18 watt fizz: please help me diagnose this?

[JdJ]

Okay - this is getting silly!

Today I tried:

- twisting the plate/ot primary leads + 150ohm resistors in series
- 100k PT grid blockers
- replacing the ground refs (470k), Coupling caps and PI plate resistors
- adding 2 470ohm screen resistors stemming from the existing 100ohm to isolate the screens
- removed all wires from the OT secondary and connected new wires directly to the speaker jack (bypassing the selector)
- reduced V1 plate resistor to 68k (was 100k)
- added a Zobel using a 10ohm and a 1uf Cap
- tried a new cab & 16 ohm speaker

What did I find? The Zobel helped tame the "blaaaat" fizz, but I can still hear it in the background. Nothing got rid of the leading edge spike on the scope when the volume goes above 7.

I'm thinking my OT is questionable, which I find hard to believe, but what else can be going on!?! Paging Greydon...

[ontariomaximus]

Add another 1uf or use a 2u2 in the Zobel.
The 10ohm + 2u2 would be average values to counteract the speakers VC inductance, but it can be tweaked to taste.

[kleuck]

Here, a Zobel calculator.

[rock_mumbles]

I don't have the fizz problem with my amp but, who knows what I may experience in the future. Because of messing with PA speakers and crossovers I am somewhat familiar with Zobel networks.

When I used the above linked Zobel calculator for a speaker that has a Re = 6.4 and Le = 0.8 (1kHz) which may or may not be typical for a guitar speaker in general, but are "close" values to a speaker I have:

The calculated results: Rz = 8 ohms Cz = 12.5uf

So OM's suggested 2.2uf is 18% of the calculated value... I have some 4.7uf 400V film caps I guess I should do an experiment today to see I can hear the effect of Zobel or no-Zobel ...

[kleuck]

0,8 is not really a typical inductance for a single loudspeaker i think
I have a lil'buddy, wich have a pretty high inductance, and it's only 0.46 mH
And the impedance to use is not -imo- the real speaker impedance at 1Khz, but the "normalized" one you want your output to see with the Zobel mounted.

So for me it would be 10 Ohms/4,7 µf

[rock_mumbles]

I used the published data for the speaker and they listed Le @ 1K...

So I just put the 8 ohm + 4.7uf Zobel on my new (not at all broken in) $36 AES closeout Sica Solid 12-100 and since my 18W head is loaned out I used my cascaded preamp TMB 5E3 with drive on full, volume at noon, treble about 3 O'clock, mid and bass about 9:00 O'clock and the PPIMV at about 3:00 O'clock.

Without the Zobel the high end is way too "sharp" with my Tele, with the Zobel the sound is smooth, all of the high end "musical content" is there but the harsh biting high end is gone.

Now I'd like to try OM's recommended 2.2uf, (BTW, I should know by now that OM knows his stuff)

[zaphod_phil]

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.

Take a look at the WEM Dominator's interesting Conjunctive Filter - kind of a differential or "push-pull" variation. As we all know the Marshall 18W design was "influenced" by the Dominator, much as the JTM45 was by the 5F6A Bassman. It also makes you wonder why Wattkins thought the filter was necessary.

Lastly, the 18 watt is a very toppy amp, and its been a bit of an adjustment to get used to it.

That's the EL84s for you. But if you want to really hear a toppy amp, you should try a Hiwatt.

Unfortunately it is made more complicated by the GDS Vintage OT

I suspect those particular GDS OTs may be slightly more hi-fi than the regular ones.

To actually treat the blocking itself you need big grid-stoppers- the bigger the better, especially with EL84s. Even 100k isn't unreasonable

I'm currently dealing with this issue in a Superlite TMB, last night I upped the grids from 8K2 to 47K, and trailing ring/fizz was lesser but still there. I'll try 100K and see...

I have wondered why we didn't see higher values here, and as you say how other resistors are tweaked all over the place but no-one seems to change the grid blockers by much at all.

Unfortunately you run the risk of killing some highs if you make the grid block too large. So it's a matter of trying to find the best compromise.

I think people get frightened that it will kill the high end

Yes, it inevitably will, since that rolling off high end is the other useful function of a grid blocking resistor.

I'm thinking my OT is questionable, which I find hard to believe, but what else can be going on!?!

The OT is the last thing we want to blames, especially when it's a known high-quality product. But realistically we have to accept the problem can sometimes lie in the OT.

Without the Zobel the high end is way too "sharp" with my Tele, with the Zobel the sound is smooth, all of the high end "musical content" is there but the harsh biting high end is gone.

That sounds very promising!

[ontariomaximus]

I used the published data for the speaker and they listed Le @ 1K...

So I just put the 8 ohm + 4.7uf Zobel on my new (not at all broken in) $36 AES closeout Sica Solid 12-100 and since my 18W head is loaned out I used my cascaded preamp TMB 5E3 with drive on full, volume at noon, treble about 3 O'clock, mid and bass about 9:00 O'clock and the PPIMV at about 3:00 O'clock.

Without the Zobel the high end is way too "sharp" with my Tele, with the Zobel the sound is smooth, all of the high end "musical content" is there but the harsh biting high end is gone.

Now I'd like to try OM's recommended 2.2uf, (BTW, I should know by now that OM knows his stuff)

Glad it's workin' for ya. You may prefer 4.7uf over 2.2uf, which I consider an all purpose non-intrusive value.
Hate to keep beating the drum, but rising impedance due to voice coil inductance should be addressed. With EL84's high output z, I don't see how 30k primary load at HF wouldn't cause problems, like fizz.

[zaphod_phil]

Yet there must be hundreds (or more?) Marshall 18W type amps out there, including Marshall's own 1974X RI version, 99% of which don't exhibit this fizziness problem. Really makes you wonder....

[stevesuk]

Yet there must be hundreds (or more?) Marshall 18W type amps out there, including Marshall's own 1974X RI version, 99% of which don't exhibit this fizziness problem. Really makes you wonder....

Zp,
How can you possibly know this ?
A lot of people love the sound and also it only occurs with very high output humbuckers on a overdrive channel. It also only occurs with the guitar up full and at high volume, or with the relatively recent innovation of VVR's at high volume, but low VVR setting. This can accentuate the issue. Many users never get to that high volume situation, particularly the bedroom players.
Many users using tremelo, tmb and all the other variants only ever use strats or teles which are not going to drive an amp into overdrive territory.
Steve UK
www.valvepower.co.uk

[zaphod_phil]

How can you possibly know this ?

Good questions. Like I said it really makes me wonder. No hard and fast statistics, but over all these years that the 18W group has been around, including the yahoo group days, you would think it might have been mentioned at least a few more times than it has. I've worked on quite a few 18W amps myself, whether building or repairing them, and never actually heard the fizz problem on any of them in real life (using normal Les Paul H/Bs - not super hot ones). With VVRs it's a different matter, as other factors come in to play at low voltages, and yes Ii have heard fizz with a VVR turned low - and not only in EL84 based amps. Also please place my comment within its correct context of a response to Randy aka OM, saying he doesn't see how all EL84 amps wouldn't have this problem. I neither disbelieve Randy nor you, but stuff like this does make me wonder what's been going on all this time...

[stevesuk]

Some Quotes from a Google 'EL84 Fizz' search.
EL84 Fizz? Forgot how much I loved that tone…
My amp is actually a tweaked version of the stock ***** as it sports an EZ81 tube rectifier, plus a tweaked circuit that adds a bit more sag and sustain. The result is just a gorgeous overdriven tone that really brings out the best of the EL84 power tubes.

One thing about EL84 amps is that if they’re done right, they have a distinctive overdrive tone that creates a subtle top-end fizz when they’re overdriven. I’ve played others that drive the power tubes too much, and they sound very harsh and incredibly compressed. Jeff did this amp right, and while the power tubes do indeed compress a bit, the overdrive tone retains its open character, while adding that nice top-end fizziness that EL84 amp lovers have come to appreciate.

This one is so insulting from a Tonequest quote that I will only list the link.
Do not click on the link if you are of nervous disposition.
http://www.thegearpage.net/board/showth ... p?t=314014

There are many google links for EL84 fizz.
Steve UK
www.valvepower.co.uk

[zaphod_phil]

Yes, but there's not a great deal on 18watt.com about fizz, and there's quite a few folks on this site who really care about good tone. The debate on The Gear Page was (a) about PP EL84 amps in general and (b) seem to have a number of people with an existing anti-EL84/18W bias. So I couldn't really see a whole lot of value in what was being said - plus I personally tend to think that "lack of headroom" is a big plus point.

Edit: There's also the problem of symantics. What some may call fizziness maybe what I think of as the creaminess you get in the 18W tone. It's not always easy to tell if everyone is talking about the same thing. But there's no doubt there's a bad kind of fizzy/distortion that affects some 18W amps. Going by OM posts, maybe we should be looking at the speaker(s) as a possible culprit.

[ontariomaximus]

How can you possibly know this ?

Good questions. Like I said it really makes me wonder. No hard and fast statistics, but over all these years that the 18W group has been around, including the yahoo group days, you would think it might have been mentioned at least a few more times than it has. I've worked on quite a few 18W amps myself, whether building or repairing them, and never actually heard the fizz problem on any of them in real life (using normal Les Paul H/Bs - not super hot ones). With VVRs it's a different matter, as other factors come in to play at low voltages, and yes Ii have heard fizz with a VVR turned low - and not only in EL84 based amps. Also please place my comment within its correct context of a response to Randy aka OM, saying he doesn't see how all EL84 amps wouldn't have this problem. I neither disbelieve Randy nor you, but stuff like this does make me wonder what's been going on all this time...

Hi Zaph, I should elucidate a little more. I didn't say all EL84 amps have the problem. It's clear they all don't have the problem. Some do, and when it happens I am pointing the finger at the rising impedance of speaker - which can vary quite a bit, coupled with the no-feedback EL84's. I haven't measured the output impedance of an 18W amp (I guess I should but I think it's tricky with pentodes) but I'm guessing it's high. FWIW, I just set up an 18W with a 8 ohm load, and applied a test signal so I had .5VAC on the output. I then switched the load to 30 ohms, and the output rose to 1.5VAC, 3X.
The frequency response of the output stage will track the speakers impedance. If that's not an impetus to be wary of the load you present to EL84's, I don't what is.
In a general sense, I am saying it is always a good idea to at least considering neutralizing the impedance rise, because the issue could become a nuisance.

[Merlinb]

I haven't measured the output impedance of an 18W amp (I guess I should but I think it's tricky with pentodes) but I'm guessing it's high.

Measuring output impedance is easy peasy!
[/img]

[katopan]

The other way to think of a Zobel in this application is it's just a filter putting the resistor in parallel with the speaker after a certain cutoff freq, and capping the max impedance. There is a theoretical value that will cancel the inductance of the speaker, but to me the extra brightness (and extra bass around resonance) is the big plus of a high output impedance amp (ie. valves with no feedback). You don't want to kill all of that effect! But the speaker range is typically up to 4 or 5kHz.

From Josh's original CRO shot of the overshoot, the width of the spike is around 0.15 div on the scale, and the wave period about 5.3 div. He said the test waveform was 333Hz. If the overshoot spike is indicative of half the period of an oscillatory frequency, it's in the ballpark of 5.3/0.3 x 333Hz = 5.9kHz. Setting the RC to have a cutoff freq around or a bit below this will reduce the spike but not have too much effect on the upper end of the speaker response. With a 10ohm resistor that would suggest a 2.7uF or 3.3uF cap.

[jhaas]

Is a polarized cap OK in a Zobel filter?

Web searches show it with both polarized and non-polarized caps, and when polarized, orientation is inconsistent.

[ontariomaximus]

Yes I used polarized caps, and have the negative toward ground.

[zaphod_phil]

Surely, with the AC coming out of the OT secondary, you should be using a non-polarised cap in a zobel filter.

[ontariomaximus]

Far less AC voltage there than the ripple at the first cap in a PS. No problem using typical electrolyics.