6SJ7 for input stage - values?

[EFK]

I have a socket wired up for a 6SJ7 to use in a two channel amp - both channels use the standard Fender brown deluxe LTPI, although the amp is cathode bias. One channel is the standard 12ax7 first stage, but I'm trying to implement a 6SJ7 for the first stage in the other channel. I have a GA20 w/ these tubes, but that's a grid leak bias and I want to cathode bias this one. Looking at GA25, GA30, GA50 etc. schematics, values are all over the map! I have B+ feeding this tube at 306 VDC. For the moment I have a 220K plate resistor giving 193 V at the tube plate, a 2.2K cathode resistor bypassed w/ 25 uf for 1.5 V, and a 2.2M screen resistor w/ .05 bypass to ground. This gives a screen voltage of only 25.5 VDC. Input topology is 33K stopper /1M to ground, tube output currently is through .02 cap into a 1M volume pot. Here is my issue - I have read in a few other places that for this tube the screen should be 2/3 voltage of the plate. Obviously, I am quite far removed from that. As this tube is capable of a tremendous amount of voltage gain, which I *don't want* being as it's feeding the brown deluxe second stage, I'm wondering what effect changing the screen voltage will have. I know most of the old Gibsons ran the plate voltage a lot lower than 193 V. Maybe I should up the plate resistor to 270K, and lower the screen to 1M? I'm out of my element here and I'm not clear on how changing values will affect gain. Just looking for some advice, anecdote, experience, thoughts etc. Anyone with ideas on how changing these resistor/cap values will affect things? Thanks!

[overtone]

Still being new to all this I can only offer you some sample voltages from a 1954 GA-40 LP with three 6SJ7s that I am currently finishing sorting out. Please do not take these voltage values as gospel – they look very low to me! As I understand it, this is not a grid leak input. When I got this amp there was no headroom and it only distorted, I think because the B+ was way too low.

The 6SJ7 plate resistors have drifted from 270K to a little over 300K, I have not replaced these.
The 6SJ7 cathode resistors have drifted from 1K to a little over 1K5, I have not replaced these either.
The V2 and V3 6SJ7 screen resistors have drifted from 1M to 5M, so I decided to change these.

On replacing those screen resistors the V2 and V3 plate voltages dropped by about 110V to 30 and 40 volts respectively!

The 6SJ7 voltages seem way too low, but I have thick cleans and it breaks up perfectly when I dig into the strings. Just how I like it. It is one of those amps with so much sustain you can go and drink a beer between notes.

Yesterday I tried putting a 2M2 pot, in series after the screen resistor on V2, to see if that made any difference, perhaps more distortion? The feel and response was better at the 1M setting, so I think that was a good call to change out the screen resistors.

Click the images below for a quick schematic of the input section typology and the voltages as measured today.


Please let us know how you get along and what voltages you end up using,
all the best

[dchang0]

You need to draw a load line for your own particular tastes. Go to Merlin's website at How to design valve guitar amplifiers and read the first chapter of his preamp book, which is available for free in PDF form, if you need help drawing the load line. He goes through the deliberate selection of specific component values, step-by-step, in plain everyday language.

I like tons of clean headroom, so I draw different load lines than the "default" or "commonly-used" ones.

[EFK]

I'm still messing around with this - I;'ve had some wicked 120 hz hum problems to work out, and it's tough to simply star ground something when it's an old two-chassis shell I'm working with (power amp in the bottom of the cab, then cable up to preamp in top section). I've got it mostly worked out - the straight 12ax7 6G3 channel is pretty quiet. The 6sj7 is noisier, but in my opinion after having messed with some old Gibsons, the 6sj7 is an inherently noisy tube.

Anyway - I played around with some screen resistor values, and as a preliminary conclusion, I think the voltage on the screen doesn't seem to effect much other than compression. As I originally had it set up, it was quite gainy and quite compressed overall, basically as soon as you began nudging the volume up. Some serious gain. I paralleled some different screen resistors in, and while different values don;t really seem to effect the overall sound that terribly much, more voltage on the screen lowers the plate voltage but seems to de-compress the sound a bit. Perhaps a tiny bit less gain, hard to notice. But the difference in compression and feel is quite noticeable. I've settled on the original 2.2M resistor paralleled with another 2.2M, so the screen resistor is effectively 1.1M. This still seems in line with some of the old Gibson amps - I notice between the GA20, GA25, GA30 and GA55 they varied the plate and screen resistor values quite a bit. right now I'm experiementing w. different tubes - the metal case RCA tubes are quieter in terms of hum but are more microphonic, while the glass JAN tubes are less microphonic but more hum. I make up little aluminum tube shields for the glass tubes which ground tightly to the metal strip around the base, though, and this cuts the hum way back. Ugly, but effective.

[dchang0]

It's good that you're going the "empirical route" by playing with values firsthand. However, a little theory goes a long way in what you're doing, so it may behoove you to go read Merlin's article.

As for the hum, I find that Aiken's article on grounding is the best I've seen. Better than Hoffman's, I think, because whenever I followed Hoffman's diagram, I still had hum problems. Pay special attention to the way Aiken explains bus grounding and also to the best "local star ground points." Once I learned to break apart my layouts into locally-grounded circuits, I was finally able to eliminate all the hum in my builds, and I thank Randall Aiken for explaining it so well.

[octal]

If you want to reduce the gain, you can leave the cathode resistor unbypassed.

Reducing the value of the plate resistor will also reduce gain, but obviously you will have to re calculate your operating point and also adjust the cathode & screen resistors.

I built a preamp cribbed from the Ampeg Dolphin II schematic- 100K plate, 470K screen, 560 ohm cathode running from a B+ of 250-300 v and it seemed to work pretty well (it was in an oddball amp with a transformer PI, so I don't know exactly how it relates to your scenerio.)

If you're not up to calculating load lines and such, you can also look up the old R-C amplifier charts in a tube manual. That will list various operating points and gains you can run them at as sort of a 'cookie cutter' approach.

Good luck,
Nathan

[ThermionicScott]

If you look in the "Resistance-Coupled Amplifiers" section in the back of the old RCA manuals (RC18 and RC19 for example), there are lots of example set ups.

- Scott

[surfsup]

Click the images below for a quick schematic of the input section typology and the voltages as measured today.
[ATTACH=CONFIG]12203[/ATTACH]
[ATTACH=CONFIG]12204[/ATTACH]
Please let us know how you get along and what voltages you end up using,
all the best

the difference in those two input stages is a coupling cap value. I would save a tube and just use a bright switch to go from .02 to .01 and back...

[jazbo8]

If you look in the "Resistance-Coupled Amplifiers" section in the back of the old RCA manuals (RC18 and RC19 for example), there are lots of example set ups.

- Scott

This is a very good suggestion, had forgotten how easy to design with the RCA Tube Manuals... The capacitors set the low frequency roll off at 100Hz, which is about perfect for guitar amps, so you can use them as is without change (unless you want to of course).



Jaz