Grid stoppers, clamping, blocking and crossover distortion.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In my homebrew guitar amp I’m using a LTPI to drive 2- 6v6 tubes in PP. They are cathode biased running ABx at 12watts idle each. The plates are at 350VDC. I’m wondering about the grid stoppers and if they are adequate at 1.5k. I sometimes think I’m hearing a bad distortion in addition to the good distortion. No doubt the PI is driving them very hard. The PI doesn’t clip on it’s own until after the output is clipping.

I’ve read these to get educated:

http://www.aikenamps.com/BlockingDistortion.html

http://www.freewebs.com/valvewizard/gridstopper.html

http://www.tubelab.com/powerdrive.htm

I understand the anti-oscillation reasons and the possible reduction in higher frequencies from the miller effect of grid stoppers. I don't fully understand blocking distortion yet. I do understand that the grid stopper curtails grid current that can occur when the grid is positive.

I think I understand the clamping that occurs when the grid is driven positive. (I know I see it on my scope.) But I’m just barely understanding the references to “time” and the charging/discharging of the coupling cap and how that is related to all this.

I definitely see crossover distortion on the speaker output when driven somewhat past the onset of the output clipping, but is that normal? Does a strongly overdriven PP output have to experience crossover distortion?

Can blocking distortion be recognized on a scope? Is the crossover distortion the manifestation of the extreme clamping? Is that essentially the “blocking distortion” revealed? If not, does anyone know of a link to blocking distortion on a scope?

I will switch to 5k or 10k resistors on the grids as a test next time I open it but I’d like to fully understand 1st. :spin:
 
As I understand it, with blocking distortion, grid current causes the coupling cap to charge, biasing the tube more negatively and causing crossover distortion.

The amp I built has DC coupled source followers and it was strange that I could see the clipping on a scope before I could hear it (playing music). I just figured clipping on a hifi amp would sound really bad, but it is amazing how much clipping can be tolerated if you can get rid of the blocking distortion. I don't know exactly what makes good guitar amp distortion, but I would imagine blocking distortion would not be it.
 
It's starting to all make sense because I leaned towards higher value caps in an attempt to maximize potential bass. I'm thinking this is contributing to the chain reaction and crossover distortion. I recall watching the period of cutoff increase on each 6v6 when overdriving harder. Now that mystery is starting to make sense if increased grid current leads to the bias cooling. I need to get back in there!

It's a modified Trainwreck Express. There is more than shown but it is about time I started to make up a proper schematic. Here is my quick and easy schematic:

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
I just built two Trainwreck Express clones with tweaks :)

Those coupling caps to the 6V6... need to be 0.022uF.

A Trainwreck is a bright circuit to begin with, so changing the cap between your fist and second 12AX7 (the 0.0025uF one) is where you want to look. Try 0.01uF.

Cheers!


(PS - your NFB resistor is too big... try 100K)
 
Ex-Moderator
Joined 2004
If you "maximise" the bass in a guitar amp, it's possible you may produce too much at low frequency for the OP transformer to manage, causing distortion.

Blocking distortion is caused when the OP tube grid is driven by cap coupling and tries to go positive (i.e. overdrive).

Crossover distortion is a separate thing, normally due to running the OP tubes too lean and is unlikely to happen in your case because you have caqthode bias.
 
Crossover distortion is a separate thing, normally due to running the OP tubes too lean and is unlikely to happen in your case because you have caqthode bias.

The two phenomenae are related since the extra grid bias can push the finals more towards Class B, if not Class C. That will indeed lead to lots of nasty x-over distortion until the excess negative charge leaks off the coupling capacitors. That's bound to be hideous.
 
The two phenomenae are related since the extra grid bias can push the finals more towards Class B, if not Class C. That will indeed lead to lots of nasty x-over distortion until the excess negative charge leaks off the coupling capacitors. That's bound to be hideous.

I'm getting closer to understanding. Sounds like the coupler discharges somewhat during the clamping time then recharges thru the grid during the "recovery" time. (The whole concept of grid current is not something I've thought about much.) Since the recovery (recharging?) time is "long" it pushes the bias colder during that time which moves the tube towards class B or C. I guess because the grid is made “more negative” while it is supplying current back to the coupler for it’s recharging.

I'm not 100% clear on all of it but I'm close enough I guess. The bottom line for me is that increasing the grid stoppers should/could reduce the crossover distortion I get when driving hard. I’m hoping this is the “bad” distortion I’m hearing with the good distortion. From what I’ve read, crossover distortion is bad sounding, even in guitar amps. Once I've finished with the stoppers I'll see if I need to reduce the couplers. I don't want to sacrifice bass if not needed. There is not a surplus of bass going on as it is.

Thanx for all the feedback.

Here's a pic of it:
 

Attachments

  • Out-DSCF0014.JPG
    Out-DSCF0014.JPG
    52.2 KB · Views: 594
Last edited:
The waveform is distorted, but symmetrical, so both 6V6s seem to be pulling & pushing OK. That is very nasty crossover distortion though.

does it dispappear at lower power output?

If not, is the impedance match correct (not an 8 Ohm speaker on a 4 ohm tap)? Is the transformer correctly rated? - adequate power handling, and 5000 to 8000 ohm anode-anode load?

The 500 Ohm resistors on the 6V6 cathodes should have about the same 12V on them. If one is wildly different, change that valve.

I agree with Geek that 0.022nF should be used for gate coupling. "Blocking" happens when the grid starts taking current. Normal operation means they take only 1uA or so, but when the driving signal comes near to the cathode voltage (say, grid = +11.5V) mA-level current is taken and this charges up the coupling cap! As this grid voltage gets above 12V the current really flows, if not limited by a stopper. Now, when the signal (grid) voltage starts going down again, the charged cap causes the grid to be more negative than the driven side of the coupling cap, and effectively biases the 6V6 in the OFF direction! As Miles says, this bias shift can result in crossover distortion. But if it's there without overdriving the amp, suspect transformer or 6V6 parts trouble.

This is one reason you use a small coupling cap in a guitar amp. You can eliminate blocking completely by using a power MOSFET to drive the grids. Circuit have been put up on here I believe.
 
The waveform is distorted, but symmetrical, so both 6V6s seem to be pulling & pushing OK. That is very nasty crossover distortion though.

does it dispappear at lower power output?

If not, is the impedance match correct (not an 8 Ohm speaker on a 4 ohm tap)? Is the transformer correctly rated? - adequate power handling, and 5000 to 8000 ohm anode-anode load?

The 500 Ohm resistors on the 6V6 cathodes should have about the same 12V on them. If one is wildly different, change that valve.

I agree with Geek that 0.022nF should be used for gate coupling. "Blocking" happens when the grid starts taking current. Normal operation means they take only 1uA or so, but when the driving signal comes near to the cathode voltage (say, grid = +11.5V) mA-level current is taken and this charges up the coupling cap! As this grid voltage gets above 12V the current really flows, if not limited by a stopper. Now, when the signal (grid) voltage starts going down again, the charged cap causes the grid to be more negative than the driven side of the coupling cap, and effectively biases the 6V6 in the OFF direction! As Miles says, this bias shift can result in crossover distortion. But if it's there without overdriving the amp, suspect transformer or 6V6 parts trouble.

This is one reason you use a small coupling cap in a guitar amp. You can eliminate blocking completely by using a power MOSFET to drive the grids. Circuit have been put up on here I believe.

Thanx for the comments. The OPT is very well built and rated for at least 50watts, perhaps more. According to my calculations I'm getting about 9watts output just before clipping. I'm using the 6600 primary with an 8ohm load on the 8ohm winding. 500ohmers are each 499ohms. There is no crossover distortion until a little after the onset of clipping. I won't be able to consider mosfets since the tube guitar amp police would quickly arrest me. :)

Here's the output and one of it's inputs at the onset of OP clipping and some clamping of the input. No crossover distortion on the OP until driven a little harder than this:
 

Attachments

  • op-pi-sm.jpg
    op-pi-sm.jpg
    50.3 KB · Views: 760
For a solid 350V supply, you should expect at least 15W without clipping.

The cathode voltage on each 6V6 should be about the same - 20 -24V perhaps. any more could be a problem.

It's probably blocking. Try .022uF coupling caps, or reduce the 220K grid leak resistors to 43K, which will give an approximation.
 
Use more MOS

At the MOS information council, we are very concerned to hear of instances of failure to take up MOS product in any authorised circuit position.

We remind you that MOS is availailable in stealth packages styles, to maintain complete protection from those perfidious Tube Police.

Would you please direct yourselves to the appropriate literature:
 

Attachments

  • NotMOS.jpeg
    NotMOS.jpeg
    67.8 KB · Views: 733
Got back in there today. I started with the grid stoppers since increasing them to 5.6k should not affect treble much in a guitar amp. I saw no improvement in the crossover notch. So I disconnected the NFB, also with no improvement. Then I changed the cathode resistors to 470ohms each and this brought to idle up to 13watts from 12. (13 is the same idle as a vintage 6v6 Tweed Fender Deluxe.) This eliminated about half of the notch. I then changed the couplers from 0.1u to 0.047u. (I don’t have 0.02u at the moment.) This did not trigger any noticeable reduction in the notch. I then paralleled some 220k’s on the existing 220k’s grid leaks for a net of 110k each. This almost eliminated the balance of the notch. So I replaced the grid leaks with 82k’s. Now the notch is just barely there when maxed.

I put the 0.1u couplers back in since I noticed some reduction in bass and wanted it back. I guess increasing the stoppers more may have resolved the notch but I recognized when designing my approach that using a LTPI on this particular 6v6 output was possibly going to be overkill since an original 5e3 Fender would use a cathodine. So I’ve now backed down the drive enough to prevent the notch but still have clipping of the 6v6’s when wanted.

I’m not worried about the 13watts idle since the JJ 6v6’s are supposed to be over designed and, for whatever reason, I’m only seeing about 9watts total output just before clipping. The volume potential is definitely adequate and to my liking with the apparent “9watts” thru the 4-speaker cabinet. :eguitar:

Or my math is wrong. :ashamed: I see about 24vac p-p on the 8ohm output on my scope. Can someone check me on this:

24vac pk-pk = 12vac pk

12vac pk x 0.71 = 8.5vac rms (also confirmed by my Fluke DMM)

RMS watts = RMS volts x amps

Using ohms law:

RMS watts = RMS volts x (RMS volts / ohm)

RMS watts = 8.5 x 8.5 / 8 = 9watts RMS

Am I doing this correctly?
 
Grid Current flows when the grid of the 6V6 goes positive with respect to the cathode.
This happens on positive peaks of the drive signal.
If you look at the current path - from the driver anode, thru the coupling cap, thru the grid stopper, into the 6V6 grid, out the 6V6 cathode and thru' the cathode bias resitor and/or bypass cap.

The coupling cap will charge to a higher voltage so it will try to push the grid more negative thus acting to reduce the tube current (biases it more toward cut off). At the same time the cathode bypass cap will charge up increasing the cathode voltage which acts to increase bias and push the tube toward cut off again.

Increasing the grid stopper restricts the grid current that can flow. Taking it from 1K5 up to 5K6 means you have reduced the grid current and both cathode bypass capacitor and coupling capacitor will charge up less reducing the 2 things that are trying to drive the tube toward cut off. This affects both blocking distortion and cross over distortion.

The change in the grid leak resistor will help bleed away the charge on the coupling capacitor faster and so the recovery from overload will be better.

The cross-over distortion on the CRO traces above will generally ONLY happen when in continuous overdrive, momentary overdrive will generally give you the blocking distortion without the cross-over distortion.

The best way (in a guitar amp) to mitigate the blocking distortion is to run high values of grid stopper on the 6V6. Don't be afraid to try 10K or even 22K. Go high enough and eventually you will start to lose high frequencies since the grid stop creates roll off with the Miller Capacitance. Remember that Miller Capacitance depemds upon the gain in the tube and when you run into overdrive the tube gain "disappears" and teh capacitance will be reduced, so set the grid stop as high as you can consistent with CLEAN sounds.

Note that "blocking distortion" can occur in the preamp section too if you overdrive a gain stage too hard.

Another guitar amp trick - if you want some power amp compression at lower power (volume) levels then increase the value of your screen resistors on the 6V6. I run 1K as standard for screen resistors but you can run 2K2 or even 4K7 for earlier "break up".

The Power Calc you did is correct - Power = V squared divided by R (V in RMS Volts).

My "two bob's worth"

Cheers,
Ian
 
hey gingertube..

i think i have come to the right place, you are an expert, and others here also.

i own a vox nighttrain and i experience some trouble.. mainly strange distortion, maybe like you explained here..
the sound is good until i turn the gain past 12 o clock..
then when the sound breaks up i hear other stuff a little later.. kind of rattling but it is not rattling.. anyway.. people suggested that i install a grid stopper between pin 2 of the v4b (in my shematic of the amp the cathodyne phase inverter is v4 and power tubes are v1 and v2)

people suggest 470 k or 1m.. like tube wizard merlin does.. and suggests for cathodyne phase inverters..

but how i can i set it not to get rid of too much highs in the sound?

and i would LOVE to get some more power tube compression at lower levels and earlier breakup, you mentioned how that works, but i am ont familiar with wiring diagrams yet.. i will post the shematic here, and please suggest what needs to be done for the grid stopper and the earlier power tube compression. then the amp would be a dream.
 

Attachments

  • vox_night_train_nt15_h_121.pdf
    47.5 KB · Views: 176
crossover distortion fix

As I understand it, when the push-pull output tubes are pushed into clipping hard enough to cause grid current to be drawn, not only does this cause a charge to develop on the coupling cap coming from the phase splitter, but because the grid suddenly has a dramatic reduction of impedance, it clamps, or shifts, the DC voltage at the coupling cap to approximately the cathode voltage of the output tube. This shifts the operating point of the output tube into clipping for just long enough to cause what's called blocking distortion, which is about recovery delay at the grids of the output tubes. When both tubes of the push-pull pair do this, it creates crossover distortion, which adds a shrill fizzle component to the sound, which isn't musically pleasantly related to the audio signal.

There are several ways to reduce this (smaller coupling cap, larger grid stopper R's on output tubes, drive output tubes from cathode followers or an interstage transformer, limiting drive from phase splitter resistively, adding a "conjuctive filter" across the speaker terminals to limit speaker impedance rise at high frequencies which causes transformer ringing - this should be done regardless of what else you do), but the fix I like the best is the Paul Ruby fix.

The Ruby fix is where you put a zener in series with a regular diode, cathode to cathode, tie the anode of the zener to the grid of the output tube (before the grid stopper R), and tie the anode of the regular diode to ground (this is for an output stage that uses cathode bias). Do this for each output tube. I've personally verified that this works (see photos). You want to pick zener voltages that are slightly above 2 times the cathode bias voltage of the output tubes, because of the clamping effect DC shift (maybe not true if circuit uses "fixed bias" - I haven't tried that yet). You let the output tube draw some grid current when overdriven, so you don't introduce a diode clipped sound, but you limit how far into class AB2 you let it go, so the crossover distortion will be greatly reduced and not a significant issue anymore.

In addition to that, I'm using 27k grid stoppers on the output tube grids (no rolloff problems at all with EL84's), and a conjunctive filter across the speaker terminals (a 20R 5W in series with a .68u cap), since the impedance of the speakers rises considerably when you go above about 10kHZ, and any residual crossover or clipping distortion will have energy way up in frequency.

With these mods, it may be arguable that push-pull is no longer worse than single ended.

It's also been suggested that single ended preamp (or poweramp) stages can also "fart out" from blocking distortion. I haven't tried it yet, but the Ruby mod should work there too. Not sure what voltage zener would be right there.

The photo on the right is with the Ruby mod. I've actually driven it harder into clipping than the middle photo which has no mod and is very typical of all output stages in guitar amps.
 

Attachments

  • My amp wb.jpg
    My amp wb.jpg
    521.4 KB · Views: 397
  • IMG_2212 normal wb.jpg
    IMG_2212 normal wb.jpg
    115.7 KB · Views: 373
  • IMG_2215 paul ruby wb.jpg
    IMG_2215 paul ruby wb.jpg
    114.8 KB · Views: 345
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.