Always trust your meter

[MR COFFEE]

Hi all,

I have personally experienced the hassle of trying to find replacements for burned out output trannies (mostly Marshalls), and I can see why I might seek out a solution like RG's tube saver circuit to try and avoid that, but it seems like a rather elaborate circuit protection scheme, especially if you don't happen to already have a low voltage power supply in the amp, just to provide somewhat better protection than what a couple of carbon comp resistors will for fifty cents with a lot less complexity.

Carbon comp resistors perform fairly well in that short-sensing role most of the time, although maybe a higher wattage metal film might have a better time constant and fuse a tad quicker for increased protection against tubes shorting out..., the difficulty is avoiding a bad tradeoff with nuisance tripping like Aron had, where it fused without a clearly shorted-to-stay tube. Who knows what the best tradeoff would be between protection against long-term damaging shorts and those damn brief ones that clear without anyone even noticing... I sure don't  :lol:  

Even with comparators and RC delays, I'd just be guessing where to set the current limit on the B+ cutoff to protect the output tranny against shorting without causing "nuisance tripping"... and damn if most guitar players I know don't think their amp only sounds good when it's on the edge of self-destruction anyway...  :wink:

I used to replace blown resistors just like Doug and others have suggested with oversized wirewounds that DON'T fuse to protect the output tranny until I read the same rap Aron was talking about - maybe in one of Aspen Pittman's books or somewhere, and thought, "oops, maybe keeping that resistor from burning up isn't such a good idea after all!").
It's a simple solution some engineer (or maybe a tech?) came up with that works pretty well most of the time. The only reason I can think of to put those resistors in the cathode circuit is to act as fuses. I can't imagine they do much to promote current sharing with mismatched tubes or anything else in a 4 output tube power amp stage, and they have no useful pupose by any stretch of the imagination in a plain push pull stage with 2 tubes. Putting a fuse in series with a big 1 ohm resistor just means I didn't know why the design engineer put it in the circuit in the first place. I didn't used to.  :oops:

I agree with RG about the reliability of solid state electronics, in general, but I haven't been too impressed with the reliability of solid state electronics in high temperature environments, be it under the hood of a car, or in a hot tube amp chassis. And to me, anyway, part of what can make tube amps so reliable is their simplicity of design, and as long as they are maintained, (which means more than replacing and biasing tubes, it means pulling the chassis and looking around, replacing electrolytic filter caps and things like those carbon comp resistors in the cathode circuit if the value has drifted more than about 20% or if they just LOOK like they're too old or have gotten too hot, etc.), tube amps are equally, if not MORE, reliable than most solid state amps, especially in terms of catastrophic failure. Just my opinion.

and as RG says, YMMV

[puretube]

Rc`s make biasing easier, by measuring the voltage across them...

[Eric H]

Rc`s make biasing easier, by measuring the voltage across them...

Haha...the thread comes full-circle. I remember a thread at Ampage where the consensus was to use 10ohm 1/2 watters because (in addition to the ease of bias calculation) they --might-- offer some protection from voltage spikes as a side benefit.. The main use was for biasing, though.

This is all current for me since I just completed an amp build where I incorporated --some-- of the protection ideas I've seen discussed at Ampage:
AC input filter with MOV's across it .
Fused mains, secondary, and heaters.
high-voltage cap and diode to lessen arcing across standby switch.
RG's voltage-bucking transformer method to reduce incoming voltage to 110 (I'm using a 45 year-old power-transformer)
10 watt 250ohm resistor across output-transformer-secondary.
The 10ohm resistors being discussed.
a few things I've forgotten ;-)

Between all that stuff, and some feature creep: switchable cathode/fixed-bias (with individual bias-pots) etc. I've turned a simple push-pull 6v6 amp into a complex bunch of wiring. Some of this stuff probably alters tone. I can see now, that when I have future problems, the diagnosis will likely be more difficult , since I've actually increased the potential failure points. Where do you draw the line? Transformers are expensive, but in the rare event of a blown transformer in ,say, a fender deluxe the diagnosis is fairly simple, and wastes some money, but little time. My time is valuable, too.
As Bart noted, much of the beauty of tube-circuits is in their simplicity (and, certainly, some of the tone) and this amp is no longer simple.

Fun stuff, huh ;-)
I wonder how much hair RG (or Puretube, etc. etc.)  has left after weighing pros and cons for those commercial amp designs .

-Eric

[Doug_H]

I can see now, that when I have future problems, the diagnosis will likely be more difficult , since I've actually increased the potential failure points. Where do you draw the line? Transformers are expensive, but in the rare event of a blown transformer in ,say, a fender deluxe the diagnosis is fairly simple, and wastes some money, but little time. My time is valuable, too.

Eric, I run into this kind of stuff with embedded s/w a lot. E.g. people want triple redundancy, voting logic, etc, things that supposedly make the system more robust. It works to a point but above a certain threshold the cure is worse than the disease. If overdone, these kind of schemes can actually make the system more vulnerable to failure (less robust) due to overcomplication and more difficult to maintain because with automatic switching schemes and so forth the actual behavior of the system can be much more difficult to trace. Sometimes it just makes more sense to keep it simple and clean, build it right, and then test the hell out of it.

Not sure how this all translates into simple vacuum tube circuits, but before implementing a lot of this stuff you need to at least run an informal risk analysis in your brain. What chance is there a particular failure will occur? What is the cost (time & money) of implementing 'protection'? If I add protective circuitry will it make the overall amp harder to maintain and more vulnerable to failure?

For example, fusing a B+ line is pretty cheap and simple. It shouldn't overly complicate maintenance and may save an output transformer if an output tube fails, which is not a completely unlikely event. OTOH, complex circuitry that guards against low probability failures may not be worth the effort. Another consideration is that commercial gear will have a much higher probability of risk for some failure modes than a piece of gear you build for yourself. As the designer/builder you are a much more "educated consumer" .:wink:

Where do you draw the line?

[...]

My time is valuable, too.

Well it's different for everyone, depending on the circumstances, but I think you just answered your own question... :wink:  :wink:

Doug

[Eric H]

Not sure how this all translates into simple vacuum tube circuits, but before implementing a lot of this stuff you need to at least run an informal risk analysis in your brain.  

Yep --that's where the problem started ;-)

Well it's different for everyone, depending on the circumstances, but I think you just answered your own question... :wink:  :wink:

Doug

haha...for today anyway

-Eric

[aron]

Funny you should say that Eric. I've been in the process of tearing out all the features from my old Bassman amp.  Now only the left input works (along with the controls) and it's a single channel + 2X6L6 amp.

It still has problems (no doubt do to existing crap inside) so I am trying to take the rest out.

It starts oscillating when I hit only low notes and it's a horrible grating oscillation. It only happens "sometimes" and that makes it annoying.

I'm going to tackle it again very soon.

Thank goodness I have my Traynor amp for gigs.

[will]

Hi,

Just wondering, if the surge that killed the cathode resistor is due to high voltage turn on surges before the heaters get up to proper temperature?  Is this what the standby switch is designed to prevent?

Regards,
Will

[R.G.]

Eric, I run into this kind of stuff .... It works to a point but above a certain threshold the cure is worse than the disease. If overdone, these kind of schemes can actually make the system more vulnerable to failure (less robust) due to overcomplication and more difficult to maintain because with automatic switching schemes and so forth the actual behavior of the system can be much more difficult to trace. Sometimes it just makes more sense to keep it simple and clean, build it right, and then test the hell out of it.

Not sure how this all translates into simple vacuum tube circuits, but before implementing a lot of this stuff you need to at least run an informal risk analysis in your brain. What chance is there a particular failure will occur? What is the cost (time & money) of implementing 'protection'? If I add protective circuitry will it make the overall amp harder to maintain and more vulnerable to failure?

True. There's not much in the way of making things more reliable that can be done successfully without using your brain.

There are four classes of "reliability" enhancers.
(1) Do nothing except make it tough ( I class "test the hell out of it" in here")
(2) Put in some first order protection stuff.  This is in the "one fuse and pray" league.
(3) Do a number of well thought out cost-benefit evaluated catch measures
(4) Do stuff the space programs will buy. How the heck will they know that the protection stuff is what killed it?

Engineering from the now-kaput Soviets tended to be of type 1 and 2. Done properly, it's great. Golden Age tube equipment often did type 0 (that's nothing at all) or type 1 if someone was on the ball. Fusing B+ was controversial in the day. Useless frippery, in some opinions.

(3) is where we get to with a lot of modern stuff. In the 60's, hifi/stereo amps most commonly turned to flame and smoke if you ever shorted the output, even momentarily. We now know that the two transistors with several Rs, Cs and diodes that limit or fold back short circuit currents are a good investment. But the tweako hifi crowd pulls this circuitry OUT because it somehow pollutes the music. They use a lot of output transistors in their quest for pure sound.

IMHO there is a continuum of protection you can do. If you don't understand what you're doing, you ought to stay out of it. I'm sure you've run into this from your description. However, the level of active protection gets more cost effective as we go. From where we stand now, it's hard to imagine being able to sell a stereo amp that will self destruct just because a whisker of speaker wire touches the other terminal.

I'd like to see someone do a cost benefit analysis on the cost of active limiting for output tube current versus the price of replacement parts weighted by the observed failure rates of these parts and figure out whether it's cost justified. If anyone wants to take that on, I can make a really good guess as to whether they're close or not.

Unless we get to the Holy Grail of a tube sounding solid state amp or get so bored with tube sound that we let them die, tubes will someday be so expensive that using a tube or two worth of circuit to protect a tube will sound like a good deal.

Protection stuff is fairly deep design water. How do you know you got it right? Knowing you got it wrong is easy. But right? That's harder. Like the Three Mile Island tee shirt - "I survived Three Mile Island - I think...)

For what it's worth - the failure rate of the parts to do tube current sense  times their cost is about equal to the product of failure rate and cost for the transformers.  The bet you make is that on average, you save one more transformer per X units in the field than the users have to spend on the circuit. And then you have to bet that your users will believe that the protector protected them. That's a hard one. Users without an exemplary disaster on their hands won't believe that the protection stuff has operated even once.

It's a funny game.

[Eric H]

IMHO there is a continuum of protection you can do. If you don't understand what you're doing, you ought to stay out of it. I'm sure you've run into this from your description.

Heck, it was a long, frustrating build. I'd been without an amp for 2 years (this is the third time I've rebuilt this one) and I was getting anxious. Now that it's working (and sounds good) I'm happy I added the protection I did. I'm also confident in the mods --I haven't missed too many posts on the subject at ampage since you helped me rebuid this amp the first time in '98 .

Unless we get to the Holy Grail of a tube sounding solid state amp or get so bored with tube sound that we let them die, tubes will someday be so expensive that using a tube or two worth of circuit to protect a tube will sound like a good deal.

Great point

[Doug_H]

Engineering from the now-kaput Soviets tended to be of type 1 and 2. Done properly, it's great.

Yes, (and as an aside) if they survived the development/production cycle. Literally. Ever seen photos of some of their shops and labs back then? Yikes!

(3) is where we get to with a lot of modern stuff. In the 60's, hifi/stereo amps most commonly turned to flame and smoke if you ever shorted the output, even momentarily. We now know that the two transistors with several Rs, Cs and diodes that limit or fold back short circuit currents are a good investment.  

That's an excellent analogy. I remember when hi-fi output sections were that sensitive. I remember the first time I saw fused protection there, one fuse for ea channel.

IMHO there is a continuum of protection you can do. If you don't understand what you're doing, you ought to stay out of it. I'm sure you've run into this from your description. However, the level of active protection gets more cost effective as we go. From where we stand now, it's hard to imagine being able to sell a stereo amp that will self destruct just because a whisker of speaker wire touches the other terminal.

Yes, and you are developing a consumer product, as opposed to a hobbyist who is building a one-off for his own use. The emphasis is a little different, as I mentioned. I agree it is a continuum. Where your threshold of pain vs. gain is set is where it gets interesting.

Protection stuff is fairly deep design water. How do you know you got it right? Knowing you got it wrong is easy. But right? That's harder. Like the Three Mile Island tee shirt - "I survived Three Mile Island - I think...)

Well as I mentioned, my experience with this is from a software perspective. "Knowing I got it right" has haunted me my whole career. Determinism is a wonderful thing.  Overzealous protection schemes in s/w can reduce determinism and fault traceability considerably.

Like I said, I don't know how this translates to hardware or where you draw the line.  I like seeing this emphasis put on reliability though. Generally speaking, I have not seen this kind of thought or solid engineering put into reliability in guitar amps, at least on the boutique side of the business.

Doug

[R.G.]

I like seeing this emphasis put on reliability though. Generally speaking, I have not seen this kind of thought or solid engineering put into reliability in guitar amps, at least on the boutique side of the business.

I guess I'm ... gulp... about to find out how it works.