All DHT Circuit Advice

[RhythMick]

Usually, the concern in fixed bias is the adjustment pot since the wiper arrangement is a weak point. The bias power supply if designed properly can be considered just as reliable as any p.s. The issue then comes down to keeping sufficient negative voltage on the power tube grid to prevent tube runaway if the adjustment pot were to become open. The easy solution is to simply put an appropriately sized resistor from the pot input to the wiper output. If the wiper loses contact and becomes open the resistor will supply a high negative voltage on the grid putting the power tube in cutoff. The only issue would be the adjustment range of the pot will be less with the safety resistor.

I'm redrawing the bias circuit - does the attached match what you were suggesting ? So if the trim pot fails (goes either open circuit or shorts) the full -120V will be presented on the grid ?

 

[analog_sa]

After some brief listening i went from an adjustment pot to fixed resistors. Imho, a substantial sonic improvement.

 

[analog_sa]

Surely, there is a final bypassing cap at the pot wiper, right?

 

[20to20]

I'm redrawing the bias circuit - does the attached match what you were suggesting ? So if the trim pot fails (goes either open circuit or shorts) the full -120V will be presented on the grid ?

Have you ever seen the bias supply circuit of a Dynaco ST-70? You don't need all that choking in the filtering. The bias current is small and non-existant through the grid. A half-wave rect. and 50uF + 50uF will get the job done, and why have all those individual resistors in series?

 

[RhythMick]

The individual resistors were chosen to give me the trimmable range I wanted (30 to 50 v) with the pots I have (1k 2W), while keeping each resistor within 25% of it's dissipation rating.

I look forward to simplifying the filtering on the bias later to see how it changes it. I may have overdone it but I can't help thinking that bias supply is right on the grid of the output tube ...

 

[RhythMick]

Surely, there is a final bypassing cap at the pot wiper, right?

Thanks - I'd missed that from the diagram. Will be a Tubecap 30uF.

 

[Brice]

The problem of loosing the bias pot have been solved long time ago.

Please look at the attached schematic. If you lose the balancing or bias pot, the tubes won't run away. The point 4 is the negative bias supply.

 

[RhythMick]

Sorry but I don't understand what that is showing me, nor how I could apply it to my circuit (which only has 2 triodes, a 5842 and a 45, and no pentodes). I'm quite prepared to accept that there is a solution there, but at present I don't understand it.

The attached is from the KT-88 circuit referenced earlier. If it's that then it seems to me it's another use of a relay. If I'm going to use a relay I think I'd rather just cut the heaters on the main rectifier.

 

[DAK808]

Mick, Regarding your bias scheme; The essential idea is correct although i am not sure if your value is good. In my circuit i used a 36K. But that was a different circuit with a 50k pot. I did have -120vac from a reverse installed filament trans. that 120 was dropped with a 25H choke to a series of RC filters then to the 50k pot. I believe my voltage range is from around -30 to -55vdc, although i don't remember what the value of the final dropping resistor is before the grid. That resistor should be carbon comp. Also the cap on the wiper of the pot should be grounded to the ground of the other filter caps in the bias circuiit. The voltage rating of your filter caps in the bias circuit only needs to be about 200vdc. The Mundorf tube cap you have there is kind of a waste in that spot, unless, you have a whole box of them. cheers.

 

[analog_sa]

Have you ever seen the bias supply circuit of a Dynaco ST-70?

It's an entry level amp...The equivalent of a NAD3020, only with tubes.

 

[analog_sa]

Thanks - I'd missed that from the diagram. Will be a Tubecap 30uF.

At the time i ended up with a similar bias circuit, input Tubecap, Lundahl plate choke, resistor, Mund MKP, resistor divider and Mund EVO. Replacing the second and third Tubecaps in my setup restored a normal tonal balance. For some reason Tubecaps sound a bit upward tilted to me.

 

[MagicBus]

Something more practical? I don't think it would work with tube rectifier though...
Since safety measures can be taken for trim pots that makes the diodes the weakest link. Assuming that they won't fail altogether at the same time, if one fails either open or short > fuse open then the psu would turn from full wave to half wave still providing the same voltage. Fuses only have to act against shortcircuit so they can be oversized well above inrush current. Sensitive ears should be able to detect the increaced ripple in the bias voltage.

Edit; I must be carefull posting late in the night! I forgot to add my worries about one diode taking the other with it...

 

[RhythMick]

Mick, Regarding your bias scheme; The essential idea is correct although i am not sure if your value is good. In my circuit i used a 36K. But that was a different circuit with a 50k pot. I did have -120vac from a reverse installed filament trans. that 120 was dropped with a 25H choke to a series of RC filters then to the 50k pot. I believe my voltage range is from around -30 to -55vdc, although i don't remember what the value of the final dropping resistor is before the grid. That resistor should be carbon comp.

Since I'd chosen to go with LCLC filtering I needed to make sure I had at least enough current to satisfy the choke minimum current. For that reason I went with a 1k wiper and a total stack of around 5k5, giving about 20mA current. I also have a liking for Allen Bradley resistors (although sadly I don't seem to be using any of them in this amp) and I saw some Allen Bradley 2W 1k linear pots for sale. Following the suggestion I figured the failsafe resistor should be a high value to avoid changing the trimmer stack significantly, so I decided 100k for no scientific reason.

Also the cap on the wiper of the pot should be grounded to the ground of the other filter caps in the bias circuit.

So the cap goes to 0v ? I'm basing the circuit on an SJS amp design, the snip attached shows it going the other way ?

EDIT: No it doesn't. Sigh. Thanks for pointing out.

The voltage rating of your filter caps in the bias circuit only needs to be about 200vdc. The Mundorf tube cap you have there is kind of a waste in that spot, unless, you have a whole box of them. cheers.

This is my attempt to build the best amp I can and I decided to avoid electrolytics completely. I'd rather get lower values of good caps than high values of crap ones. I could be completely wrong. I'll find out soon.

Thanks for your help

 

[RhythMick]

Something more practical? I don't think it would work with tube rectifier though...
Since safety measures can be taken for trim pots that makes the diodes the weakest link. Assuming that they won't fail altogether at the same time, if one fails either open or short > fuse open then the psu would turn from full wave to half wave still providing the same voltage. Fuses only have to act against shortcircuit so they can be oversized well above inrush current. Sensitive ears should be able to detect the increaced ripple in the bias voltage.

Edit; I must be carefull posting late in the night! I forgot to add my worries about one diode taking the other with it...

Yeah I thought about going diodes for rectification and doubling up for redundancy, but you can never eliminate the chance of something failing, so back to the relay I think.

 

[trobbins]

SS diode in series with each valve diode anode provides fault alleviation.

 

[DAK808]

This is my attempt to build the best amp I can and I decided to avoid electrolytics completely. I'd rather get lower values of good caps than high values of crap ones. I could be completely wrong. I'll find out soon.

Thanks for your help

I know there is a lot of advice being thrown your way but all the more reason to be conservative about your component choice at least until you get the amp up and running. Since you are designing your amp and not copying something that has an established track record. So, using "cheap" electrolytics for now would make sense until you get the amp functioning properly after which point you can subbing in the better caps you have ready. Just be sure to leave enough room for your replacement parts.

 

[Shef]

I'd summarize it as "keeping all tube in audio path" except maybe the CCS for balanced stages and I stick to it too, anything else - up to what makes it the best sense and outcome.

Once in awhile I see someone's insane "pure" project with making even the filament rectifiers on tubes. .. they are ALL super-ugly.

 

[prairiemystic]

Are you knocking my 866's lol

Bias supply hum gets somewhat cancelled in a PP amp, I think this is for a SE design? Then it should be very low ripple.

Relays have a lot of inductance, even small ones measure 70-175mH. It goes up once the armature pulls in. This is too much on a cathode, it would roll off the high-end significantly IMHO.

 

[MelB]

There has been a lot of discussion about loss of bias and tube runaway but my experience is that the bias remains intact but the tube runs away because it becomes old and gassy. That is to say in my experience Chinese 300B's die that way.

 

[RhythMick]

Are you knocking my 866's lol

Bias supply hum gets somewhat cancelled in a PP amp, I think this is for a SE design? Then it should be very low ripple.

Relays have a lot of inductance, even small ones measure 70-175mH. It goes up once the armature pulls in. This is too much on a cathode, it would roll off the high-end significantly IMHO.

SE design yes so I'm minimising the bias ripple as much as possible. Bias through Tribute interstage transformer. Can't wait. The current design was a revelation (very simple signal circuit, 5842 driving 45, cap coupled) but I've not heard an interstage and a Tribute should be something special.

The relay trigger will be taken off a separate bypass leg (not the trimmer stack ) and it will engage / disengage the 5v heaters on the main rectifier. That way I should get both a slow start and a cutout if the bias fails. The relay itself will be located near to the mains transformer but underneath so not right next to it. I figure any noise introduced there has a much bigger chance of being smoothed.