Screen resistors on ultralinear tap (Also bias question)

[Cliff Schecht]

The bass amp I'm working on is using an ultralinear output section, something I've never used in a design before. I've seen in most designs the ultralinear taps are hooked up directly to the screens but in some designs, there is a resistor between the pin and the wire (something like 470 Ohms).

Since I'll be running a quad of KT-120's at 500-550V, I'm not quite sure if I actually need this resistance. These tubes are tough from what I've read and seen in the datasheets but I'm having a hard time finding the ultralinear curves for KT-120's. Most of the datasheets I've found have no curves at all and the few I have seen are at very low plate voltages (vintage datasheets for essentially a different tube). I'm honestly not even sure what my bias will need to be without these frickin' curves!

Any thoughts on what I should do with the screens in this scenario? Use resistors for added protection? What value/wattage should I use?

How about the bias for these tubes with a 1900 Ohm CT primary?

[Reeltarded]

Would it help at all if I said a Major is like 620+ volts 1400 ohms and like mid -70s with KT88s afaicr..

UL also.

[pdf64]

I don't have any experience of KT120s but screen grid resistors are always a good idea. The screen grid is also a control grid, ie the voltage on it affects plate current, and so something there to act as a grid stopper is beneficial in mitigating for oscillation.
But if the amp may get overdriven then they are necessary to limit screen grid current / overdissipation.
With a high VB+, 1k 5W is probably a reasonable starting point.
Pete.

[martin manning]

As I recall some people will increase the size of the screen resistors on Marshall Majors to kepp them from eating tubes. The schematic shows 250 ohms 5W for the KT88's, but I think I recall an increase to 1k was recommended.

On the bias, will your amp running fixed or cathode bias? Either way I don't think the ultralinear connection matters, just get them to 60-70% of max Pa at idle. How did you arrive at the plate load value of 1900 ohms?

[cpollack]

+1 on Pete's take. Most UL amps are for HiFi (or derived from HiFi circuits) where we never intentionally overdrive the output section, though some of the GEC (UK) hiFi UL designs do use 100 Ohm screen resistors. Something 470 Ohm - 1kOhm is prudent and your tubes will thank you.

[Cliff Schecht]

The primary impedance is 1900 Ohms CT for the Hammond 1650T that is going in this project. I didn't choose this, it chose me!

[martin manning]

Just looking at max power on the load line 1900 ohms seems good for a quad-KT120 output running at 550V. Idle current would want to be 70mA.

[Cliff Schecht]

I guess the other variable I'm trying to figure it is the range that I need to be able to adjust my raw bias to. I'm using a post PI cathode follower in this amp (from the SSS schematics) which is sort of where my confusion lies. I understand that this circuit needs about -175V of raw bias and have a way to provide this, but I don't know how much drop I need from the CF output to achieve the actual negative grid bias voltage to the power stage.

Since I don't have ultralinear curves I'm not sure how to guess the negative bias the power tubes will need from the triode/pentode curves. How is this done typically? I've seen someone calculate the negative bias needed without curves before but I don't remember how..

Usually I'd at least try to figure this out on my own but I have been very pressed for time as of late. I still have to produce a thesis within the next three months in time to graduate!

[gingertube]

Cliff - Think IDLE (No signal). For Ultralinear at idle the screen voltage will be VERY close to the anode voltage. Use the triode connection bias value with some confidence.

Then for the output tube grids DC coupled to the cathode follower (only grid stop resistor in between) and bias applied to the cathode follower grid --

The load resistor on the cathode follower needs to be returned to a negative rail. The "Rule of Thumb" for that negative rail is to use approximately three times the bias voltage at idle. This gives sufficient "head room" for output tube grid to be driven (pulled down) sufficiently negative to cut off the tube (for Class AB operation).

If you want to get "fancy" and impress the prof. then you can get better noise performance (blacker background) by replacing the cathode follower load resistor with a current source. Any residual negative rail noise gets divided according to the ratio of the cathode follower load resistance and the impedance looking into the cathode of the cathode follower but that divided residual is indestiguishable from signal and will be amplified by the output tubes and appear as hum and/or buzz. A current source has much higher impedance compared to the typical required resistor and so that noise due to residual negative rail noise will be MUCH lower.

A simple current source will do. A "Ring of Two" Transistors gives good results, use a high hfe small signal transistor on the bottom and an MJE340 (for the voltage withstand) on the top.

I would use screen resistors. These act as grid stops just like the grid1 grid stop and suppress any tendency to parasitic oscillation. Screen stops are much more important with Ultralinear connection than with either triode or pentode mode. In an parallel push pull EL34 circuit with that same Hammond output tranny I found that 150 Ohms was Ok for Triode and Pentode Mode but in Ultralinear I had to take that resistor value up to 1K to get rid of parasitic oscillations. These oscillations were not the usual rail to rail thing you normally see but showed up as a brief burst of ultra sonic frequency on the trailing edge of the sine wave I was testing with (just after turn over) most likely (bit of a guess) at the point where one side of the push pull goes into or comes out of cutoff. Use 1K 5W. These should be mounted with the resistor body as close to teh screen pin on the tube socket as possible (just like grid 1 stoppers).

Cheers,
Ian

[Reeltarded]

i was just about to say the same thing

[gingertube]

A scribble of what I said above.
Cheers,
Ian

[Cliff Schecht]

Hey Ian that's fantastic info, thanks a bunch. I will probably stick with the resistor in the CF simply because this is a clean amp and any noise contribution at this stage should be negligible (or close enough).

Also FWIW this is not for my thesis, I do my research over integrated envelope tracking power supplies. I probably would enjoy grad school a lot more if I was doing tube amp related research..

[gingertube]

We really are going to have to start educating these professors so that grad reasearch projects can be tube based - after all tubes are still the most linear amplifing devices ever devised by man.

Cheers,
Ian

[martin manning]

Cliff, here are the characteristic curves for the TS KT120. The pentode curves are all at rather lower voltages than you are interested in, but the triode curves are there for your bias voltage estimate... -70V looks about right.

[Cliff Schecht]

I knew I could count on you guys. Appreciate the help!

Ian being that I use my real name I won't badmouth my professor or my work too much on here but there is a research group at my school that works with ultra-high voltage GTO's and are always working on new drivers for these devices. I've consulted for them quite a few times for either tube based drive circuits or high voltage switchers. It's interesting work but I think the hands on IC design experience I get in my group is invaluable. Now I just need to GTFO of school and get a real job so I can fuel my tube/amp/guitar addiction more readily (for example, a real deal '53 Tele came into the shop where I do repair work on the side - If I would have had $12k to cough up I could have easily added it into my collection.. ).