18 watt fizz: please help me diagnose this?

[katopan]

Then I replaced the PI tube with a lower gain 12ax7-wa (it had a new production Mullard in there prior). The spike is less pronounced but still quite present - same with a JAN 5751.

That makes sense. A lower gain PI is driving the power stage a bit less, just like turning down the volume control a bit.

Could it be coupling caps between the PI and PT grids?

I'd be grasping for any straw I could try. Would be cheap to try a replacement or different type of coupling cap. You've seen that the zener fix does what it should but doesn't fix your problem, so it's not the Ruby buzz. It's normal to see inductive characteristics as unbalanced because each side of the primary has the same no. of turns but not the same inductance. And your GDS is one of the best you could buy. If you did want to try a cheap test OT it would be possible to put in a 20W line P.A. transformer with the right turns ratios. It won't sound anywhere near as good but would rule it out for $20. I was also thinking output valves, but I read back and you said you swapped them out. Short of the stuff we've already covered I'm running clean out of ideas too. If you do work it out I'd be interested to know. Good luck!

[guitarmike2107]

I have being following this thread and others about the PR-buzz

The frustrating thing about the zener fix is its just that .. its a fix rather than understanding what’s causing it.

Paul said that the PI in both amps were the same and changing the OT didn’t make a difference, and changing power valves doesn’t make a difference so surely its got to be something between the two sockets, if not the sockets themselves.

one amp which had this problem recently had the OT primary wires running quiet a long way all three twisted together, rather than just the two out of phase wires, I wondered if this had anything to do with the kickback/slow recovery. It would be really good to get a proper understanding of this one.

Sorry I know that doesn’t help you guys.. just voicing my frustrations with my lack of answers!

[VTAMPSUSA]

I agree with Mike - this seems quite unsettling as some 18Wers seem fine with HO p/ups and don't get the fizz, some do...

Paul's document never really honed in on what exactly could be the root cause, which I suspect has to be something from the PI load resistor to the EL84 grid... something in there changed/changes and causes it. I think this is what Mike was pointing out.

I do recall a post on AX84 where Paul mentioned how important cap type was on EL84 output sections from the PI. It is most definitely the charging of the cap that is causing the issue, by why on some and not the others?

IIRC, Paul had mentioned, or showed pictures of some of his Rocket builds using PIO caps and when asked why he used them it was to 'cure' the fizz.

I know from my own experience that PP caps of certain sizes and shapes can promote unwanted, and fizzly distortions when used in the wrong areas. Most tubular PE caps seem to be better suited.

It would be interesting if we could isolate two examples of fizz/no fizz 18w'ers like Paul did but try to isolate the problem better instead of tacking a Zener on the grid leak - it seems that this SHOULDN'T be necessary if all other things are done correctly... we just don't know exactly what that is

[JdJ]

In my case I think I have 2 things going on. The ruby buzz and the fizz which is represented in the big overshoot on my previously posted oscilloscope traces.

AFAIK - in the worst case, with all other things equal, an amp with a cold bias on the power tubes and a high output pickup is going to sew the seeds for the Ruby buzz since there is an increased chance of grid conduction charging the coupling caps. I was able to eliminate this possibility by installing the diodes. Aside from that solution, I'd imagine there has to be a way to keep the PTs in or close to class a while maintaining a healthy (but not too healthy...) signal to the pt grids. Aiken on his site, and zaphod_phil here have mentioned decreasing ground reference resistors and/or increasing grid blocking to help keep the PT's from premature cutoff.

I'm beginning to wonder if I have some PP imbalance that is resulting in the overshoot. I'm going to try replacing the coupling caps and all of the grid blockers/refs for the power section - there seems to be a small difference on my ohm meter between both sides...

[guitarmike2107]

Paul's document never really honed in on what exactly could be the root cause, which I suspect has to be something from the PI load resistor to the EL84 grid...

I do recall a post on AX84 where Paul mentioned how important cap type was on EL84 output sections from the PI. It is most definitely the charging of the cap that is causing the issue, by why on some and not the others?

IIRC, Paul had mentioned, or showed pictures of some of his Rocket builds using PIO caps and when asked why he used them it was to 'cure' the fizz.

Makes you wonder if Paul actually knows more than he is telling about the cause of the buzz, the rocket builds will be slamming the power valves too.

What would cause the slow recovery shown between the two amps on Pauls bench, I wonder whether using low ESR caps for coupling caps, power supply and Power valve cathode bypass would help?
It may just be the OT causing the slow recovery though…

Josh can you upload a gutshots picture so we can see what in your amp and what the layout looks like too? and let us know what you change…

Mike

[JdJ]

Josh can you upload a gutshots picture so we can see what in your amp and what the layout looks like too? and let us know what you change…

Here are a couple pictures from before I started mucking around. Sorry for the blurryness. Currently everything is the same except I have the 15v zeners/4007s in there. I'll try to take a shot later showing those. The PI to PT Coupling caps are 103k (.01uF) Sozos. I have a range of Mojo Dijons coming too. It is a Ceriatone Tremolo with Sozos and GDS Vintage iron.





Thanks for wanting to find a solution to the fizz guys!

Josh[/img]

[snoof]

Have you tried a 10k,.001uf conjunctive filter yet just for grins?? It's still a band aid, but of a different kind.

[stevesuk]

According to a Tonequest article interviewing Dr. Z, they fit a conjuctive Filter in every 15/18W EL84 amp as standard

Steve UK
www.valvepower.co.uk

[ontariomaximus]

Every EL84 amp I do has a Zobel, which is a conjunctive filter but on the secondary. 10 ohm 5 W R in series with a 2u2 cap across the 8 ohm connection. Contrary to what some believe, it ain't a bandaid. What it does is alter the load presented to the amp at high frequencies. You wouldn't use a output trans with a 30K primary would you? Well that could be load presented to the amp at high frequencies due to the the speakers voice coil inductance. The Zobel counteracts the rise in impedance due to that inductance. I'm not saying that is the cause of the fizz 100%, but I never hear it. And perhaps that's why some people have it and some not - it depends on the impedance of the speaker.
You have to remember EL84's with out feedback have a very high output z, so a high z load will likely sound like fizz.
JMHO

[JdJ]

So what are the differences between a conjunctive on the primary and the Zobel on the secondary? I did notice a small change in the spike depending on speaker which leads me to believe it is some sort of flyback or inductance issue - it also changes depending on what OT tap I'm using. Right now I have a selector wired in, so I'd imagine putting a conjunctive filter on the primary would be the simplest to execute in my situation...

[zaphod_phil]

The frustrating thing about the zener fix is its just that .. its a fix rather than understanding what’s causing it.

Paul's document never really honed in on what exactly could be the root cause...

I don't think I can completely agree with those comments. Paul Ruby did a thorough analysis of that one particular problem and accurately pin-pointed the actual cause. He then proposed an effective remedy for that problem which goes directly to the source of the problem but has no impact on the amp's gain or overall voicing - all the R and C values stay exactly the same. I think that's pretty clever.

The only thing that Paul was unable to figure out was why in some amps you didn't hear the buzz although it could be seen on the 'scope trace. That's the remaining mystery.

In my case I think I have 2 things going on. The ruby buzz and the fizz which is represented in the big overshoot on my previously posted oscilloscope traces.

Yes, there's this separate issue of a general tendency to a bit of fizziness in some 18W amps, which Paul didn't address. I have seen some reports of this issue being affected by choice of OT as well as power tubes. With regard to the OTs I wonder if primary winding resistance is an important factor. AFAIK early RS 18W OTs had a higher primary winding resistance than later ones (650 ohms vs 480 ohms), and I suspect that perhaps the higher resistance in some OTs may help snub overshoot spikes that contribute to a fizziness in the tone.

I'd imagine there has to be a way to keep the PTs in or close to class a while maintaining a healthy (but not too healthy...) signal to the pt grids. Aiken on his site, and zaphod_phil here have mentioned decreasing ground reference resistors and/or increasing grid blocking to help keep the PT's from premature cutoff.

Reducing the reference resistors simply reduces the effective gain of the PI and hence drive to the power tube grids. Increasing the size of the grid blocking resistors limits grid current and also rolls of more upper frequencies. Also FYI the EL84 design guide says to use a minimum value of 10k for grid block resistors. So I've often wondered how come the Marshall 18W (or Watkins Dominator) uses 8.2k and Vox/Matchless amps use 1.5k.

I'm beginning to wonder if I have some PP imbalance that is resulting in the overshoot. I'm going to try replacing the coupling caps and all of the grid blockers/refs for the power section - there seems to be a small difference on my ohm meter between both sides...

I would really doubt that. And in any case I always maintain that some imbalance is desirable to help all the nice even harmonics not get heavily attenuated.

Makes you wonder if Paul actually knows more than he is telling about the cause of the buzz, the rocket builds will be slamming the power valves too.

IIRC TW Rocket amps use the same 220k grid reference value as Vox/Matchless amps. So they aren't really slamming the power tubes .

According to a Tonequest article interviewing Dr. Z, they fit a conjuctive Filter in every 15/18W EL84 amp as standard.

The Watkins Dominator, which the Marshall 18W derives from, also has a conjunctive filter. But I thought they were only meant to be used to correct the frequency response of cheap OTs.

[guitarmike2107]

I don't think I can completely agree with those comments. Paul Ruby did a thorough analysis of that one particular problem and accurately pin-pointed the actual cause. He then proposed an effective remedy for that problem which goes directly to the source of the problem but has no impact on the amp's gain or overall voicing - all the R and C values stay exactly the same. I think that's pretty clever.

Yes he pin pointed the cause of the fizz from his Scope and Excel calcs but like you said he didn’t find or report the root cause of the differences between amps.. that’s what would be good to know with the aim of building more consistent amps.
The buzz fix is pretty clever, I was not meaning to slag up Paul, Sorry he has saved many an 18watt build from certain death.

IIRC TW Rocket amps use the same 220k grid reference value as Vox/Matchless amps. So they aren't really slamming the power tubes .

I guess that’s just terminology, I would say any amp that gets the power valves distorting before the pre-amp is slamming the power valves, maybe my terminology needs corrected

AFAIK early RS 18W OTs had a higher primary winding resistance than later ones (650 ohms vs 480 ohms), and I suspect that perhaps the higher resistance in some OTs may help snub overshoot spikes that contribute to a fizziness in the tone.

That’s a good point, I have an SE OT on my bench that has significant overshoot when the power valve is driven hard and very low DC resistance on the primary, would placing a resistance in series with the primary be a good enough simulation of a higher primary resistance?

According to a Tonequest article interviewing Dr. Z, they fit a conjuctive Filter in every 15/18W EL84 amp as standard.

Maybe that’s why they call them “Doctor” Z amps .. every amp gets a fix

I can't help but wonder if the fizz issue is just a result of using a HiFi OT with the extended frequency response and EL84's, If you listen to the video Tom Olsen posted in another thread of the new Marshall modelling amp the 18watt tone they have modelled is quite fizzy compared to the other stock Marshall tones. Also most 50w/100w Marshall’s OT’s roll the highs off earlier than the18watt RS trannies. Is the conjunctive filter not just doing the same thing .. or maybe a simple cut control would help a bit too


Mike

[conger]

As an observation, you don't seem get these type of conversations going on at the Vox, Lightning (Trinity) and other forums that discuss PP EL84 derivatives. I know that some will say that the Vox PI introduces a fizz of its own, but it may be significant that driving the EL84's with a pair of 220K's is an important factor in avoiding the buzz/fizz.

Just thinking out loud really and throwing this into the pot.

[JdJ]

...but it may be significant that driving the EL84's with a pair of 220K's is an important factor in avoiding the buzz/fizz.

At least in my particular case 235k grid ground reference resistors did not get rid of the fizz...

[zaphod_phil]

Yes he pin pointed the cause of the fizz from his Scope and Excel calcs but like you said he didn’t find or report the root cause of the differences between amps.. that’s what would be good to know with the aim of building more consistent amps.

Sorry to harp on about this, but actually IMO he got down the very root cause, and provided the cure. IIRC he couldn't figure out why in some cases the problem wasn't audible, when it was still actually *present* in the amp - which is a somewhat different matter.

I guess that’s just terminology, I would say any amp that gets the power valves distorting before the pre-amp is slamming the power valves, maybe my terminology needs corrected

In a typical Vox or Matchless amp the PI starts clipping before the power tubes, which is when they can start to sound fizzy. The power tubes also aren't driven very hard in comparison to a Marshall 18W. You need to turn the volume all the way up on an AC15 or AC30 to get the Brian May overdriven EL84 sound.

That’s a good point, I have an SE OT on my bench that has significant overshoot when the power valve is driven hard and very low DC resistance on the primary, would placing a resistance in series with the primary be a good enough simulation of a higher primary resistance?

Yes, the use of snubbing resistors in series with the OT primary is a known technique. It might be interesting to try some different resistor values.

I can't help but wonder if the fizz issue is just a result of using a HiFi OT with the extended frequency response and EL84's, If you listen to the video Tom Olsen posted in another thread of the new Marshall modelling amp the 18watt tone they have modelled is quite fizzy compared to the other stock Marshall tones. Also most 50w/100w Marshall’s OT’s roll the highs off earlier than the18watt RS trannies. Is the conjunctive filter not just doing the same thing .. or maybe a simple cut control would help a bit too.

You have a good point. There are quite a few posts from the past about certain 18W OTs which have a slightly rolled-off high end being less prone to 18W fizz - which I guess means they're less hi-fi-ish. I wonder if those ones have a higher winding resistance...

As an observation, you don't seem get these type of conversations going on at the Vox, Lightning (Trinity) and other forums that discuss PP EL84 derivatives. I know that some will say that the Vox PI introduces a fizz of its own, but it may be significant that driving the EL84's with a pair of 220K's is an important factor in avoiding the buzz/fizz.

Yes, that's something I've noticed as well. Since the power valves aren't driven so hard in these amps, they get less grid current related issues.

At least in my particular case 235k grid ground reference resistors did not get rid of the fizz...

Which suggests in your amp it's not a grid current issue. Using something like 22k grid blocking resistors would also help confirm if that's the case. Maybe the problem in your amp is more at the OT end...

[JdJ]

Which suggests in your amp it's not a grid current issue. Using something like 22k grid blocking resistors would also help confirm if that's the case. Maybe the problem in your amp is more at the OT end...

I'll try the 22k blockers just to be sure, but 18k2 didn't seem to do much. So if I were to try some snubbers or a conjunctive filter, what values would be a good start? Does more primary winding resistance = more HF roll-off? 5W on the plates adequate?

[VTAMPSUSA]

Might just want to try putting a couple 100R resistors between the primary and each plate and see if that damps it out before you go to the Conj Filter or Zobel.

IMO the Zobel is probably a better tone preserver because it is merely altering the reflected impedance and not messing around with the interactions of the primary and secondary.

I would use the values Randy (ontariomaximus) stated of 2u2 in series with 10R in parallel with the 8R tap (IIRC) - his EL84 amps sound good to my ear. Couldn't hurt.

[JdJ]

Might just want to try putting a couple 100R resistors between the primary and each plate and see if that damps it out before you go to the Conj Filter or Zobel.

Thanks for the suggestion. I have a couple of 150ohm 6.5 watt resistors here - is that rating okay, or should they be more robust? I am running at max dissipation right now...

[VTAMPSUSA]

That should be fine. I doubt an EL84 will draw over 200mA!

[katopan]

I can't help but wonder if the fizz issue is just a result of using a HiFi OT with the extended frequency response and EL84's....

Very interesting point. The direction of this thread is the first plausible reason that has surfaced. Maybe our beloved 18W is that close to this fizz being audible or not, from all the reasons mentioned, that it doesn't take much with tolerances to balance one way or the other. OT primary resistance, speaker selection and even EL84 plate resistance will have an effect on the dynamic characteristics of the reflected inductive load.

And with all the other tests you've done Josh, you've ruled out everything else we could think of.

So what are the differences between a conjunctive on the primary and the Zobel on the secondary? ..... Right now I have a selector wired in, so I'd imagine putting a conjunctive filter on the primary would be the simplest to execute in my situation

A filter on the primary is subject to the valves cutting off through the class B part of the signal swing. On the secondary conduction is all the time. I have no idea on what difference this would make to the sound. You can still fix the filter to the 8 ohm secondary even if you are switching the speaker between taps. The idea is just to reflect a load in parallel to the speaker at higher frequencies to 'cap' the max impedance, and this can be on a separate winding.