El84 snubber - Anyone try this?

[jjasilli]

I've had problems with fizzy el84's. Tried large grid stoppers which helped.  I just blundered into thjis:  http://www.tubesandmore.com/products/K-SNUB1  and http://www.guitarplayer.com/article/smooth-your-sonic-ride/5505

Any experience with this?

[Tone Junkie]

I've had problems with fizzy el84's. Tried large grid stoppers which helped.  I just blundered into thjis:  http://www.tubesandmore.com/products/K-SNUB1  and http://www.guitarplayer.com/article/smooth-your-sonic-ride/5505

Any experience with this?

jjasilli I have looked at this many times I dont build much with EL84,s but was getting ready to use it in a rocket build Im starting.
Bill

{EDIT - un-tangled quote --PRR}

[PRR]

The numbers in the guitarplayer article don't add up?

Two octal tubes = two 0.01uF capacitors
Four octal tubes = two 0.22uF capacitors

Double the tubes leads to twenty times the capacitance??

Probably a typo.... 0.022uFd.

The original point of this scheme was: loudspeakers have inductance, their impedance rises in treble, and with pentode power (*assuming no NFB!*) this causes the treble response to rise. The speaker sounds very bright.

The suggested values do almost nothing up to about 10KHz, then stick roughly Nominal Load on the tubes despite the speaker impedance rise.

Large speakers (even an Eight) have a rise starting nearer 1KHz. So a 10KHz "conjoint" doesn't seem complementary.

Actually the speaker "inductance" is very "impure" so the impedance does not rise all that fast. I had to plot it out. The snubber has very little effect to 2KHz, but it keeps "8 ohm" system impedance below 15 Ohms, falling to 9 ohms at 10KHz. With no-NFB power pentode(s) the response is shaved in the 2KHz-4KHz zone (and above, but >5KHz is SO far down from the 2.5KHz peak that the ear can't tell).

So if that last octave is irritating you, this is one way to trim it.

However "most" good guitar speakers ARE designed around the amplifier interaction. Their tone-testers have balanced impedance against cone-cry and cone-angle to give a "good" response. And while snubbers are useful things and good gadgets to sell, they have NOT traditionally been used on guitar-amps.

And that capacitor has to be rated for BIG voltage if you play LOUD and LONG. Spec same as B+ will work for a while, but in my experience a HARD-played 110V radio will kill a 100V snubber cap in a few years.

So if you think you need it, re-consider your overall treble response. (It IS useful to have the setup to plot amplifier frequency response, on dummy-load and into a speaker.)

[DummyLoad]

try a different tube?

RicharD and l found a couple mods addressing EL84 fizz and sparkler sounding distortion: don't use a bypass cap + a LARGE-ish screen resistors - around 2K or so per valve, those two mods and 10k grid stoppers also helped. overall, i shy away from designs that use EL84 for this reason.

mark huss built a plexi with 6V6 finals that sounds nice. have any thoughts about something like that?

--DL

[phsyconoodler]

I've had the fizzy EL84 sound on some amps I've built and it's always when I use a different OT. I've used the snubbers with great success,and yes,it's a .022uf cap.
   They take some high frequencies off and that stops the fizzies in every case I've tried them with.
I've tried grid stopper resistor increases and they head in the right direction but don't work completely.
   I don't get any fizzies using hammond OT's or Heyboer. I built a 36 watter once with a Marshall JCM900 OT and it was a nasty fizz monster.The snubbers got rid of that.

[jjasilli]

PRR:  The numbers in the guitarplayer article don't add up?  I agree that the numbers seem weird, but the circuit originates from a venerable source and seems to work empirically. 

try a different tube?  RicharD and l found a couple mods addressing EL84 fizz and sparkler sounding distortion: don't use a bypass cap + a LARGE-ish screen resistors - around 2K or so per valve, those two mods and 10k grid stoppers also helped. overall, i shy away from designs that use EL84 for this reason.

mark huss built a plexi with 6V6 finals that sounds nice. have any thoughts about something like that?

--DL

My EL84 issue does not involve a scratch build.  I tried large grid stoppers and NFB.  The amp gas a fixed / cathode bias SW. It's fizzy in fixed bias too, so I haven't tried to eliminate the bypass cap.  Besides cathode biased power tubes are supposed to share a bypass cap for stability.  The steps I've taken have helped but not eliminated the issue. In 6V6 amps, different speakers or enough NFB cured the fizz.

I've had the fizzy EL84 sound on some amps I've built and it's always when I use a different OT. I've used the snubbers with great success,and yes,it's a .022uf cap. . .   They take some high frequencies off and that stops the fizzies in every case I've tried them with.
I've tried grid stopper resistor increases and they head in the right direction but don't work completely. . .   I don't get any fizzies using hammond OT's or Heyboer. I built a 36 watter once with a Marshall JCM900 OT and it was a nasty fizz monster.The snubbers got rid of that.

Thanks Psycho, that't right on point.

[jojokeo]

I listed most of the article which explains the "fizz" aka "transformer ringing" here if it helps? http://www.el34world.com/Forum/index.php?topic=16170.msg158016#msg158016
 
Eliminating the cathode bypass cap on cathode biased power tube(s) definitely removes gain, high end, & ultimately changes  the tone and response of the amp. This may not be everyone's cup of tea to remedy the fizzies but does give it a more "old school" feel and vibe? Same with high grid stoppers as far as removing high end and dulling down the sound. Using too large of screen resistor values and you start to lose output power and even can lower by a lot the amps headroom causing early break-up which probably isn't wanted either? I would suggest to always try changing the tube(s) first as some EL84s can be more sensitive to this effect than others. Also try your best to determine that you're not hearing parasitic mishaps too?

In reading about others' experiences regarding the snubbers which I believe one that comes to mind was by Mark Huss - he originally used a lot of same tweaks as tubenit's "smoothing" techniques trying to get his amp just right as it was bright and shrill. I don't remember the fizzy part being an issue or not? But after relenting and installing the snubbers he was able to remove all of his various smoothing caps in all the usual places along with any other tweaks he tried and was able put the circuit back to the original basic stock condition which resulted in much happiness.

[Merlin]

I just blundered into thjis: http://www.guitarplayer.com/article/smooth-your-sonic-ride/5505

Ah, another example of GuitarPlayer not have the faintest idea what they talking about. Brick wall diodes? Shock absorbing OTs? What a time to be alive.

[HotBluePlates]

Ah, another example of GuitarPlayer not have the faintest idea what they talking about. ...

Dave Zimmerman of Maven Peal wrote the piece, though I hear ya that Guitar Player won't be stepping in with any tech editing.

[jojokeo]

I'm not trying to stick up for the magazine or the writer of the article but it could just be poor analogies in trying to explain something very technical to the mostly non-technical public in helping not only to gain a basic understanding but also to try to generate business for those in the industry (& those with invested interest directly with the mag) in doing so, which is one of if not THE main objective and business strategy of GP.

As with most things seen & read in life, take with a grain of salt - (and a shot of tequila if necessary)

[HotBluePlates]

The bottom-line is that the circuit alteration the article is proposing is functionally no different than the "conjunctive filter" or "corrective filter" re-popularized by Dr. Z and proposed by RCA in their tube manuals as a way to correct the frequency response of an OT when negative feedback is not used.

The reason for the issue is as PRR stated: speaker load's impedance rises with frequency, resulting in greater output voltage for highs. A cap and series resistor in parallel with the OT primary causes a falling impedance with rising frequency, counteracting the natural rising impedance.

RCA's method recommends a fixed amount of resistance proportional to the OT primary impedance (they suggest 1.3 times the primary impedance), and suggests you experiment with the cap value to find a value with equal output on a speaker load (NOT a dummy load resistor) at 400Hz and some higher frequency (1kHz or higher).

The suggested procedure merely breaks the resistor into 2 parts (of half the total composite value) and the cap into 2 parts (of double the composite value and half the total voltage rating) with low-wattage resistors to balance the voltage across each cap. Not a bad idea, as a single cap with the needed voltage rating might be cumbersome.

All that said, there are some statements in the article that are technically wrong. At first, I was on-board with the "it's for the non-tech audience" but they're gonna be soldering and poking their amp right at the highest voltage in the amp, so they need at least a little tech-knowledge.

Also, the sonic result will be what he claims, so I didn't jump on it at first. If it wouldn't do what he claimed, I would've pointed that out earlier. It's just that the harshness it kills doesn't happen for the reasons he says.

And I remember how much stuff I had to unlearn that was presented in the first books and articles I read when I first got into this stuff. I argue that Merlin and PRR show you that "technically accurate" and "easy to grasp" aren't mutually exclusive, even if you don't get all the implications until years and many more books later.

[Merlin]

I'm not trying to stick up for the magazine or the writer of the article but it could just be poor analogies

Not just poor analogies: Removing diodes might put your Silverface back to Blackface specs, but it will also have precisely no effect on the sound at all. It just removes what meagre protection they provide, and removing harmless protective features is bad advice if you ask me. They don't prevent saturation, or whatever the articale implies. In fact very few guitar OTs saturate anyway- it is a myth.

[labb]

This is a pretty good article by Paul Ruby on EL 84 fizz or buzz.    http://www.18watt.com/storage/18-watter_buzz_info_311.pdf

[jjasilli]

Thanks, Labb!  Great in-depth article.  Clearly this issue is widespread.  Seems like the Paul Ruby zener circuit prevents the issue, while the snubber circuit addresses it after the fact.

Another aspect is NFB.  The info on hand seems at least to imply that the snubber circuit should not be used with NFB (I'm not clear on this).  The Ruby schematic does not show NFB, but that may be irrelevant to the topic. Ruby ID's the culprit as crossover distortion.  So the use, or non-use, of NFB may not matter to the EL-84 fizz issue.  

Any thoughts on this?:  This leads to the tentative conclusion that the Ruby fix is more flexible, because it leaves open the option for NFB if wanted for some other purpose, like maybe a presence control, whatever.

[labb]

There was a lot of discussion about this problem on the old 18watt (www.18watt.com) site a few years ago. That site crashed but I think that is is back up and operational now. Zaphod_Phil and Graydon are two of the mods over there and they are both helpful. Might try asking them. Been sometime since I have been on that site. It is specific for the 18 watt Marshall only.

[phsyconoodler]

I've tried the Paul Ruby fix with a couple of amps.It didn't work as well as the conjunctive filter.Especially with VVR.
 I had a Reinhart 18 watt amp in that was fizzy and it had the Ruby fix.Didn't work at all.Removed it totally and used the filter.Worked perfectly.

 Crossover distortion is what gives the EL84 it's unique tone.

[HotBluePlates]

... Another aspect is NFB.  The info on hand seems at least to imply that the snubber circuit should not be used with NFB (I'm not clear on this). ...

You can use the "snubber" with NFB. It's just redundant.

"Snubber" is a wrong name, taken from a similar circuit. A snubber suppresses inductive voltage spikes that occur when a coil is attached to the output of a diode, like you might find in a choke-input power supply. Voltage spikes could happen when the diode switches off, and a snubber combats those spikes.

This type of "corrective filter" (to use RCA term) counteracts the rise in load impedance due to speaker inductance, which would cause increased output for high frequencies. The filter presents a falling impedance at highs, so it attempts to make the overall impedance presented to the tubes more-constant. In a way, it is a fixed tone control.

RCA suggested using it if NFB wasn't applicable around the output stage. That's because if you have a rising impedance for high frequencies, and therefore a greater output to the speaker for highs, then you also have a larger source voltage at highs for your negative feedback, and thereby more feedback for high frequencies. That in turn knocks down the gain of highs more than for other frequencies, and tends to even out the response.

So it's not that you can't use a "snubber" with NFB, just that NFB alone will get the same job done. [Now when you create a big enough output signal that the stages inside the loop run out of headroom and the feedback collapses, you'll have an abrupt shift from balanced output to treble-accentuated output.]

... Ruby ID's the culprit as crossover distortion.  ... This leads to the tentative conclusion that the Ruby fix is more flexible ...

There's more than one possible cause. So there's more than one possible fix.

Crossover distortion could be one cause, or rising impedance due to speaker inductance, or simply distortion and too much response at treble frequencies, or ...

Most of what Ruby keeps calling crossover distortion seems to be about blocking distortion, and he even notes both amps show crossover distortion in their traces. Further, he said of the circuit to rid "crossover distortion", "I added the same circuit to the PAM and it has no effect on sound, despite visibly getting rid of cross-over distortion. How can this be???"

That seems evidence that the root problem wasn't identified, given that he added the circuit to the good amp (PAM) which he said didn't have a buzz issue to begin with.

[jojokeo]

Most of what Ruby keeps calling crossover distortion seems to be about blocking distortion, and he even notes both amps show crossover distortion in their traces. Further, he said of the circuit to rid "crossover distortion", "I added the same circuit to the PAM and it has no effect on sound, despite visibly getting rid of cross-over distortion. How can this be???"

"MAGIC!"

[HotBluePlates]

It occurred to me after posting... The reason for overshoot/undershoot in his scope shots for the "bad amp" is simple: when the signal is nearly-square and you get overshoot, you have oscillatory response. Meaning oscillation.

Dunno what is the cause of his oscillation; we don't see as-square signals with the resistor load so maybe it only appears with the addition of the speaker load, and might be due to the OT or its wiring.

"MAGIC!"

 

That is what he said! Maybe the magic was the good amp didn't have a parasitic oscillation.   

[sluckey]

The reason for overshoot/undershoot in his scope shots for the "bad amp" is simple: when the signal is nearly-square and you get overshoot, you have oscillatory response.

Could be as simple as a scope probe that has not been adjusted properly. Nothing in that presentation makes me think he found the cure for the fizzies.

[jjasilli]

Hotblue:  good post!  But note -- So it's not that you can't use a "snubber" with NFB, just that NFB alone will get the same job done. The problem remains that many people, including me, are finding that even a big dose of NFB is not fully curing the EL84 fizz, even with large 36K grid stoppers.  However, I just realized the screen resistors are 470R.  KOC recommends 2.2K I think.  I messed with screen resistor value during the re-build / biasing process; but not with regard to the fizz which I realized later.  If I remember, the 2.2K value was tamping down the performance and tone of the amp.  Got to keep better records!

Also, Psycho's practical experience seems compelling.

[Shrapnel]

There was a lot of discussion about this problem on the old 18watt (www.18watt.com) site a few years ago. That site crashed but I think that is is back up and operational now. Zaphod_Phil and Graydon are two of the mods over there and they are both helpful. Might try asking them. Been sometime since I have been on that site. It is specific for the 18 watt Marshall only.

Yupp. DO NOT make references to it outside of a Marshall 18 Watt or Watkins/WEM circuit. It will get the conversation (if in the public forum) Locked and deleted for being off topic.  If the circuit isn't Watkins/WEM or Marshall 1958, 1973, or 1974 related, in their opinion, it is off topic (for the website) and don't belong.

Other than that, they do seem pretty good guys over there.

[jojokeo]

 

That is what he said! Maybe the magic was the good amp didn't have a parasitic oscillation.    

I have a close friend that's a great magician, and slight of hand is one of the oldest tricks in the book. O'bama is also a very good "political" magician, i.e - everyone lets talk about & look at something over here (while he's speaking & doing something very different somewhere else). I won't go into political argument just wanted to make a current & relevant example about his M.O. and usual tactics when things don't go as planned or designed. He can squeeze through some mighty small holes with those big ears of his.

the 2.2K value was tamping down the performance and tone of the amp.

Yes, be careful w/ high values for screen resistors. A while back when using higher voltage PTs and mixing & matching various power tubes, I worried too much about exceeding screen pDiss max ratings and learned this lesson.


*Regarding the Ruby mod: he speaks of finding the right zener voltage just past the clipping point after the pi which he thinks gets rid of the fizzies - if you had a ppimv I would think you could dial-in a setting that would essentially be accomplishing the same thing? It is morning here while sipping my coffee so not totally awake & a bit fuzzy from last night but this could be a simple way to test my fuzzy-fizzy theory???