question about Morph 5879 screen voltage

[luthierwnc]

Hi guys,  I've got an amp in the tweak phase using a 5879 pentode setup with the Blencowe morph design.  I've done these before but am getting a different result.  I've got 220VDC at the last rail node.  A 56k resistor across a .001 cap goes to the plate and I'm getting 149 volts.  A 470k resistor goes to the screen and I'm getting 61 volts there.  I remembered those being much closer.  I've tried two different tubes and they are within a couple volts of each other.  Volts don't change much on either end of the 1m pot sweep.  In the absence of ideas I'm thinking of increasing the plate to 100k-ish 

Any thoughts about techniques or results would be appreciated.  Cheers, sh

[tubeswell]

Got a schematic?

[luthierwnc]

The new red voltages on the front end are correct.  The older, smaller ones elsewhere were targets.  The PI plates are only 174 and 170 so I might change the first dropping resistor.  Amps sounds pretty good as is but I'll probably throw in fixed-bias to clean it up a tad more.  Thanks for looking, Skip

[tubeswell]

Great set of notes there.


You've got:

0.8V/470R = 1.7mA cathode current

(225-149)V / 56k = 1.35mA plate current (Ip)

(225-54)V / 470k = 0.36mA screen current (Ig2)


So Ip:Ig2 = 3.75:1

Datasheet (https://frank.pocnet.net/sheets/049/5/5879.pdf) says 4.5:1, so the ratio in yours is a little low by comparison (i.e. you have a higher proportion of Ig2), but it's not toooo bad.

What does it sound like?

If you want, you could either:

1) raise the bias voltage slightly (e.g. try 680R, 820R or 1k) and see where you land (this will lower overall tube current but won't change the Ip:Ig2 ratio), or
2) decrease Rp to (say) 47k, which will increase Ip (and steepen the load line) and therefore bring the Ip:Ig2 ratio up a bit (this will also increase the overall load impedance on the stage), or
3) increase Rg2 to 680k, which will lower Vg2 a bit (and also bring the Ip:Ig2 ratio up a bit)
4) some combination of the above

I'd try either 2) or 3) first - YMMV

Don't forget to try things iteratively with voltage notes to see how the different aspects interplay


(Edit: the Tung-Sol 5879 datasheet has some really helpful suggestions for playing around with the gain structure in the table on page 3. https://frank.pocnet.net/sheets/127/5/5879.pdf)

[Backwoods Joe]

I've got a similar amp. 5879: cathode R 2.7k @ 3.8V, plate R 100K @ 176V, screen R 560K @ 118V.  7591: cathode R 270ohms @ 19.5V, plate V 430. 

[shooter]

72mA is a good place, sound good?

[tubenit]

Thanks for sharing the schematic!  That looks like a very cool amp.

IF you have soundclips, I'd love to hear one.

With respect, Tubenit

[luthierwnc]

Thank you my friends!  Tubenit: this is mostly yours and Geezer's.  I need to get rid of some distortion when the amp is turned up.  Suspicion falls to the low plate volts on the PI.  I've got an AT in there now.  That could change.

Tubeswell; I ran through the three steps and voltages were little changed -- maybe ten volts and the ratios stayed the same.  I've attached an updated schematic with the changes and voltages in Carolina Blue.

Backwoods Joe; haven't gotten to your ideas yet but that may be coming.
Also attached is a picture of a 1k sine wave at the .0047uf coupling cap leading to the volume.  Morph, volume and MV all at halfway.  8 ohm dummy load.  I've only tested this with the basement Strat, no FX in the loop with both pots set at halfway.  It is very quiet as far as hums, buzzes and shhkghkkksh's
Cheers, sh

[tubeswell]

I see you changed Rp and Rk (Which has increased plate and screen voltages slightly) but not Rg2. Try 680k there

[luthierwnc]

I did but it only widened the gap (5 volts) so I put the 470k back in.  Sorry for not mentioning that, sh

[tubeswell]

Did you try the 680k with the original cathode resistor? The combination of hotter bias but lower screen voltage should start to decrease the Ip:Ig2 ratio


Try other combinations. The screen and bias both affect tube current but the screen voltage more so. If you tweak it so that you can get ‘as pentode mode as possible’, the morph control will have a bigger range of tonal effects.

[luthierwnc]

No, I had already changed the K resistor before.  I'll put it back to 470R after dinner and report.  sh

[luthierwnc]

I sneaked downstairs and put the 470R back on the cathode with little change in the plate to screen ratio.  For yuks I put 120k on the plate which gave me 104v plate, 54v screen, 1.0v K.  It made the bottom of the waveform a little rounder.  sh

[tubeswell]

I sneaked downstairs and put the 470R back on the cathode with little change in the plate to screen ratio.  For yuks I put 120k on the plate which gave me 104v plate, 54v screen, 1.0v K.  It made the bottom of the waveform a little rounder.  sh

Its the current ratio that's important for ensuring optimal screen grid dissipation. You want the screen voltage as high as possible without risking over-dissipation, and the plate load just through or slightly below the knee of the Vg=0 grid curve.  So you adjust the screen voltage and/or plate load to get to that point. If you aim for a bias that gives you the hottest possible tube current while maintaining a 4:1 plate-to-screen current ratio, you can then manage the load line and the knee.


I assume you've still got 225V HT?


(225-104)/120k = 1.0mA at the plate
(225-54)/470k = 0.3mA at the screen


so Ip:Ig2 = 2.75:1 (this is going the other way from 4:1)


But 1V/470R = 2mA at the cathode, so something doesn't add up (unless I don't have the up-to-date voltages?)

[luthierwnc]

As of right now, the best sounding combination is 470K on the screen, 68k across a .001 cap on the plate and 820R on the cathode across a 10uf cap.  Volts are 53, 137 and 1.33 respectively.  sh

[PRR]

....optimal screen grid dissipation. You want the screen voltage as high as possible without risking over-dissipation,....

There's no way this screen can over-dissipate, 200-some volts and half a Meg in series. 0.020W max?

High screen volts is for high Power Output. Raises maximum cathode-plate current. (Not that we can pass more than about 3mA in 56k...)

Low screen volts for high voltage gain.

High voltage gain with hot guitars is gonna distort.

[tubeswell]

I wasn’t meaning to imply the screen would over dissipate the way it is set up- Sorry, my bad choice of words.


I was thinking about how to setup up the signal pentode in V1 so that there is high gain in pentode mode compared to triode mode. I wasn’t thinking about getting distortion from V1 on its own, but from how changes in signal gain there interacts with the rest of the amp. And of course, that may not be the goal. I should just shut up 🤐

[DummyLoad]

you are biasing too warm. 5879 works cleaner with 250V-300V B+ and a cooler bias. to increase gain make Rg2 larger, 1.5M is about the useable limit with all others the same. 

my 5879 ckt. values are: 100K Ra, 560K Rg2, 1.8K Rk, 22uF Ccath. 100nF Cg2
B+ is 260V, Va 148V, Vg2 90V, Vcath 2.6V

1Vpk in, I get ~60Vpk out clean.

--pete

https://www.dropbox.com/sh/kj884fz9gmn9uuv/AACDzqBMKoZfwLCdPN8jCNVza?dl=0

scope setup - Ch2 is 1V and input monitor ; Ch1 is 20V and output monitor.

[DummyLoad]

your long-tail PI seems sick if the posted values in your latest schema are correct. with those values there should be more than 18V across the 39K - something miswired? values you're using are really better suited for 400V+ supply.

values on BB ckt.:  100K Ra1, 100K Ra2, Rk 1K, Rtail 39K, Rgf1 270K, Rgf2 270K.
Supply 330V, Va1 213V Va2 224V, Vk 85.3, Vese of Rtail 82.9V

values on BB w/ your ckt.: 100K Ra1, 100K Ra2, Rk 470R, Rtail 39K, Rgf1 270K, Rgf2 270K
Supply 330V, Va1 194V, Va2 200V, Vk 91.5V, Vese of Rtail 85.7V

Vin with clean 60Vpk output (just before onset of clipping) was about 3.8Vpk.

--pete

[PRR]

Fully agree about the Ip/Ig2 ratio giving insight, and the datasheet always a good start.

> ...high gain...getting distortion...how changes in signal gain....may not be the goal....

That's about where I got to. Is there a goal? Is there a problem?? I see some observations and "I remembered those being much closer". We see some "rounding" on the bottom. Well, it's gotta get bent somewhere. And it looks like good strong signal before it goes bent, and it does not look like a stinky clip. Is it getting booed off stage? Does the morph knob do anything?

[PRR]

Thanks for the datasheet link https://frank.pocnet.net/sheets/127/5/5879.pdf

The two pentode datasets neatly guide us to our goal, whether max gain or max output. Many g-amp stages tend to a 200k-270k audio load, and 100k plate resistor fits that. The max gain condition needs about a half-Volt input to clip; the max output can stand near 2V input.

[luthierwnc]

Sorry to be late with thanks.  A big pine came down across the street and crushed the utility pole.  Six bucket trucks in the middle of the night!  Internet is back up as of ten minutes ago.  I'll go downstairs and have a look at the amp tomorrow.
To be honest, I'm not entirely in love with such a low B+.  I think this thing would be  happier in the 400-410 range.  And yes, 175VDC on the PI plates is not what I was shooting for.  The amp sounds pretty good at medium settings but it gets dirty much too fast. 

cheers, sh