Long Tailed Pair w/o The Tail, Explanation?

[rp]

I'm looking at the Marshall 20W and curious why no tail resistor on the PI?

[pdf64]

The 8k2 is the tail resistor.
See http://www.valvewizard.co.uk/dcltp.html for analysis.
I think that there may be an error on the Marshall schematic, as the LTP input needs a dc reference above the 0V that it's got.
See http://www.vintagehofner.co.uk/britamps ... ust15.html for an example of that.
Or it may be that Marshall decided to bias the LTP really cold (perhaps to limit gain), and that it's grids really are referenced to 0V, with the 8k2 serving as both bias and tail resistor.

[rp]

The 8k2 is the tail resistor.

Got it, I should have said the (usual) bias resistor. I had read the V-Wiz in the past and just reread, and noticed the below passage, which helps - a bit. I think this explains the WEM too, though I have trouble wrapping my head this stuff. I need to change my question, why this way rather than the more typical Vox, Fender, 18W, with the bias + tail resistor?

Because of this large 'tail' resistor (or current sink) the differential pair is affectionately known as the "long tailed pair". There are several versions of the long tailed pair (also known as the Schmitt inverter), and the one requiring the least components is this DC coupled version.

I think that there may be an error on the Marshall schematic, as the LTP input needs a dc reference above the 0V that it's got.

All three 20W schematics (Lead, Bass, PA) are drawn like this. So is the Handwired Reissue (for $2K!!) which someone drew up a schematic for. It's the same as the originals. Also have a pict of an orig and it looks like the schematic.

Or it may be that Marshall decided to bias the LTP really cold (perhaps to limit gain), and that it's grids really are referenced to 0V, with the 8k2 serving as both bias and tail resistor.

Odd, as I've met people that much prefer the 20W to the 18W as the more aggressive amp. Since the 18W does have the usual Vox, Fender type with the bias and the tail resistor, it might be interesting to mod the 20W PI to the 18W PI and see what happens.

Thanks to whoever's pict that is.

[tubeswell]

This type of LTP is covered in the original edition of Merlin's preamps book FWIW (although the examples shown there are fixed bias)

[katopan]

Huh, you know I've never noticed that before. Interesting!

[jazbo8]

Guys, that circuit is as old as dirt, and came straight off the pages of the Mullard application notes and popularized by the 5-20 amp design...

[Malcolm Irving]

Good spot Jazbo. Here is a nice article about the Mullard 5-20 from the UK 'National Valve Museum':

http://www.r-type.org/articles/art-003d.htm

but there are some differences to the Marshall 20W. In the 5-20 the PI is DC coupled to the previous stage and its tail resistor is 82k rather than 8.2k.
(Which I think makes sense given the DC level from the previous stage.)

Maybe Marshall were 'inspired' by that circuit, but with their PI being AC coupled they thought 82k/10 = 8.2k - let's give that a try, and it worked!

[jazbo8]

Maybe Marshall were 'inspired' by that circuit, but with their PI being AC coupled they thought 82k/10 = 8.2k - let's give that a try, and it worked!

Yup, it was an easy day at the shop, just flipped through some magazines, changed a few values here and there and called it a day...

[Malcolm Irving]

The schematic for the Marshall 20W in rp's second post shows 4.5V on the PI cathodes. That is really 'cold' for a 12AX7. Even with the relatively high plate voltages at the PI (330V and 327V)
most 12AX7 will just about cut-off at that bias level.
The region of cut-off varies quite a lot among tube samples - so I think tube rolling in the PI position here could have significant effect.

[Malcolm Irving]

With the B+ of 365V at the PI, one triode is drawing (365-327)/100k =0.38mA and the other (365-330)/150k =0.23mA.
Pretty close to being cut off!
Check on the bias voltage: (0.38+0.23)8.2= 5V (OK maybe close enough to 4.5V)

[rp]

Rereading the articles made me understand the LTP better, but I'm still puzzled at the 20W. As much as I can "hear" a schematic the 18W, AC15, Matchless Spitfire, WEM, look like much better amps, which is strange as people really dig the 20W. I never played either, 18W'ers were always very rare and were always collector amps whereas players sought out the little 20W heads enthusiastically. Maybe it was only because the 20W'ers were so much easier to find and cheaper, but people raved about them.

So an other fortuitous guitar amp engineering mistake, like the Vox tone controls or the cold clipping stage? Maybe the 20W started out as only a PA amp where someone intentionally toned down the PI, later when they tried guitar it sounded cool, they stamped Lead on it, called it day w/o ever revisiting the circuit? The cold PI somehow gives the 20W the 50W Marshall grind people like about it over the 18W bluesy warmth?

People used to do a lot of gain mods to the 20s like cascading inputs when maybe all it needed was a quick change at the PI. If anyone is bored and has a 18W/Spitfire/20W or clone sitting around and wants to take half an hour to mess w/ the PI tail and report back, this post will still be here.

[Malcolm Irving]

With the quiescent plate voltages on the PI at about 330V, they have quite a limited upward swing towards the B+ of 365V as they progress further into cutoff.
Meanwhile output tubes are EL84 and are biased at say 12v (just a guess - it's not shown on the diagram). So the EL84 only need +12V swing on the grid to push them up to grid limited clipping.
The well known problems of 18 waters are cross-over distortion, 'buzz' and blocking, which arise from the PI putting too much swing onto the EL84 grids (positive going swing in particular).
So, just thinking aloud here, perhaps this very cold PI is a good way to drive the EL84s if you are happy with lower gain and want to avoid 'buzz' distortion.

[M Fowler]

I built a 2061 20w head and recently sold it. Good sounding circuit with some nice break up when pushed.

[roberto]

Fowler, do you recommend a MVPPI after that PI?

[M Fowler]

No I prefer to keep it as Marshall built it.