Marshall 20W Phase Inverter

[Leevi]

Can anyone explain the idea of the PI of the Marshall 20W Lead amp.


It differs from the traditional PI used in the Marshall amps for instance what comes to Grid leak resistors.

http://el34world.com/charts/Schematics/files/marshall/Marshall_20w_2022.pdf


/Leevi

[pompeiisneaks]

Looks like a DC long tailed pair to me, Merlin explains it better than I could: http://www.valvewizard.co.uk/dcltp.html

~Phil

[Leevi]

Thanks for the link Phil
There is a good explanation for it.
/Leevi

[DummyLoad]

that LTPI is not DC coupled. follow the signal path: there are two .022uF coupling caps. as typically implemented AC coupled there is another cap after the summing resistors (680k pair) and a Vdivider from B+ to ground for ~1/4 B+ Vreference to bias the stage. as presented the circuit is flawed - not saying it doesn't work, just works flawed.

--pete

[pompeiisneaks]

sorry I meant the AC Coupled Long tailed pair :)  http://www.valvewizard.co.uk/acltp.html

but oddly it seems more on the cathode side like the DC form, but I digress.  You're 100% right it's got the coupling caps that block the DC coupling :D my bad.

~Phil

[Tony Bones]

I don't see how the amp can work well as drawn. The PI grids are ground referenced through the volume controls with a shared 8.2k cathode resistor. I suppose it would make sound, but clean would stop at 0.5 on the volume knob and just get dirtier from there.

I suppose the drawing could be wrong and the 0.022 coupling caps are really under the volume pots, but that would put some high DC voltage on them, which seems like a bad idea.

[PRR]

> don't see how the amp can work well as drawn

I think it will work.

I do suspect it is a mis-wire mix of other circuits.

V2 cathodes at 3.5V, in 8.2K, is 0.42mA. Half that in V2a's 100K plate resistor makes 21V drop, <21V peak swing.

However the power bottles bias with 12V on cathodes, do not need >12V of drive. So V2a can drive V4 OK.

V2b gets V2a's dynamic swing, minus what sneaks out through the 8.2K. At 0.21mA, the Gm is maybe 500uMho, or like 2K resistance. So V2b takes 80% of dynamic current. But V2b plate resistor is 50% larger, 150K. So at a glance it seems that V2b will also be able to smack its V3 power bottle. Balance probably not 50:50 exact, and do we care?

A particular point is that the output bottles will not grid-block as much as some other amps. It won't "faint" as much after a HARD strum.

[PRR]

> 0.022 coupling caps are really under the volume pots

That would cut treble but not bass. "Zero" would "boom" with no zing. When you put a cap under a volume control you want it BIG. So big that you hardly ever see this done for a Volume control. (It can be a Loudness control for home hi-fi, but there's better ways, and this is no hi-fi amp.)

[Leevi]

It looks for me still like the DC Coupled Long-Tailed Pair
http://www.valvewizard.co.uk/dcltp.html

I'm going to build that amp, let's see how it works and if there is need to change the PI.

/Leevi

[Tony Bones]

> 0.022 coupling caps are really under the volume pots

That would cut treble but not bass. "Zero" would "boom" with no zing. When you put a cap under a volume control you want it BIG. So big that you hardly ever see this done for a Volume control. (It can be a Loudness control for home hi-fi, but there's better ways, and this is no hi-fi amp.)

OF course, you're right. Not sure what I was thinking. 

[2deaf]

What if you put a 12AT7 in there instead of the 12AX7?  It would take more signal to drive it, but it would swing a bigger signal.  Yeah, I know, a 12AX7 in there can muster enough voltage to drive the EL84's, but just barely.  The signal from the P.I. is probably rounded-off before the EL84's clip, at least on one side.  It looks like there is plenty of signal coming off of V1 to drive a 12AT7 P.I. even with the worst case voltage divider.  The 12AT7 P.I. output should be far from rounding-off at the EL84 clipping point.  Of course, this may not be what you are trying to achieve.

Why is the 1M resistor on pin 2 (grid 1) going to pin 7 (grid 2) when there is a 1uF cap. there that is going to shunt all signals to ground?  Sure, it references grid 2 to ground, but why not connect the 1M resistor on grid 1 to ground and hook grid 2 directly to ground eliminating the 1uF cap.?

   

[PRR]

> The signal from the P.I. is probably rounded-off before the EL84's clip

I say it is adequate, just-barely. And just-barely may have been a design goal, so it won't grid-block to near-silence after a hard strum. But I am not inclined to build it and see, even in simulation.

> Why is the 1M resistor on pin 2 (grid 1) going to pin 7 (grid 2) when there is a 1uF cap.

Indeed. I think this plan "just happened", and nobody really thought it through.

[Leevi]

> Why is the 1M resistor on pin 2 (grid 1) going to pin 7 (grid 2) when there is a 1uF cap. Indeed. I think this plan "just happened", and nobody really thought it through.

I refer here to the statement in http://www.valvewizard.co.uk/dcltp.html

The second grid must also be at exactly the same DC potential. This is applied from the first grid via a high value grid resistor or "pull-up resistor" (Rg), usually the same value as the grid-leak in a normal gain stage, or 1Meg in this case.

/Leevi

[Tony Bones]

Well, yeah. They copied a direct coupled PI even though it's not direct coupled.

Study enough schematics and you'll eventually find some head scratchers. Like Fender using opamps to drive relays in the deluxe / deville series.

[pompeiisneaks]

Well, yeah. They copied a direct coupled PI even though it's not direct coupled.

Study enough schematics and you'll eventually find some head scratchers. Like Fender using opamps to drive relays in the deluxe / deville series.

I've got one, and I didn't recall seeing that, they had opto couplers doing that no (the relay switching, used to decouple the relays power side from the rest of the system)?  The opamps were used to drive the reverb and recover it. 

~Phil

[2deaf]

I refer here to the statement in http://www.valvewizard.co.uk/dcltp.html

The second grid must also be at exactly the same DC potential. This is applied from the first grid via a high value grid resistor or "pull-up resistor" (Rg), usually the same value as the grid-leak in a normal gain stage, or 1Meg in this case.

The first grid's DC potential is zero as in ground.  Grounding the second grid would put it at the same potential as the first grid.

They just copied a DC coupled splitter without understanding how it works, sort of like that 59 Bassman thing.

[DummyLoad]

for AC coupled see attached. if there is a tone stack, level control, etc. then insert BEFORE C2. how marshall should have built it. 


--pete

[Tony Bones]

Well, yeah. They copied a direct coupled PI even though it's not direct coupled.

Study enough schematics and you'll eventually find some head scratchers. Like Fender using opamps to drive relays in the deluxe / deville series.

I've got one, and I didn't recall seeing that, they had opto couplers doing that no (the relay switching, used to decouple the relays power side from the rest of the system)?  The opamps were used to drive the reverb and recover it. 

~Phil

Here's an example of what I'm talking about. Everything below the first two gain stages and the tone stack is just to drive two relays.

It takes a dual rail supply (+16V and -16V) and a bunch of saturated opamps when a single supply and two $0.25 transistors would have worked just as well. Maybe I'm dumb and I just don't see why they did it that way, but it looks to me like the designer had no idea how to drive a couple of relays and wasn't smart enough to copy a circuit that they know works. (Something every good engineer does when they don't know what else to do.)

http://el34world.com/charts/Schematics/files/fender/Fender_BLUES_DELUXE.pdf

[pompeiisneaks]

Well, yeah. They copied a direct coupled PI even though it's not direct coupled.

Study enough schematics and you'll eventually find some head scratchers. Like Fender using opamps to drive relays in the deluxe / deville series.

I've got one, and I didn't recall seeing that, they had opto couplers doing that no (the relay switching, used to decouple the relays power side from the rest of the system)?  The opamps were used to drive the reverb and recover it. 

~Phil

Here's an example of what I'm talking about. Everything below the first two gain stages and the tone stack is just to drive two relays.

It takes a dual rail supply (+16V and -16V) and a bunch of saturated opamps when a single supply and two $0.25 transistors would have worked just as well. Maybe I'm dumb and I just don't see why they did it that way, but it looks to me like the designer had no idea how to drive a couple of relays and wasn't smart enough to copy a circuit that they know works. (Something every good engineer does when they don't know what else to do.)

http://el34world.com/charts/Schematics/files/fender/Fender_BLUES_DELUXE.pdf

Oh yup forgot about that, I was thinking about the top right section with the TL082's used for the reverb.  That is silly heh.   (I think i'm thinking of a Splawn I worked on recently where they used optocouplers for the entire switching networks. some cases they enabled/disabled a ground, others they triggered the relays.)

~Phil

[Leevi]

for AC coupled see attached. if there is a tone stack, level control, etc. then insert BEFORE C2. how marshall should have built it.

Would the following be a suitable mod maybe with bigger grid leak resistors?

http://www.geocities.jp/e08120/M2061X/PA20-MOD-SC-ALL.jpg

/Leevi

[DummyLoad]

for AC coupled see attached. if there is a tone stack, level control, etc. then insert BEFORE C2. how marshall should have built it.

Would the following be a suitable mod maybe with bigger grid leak resistors?

http://www.geocities.jp/e08120/M2061X/PA20-MOD-SC-ALL.jpg

/Leevi

not a clue.

this circuit needs a DC reference at the input grid for proper bias: how does the proposed mod provide that?

--pete

[Leevi]

The idea of this mod was to change the PI like done in the 2061.


http://music-electronics-forum.com/attachments/6463d1254375898-marshall-2061-reissue.pdf


/Leevi

[pompeiisneaks]

If you're thinking of doing about a 20W marshall, could you just do Hoffman's 18W plexi?  Or is the preamp of this circuit different and you want to give that a try? 

~Phil

[Leevi]

If you're thinking of doing about a 20W marshall, could you just do Hoffman's 18W plexi?  Or is the preamp of this circuit different and you want to give that a try? 

The decision of the circuit is not in my hands


/Leevi

[pompeiisneaks]

Oh is this an existing one you're working on then I guess?  Are you thinking or needing to mod it to be more 'sane' or just generally curious about it?

~Phil

[2deaf]

The idea of this mod was to change the PI like done in the 2061.

Once again, I fail to see how a 220K in parallel with a 100nF cap. references pin 7 to ground any differently than a straight piece of wire.

[Leevi]

I got the amp wired. I did it according to 2022 (a better version of the schematic below).


IMO it sounds like a Marshall regardless if the PI is correctly designed or not. I don't see any reason to change the PI.


http://marstran.com/2022.gif


/Leevi