Can you help me understand this phase-inverter section (Matchless guitar amp)

[Mbira]

I keep coming back to this Spitfire clone because I love the sound so much.  I want to better understand what is going on.  So specifically, what is going on with the bottom half of that 12ax7 and v4?  I know that usually the negative feedback would be feeding that tube, but not here, so I don't understand how that V4 is getting any signal?  Please forgive my ignorance here!

 

[PRR]

> I know that usually the negative feedback would be feeding that tube,

Not "feeding it signal", but "feeding a correction signal".

> I don't understand how that V4 is getting any signal?

It is a Long Tail Pair. Any signal to -either- grid gets to BOTH plates.

Signal at V4a grid upsets V4a grid-cathode voltage, changes V4a current, result appears in V4a plate, of course.

But signal at V4a -also- appears at V4a cathode, across the very large 48.2K total cathode resistance.

So V4a cathode bops up and down.

And since they are tied together, V4b cathode bops up and down.

Meanwhile V4b grid is held steady by the 0.01 cap (which is too small but skip it).

So V4b grid-cathode voltage MUST vary, V4b current varies, a signal appears at V4b plate.

And if you keep track of all the phase relations, V4b plate swings opposite to V4a plate.

And under reasonable conditions, the two plate swings are near-equal.

You can instead see V4a as a Cathode Follower (which also has a plate load), and V4b as a Grounded-Grid stage. We don't find many GG stages in audio; this is one.

If you know any transistor opamp topology, the input-pair works the same.

Details.

To get decent coupling from V4a to V4b, the common cathode resistor must be "much" larger than the cathode impedance. Small-triode cathode impedance is roughly 1K. We need 10K to 50K resistor to get good coupling and near-equal signal at both plates. "Long Tail". (The image is clearer if you draw the triodes side-by-side and have the one tail-string coming down between.)

If we just tied grids through 1Meg to ground, as we usually do, the couple-volts cathode-grid across 10K-50K resistor would give uselessly low tube current. We normally use a ~~1K cathode resistor. We have to hold the grids at some large-ish positive voltage to get useful current. Some amps have a +75V bias-rail, some DC-couple to the previous plate biased very low. Premier and later Fender used this trick. Take the usual ~~1K cathode resistor, double-down to 500R for two tubes, and connect the 1Meg grid resistor to the bottom of that. Tube bias is OK. Now stand the whole shebang on your 10K-50K resistor. Bias stays OK but V4a-V4b coupling is shunted by the "high" resistance instead of a low resistance.

If you let V4b grid follow the common-cathode node, it gets no signal and makes no output. That's why you have the 1Meg to bias and 0.01uFd to ground (or NFB), to hold V4b grid steady (or a correction signal).

The 48.2K resistor "wastes" a lot of voltage, in a stage which could be the highest-signal point of any stage except the 6BQ5 plates. Here it wastes 62V of an available 310V. Since you need a specific signal to flog 6BQ5 grids (here 11.2V), "wasted" supply means higher distortion, in a traditional single-triode stage. But the Long Tail Pair has many of the virtues of Push-Pull. 2nd harmonic distortion largely cancels. This stage will make as much drive as a gain-stage plus cathodyne, with less 2nd. There's a 3rd but the signal level gets quite high before that hurts. And with less-sensitive tubes than 6BQ5, it is easy to "lame" the LTP so that it symmetrically clips both sides at a point 120%-200% higher than output grid overload point, so you can over-load without going into grid-blocking (or no further than you want to go). (That's not true here: even with all the slop, you can get 50V at each V4 plate, four times more than 6BQ5 needs to get hammered, and enough extra to grid-block all to hell.)

 

[PRR]

The bias conditions around the 6BQ5s look way too hot. My thumbs say 17W Pd.

11.2V across 120R is 93mA total cathode current.

6BQ5 Screens will pull as much as 15% of plate current, so plate current is 81mA, 41mA per plate.

336.8V plate-cathode.

337V*41mA= 13.8 Watts idle Pd.

OK, I'll accept that for a stage-amp, IF you measure and verify it is no higher. Any healthy recent EL84 will stand 14W Pd... not for the entire life of your grandmother and her radio, but many-many-many paying gigs. Also, the louder you play, the cooler the tube run, and small stage amps average well over 10% max-power on average. You'll be under the 12W rating most of the night.

 

[Gus]

At the older Ampage site the Lightning schematic and matchless output transformer was posted about was measured and it measured a 4K at the P for two el4,6bq5 in a lightning,"normally" you find 8K to 10K for two el84s.

 

[Mbira]

Thank you PRR.  I'm going to sit down for a good while and digest your info.  As far as running too hot and that sort of thing, I don't know.  What I do know is that I've had this amp for I think 6 years now in constant gigging use.  It has performed wonderfully!  It has been the most stable and consistent amp I've had.  It is a tad bright, and so I have the tone fully counterclockwise, but I may install a different value pot.  I've tried a few different cap values in the tonestack with no real noticeable difference.  For whatever reason, the response out of this thing is amazing.  I have had more than one "big name" guitar players plug into it and tell me it is the best sounding amp they have ever played.  It is a 'one trick pony', though, but I'm a one trick rider so it suits me well.

 

[Mbira]

Hi Gus,
I'm using a Hammond 1650E, which is 8K center tapped into an 8 ohm one 12".

 

[dabo]

Check out http://www.aikenamps.com/LongTailPair.htm

 

[Mbira]

That's the exact article I was reading before I posted this because I was confused about the lack of NFB.
;D

 

[analag]

Meanwhile V4b grid is held steady by the 0.01 cap (which is too small but skip it).

The size of this cap is important in matching the phase inverted frequency resp. coming off the plates.

 

[Mbira]

That adds to my confusion. ;D  In the Aiken article, he states:

If C1 (the one feeding the first grid) is made small (less than .01uF or so, with 1Meg grid resistors), it will improve the low frequency response balance between the two output phases if the second coupling cap, C2 (the cap in question), is made at least ten times larger than the first cap, C1.

But here they are the same value...will I blow stuff up if I start tweaking that value?  I was thinking of trying to make C1 .001 and keeping the second at .01

 

[abbey road d enfer]

That's the exact article I was reading before I posted this because I was confused about the lack of NFB.
;D

Several tube guitar amp designers  advocate NOT to use NFB, the main argument being that guitar amps being meant to be overdriven, they have better recuperation. Recuperation is how the amp returns back to normal after being overdriven. Some amps tend to drift seriously off their operating point when overdriven and take some time to return. These designers think that an amplifier with NFB does not recup as quickly as one without, attributing that to the fact that when the amp is out of wack, the error voltage is very large and it overloads the stage where the NFB returns, leading to the amp going out of wack in the other direction, creating another large error voltage, and so on... This is a debatable subject, very hard to SPICE.
Although almost all Fender amps of the Leo era haf NFB, one of the new designers at Fender, Zinky, is a proponent of this scheme; his amps sure sound good, but not significantly better, nor different than those with NFB. The Studer amp (A78 IIRC) was designed according to this concept, with very little global NFB (but a number of nested FB loops) and the results were superb. Was it because of that or the general design and good choice of components and superb build quality?

 

[walter]

I understand the P.I.plate resistors in the fender blackface had one 100k and one 82k, this was to set it off ballance slightly for a meaner tone. I'm curious how that Master volume would work with a setup like that. How do you like the master volume? It works by phase cancelling the signal. Try a .002 for the signal, and a .1 to ground.

 

[Butterylicious]

Joel,

You need to come over and have an amp building party.  I've been getting medieval w/ 6BQ5's lately. 

 

[nickt]

I understand the P.I.plate resistors in the fender blackface had one 100k and one 82k, this was to set it off ballance slightly for a meaner tone.

Nah - it was actually to make it MORE balanced! Rock over to the ampgarage.com and check out the Dumble forum for more info.

 

[Mbira]

I understand the P.I.plate resistors in the fender blackface had one 100k and one 82k, this was to set it off ballance slightly for a meaner tone.

Nah - it was actually to make it MORE balanced! Rock over to the ampgarage.com and check out the Dumble forum for more info.

That Aiken article above discusses that issue...

As far as the master volume-yeah I like it.  I love this amp! 

 

[Mbira]

I'm seeing that what I am lacking is an understanding of cathode followers...more studying now.

 

[Mbira]

I found use in this article:
http://300guitars.com/index.php/articles/article-demystifying-the-phase-inverter/

 

[PRR]

> But here they are the same value

Yeah, that don't smoke neither.

The input grid only has to be driven "sorta good". The off-side grid must be NAILed to reference; even a slight signal here upsets your balance at lower frequencies.

Which is NOT bad for guitar. That isn't about "perfect balance". You could try 0.1u, but the 0.01u may be giving a phatt tone on the lowest guitar notes which may be good.

I have no idea what Analag's point is.

> guitar amp designers  advocate NOT to use NFB

There's many sides to this. A BIG point is what kind of speaker you use. Naked pentodes driving open-back cabinets with stiff cones are liable to bass-slap. I think the original point was to add just enough damping to get the bass-peak down so it did not make rude noises on your lowest note. Later fashion in Hi-Fi design and also cabinets tends to favor more NFB, right up to early Sunn with Hi-Fi damping. Of course transistors damp great and clean, and there is some rediscovery of loose no-NFB amps.

Make the NFB variable. Understand that the "best" degree of NFB depends on your speaker-rig, as well as your style and religion. I tend to worship high-NFB, but I must admit the zero-NFB Ampeg VT-40 is a heck of a guitar voice.

> P.I.plate resistors in the fender blackface had one 100k and one 82k, this was to set it off ballance slightly for a meaner tone.

They "improve" balance. The nearside plate makes more signal than the offside plate. The 82K reduces the output of the nearside. The overall NFB into the tail also seems to improve balance, though it was probably a mistake which worked good.

http://300guitars.com/index.php/articles/article-demystifying-the-phase-inverter/

Yes, a good re-drawing, simplified. The drawing lacks stuff to get a good operating point. The text mis-states the reason balance is flawed... at least using Fender-like values, the imbalance is because the tail resistor diverts signal current. But it is NOT useful to think too hard about gitar amp balance. PUSH-pull works fine, and may be sweeter than exact-balance push-pull.

 

[Mbira]

Thanks for all your help...before last night, I didn't even know you could take a signal off of the cathode...It's amazing the little things I miss and the amount I still have left to learn here!