Ghost in the (Matchless) machine – a call to armchair quarterbacks

[cjrtrio]

Statesmen-

I come before you with the following regarding ghost notes in my recently-acquired and recently de-modded 1994 Matchless HC-30:

Facts: Had an experienced (retail!) amp guy in Dallas retube and de-mod it – he worked with Matchless to return it to original specifications using parts from Matchless including three 22mF 450V SAMHWA caps, three 33mF 450V BMI caps, one 250 mF 100V BMI cap. Ruby tubes in triode channel are super quiet and musical and I’m still plumbing the excellent sound of that channel.

Frustratingly, however, I now plainly hear 120hz-induced ghost notes at medium or high volume settings (e.g., 293.7 hz brings along ~123.5 hz – the usual 60/120 sneaking through). Same result regardless of cabinet, output tubes, recto tube configuration, channel, run time, instrument, polarity switch, etc. On low power setting, there’s less ghosting but it is still noticeably present.

Opinion: Tesla’s gift of alternating current does present challenges for an amp design featuring sag, but ghost notes in a fine instrument are unacceptable, period (especially in an amp that sounds this great in all other aspects).

Goal: Rather than sell the amp, I’d like to improve the design and arrive at the same sound (same sag, squish, etc.) – just void of ripple. The message I’m getting is that the Matchless is indeed built like a tank on the inside too – it is not altogether elegant and can be improved in certain respects, with filtering and isolation being notable examples. Can’t get any of the amp techs I know to show excitement about the undertaking. I think they’re getting stuck when I say “I don’t want the sound changed noticeably – just nix the ripple”.

…so,…

Petition: I humbly request some replies to this email containing artful “success stories” or specific instructions regarding removal of 120hz ripple from this amp or a similar one. So that I don't have to expose my current paucity of amp savvy, please assume the amp is brand new and all parts are running to spec (I played a neighbors' brand-new unnamed nice/expensive boutique 2XEL84 amp on Holloween - ghostorama - Poltergeist, even - so new amps apparently aren't immune).

I’m attached a pic of chassis if that will help.

Thanks. I'm jealous of all of you amp geniuses.

[Steve Conner]

A reasonable way to do it might be just to make some caps bigger:

1) Find the point(s) where the ripple is causing trouble- most likely to be the power tube screen and/or plate supply voltages. You can do that by temporarily hooking up a big-ass capacitor across the existing one at each point (but taking care not to put too much capacitance directly on the output of the rectifier tube, if you've got one)

2) Once you find the problem spot, try different capacitor values on it till you find the smallest one that gives an acceptable level of ghost noting.

If you're now unhappy with the sag characteristic, it starts to get messy. An alternative plan that may affect the sag less (or in a way that you prefer) might be to add a filter choke somewhere if the amp doesn't have one already.

This work involves dangerous voltages so we assume you know what you're doing.

[cjrtrio]

Steve-

1st off, This will be a directed study with me alongside very experienced amp guys - I will approach it like Bonaparte (1 hand...). One of the aforementioned techs has the amp now, so I'll be sifting this thread and communicating with him.

Thanks for a good start. I always hear folks extolling the virtues of their chokes - if my HC30 doesn't have any, I'd consider that.

I've also heard the following mentioned as possible culprits:

- gain stages that aren't isolated from eachother in the pwr supply
- Unless someone else tells me mine are appropriately isolated, I'd be in favor of mods there

- under-filtered "nodes" of the B+ supply that don't allow the capacitor to "recharge" quickly enough under heavy loads
- Is there a "faster recharging" cap solution, or is this just another way of saying "up the cap values"

English major question - does choking allow the PS to "sag without a ripple"? If so, that sounds like the drill. Whatever is most faithful to the original sound...

Isolation, chokes, raised cap values?...Calgon, take me away.

Chris

[Steve Conner]

It goes without saying that "Most faithful to the original sound" is a case of leaving the sucker alone. Many people consider ghost noting to be a legitimate part of the vintage amp sound. Especially the guys selling boutique amps with undersized filter caps, go figure

A choke can be very effective at getting rid of ripple, far more so than capacitors can do on their own, but shouldn't change the sag too much. You may need to add an extra capacitor along with it, or change one of the capacitors for two caps half the size with the choke slung in between them, to form a CLC filter.

[cjrtrio]

Steve (everyone)-

Ready for a laugh? I read your last post, figured I'd read up on the finer points of a CLC filter, and did a google search on "choke" and "slung"... (Search result links contained descriptions of physical violence with no discernable relationship to tube amplification).

I don't think I'll ever own a ghosting amp...unless (dreaming now) my executive producer wants me to.

I sent my amp tech this thread link - he didn't comment yet but he is going into the amp and I trust he'll consider the CLC approach. He wants to check the performance of the choke that's there first. I'll post details of his approach and the results by early next week.

Thanks again.

Chris

[stokes]

!20hz comes from the ac (60hz) being rectified.The job of the filter caps is to eliminate this 120hz ripple.It is quite possible that a simple recap is in order.Caps tend to lose their effectiveness when they are over ten years old,not a hard and fast rule,some last longer some go sooner.The caps I see in that picture look like the jap variety,if they are original I would say to start there.The jap caps are notoriously crap.You can replace the old caps with quality (Sprague and F&T's are my favorites) caps of the same value without changing the sag or tonal qualities of the circuit and most likely improve the 120hz elimination.Putting larger caps in your main and screen positions will tighten up the response and stiffen up sag to some degree,but weak,old caps will induce the 120hz ghosting you describe.I am surprised your techs didnt suggest a simple cap job.That has been a basic repair technique since the first amps were made.

[cjrtrio]

Unrippled sag, same sound/no ghosting

Terms like "unrippled sag" and "same sound/no ghosting" had us in a holding pattern (ie he was thinking "what's this guy on"?). After a few discussions, he convinced me about the "simple cap job" you speak of, based on following:
* He assured me that his spragues were better than the stock samhwa caps which I had I opted for in a bid to return the amp to "stock" condition
* Radical changing of the filtering wasn't the right idea for this amp
* He could cycle through caps until he found the right values to solve the problem (he was guessing 30u, 500V)


When I build an amp (one fine day), I'll be going for these holy grails:
* sag without ripple
* high gain with low noise
* short signal path
* some sort of convection-based heat management

[stokes]

Sag and ripple are two different issues.If you want the sag you use a tube rectifier.With a tube rect.your main or first cap has to be within the spec for that tube,you can go a little higher but it isnt going to reduce the sag all that much.The ripple is kept in check by fresh,properly functioning caps.If you are hearing the 120 cycle hum,it is because the cap is not functioning properly and needs to be replaced,with a value that wont damage your rectifier.

[Steve Conner]

I and a few others have built amps with regulated power supplies that get rid of both ripple and sag completely. (At the cost of a serious infestation of silicooties!) You can then reintroduce as much sag as you like, without the ripple, by adding resistors and caps, or by building it into the regulator circuit.

Maven Peal make an amp like this I believe, with variable sag. They also made the B+ voltage variable, so their 100 watt head can be adjusted down to 1 watt. Maybe you should check one of their amps out.

In my own design, I just left it at fixed voltage with zero sag. I never bothered working too hard on it once I found out that Maven Peal already patented the idea and so I wouldn't be able to make (much) money out of it. Also, I found it was easy enough to make ghost notes and hum inaudible without the regulator, by just using dirty big filter caps and a choke. (With a solid-state rectifier, since the large first cap would kill a rectifier tube, like stokes said)

Then again, Randall Smith patented the half-power switch and that never stopped everyone else using them

[stokes]

Steve,the sag you describe is introduced in the supply by the resistors.Its an old trick to induce sag to a SS rectifier,just put a resistor on the output of the rectifier.In a tube amp there is no such thing as zero sag,since all the tubes will sag to some degree.You can have less sag,but not zero,unless you go SS throughout and get rid of your tubes altogether.But you can get close to zero 120hz hum and have lots of sag.To "adjust" the amount of sag in a power supply you would do better to alter the size of the dropping resistors in the supply.

[Steve Conner]

Stokes, sag is defined as how much the B+ voltage drops (sags) when loud notes cause the power tubes to draw more current. In a regular P-P pentode amp, the screen voltage is probably the most sensitive to sag, since it modulates the gain of the tubes.

So by definition, a power amp with regulated plate and screen voltages has zero sag. The tubes themselves do have less gain for loud signals than quiet ones, partly due to their own characteristics and partly due to the screen resistors. But you would call that power compression, or distortion, not sag.

Also, for sag to be noticeable as such, it really needs to have a R-C time constant. Putting too-big screen resistors on your power tubes will make them compress drastically, but because it happens instantaneously with the signal waveform, that's classed as distortion. To get something that sounds like sag, you'd need to hook a capacitor from the screen to ground too, big enough that the resulting time constant was within the 0.1 to 1 second kind of range.

IMO.

[stokes]

So what you are saying is that in your regulated plate and screen supply your tubes voltage and current remain constant when you put a signal thru it?I know this can be reduced to some degree,but I wouldnt call it zero sag.Zero sag would produce a sterile, solid state,effect-ugh!

[cjrtrio]

Sagging, Ghosting, and age

As I age, it is clear to me that I'm sagging more and more - and I want to do as much sagging as possible before I cross over into the next plane and begin issuing ghost notes from my electric lyre.

Update-

* We did a recap on the HC-30
* Ghosting was partially addressed - I could have lived with it, but...
* I was told that solid state rectification could further improve ghosting
* I read that Komet Constellations are solid-state biased and not ghosty
* I played a Constellation, and...
* Sold the Matchless (will be getting a Connie soon)

Thanks for the thread. Information is most excellent.

CJRTrio

[cjrtrio]

Komet Constellation Praise

I have my Komet Constellation (I like "KCon" better than "Connie" - I never knew a really ravishing hot femme named Connie).

It ghosts just a bit, but do I care? (If you know about how transcendant this amp is, then you know what a state of bliss the new owner finds him/herself in).

I may buy another one - it is worth every dollar.