coupling caps n DC

[shooter]

I get this, “crack” sound when V2b plate hits clipping, (asymmetric top)
At the same time I get a momentary -5ish VDC on the right side of both coupling caps for V2, (v2a does not clip)
below clipping I get maybe -50mVdc on the meter.
I know the answer is change caps, but my ?; 

could the circuit “design” cause the -5vdc, like grid drawing current?
Bypass cap not handling the swing?

The “problem” was there pretty sure from the beginning, so the split load(v2b) was added “after”

[sluckey]

How does a 12AX7 behave in that circuit?

[shooter]

I tried an AT, but tube was bad, tried another AU, same, I should be able to plug in an AX sometime today n see.

[shooter]

After a day que’d up with the masses, I plugged in an AX7, then another, and another, after 4, plus the AT from last night, the results were, unusual!  All the non AU tubes had between 1 and 10vac noise floor, the AU’s 50mV!

I figured out the “noise” was >100khz, and into a speaker it was horrendous squeal, squelch.
The AU’s are quiet, like, oh S#$T I left the amp on.

I was able to test the DC at the caps, ~ -.5 max, generally mV range.  I installed a different AU than original, and same, -5vdc momentary at clipping, put the Fluke to min/max and it tripped out OL, but I was in manual range 0-6vdc.

EDIT: here's a link to the whole schematic;
http://el34world.com/Forum/index.php?topic=20705.0

also, when I tried looking at the 50mV noise of the AU, there is a high freq sine-ish wave there, but it's below my scopes ability to discern any intelligent info

I think I’m confused, maybe a couple more hours que’d up with help

[shooter]

I was using the guitar as source, I switched to sig-gen, same/worse since I could grossly distort the pre and the DC went up as top clipping got worse, 1st coupling cap 6vdc, 2nd, north of 14vdc.  Think i'll change caps, both are new orange-drops.

[shooter]

swapped both .047s but only had .02s on hand.  basically the same -vdc, only tested with guitar source but still got to about -3.5vdc.  Tone did change some, expected, cracking sound still there, but takes more drive to become apparent, 6ish on gain with orange caps, 7ish with mojo yellow.

could I be seeing/hearing blocking distortion?  I get the jist of what it is, just have no experience with what it looks or sounds like.

[shooter]

I went surfin and found this on pre-amp blocking distortion;

http://hiwatt.org/2_mod.html
the -14v reference seems to be what I'm experiencing

since I see the "effect" at output of V2a, do I go to V2a's grid and change the 120k (inside the TS)?

I didn't listen to the clips, my 'puter is mute at the moment!

[terminalgs]

At idle, what is the VDC of your plates, cathodes, and B+ supply?

[HotBluePlates]

... -5vdc momentary at clipping ...

EDIT: here's a link to the whole schematic;
http://el34world.com/Forum/index.php?topic=20705.0 ...

I went surfin and found this on pre-amp blocking distortion;

http://hiwatt.org/2_mod.html
the -14v reference seems to be what I'm experiencing ...

I've been meaning to reply for a couple days... Yes, you're almost certainly experiencing blocking distortion.  The tip-off is a grid which should be at 0v now has a negative voltage on it (and pretty big compared to the tube bias).

You're slamming V2-b with a too-big signal.  You have an output section which only needs ~15v peak, and also have 4x 12AU7 gain stages which should be able to manage a gain of ~12 per stage.

You've got a number of gain-loss elements.

• There's a Master Volume; I'll assume it's at max throughout the below.
• The split-plate-load in V2-b means you probably have a gain of 6 at that stage.
• I'll arbitrarily guess the James tone stack is -20dB, or a "gain" (loss) of 0.1.
• I'm not gonna analyze the impact of the 47kΩ and 0.01µFat the plate of V1-b, except to say it drags down the output of that stage (adding to the loss due to the tone stack) and probably causes V1-b to distort earlier. (requires loadline investigation, and I'm tired...)
• The Guitar channel volume is a loss of ~0.25 at a rotation between 7 & 8 (out of 10).
• The 470kΩ on the Guitar channel's wiper has a confused interaction with the Aux channel volume and its 470kΩ & 220kΩ resistors.  If the Aux channel volume is at max, the other resistors impose an additional loss after the Guitar volume of ~0.38.  But if the Aux volume is at 0, the loss is ~0.24.  The extra resistors don't provide the isolation you appeared to have wanted...

So working output to input, you need:

• 15v peak / 6 = 2.5v peak at the grid of V2-b for full output (MV at max)
• 2.5v peak / 12 = 208mV peak at the grid of V2-a for full output
• 208mV peak / 0.1 = 2.08v peak at the output of V1-b (+/- depending on real effect of the tone stack loss & 47kΩ to ground)
• 2.08v peak / 12 = 173mV peak at the grid of V1-b for full output
• 173mV peak / 0.24 = 722mV peak at wiper of Guitar channel Volume for full output, with Aux at 0.
• 722mV peak / 0.25 = 2.89v peak at plate of V1-a for full output with Guitar channel Volume at ~7.5.
• 2.89v peak / 12 = 241mV peak at grid of V1-a/Guitar input jack.

Confused yet?  Me too...  It sure looks like you've got a whole 12AU7 you don't need to use because of haphazard boosting/cutting/loading, because the input jack is only as sensitive as V2-a's grid.

If you haven't gone through the above, you also don't know if you're slamming the input jack with a much bigger test signal than the output stage can handle.  I don't recall seeing you mention your test signal level...

So what to do?  You may want to ditch the funky arrangement of the Aux channel, unless you're certain you need (a variable) loss there to the Guitar channel.  Having both the split-plate-load of V2-b and the Master Volume seems silly (2 loss methods at the same spot in the circuit?), when there is no voltage divider (or other loss mechanism) between V2-a and V2-b.  I'll leave it to your taste whether the 47kΩ and cap leading into the tone stack is useful.

Normally, a guitar amp would distribute voltage dividers between each stage (kinda like fixed volume controls) to keep late stages from being driven hard enough to begin grid blocking. See this Marshall amp; there are voltage dividers (loss networks) between the 1st three gain stages, which also shape frequency response.  They prevent grid blocking in the 2nd & 3rd gain stage, and also cut bass to keep the distorted sound punchy.

Mark Huss' solution was to slap large series resistance to slow the charging of the coupling caps when slammed with big signals.  You could do that or use the voltage divider method above to ensure signals never get big enough to cause grid blocking.  Regardless of the method you choose, make sure you're not creating flase gremlins by using an unrealistically-big signal at the input jacks.

[shooter]

Confused yet?  Me too

ok, with 30mVrms in (at guitar), TS centered, aux 0, gain ~ 6, I get just to clipping at top of MV pot, ~ 15Vac rms.  my bias is @ 16.1vdc.  max'd out gain gives me about 20Vac rms @ top of MV
You are correct, most of the gain comes from V2, hence the split load mod attempt.

the amp is my best tone build yet, till you OD the pre!  I guess the biggest question I have since I know what the problem is, how do the hi-gain, multi-stage pre's that produce square waves, or diode pseudo-square waves avoid blocking distortion?

At idle, what is the VDC of your plates, cathodes, and B+ supply?

If you follow the link to the schematic, I have all the DCV's on the schematic

[HotBluePlates]

... most of the gain comes from V2, hence the split load mod attempt. ...

If you want distortion while reducing the risk of grid blocking, you'd put your loss network leading into the grid of V2-b (because the Master Volume is already available to reduce drive to the output tubes).

...  I guess the biggest question I have ... is, how do the hi-gain, multi-stage pre's that produce square waves ... avoid blocking distortion? ...

Design output to input (to make sure enough drive is available to get all the power from the output tubes), add too much gain (or too many stages), then re-design input to output (to know where a stage is likely to get slammed).

They usually don't look for a single triode to be slammed to each rail, but a little distortion from each triode as it amplifies a signal it can handle.

But the output of a triode is probably bigger than the input of the next-cascaded-triode can handle, so a voltage divider is added between stages to knock the signal down (As shown in that Marshall schematic I linked).  Build up, knock down, build up, knock down, build up while adding a little bit of THD at each stage.  Yes, you'll get more noise that way, but that's the way it works.

For those designers who want to treat tubes like solid-state distortion pedals (or maybe get lots of clipping with less noise), diodes-to-ground can clamp/clip a signal between the output of one tube's plate and the next tube's grid.

I haven't investigated the loadlines for each stage of your amp to see if it even makes sense as an "overdrive amp".  It did, after all, evolve from an RCA application to cleanly add tone controls to an existing amplifier circuit using 12AU7's (while "wasting" a triode before and after the tone circuit in cathode followers).  It just seems like there are easier ways to get preamp distortion with higher-gain tubes...

[shooter]

Build up, knock down

That was my guess,  I really didn't design for pre distortion, I did hope for enough umph to bend the 6V6s into that 5E1 champ-ish sound when dimed, hence the reverb OT trannys instead of an edcor.

For those designers who want to treat tubes like solid-state distortion

I want the tubes happy bending themselves without authoritative diodes forcing the issue

I got a couple paintings in the queue, but while the oil dries, I'll get the amp back on the bench n see if I can re-balance the stages.

Thank you for the insight n time

[shooter]

I was reading Merlin and came across this statement;

"Coupling capacitor C prevents the negative voltage on the grid from interfering
with previous circuitry by blocking DC"

I have always assumed the opposite, the cap protected the grid from the plate DC volts
can I draw from merlin's statement that I "should" find DC on the "grid side" of a coupling cap?

[sluckey]

I "should" find DC on the "grid side" of a coupling cap?

There are plenty of cases where you will find dc on the grid. Cathode follower and LTP PI are two common guitar circuit examples. A less common circuit is a grid leak biased circuit where the cathode is connected to ground. Look at the trem oscillator or CHNL 2 preamp second triode in the Magnatone M10A for a couple grid leak examples...

     http://el34world.com/charts/Schematics/files/magnatone/magnatone_m10a.pdf

[shooter]

grid leak biased circuit where the cathode is connected to ground

Thanks, more spice in my pudding

I've got the paintings drying so I'm gonna open up the amp an see if I can get the pre re-balanced to keep enough PA drive umph, also distort just enough NOT to insert annoying stuff!

thank you as always

[PRR]

> Merlin and came across this

You take that out of context, it sounds like a universal rule.

You can read the full pages on Google Books, or in your paper-copy. I will risk Merlin's wrath by attaching snips of the relevant pages.

Designing Valve Preamps for Guitar and Bass, Merlin Blencowe

He is specifically explaining Fig 1.11, the most basic way to bias a tube to linear zone.

We are only on page 11. Merlin is still setting up some VERY basic introductory lessons. At this point (where the previous circuitry is not known) he's merely deflecting the wise-guy objection "your battery will fry anything connected!" The full story takes more pages to unfold.

Put the book in the bathroom and study a few pages every night. It is worth digesting thoroughly.

[shooter]

Put the book in the bathroom and study a few pages every night

I do keep going back, n back, n I feel like software caught in an endless for next loop

[sluckey]

And I feel myself a cog in somethin' turning

[PRR]

The Roman Empire expanded to control most of (what we now call) England. But when it reached (roughly what we now call) Scotland, around 122AD, Emperor Hadrian visited the area and instead built Hadrian's Wall.

Was Hadrian's Wall to keep the Scots in? Or to keep the Empire out?

While there is no good back-story, probably Hadrian told the Romans it would keep the nasty Scots in.

However the Scots made no serious objection, and probably were happy it kept the Empire out of their turf.

So is the coupling cap keeping the Empire's grid-battery out? Or keeping nasty Scots (high voltage from previous stage) in? Both. However at page 11 of the story we have not yet introduced enough background to cover both sides.

BTW, Hadrian's Wall was largely abandoned some years later. It is clear that trade went through the wall both ways, so it may be more a tax-stop than a barbarian barrier. But that's outside the topic of Coupling Caps.

[shooter]

more a tax-stop than a barbarian barrier

you're killin me, I was soldering, laughing at the visual in my head and when I smelt burnt flesh!!!!!
anyway;
I did prove? your theory, when the signal goes into clip, the right side of the CC leaving the pre goes negative (~0 to -8) at about the same volts as the right side goes positive (~ 182 - 188).

I have the pre tamed enough not to clip, but there's not enough umph left to push the 6V6's over to the dark side.  once the 2nd degree burn quits hurting i'll get back to it

[shooter]

Ok, I screwed up, there might be enough umph to bend the 6V6’s.

I can hit 14Vacrms @ onset of clip, 39VacP-P, the data sheet shows Grid Peak signal @13V, (26p-p)?

Here’s 3 scope shots, 2 FFT @800hz, -1vdc and -9vdc., plus clipped sin at -9vdc
As I can see, the -9Vdc shifted the spectrum up and tweaked the 1st few harmonics.

How many harmonics out is “important”?  If I see it I hear it?

(All the pics are taken @ right side of the cap leaving the pre.)

[PRR]

Sure looks like 39 pounds of potato in a 26 pound bag.

It "should" sound bent at that point. You have a problem with it?

[PRR]

Put the cathode voltage reference on your 'scope. Either centered, or at say the 3rd graticule line up. We "know" the grid can "never" rise positive of cathode. Grid sucks when driven positive. So up-to where drive wave not-quite touches cathode voltage should be clean-ish. Anything trying to go higher than that will actually force the cap voltage more negative, and output should be distorted. The flat-tops on your wave shape suggest this, though without a cathode reference I can't be sure this is what you are doing.

[shooter]

It "should" sound bent at that point. You have a problem with it?

no, I like the crunch, even hoping to cokes a little more down the road
Only problem, and this whole thread is I get an annoying "ckack" sound when I hit OD/distortion, opened it, scoped it, found the "sound" corresponded to the point the wave clipped.  Even after all the insight, I'm still not convinced "where" the Bad sound originates, I'm gonna follow it out the amp and see if I get some introduced transient in the PA, or anemic OTs that "show" when hit by a clipped signal.

without a cathode reference I can't be sure this is what you are doing.

I'll grab n overlay the cathode n plate actions, I'm down to 1 probe, so it might be a side-by-side

[PRR]

> I'm down to 1 probe

Cathode will be nearly a straight line. Mark with grease pencil or MSPaint.

[shooter]

I got my HV probe to play nice so no grease pencil, some MS paint though

I'm surprised with the phase shift   (It was a long day, I didn't do a probe cal)

EDIT: I added -9vdc;
I'm interested in even WAG answers, the rise n fall of the harmonic is vertical,  my brain "hears" something, but I don't know why.