulated supply - could you point at an example ?

You mean something like this ?

Capture.PNG (16.07 KiB) Viewed 6328 times

[RhythMick]

and here

http://www.ti.com/lit/an/snva583/snva58 ... 1114596772

[Nick]

Adjustable regulated supply - could you point at an example

There is a discrete one in MJ. Another choice is a 100v VR tube and potential divider from that. But again, you have the varying B+ issue to deal with. Depend on your mains voltage or if you plan a current meter in the circuit so you can tweek as required.

Depends on the value of gm for the valve you are using. No big deal with a 300b, but I would worry about a 6c33c (for example)

[RhythMick]

Adjustable regulated supply - could you point at an example

There is a discrete one in MJ. Another choice is a 100v VR tube and potential divider from that. But again, you have the varying B+ issue to deal with. Depend on your mains voltage or if you plan a current meter in the circuit so you can tweek as required.

Depends on the value of gm for the valve you are using. No big deal with a 300b, but I would worry about a 6c33c (for example)

valves I'm considering this for are all DHT : 26, 45 and 300B

[izzy wizzy]

http://www.pmillett.com/bias_supply.html

http://lyrima.co.uk/Bias/Bias-Regulator-V2.pdf

I wouldn't inject the mess from a LM317 up a grid. They sound awful even on the heater of an IDHT. IMHO of course.

[RhythMick]

Ok as I don't currently regulate HT and don't have any plans to introduce that I'm going to filter the bias as suggested rather than regulate it.

That also means I will be taking the bias from a simple voltage divider (pot) as originally planned, as opposed to an adjustable regulator method.

I'm not 100% clear on the grid resistor, though I agree the value needs to be limited (working on 50k). Looking from the grid, this would be the series resistance of part of the wiper plus the base resistor. Is that correct, or should it be the parallel path both to 0v and to -100v ?

Limiting that grid resistor also means a larger coupling cap from the previous stage to keep the frequency corner down. If the objective was to avoid the cathode bypass cap it's clear this approach merely swaps one capacitor for an increase in another. That's not my objective by the way, I want to try it and listen to compare. Interesting though.

[Nick]

Limiting that grid resistor also means a larger coupling cap from the previous stage to keep the frequency corner down. If the objective was to avoid the cathode bypass cap it's clear this approach merely swaps one capacitor for an increase in another. That's not my objective by the way, I want to try it and listen to compare. Interesting though.

Grid resistors are always limited, its likely you are just increasing the value. The solution to this is to use active drive to the grid this fixes several problems and allows A2, but it also adds complexity.

It will never be a case of just comparing fixed and auto bias, you will also always have the other changes that that brings.

Personally I try and avoid dogmatic thinking like X is bad, remove all X. Doing that will add other compromises so you never compare just X and not X

[RhythMick]

The solution to this is to use active drive to the grid ...

I'll need an explanation on that if you have one to hand Nick ...

[pre65]

Grid resistors are always limited, its likely you are just increasing the value. The solution to this is to use active drive to the grid this fixes several problems and allows A2, but it also adds complexity.

That's interesting, I thought active drive was predominantly to achieve A2 drive. I had thought of trying fixed bias on the GK-71, but it's way down the list for the time being.