Help w/ phase inverter re-design

[Geezer]

I want to make sure a PI is optimal for the most clean headroom/power.

The PI uses the triode section of 6BM8 tubes, Datasheet linked here: http://www.drtube.com/datasheets/ecl82-philips1969.pdf

I have attached the schematic (as the amp is right now)....there simply isn't any clean there. It's not the preamp, as I have run different preamps into it with the same results.

Thanks in advance, G

[phsyconoodler]

Two things come to mind for me.Increasing the PI voltage and adding more negative feedback.That would give more headroom for sure.
 

[Geezer]

Increasing the PI voltage

Well, the B+ rail is already pretty minimal, so I'll have to lower the plate resistors to get the voltage up. Sounds like a place to start (I had actually thought of that, along the lines of the SilverFace Fender amps.....) I'm thinking maybe 33k on the plates(?)

Now, if I change the plate resistors, will I also have to change the value(s) of the cathode & tail resistors?

I will try the NFB if the increased voltage isn't enough.

Thanks

G

[Geezer]

OK, I knew I saw low plate resistors on a build somewhere...it was Sluckey's "November" build......will this plate resistor/PI Rk & tail value work for my 6BM8's?

http://home.comcast.net/~seluckey/november/november.pdf

[tubeswell]

Lowering plates resistors usually increases tube current - which lowers plate voltage. This should lessen the gain - which may result in less swing going into the grid(s) of the next stage(s), however it might also be more prone to grid current on the following stage(s) resulting in blocking distortion.

AFAICT the best thing is to get the B+ up, and bias the stage for bias point somewhere in the middlish of the  grid curves on the load line

Edit - I see it is a LTP inverter, in which case increasing the tail will reduce gain and allow you to run matched plate resistors, which should also help clean it up. Maybe try a 47k tail and 100k plate resistors on both sides and a 470R cathode resistor

[sluckey]

"November" build......will this plate resistor/PI Rk & tail value work for my 6BM8's?

I don't know about that, but the low plate resistors were used to lower the drive to the EL84s to prevent them from overloading so early. The idea was to get a little closer to the 1987 Plexi PA sound. I don't know how successful it was but I sure like that amp. Give it a try and see.

[RicharD]

I still walk on eggshells with PI's but I wanna take a stab at this.  Please feel free to shoot me down.

Looking at the data sheet, the typical characteristic given show Vplate=100 with an Iplate of 3.5mA so I'm assuming Rp to be 28.5k ish.  15k plate resistors might not work.  I could be completely wrong.

I loaded the 6BM8 values into Tube Cad (not a tube that comes with) and ran the values from your schematic.  The returned results (attachment1 1) look pretty durn close so I'll make my 2nd assumption and say I loaded it correctly.  Looking at the results, the input threshold looks super low, 0.79V and assuming gain is 24, that's only 19V out before your PI is clipping.  

I think the solution is to flow a LOT more current, like 10mA per tube (attachment 2).  You can get there by lowering there plate resistors to  33k and lowering the tail to 15k.  Here's where I start to question my design.  Vbias is only 71mV  .071/.02=3.5 ohms  Kind of a funky looking tail, 3.5 ohm / 15k  It might work fine with a 10 ohm resistor.  The benefit is an input threshold of 4V * a gain of 18 = 72V out.  The down side is the current hogging.

I ran 1 more scenario dropping current back to 14mA.  Input threshold = 2.8V and clean output is 51V which should do the trick and is > double what you currently have.    The tail would be 75 ohm over 15k.

I think the rub is being limited to 270V of DC supply.  More B+ = more output swing.  Working with what you've got, this is at least something to try.

You might notice that TubeCad calculates a slightly higher Ra2.  I think this is to compensate for the loss within the cathode coupling and create a more balanced output.  This is commonly seen in some fenders where the plate resistors are 100k and 91k.  Always one more hair to split.

-Richard

[RicharD]

>Maybe try a 47k tail and 100k plate resistors on both sides and a 470R cathode resistor

Yeah!  That looks good but I came up with a 220 ohm cathode resistor.  If you want clean, you gotta get that current up with such a low B+.

[Fresh_Start]

I still don't "get" the LTPI but have been collecting schematics of amps using 6BM8s.  Don't think any of these examples solve your problem, but just in case here are some "known good" variations.

Almost half of the 6BM8 push-pull designs I've got use a driver/concertina splitter topology.  Didn't think that would be helpful.

Chip

[RicharD]

Concertina?

[Fresh_Start]

Buttery - I don't know how to interpret your chart, but here are a couple of 6BM8 concertina circuits:

Gibson GA-8T

Are either of these helpful?

Chip

[Geezer]

Thanks guys....I won't be able to experiment with these ideas (in the real world amp) until Saturday.....

[PRR]

> there simply isn't any clean there

The longtail stage looks fine.

The stand-out oddity is V2B: 178V grid 176V cathode. This can't work clean. Change the 3.3K cathode resistor to 1K or less.... what changes?

[Geezer]

> there simply isn't any clean there

The longtail stage looks fine.

The stand-out oddity is V2B: 178V grid 176V cathode. This can't work clean. Change the 3.3K cathode resistor to 1K or less.... what changes?

Ah-Ha!...I'll lower the Rk & see if any improvement is evident.

Thanks!  G

[Tone Junkie]

Geezer did you get a chance to experiment with this on sat. Sorry I read everything that comes up, And a low watt version of this amp sounds good to. Bill

[jjasilli]

Here's a univox u45b schematic; maybe some ideas there:  http://music-electronics-forum.com/t17773/

[Geezer]

Geezer did you get a chance to experiment with this on sat.

Bill, I did not get the chance.....hopefully soon.

G