Driving Power Tubes To Hard

[lkrasner]

I think I'm driving the power tubes a bit too hard in my most recent build. I believe the problem is in my effects loop circuit, which has an always active cathode follower buffer and 12au7 gain stage (there is no true bypass) Even with the preamp set very clean, the output is very high, but there isn't much headroom, I always get a lot of distortion out of the poweramp. How can I best reduce this effect? I tried removing the bypass cap on the fx recovery stage without much effect.

Schematic attached

[shooter]

driving the power tubes a bit too hard

You have both adjustable bias and a "drive" pot to control *how much*.  Find a setting you like and, walla!

If you have access to a scope you can even dial it in that way, which is how I know what *number* on my pot is max clean.  The EL34 I did last month, 7 is the magic number, after that it starts sounding like anger management gone bad!

[lkrasner]

driving the power tubes a bit too hard

You have both adjustable bias and a "drive" pot to control *how much*.  Find a setting you like and, walla!

If you have access to a scope you can even dial it in that way, which is how I know what *number* on my pot is max clean.  The EL34 I did last month, 7 is the magic number, after that it starts sounding like anger management gone bad!

do you mean the pot I have as the FX send level? If so, It is only active when something i plugged into the loop, although I could change that.

[shooter]

It's listed as RV1/RV2 and the way you have it drawn, I'd call that a master volume pot, or PA drive.  If it's at 0, I don't see how your preamp signal gets to G1, so, to my way of thinking, that's how you *control* drive to the el's.

[2deaf]

That seems like an awfully hot effects loop to me.  What if you replaced R21 (18K) with a 100K and a 10K in series with the 10K going to ground and the effects send taken off of the 10K?  If you leave C9 off of the return stage cathode, the overall gain of the loop will probably be around unity.  Also, the signal level sent to an effects device will be a lot more appropriate.

[jjasilli]

I think Shooter is referring to RV1 & 2. But, 1M is way too high for EL84 grid leak resistors.  100K to 220K should be plenty.  The smaller value pots are better not only operationally for the tube, but will give you more useable rotation to dial in a vol. setting.


EDIT:  Because RV1 & 2 are in parallel, their combined max value is 500K.  Hence, I think you should audition a pair of 500K pots.  Then jumper a 500K resistor across the outer lugs of both 500K pots to make them 250K ea.  This will yield net values of 250K and 125K.  See which you like better; then hardwire the circuit (or make it switchable).

[lkrasner]

I think Shooter is referring to RV1 & 2. But, 1M is way too high for EL84 grid leak resistors.  100K to 220K should be plenty.  The smaller value pots are better not only operationally for the tube, but will give you more useable rotation to dial in a vol. setting.


EDIT:  Because RV1 & 2 are in parallel, their combined max value is 500K.  Hence, I think you should audition a pair of 500K pots.  Then jumper a 500K resistor across the outer lugs of both 500K pots to make them 250K ea.  This will yield net values of 250K and 125K.  See which you like better; then hardwire the circuit (or make it switchable).

I'm not quite sure I follow what you are saying.  I commonly see 1M pots for MV controls, but I'd be open to experimenting with that, however  I kind of lost you with what you were saying about the yielded values.

[shooter]

Look at the tube data sheet under design center values, (I believe).  There is a max value for the G1 to ground R, I believe it's 500k max for the EL34.  I usually find 2 or 3 *commercial* schematics that use, in this case EL34's, then I read (to the best of my ability) the tube data sheet.  Then I sorta ball-park average R and C values for all that information I just studied, or just outright plagiarize values.

[PRR]

Why do you think it is the power tubes distorting? A driver tube can make similar racket.

A test for the whole power section is to run your FX out into the FX In of another amp.

How do you know the power section is even clean? SE amps are often mis-loaded, mis-biased, over-volted, etc. (I do see that you have a reasonable 300V DC on tap.)

Where did you get the bias values for the 12AU7 FX recovery/driver stage? Those are more suited to 12AX7. The cathode resistor will normally be around Rp/Mu. 12AU7's Mu is like 20, so 100K/20 is like 5K, not 1.5K. The plate is sitting quite low, and the maximum negative swing *may* be limited. I'd have to plot before making accusations. But a fast-check is to put a 12AX7 in there, snip the cathode cap, to get a well-biased and very clean booster stage.

I'm unsure about the FX driver CF stage bias. Looks awful hot. That will work fine but do you need all that power (heat) here?

The signal level at the FX return is about 0.6V, 600mV. That's groovy for studio FX (but the 250K pot seems high for driving 10K-22K inputs?). Guitar-level FX here may be very distressed by the high level.

[PRR]

> max value for the G1 to ground R, I believe it's 500k max for the EL34.

Plan shows EL84.

Max Rg for the '84 with "automatic" (cathode resistor) bias is 1 Meg. (But with fixed bias, 300K.)

He's OK there.
______________________

What's with the font-size tags? I can't make-big two places in one posting without 6 edits and manual tagging??

[2deaf]

I see EL84 and 6V6.  I also see four stages with gain that is swamping both of those tubes. 

[lkrasner]

Why do you think it is the power tubes distorting? A driver tube can make similar racket.

A test for the whole power section is to run your FX out into the FX In of another amp.

How do you know the power section is even clean? SE amps are often mis-loaded, mis-biased, over-volted, etc. (I do see that you have a reasonable 300V DC on tap.)

Where did you get the bias values for the 12AU7 FX recovery/driver stage? Those are more suited to 12AX7. The cathode resistor will normally be around Rp/Mu. 12AU7's Mu is like 20, so 100K/20 is like 5K, not 1.5K. The plate is sitting quite low, and the maximum negative swing *may* be limited. I'd have to plot before making accusations. But a fast-check is to put a 12AX7 in there, snip the cathode cap, to get a well-biased and very clean booster stage.

I'm unsure about the FX driver CF stage bias. Looks awful hot. That will work fine but do you need all that power (heat) here?

The signal level at the FX return is about 0.6V, 600mV. That's groovy for studio FX (but the 250K pot seems high for driving 10K-22K inputs?). Guitar-level FX here may be very distressed by the high level.

I believe you are backwards in your wording about testing the poweramp. I will try running the FX OUT of the OTHER amp into the FX in of mine (although that still goes through my fx recovery stage, which could be an issue).

I can also test my preamp into the poweramp of another amp with the same method in reverse. I assume it is safe to leave my OT unloaded if I pull the power tubes out?

I got my design from here (see the bottom of the page): http://www.valvewizard.co.uk/accf.html Thought that would be pretty reliable. I will try playing with the bias. Should I leave it bypassed or not?

[MakerDP]

Where did you get the bias values for the 12AU7 FX recovery/driver stage? Those are more suited to 12AX7.

I've been trying to tell him this for a week on AX84's forums. He is distorting his effects loop recovery stage IMHO. Your proposal to test by sending his pre to another poweramp will work to test that, but I think he should also just try completely bypassing his effects loop and I bet his problems go away.

[lkrasner]

Where did you get the bias values for the 12AU7 FX recovery/driver stage? Those are more suited to 12AX7.

I've been trying to tell him this for a week on AX84's forums. He is distorting his recovery stage IMHO. Your proposal to test by sending his pre to another poweramp will work, but I think he should also just completely bypass his effects loop and I bet his problems go away.

I understand, and I will investigate it. I got the schematic from a source I thought was very reliable, and it was designed for a 12AU7. As far as I can see, I am doing everything (within the FX circuit) exactly as designed.

[MakerDP]

In a reply to one of my questions on AX84 about his effects loops, Merlin told me not to use the one on his website. The one he does with a 12AX7 from his 2nd edition preamp design book is the one he recommends. It has a simpler design and it will not send too-high voltages to pedals.

[lkrasner]

In a reply to one of my questions on AX84 about his effects loops, Merlin told me not to use the one on his website. The one he does with a 12AX7 from his 2nd edition preamp design book is the one he recommends. It has a simpler design and it will not send too-high voltages to pedals.

Hmm, ok. I will have to look into that. I guess that's on me for assuming the circuit on the website was working.

I did some tests as discussed above. With my FX out going into the FX in of a solid state peavy bandit 112 (what I happened to have next to me), I get a similar distortion unless I turn my send level (on the tube amp) way down. It's not as bad, but that probably has to do with the design of the poweramp in peavy.

If I reverse this, and run the peavy into my FX return, I can get a perfectly clean tone.


The problem definitely seems to be in the FX send CF of the tube amp, I will try some of the suggestions to tame it and report back. Thanks for all the help!

[shooter]

    

Plan shows EL84.

I broke my glasses a couple days ago wearing my '90's harry potter spare

sorry for added confusion

[MakerDP]

Well, to be fair to Merlin (and you) that article is about cathode followers not effects loops. That is just an "oh by the way you COULD try something like this with the extra half of the tube" afterthought tagged-on at the end.

Your experiments with the Bandit are also not 100% valid. All you've proven is that you are introducing some sort of unpleasant artifact somewhere from the input jack to the cathode follower, it does not immediately mean it's the cathode follower's fault.

I would do this before trying anything else to prove one way or the other the problem is in the effects loop...
1) disconnect your treble pot's wiper from C11.
2) disconnect C8 from R16 and V3 pin 1.
3) connect your treble pot's wiper to C8.
4) test your sound.

No unpleasant artifact means your problem is somewhere in the V3 circuit. If it's still there, your problem is in your preamp or your poweramp.

As a side note, when I do a build, I do the bare necessities first... preamp, tone stack, poweramp. Test. Tweak. Test. Tweak, etc. Only after all is well, go back and add the icing... but only one piece at a time. Add the Reverb... test tweak test tweak. After all is well, add the effects loop... test tweak... Makes troubleshooting a whole lot easier!

[lkrasner]

Well, to be fair to Merlin (and you) that article is about cathode followers not effects loops. That is just an "oh by the way you COULD try something like this with the extra half of the tube" afterthought tagged-on at the end.

Your experiments with the Bandit are also not 100% valid. All you've proven is that you are introducing some sort of unpleasant artifact somewhere from the input jack to the cathode follower, it does not immediately mean it's the cathode follower's fault.

I would do this before trying anything else to prove one way or the other the problem is in the effects loop...
1) disconnect your treble pot's wiper from C11.
2) disconnect C8 from R16 and V3 pin 1.
3) connect your treble pot's wiper to C8.
4) test your sound.

No unpleasant artifact means your problem is somewhere in the V3 circuit. If it's still there, your problem is in your preamp or your poweramp.

As a side note, when I do a build, I do the bare necessities first... preamp, tone stack, poweramp. Test. Tweak. Test. Tweak, etc. Only after all is well, go back and add the icing... but only one piece at a time. Add the Reverb... test tweak test tweak. After all is well, add the effects loop... test tweak... Makes troubleshooting a whole lot easier!

Agreed 100%. I have done that exact test before, although there was some other stuff going on that has since been fixed as well. I'll do it again to verify, but I do believe I found that it worked well (minus the bass control issue that still exists).

Also agreed on the bare necessities point. I will not try to justify my building everything at once....

[2deaf]

When I see loop designs that you guys reference, I'm not seeing what the level is that is coming into them.  What signal level do you folks think should come out of an effects send?  I go with -10 dBV because it works well with line level devices intended for guitar. 

 

[MakerDP]

Depends... are you going to have pedals in your loop or rack mount gear? That's why some high end amps have a switchable Send level and/or a send level pot.

[2deaf]

What maximum level?

[MakerDP]

Krasner, if you are still having trouble with your tone stack I would recommend bypassing your effects loops as discussed above and leaving it that way until you fix your tone stack issue. After that go back and fix the loop problem, assuming that's where it is of course.

[MakerDP]

What maximum level?

Merlin recommends -9dBV so I should think your -10dBV is fine.

[2deaf]

What level do you think is coming out of the effects send of lkrasner's design?

[lkrasner]

What level do you think is coming out of the effects send of lkrasner's design?

For the record, I do have a send level pot to allow adjustment in this.

Making some progress overall, removed the bypass cap on the recovery and adjusted the bias resistor on the CF to be colder. Might try tweaking the load resistors next.

[2deaf]

For the record, I do have a send level pot to allow adjustment in this.

25K would be more appropriate than 250K.

[jjasilli]

I think Shooter is referring to RV1 & 2. But, 1M is way too high for EL84 grid leak resistors.  100K to 220K should be plenty.  The smaller value pots are better not only operationally for the tube, but will give you more useable rotation to dial in a vol. setting.


EDIT:  Because RV1 & 2 are in parallel, their combined max value is 500K.  Hence, I think you should audition a pair of 500K pots.  Then jumper a 500K resistor across the outer lugs of both 500K pots to make them 250K ea.  This will yield net values of 250K and 125K.  See which you like better; then hardwire the circuit (or make it switchable).

I'm not quite sure I follow what you are saying.  I commonly see 1M pots for MV controls, but I'd be open to experimenting with that, however  I kind of lost you with what you were saying about the yielded values.

Let's assume that you are right -- too much signal voltage is hitting your power tubes. 


1.  Why is this happening?  Maybe because of the effects loop, as many posts suspect.  So, follow makerdrp's suggestion in Reply #22:  bypass the effects loop to see if the problem persists.   This would be good to know.


2. What to do about it.  Regardless of the source of the "overvoltage" signal  problem, there is more than one way to resolve it.  The way I suggested is to reduce the value of RV1 & 2.  These are not only "drive" or "volume" controls.  They also serve as grid leak/signal load resistors for the power tubes.  If you reduce their value, more signal will bleed through them to ground. This is one way to resolve the "overvoltage" signal  problem.


So why not just turn the pots down?  Because with 1M pots (assuming linear pots), ea 1/10th rotation of the pot is an addition of 100K!  You probably need only 100K to drive the power tubes (though 100K might not be optimal).  With the pots on "3", the power tubes are already being overdriven excessively.  By contrast, with a 250K pot, 1/10th rotation is an increment of only 25K.  Much easier to fine-tune a vol. setting, and without excessive signal voltage at full rotation.


You can easily test this by jumpering fixed resistors across (in parallel with) your existing pots to reduce their values.

[VMS]

I would put a volume pot after the treble pot, that way it won't distort the V3b triode at high gain settings.

[tubenit]

Have you thought about a parallel loop instead of series loop?  IF bypassing  the loop gives you what you want, then maybe trying a paralleled loop might be worth considering? 

With respect, Tubenit

[jjasilli]

Yes, tubenit has posted great fx loop circuit schematics; and VMS is right about a vol pot after the tonestack.  But simply reducing the value of RV1 & 2 may suffice, without re-designing the amp.

[2deaf]

I would put a volume pot after the treble pot, that way it won't distort the V3b triode at high gain settings.

With this pot in place, you could eliminate the 250K effects send pot.

[2deaf]

VMS is right about a vol pot after the tonestack.  But simply reducing the value of RV1 & 2 may suffice, without re-designing the amp.

VMS's suggestion allows you to reduce the signal level sent to an effects device.  Simply reducing the value of RV1 & 2 does not solve the effects send problem.

I'm not really following the logic of the RV1 & 2 reduction.  It will lower the plate load for V3a (effects return stage) and lower the gain by maybe 25% or so, but I don't see the rest of it.  EL84 is happy with 1M on G₁ and 6V6 is happy with 2.2M.

[lkrasner]

I would put a volume pot after the treble pot, that way it won't distort the V3b triode at high gain settings.

With this pot in place, you could eliminate the 250K effects send pot.

Don't I want the effects before the master volume though? With some tweaks to the gain I've gotten this sounding pretty darn good actually.

Yes, tubenit has posted great fx loop circuit schematics; and VMS is right about a vol pot after the tonestack.  But simply reducing the value of RV1 & 2 may suffice, without re-designing the amp.

I ordered some other pots to put in place of the current MV, we'll see what that does.

Have you thought about a parallel loop instead of series loop?  IF bypassing  the loop gives you what you want, then maybe trying a paralleled loop might be worth considering? 

That's a good point. Perhaps I will add a switch to allow for either some time in the future.

[lkrasner]

Alright, so I improved things alot by raising R22 from 470R to 1.8K to bring the bias of the CF closer to center and removing the bypass cap on the recovery stage, however it is still a bit hot. what would you recommend to drop the gain of the effects circuit a bit.

I'd like to avoid making huge changes for now, and get this sounding good. I may modify things later to add a bypass to the loop, make it parallel, etc..

[MakerDP]

Read this post. It is a short conversation between myself and Merlin Blencowe. Pay close attention to the third post, my third question and the responses to that.

http://ax84.com/bbs/index.php?area=-1

But, be aware that my calculations there are based on his book's 12AX7 effects loop, not his website's 12AU7 loop.

If you don't have his book, you really need to get it and read the chapter on effects loops. Well, you really should read the rest of it too... great stuff and SO many of your questions you've asked here are answered somewhere in those pages.