Cathode resistor value for SE 6L6/EL34/6550

[Steve A.]

The JCA PicoValve (designed by Andy Marshall of THD) uses a 250 ohm cathode resistor with a B+ of 305v. So how does it not blow up like a firecracker? It almost did when I rewired the preamp as a Trainwreck Rocket... So I guess if you keep the voltage on the grid down low enough it will not go into meltdown mode?

I checked the THD Uinivalve schematic and it uses a more reasonable 300 ohm cathode resistor, but the other schematics I checked all used a 470 ohm resistor. Except for the Torn Sun Dark Star MkII (a SE version of the TW Rocket) uses a 200 ohm cathode resistor with a plate voltage of 355v and a screen voltage of 250v*** (it uses a 24k resistor in the power supply between B+1 and B+2 but no screen resistor). Hmmm... that could be answer to my question: the stock PicoValve design uses a 1k dropping resistor and 5.6k screen resistor (which I changed to 1K to make the 2W triode mode sound better). I guess if I put the 5.6k screen resistor back in I could use a 250-300 ohm cathode resistor.

With the 1k screen resistor and 250R cathode resistor and 6L6GC tube I was getting 305v at the plate, 301v at the screen and 3.02v at the cathode. I know that E=IR but I forget the multipliers used...

Here is a YouVideo of someone playing Torn Sun Dark Star MkII but you might want to cover your ears because the guitarist is a real Stevie Ray Wannabee (SRW?). He evidently wants to play like SRV very badly and definitely succeeds!

Torn Sun Amp Demonstration - YouTube

Steve Ahola

*** I just reread the post at AmpGarage and the voltages listed on the schematic are not correct. My bad!

OK here are the voltages for the PicoRocket:

B+ rail (270v) -> 1k/5W (261v) -> 10k/2w (212v)b-> 10k/2w (194v)

B+2 -> 1k/52 -> screen

B+3 -> 100k -> V2a plate/V2b grid (125v) (V2b cathode also 125v w/56k cathode resistor)

B+4 -> 100k -> V1a/V1b plate (shared) 86v! (V1a & V1b cathodes split 1k5 w 22uFcap and 1k5 1 1uF cap/ voltage 0.71 and 0.78)
V1a/b grids and plates tied together

One problem I just noticed (now that I could turn amp up in daytime): I love the sound but it doesn't sustain (it cuts off fairly quickly) That happens to work great with lap steel- keep bar moving for infinite sustain and it will mute shortly after you stop. But I know that indicates a problem I need to fix. [EDIT that problem has been solved.}

[Steve A.]

Schematic for Torn Sun Dark Star MKII

This is pretty much the schematic I used for PicoRocket mod:



Gotta go to class! More details to follow

Steve Ahola

[Jazz P Bass]

Time to break out the scope & signal generator.

[Steve A.]

Time to break out the scope & signal generator.

This forum is called "Theory and Design" and I was asking about the theory behind the screen and cathode resistors in a SE output stage. In particular how Andy Marshall manages to get by using a 250 ohm cathode resistor while most schematics show a 470 ohm resistor for 6L6's. Would it have to do with the value of the screen used- 5.6K instead of 1k/5W? Or is it because the voltage on the plate is fairly low- 270v?

The Torn Star Dark Sun uses a 200 cathode resistor, even more extreme. Is he able to make that work because of the 24k dropping resistor between the plate and the screen? (I have no idea what the actual voltages were in that amp.)

More theory- I remember reading about a master volume control that worked by adjusting the screen voltage. Is there some formula or chart that shows how that work?

(Yes, I did mention a problem I had noticed with the notes not sustaining but I was not asking for advice on that. I am still in the preliminary stages of fine-tuning the design.)

Steve Ahola

P.S. One other question concerning theory that I didn't get around to asking: how does the cathode resistor on a DC-coupled cathode follower affect the signal level. Traditional 5F6A and Marshall designs use 100k- what is the effect of using 68k or 56k? Do the lower values increase or decrease the signal levels? Would they increase or decrease the voltage at the cathode?

[Gingertube]

Go to Duncan Amps site and lookup the 6L6 data fior SE operation
This is what it says.
Va = 350V, Vg2=250, Ia = 54mA, Rk = 300 Ohms, Rload= 4200 Ohms, Pout = 11 Watts @ 15% THD

The schematic you posted as a reference is pretty much those values (5K load rather than 4200 Ohms, else the same).

A 300 Ohm cathode bias resitor for the 6L6GC sets tube anode dissipation to about 18 Watts so I would probably go for a 270 Ohms for a dissipation of around 20 Watts. This is a conservative value, tube not working too hard and should last many years. Cathode bias resistor dissipation is justv on 1 Watt, use a 5W "Rats Coffin" style resistor.

200 Ohms is probably OK for an EL34 (higher gm so doesn't need as much negative bias voltage for the same idle current) but if using 6L6 then Rk=200 Ohms is asking for "cooked" tubes which will wear out really fast.

6550 is also a high gm tube and has more power capability so 200 Ohms would suit it as well.

It simply is'nt pratical to have one value (of cathode bias resistor) "fits all" but if you must try this then you need to use the 6L6 value as the "limiting" value. A 270 Ohm is probably a good compromise value, Good for 6L6, sub optimum but also OK for EL34 and 6550.

If you don't mind "hammering" the 6L6 a bit you could also try 2 x 470 Ohms 2W resistors in parallel to give 235 Ohms effective but I would want to measure idle current (measure the voltage across Rk and calulate from Ohms Law) and calculate the anode dissipation.

Cheers,
Ian

[Steve A.]

Thanks a lot, Ian! With the voltages on the Dark Star schematic questionable here are the voltages I measured tonight on the PicoValve (now my PicoRocket):

With 6L6GC (I doubt if I will be using EL34 or 6550 but I would like to use 6V6 if that is at all possible) I noticed a big difference in measurements when switching from 5.6K/2w screen resistor to 1k/5W:

w/ 1k/5w
Plate 272v --> 1k/5w --> Screen 262v
Cathode 59v (with 470R cathode resistor)

w/ 5.6k/2w
Plate 280v --> 5.6/2w --> Screen 234v
Cathode 51v (with 470R cathode resistor)

The power supply is as follows:
B+1 to plate
--> 1k/5w B+2
--> 10k/2w B+3
--> 10/2w B+4

Note: the PicoValve originally had a 5k6 screen resistor and the 250R cathode resistor. In an attempt to improve the pentode/triode switch I replaced the 5k6 resistor with 1k/5w. I didn't notice any problems with the original PicoValve preamp but when I rewired and reconfigured the pcb with the Trainwreck Rocket preamp, the 250R cathode resistor got very hot and the output started cutting in and out. I wasn't sure if the TW preamp was hotter, or that the 250R cathode resistor didn't like the 1k/5w screen resistor I put in. I then replaced the very discolored 250R cathode resistor with a 470k/5w "bathtub" resistor. With the 470R cathode resistor, the amp seems to be very stable using 5k6 screen resistor but at high volumes it cuts in and out with the 1k/5w screen resistor. I like the brighter sound with the 1k/5w resistor, so I am next going to try a 2k7 screen resistor to see if that is stable.

Steve

P.S. I know that repeating the wattage of the screen resistors is kinda stupid but for me it is easier to read "1k/5w" than "1k" which kinda blends in with the text.

[Gingertube]

Steve,
Those Cathode Voltages you list above have me worried.
59V with 470R => 125mA, 272V Anode Volts => 272-57=215V across tube => 215 x 0.125 = 27 Watts dissipation, A bit high
51V with 470R => 108mA, 280V Anode Volts => 280-51=229V across tube => 229 x 0.108 = 25 Watts dissipation, better

Those cathode voltages (i.e those cathode currents) are way higher than I would expect. I think you may have another problem.

Check that Grid1 of the output tube is at 0V DC wise. If it is higher than say +2V at idle then you might have either:
1) a leaky coupling cap dragging the GRid1 volts positive
2) too high a grid1 to 0V resistance (grid stop + resistor to 0V) such that the grid1 is going positive with grid current.

If you want to put a 6V6 in at some stage then 270 Ohms will probably be OK but I would check carefully for "red plating". 300 Ohms would be safer.

Note that your measurements show that screen voltage is more important than anode voltage in determining the tube current. That is as it should be.

The screen resistor RECOMMENDATION for an EL34 in SE is 2K2, for 6L6 I would go for the same value.

When you say the amp is cutting in and out, do you think that is just blocking distortion or do you think there is something else happening. Once again checking grid 1 voltage and cathode voltage will show you if it is blocking distortion. With blocking distortion, you will see either or both :
1) grid1 being driven negative such that the tube cuts off then slowly recovering
2) cathode going more positive (such that grid to cathode voltage is more negative) such that the tube cuts off then slowly recovering.
This is best seen using an oscilloscope but a mukltimeter may show it.

You said that you have been modding/swapping preamps, be aware that you can also get blocking distortion in a preamp.

Cheers,
Ian

[Steve A.]

I wrote up an EDIT to my last post but forgot to post it! My bad!

EDIT The PicoValve is sold as a 5 watt amp (with a Triode mode that is supposed 2 watts but sounds much lower than that to me). So I have a hunch that the PT isn't up to handling something like 18 watts (which the Dark Star probably can). My Epiphone Galaxy 10 amp uses a SE 6L6 but it is only rated at 10 watts (I put a big bottle KT66 in there just for kicks and figured it brought the wattage up to 12 watts). [End of EDIT]

The amp first started cutting in and out after I played it full bore for 15-20 minutes. Turning the Gain down to Clean settings didn't really help- the amp was sick! A quick look showed that the 250R cathode resistor was badly discolored, probably caused by me replacing the 5K6/2W screen resistor with 1K/5W. I then replaced the 250R resistor (still reading 250R when cool) with a 470R/5W but left in the 1K/5W screen resistor. Although the cutting in and out was gone, the notes were not sustaining as long as they should (possibly high frequency oscillation in the preamp.) I then replaced the 1K/5W resistor with 5K6 and also toned down the preamp by replacing the 100K shared plate resistor for V1a/b with 56K and added a 51pF snubber cap between the shared plates and grids. After doing all of that the amp was very stable, but it had lost that edge that I liked so much. So I went back again to my workbench (some people would call it a dining room table) and tried out a 3K3 screen resistor and a 33pF snubber cap, and at this point it seems to be working perfectly. I just need to pick up a 2 or 3 watt 3K3 resistor to replace the 1 watter I put in temporarily.

I will check the things that you mentioned if I continue to have problems. I don't think that there is voltage on the grid of the 6L6 because it is right after the tone stack which feeds the MV which then runs into a 0.022uF coupling cap. And I obviously need to study up the theory behind guitar amp circuits. I was going to go browse through Randall Aiken's site if its still up.

Steve

[jmaf]

2cents: I've experimentally hooked up the screen to a larger resistance in the past, the amp went really compressed, and lost power. Everything I played sounded like country, I was proud, because I've never been able to play country.

Then when switching it back, the current went from 76mA to 120mA for 2 x 6L6 - glowing red eyes stared back at me from the back of the amp. So yeah, this matches your experience there: if you limit the screen current by increasing its resistor, then you can reduce the cathode resistor - and apparently vice versa.

/endof2cents

[Steve A.]

I checked the plate, screen and cathode voltages of the 6L6GC with the latest (final?) choice of resistors. 3k3 screen resistor and 470R cathode resistor:
Plate 278v
Screen 248v
Cathode 54v

I will see if I can do the math...

Steve

[Gingertube]

Steve,
54 Volts across the 470 Cathode resistor is still 115mA idle current.
Make sure you are measuring at idle with no signal applied
If so than there is something sus..
Either
1) The tube is seriously old and its gm has dropped so far that it really needs 54 volts of bias (actually if that bad it really needs chucking in the bin )
OR
2) the grid1 is not at 0V

Sorry to bug you again with this but 54V bias with about 200V across the tube should be giving you about 1/10th that idle current.

Cheers,
Ian

[tubeswell]

...I was asking about the theory behind the screen and cathode resistors in a SE output stage. In particular how Andy Marshall manages to get by using a 250 ohm cathode resistor while most schematics show a 470 ohm resistor for 6L6's. Would it have to do with the value of the screen used- 5.6K instead of 1k/5W? Or is it because the voltage on the plate is fairly low- 270v?

In short, yes its because of the low plate voltage.

In long, its the plate-to-cathode voltage, multiplied by the plate current which gives the plate dissipation. For Class A operation, you want 100% plate dissipation. So if the plate voltage is lower, then the cathode resistor (in a cathode-biased stage) will have to be lower to compensate (i.e.: to allow more current to get to the plate). Otherwise the amp'll sound sicker.

The current flowing through the cathode is the 'tube current', which is the sum of the 'screen current' plus the 'plate current'. So if you are calculating the tube current (from the cathode voltage divided by the cathode resistance), then you've got to subtract the screen current to get the plate current. You can calculate the screen current the same sort of way you calculate the tube current (to get the screen current, you measure the voltage drop across the screen grid resistance. voltage divided by resistance equals current).

So tube current = 54V/470R = 0.1149A (or 114.9mA)

Say you've got 5mA screen current (assuming a 6L6), then the plate current must be ~110mA

0.110A x (278V-54V*) = 24.64W, which is about bang on for a 25W tube

*plate-to-cathode voltage

[Steve A.]

Sorry to bug you again with this but 54V bias with about 200V across the tube should be giving you about 1/10th that idle current.

You're helping me, not bugging me- I will recheck voltages with a different meter and a different tube (and check g1 for DC voltage). I remember having problems measuring B+ with some of my meters back in the old days. I will also do some measurements with a cathode resistor around 275 or 300 ohms.

Thanks!

Steve

[Steve A.]

Damn- 2.86vdc on the grid! I hadn't been using my Fluke 16 HVAC meter because the battery was low (I thought that the battery icon meant that the battery was GOOD!) and it kept giving me goofy voltage measurements although ohms and microfarads seemed to be okay. Turning on the amp without warming it up gave me 280v plate/250v screen/54v cathode using the Fluke.

I need to find out where that DC voltage on the grid is coming from! (Yes I call g2 the grid and g1 the screen although they both are grids. With g1 and g2 I'd forget which was which- like when you use the bathroom, which is #1 and which is #2? )

score so far:
Southern Hemisphere: 2
me:-1

Thanks! You guys are really sharp. Like me when I used to be an HVAC service tech...

Steve

[tubeswell]

Damn- 2.86vdc on the grid! ... I need to find out where that DC voltage on the grid is coming from! (Yes I call g2 the grid and g1 the screen although they both are grids. With g1 and g2 I'd forget which was which- like when you use the bathroom, which is #1 and which is #2? )

If you have DC voltage on the signal/control grid (g1) then the coupling cap feeding the grid is probably leaky, and this would be contributing grid current (via the coupling cap and the previous stage's plate supply) which could well manifest itself as higher cathode current on the output tube, thus affecting the bias (since the current being 'stolen' from the previous stage's plate would be flowing through the coupling cap and through the output stage's cathode). Lift the the 'grid-end' of the coupling cap from the board and re-check the cap-to-ground DC voltage. It shouldn't have any DCV on it.