• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

SE With shared Rk in output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The reason I ask is that I am planning on building a 6LU8 SE UL similar to Kegger's but with switchable NFB and UL/Triode modes. I think I have just about everything I need on hand except that I don't have the appropriate cathode resistors. I do however have and 1800 and 470 ohm that I can parallel which should give me about the right bias if I run both cathodes through them.

I thought I might try prototyping it that way to see how it goes but if the sound quality would be seriously compromised then it wouldn't be worth the trouble. A little loss of separation is probably not a problem.
 
Little bit of a Duh here. I just realized that my PT has a 45V bias tap. I could just use fixed bias. I will have to do a little bit of reading up on it as I have not really looked at that option seriously before.

The tap is just a tap rather than a separate winding. I suppose that the load current is so small that interaction with B+ is theoretically insignificant. Also due to low current I imagine that filtering is not too difficult. I have a lot of SS diodes and bridges around as well as low voltage caps for filtering and diode snubbing so this should be very doable with parts on hand. I also get to save my power resistors for B+ filtering and screen connection for UL and triode mode. Figuring on 470ohm for the screen.

I worry a little bit about getting a hard SS kind of sound but will just have to try it and see how it goes. Has anyone ever used fixed bias for both output stage and VAS?
 
Don't take the concern over harshness too seriously. I have not had any experience in that area it is just that I have heard a lot about fixed bias guitar amps being edgier than cathode biased amps. Of course hi-fi is a totally different thing and edgy does not necessarily mean harsh. It is just something that I will be listening for.

On advantage of the "fixed" bias approach is that I can easily compare different bias settings.

On other thing. If I plan to have the NFB and mode (UL/Tri) switchable I figure that hot switching of either is probably not a good idea. Would you all concur?
 
Oh, BTW, I have been trying to find good information (or example schematics) of a grid bias supply from separate tap but haven't had much luck. Can anyone point me to some good schematics or descriptions? It seems like it should involve a half wave rectifier with at least a single RC filter and a voltage divider. It also seems wise to design so that an open wiper results in sufficient bias to prevent meltdown.

Also for calculating the maximum grid resistance (for stability purposes) the value of interest would be the grid stopper in series with the worst case parallel combination of the two legs of the voltage divider (including the filter resistor) correct?
 
I got curious and Googled grid bias supply. This was the first entry: http://www.pentodepress.com/bassman/JMP50gridBias.html

It's a good simple model for a basic supply, which addresses the issue of open wiper protection. Though it's designed from the full B+, the same principles apply to a tap. As long as the bias supply draws a couple of mA, just use the grid leak resistor specified for fixed bias.

Sheldon
 
mashaffer said:
So the output of the bias supply shown would simply be connected to a resistor (up to the max allowed for fixed bias) in series with the grid correct?

Yes.

mashaffer said:
Oh, and in my case I assume that I can eliminate R1 as I don't need to drop any voltage.

You can. If so, I would use a 47u input cap, then two RC sections with 7.5k/10uf, then your resistor shunt (RA,RB). With the shunt at about 25k you'd get ripple at about .5mV and 39V output. If you don't need ripple that low, you can reduce the input cap size (or the others). Play in PSUD until you're happy.

Sheldon
 
OK, looks like we were thinking along the same lines. I setup a CRC filter in PSUD but used 100uf and 1.8 to 2.2k and I think it was a total of about 18k on the shunt (not sure right off hand) to get about 2mA of current and 4mV of ripple with bias in the high 30Vs range. That is quite a bit more ripple than your design but it was late so I may have misread the results.

In the final design I need independent left and right bias of course so I was thinking about a single CRC filter that splits into a shunt made up of a series connection of two resistors and a pot for each side. The top resistor on each side would drop the excess voltage for that side and the actual bias would be taken at the junction of the first and second fixed resistors. The values would need to be chosen so that setting the bias pot to minimum still results in a current less than meltdown. If the wiper fails open then maximum bias would be applied thus guaranteeing safety.

Assuming that I am using the bias only for the output stage what value of ripple would you consider adequate? I was thinking that 4mV was higher than I wanted but with the output stage input signal being rather large in comparison and the stage gain pretty low it might be OK. I like you .5mV better of course.

Do you see any value in a small snubber cap across the diode?

I have lots of 1004 and similar diodes but also have some fast low noise types (don't remember the number off hand but they were ones suggested by Eli) that I got for a doubler supply (it is my habit to buy extra of the little inexpensive parts whenever possible). Any reason to go with the better diodes in this application?

And lastly (sorry for being so long winded) I have a 50uA meter that I am planning to use as the bias meter. It measures 2.2k ohms on my multimeter. If my calculations are correct I can get 1/2 scale deflection at 50mA if I put it in parallel with a 1.1 ohm cathode resistor. Does that sound right?

I plan on biasing in the 50-60mA range so I wanted that right in the middle of the meter's range. This is convenient because I can just double the reading and interpret as mA instead of uA.

Thanks for your patience!
 
mashaffer said:
I setup a CRC filter in PSUD but used 100uf and 1.8 to 2.2k and I think it was a total of about 18k on the shunt (not sure right off hand) to get about 2mA of current and 4mV of ripple with bias in the high 30Vs range. That is quite a bit more ripple than your design but it was late so I may have misread the results.

That's about right. But look at the waveform too. I like to see something close to sinusoidal. Better to use an extra RC section.

mashaffer said:
In the final design I need independent left and right bias of course so I was thinking about a single CRC filter that splits into a shunt made up of a series connection of two resistors and a pot for each side. The top resistor on each side would drop the excess voltage for that side and the actual bias would be taken at the junction of the first and second fixed resistors. The values would need to be chosen so that setting the bias pot to minimum still results in a current less than meltdown. If the wiper fails open then maximum bias would be applied thus guaranteeing safety.

I would just split after the first C. Then follow with two parallel RCRC and shunts (just take into account the paralleled values when you model). I would take the output from the top of shunt, as per the example. This is a fail safe design. If the wiper fails, resistance through the pot increases, and bias increases. I don't see any advantage to taking your output from a divider. In principle you could use a pot only. Using a resistor and a pot just reduces the current through the wiper (a good thing, but not absolutely necessary).

mashaffer said:
Do you see any value in a small snubber cap across the diode?

Might reduce HF hash, probably won't hurt. Try it and see.

mashaffer said:
I have lots of 1004 and similar diodes but also have some fast low noise types (don't remember the number off hand but they were ones suggested by Eli) that I got for a doubler supply (it is my habit to buy extra of the little inexpensive parts whenever possible). Any reason to go with the better diodes in this application?

Sure, if you have them handy and they are cheap, why not?

mashaffer said:
I have a 50uA meter that I am planning to use as the bias meter. It measures 2.2k ohms on my multimeter. If my calculations are correct I can get 1/2 scale deflection at 50mA if I put it in parallel with a 1.1 ohm cathode resistor. Does that sound right?

I plan on biasing in the 50-60mA range so I wanted that right in the middle of the meter's range. This is convenient because I can just double the reading and interpret as mA instead of uA.

Should work.

Sheldon
 
I would just split after the first C. Then follow with two parallel RCRC and shunts (just take into account the paralleled values when you model). I would take the output from the top of shunt, as per the example. This is a fail safe design. If the wiper fails, resistance through the pot increases, and bias increases. I don't see any advantage to taking your output from a divider. In principle you could use a pot only. Using a resistor and a pot just reduces the current through the wiper (a good thing, but not absolutely necessary).

Ah yes, I can see that if you have separate RC filters for each side the excess voltage would be dropped across the filter resistors eliminating the need for the extra resistor. If I had used a single filter section as I had described then the extra voltage dividing resistor would have been necessary otherwise both channels would have been at the same voltage.

In the original diagram Ra prevents one from setting the bias to zero which is a good thing so I think that I should leave that resistor in.

I will give your suggested layout a try.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.