Tubes that can handle 500VDC on screens, or tame the power supply??

[Gtr0]

Thank you for the suggestions. Much appreciated.

I had prepared a long reply to most comments and somehow the forum asked me to sign back in and only the quotes remained in my reply.... [breathe]. So I will try to just address the overall comments.

Let me take a step back and give a better overall story of what's going on. I first started learning and buying parts from Metro amp back in 2004 to upgrade my then JMP mater volume. As time went I started replacing more and more, then started building whole amps. But this is where I have done the bulk of my learning. Perhaps some of the info from the old Metro forums wouldn't apply to high gain amp techniques. Such as biasing to 70%. And I admittedly say I have always pushed those boundaries until two days ago. Now I get the idea that it is not such a good idea. Especially with higher voltages such as what I am seeing here.

Having been building amps on and off for 14 years, you do learn some things, but learning this way leaves HUGE holes in one's education. I am trying like hell to fill those gaps in but there is still some mystery to things that I am desperately trying to get a grip on (like impedance). The internet is so flooded with bad info (and good), it's tough to weed out. I have turned to books where possible... started with Merlin's books and next (?) who knows... I will cross that bridge when I get to it.

This amp initially set out to be close to a SLO. Looking at every schematic I can get my hands on though, I took some ideas from the dual rectifier (LOVE the 680pF on the treble control for example). The idea was that I would build one out, tune it to my taste, and use that as a template for what in my mind is the ultimate amp. I have been working on this amp for the vast majority of the last year and this is the third time that the amp has been rebuilt, from the ground up. So it is not this particular iteration of the amp that is blowing all of the tubes. But this iteration was involved with the latest tube blunder. Having never ever blown a tube in 30 years until last year, I had to look at what is different in this amp from the previous amps I have had in the past.

Of course there could be an error somewhere in my wiring, however I have been over it, literally, at least a dozen times, tracing through schematics and my own drawn up plans. Perhaps I could post the schem and someone could take a look to aviod any "user" errors.

In addressing comments;
I do not have any retainer clips or other mounting hardware for the tubes. Since I built this more or less as some kind of prototype I just left them out. And yes, I do gig the amp. Not a good idea to cart this thing around, but I had no other suitable (as in high gain) amp at the time for this particular brand of music. I have since finished another prototype of the amplifier while the circuit evolves... now I have two to cart around... I know it is a somewhat asinine thing to do. I am working toward correcting that, but for the moment it is what it is.

I am interested in the whole choke/resistor thing. I thought I read, or saw, in one of Merlin's books where he had a dropping resistor placed before the choke. At least I think it was one of his books and I will have to reference that. BUT is that a possibility? To receive the best of both worlds? As far as I understand the choke is far better at smoothing than a resistor. So if I could drop the supply something like 40 volts (I know that's a lot), that would help tremendously, would it not? I can make up for it with lower resistors between filter stages.

Until now, that my second amp is finished, I did NOT have a fuse in the DC side of the rectifiers in the latest version of the amp (I did in previous versions). I did this for two reasons, one was I began to model the filtering/power section from the Dual Rec where there are NO fuses after the main fuse... bad idea, I know. The other reason was that I planned to fuse up the ac secondary as per ValveWizard recommendations (I just hadn't gotten there until yesterday). My latest build I have secondary fusing on the AC side. Today I will add fuses on the DC side of the rectifiers AND a fuse on the heaters.

Do I abuse the tubes in general? I suppose so. As I said I am pushing the boundary of 70% dissipation. Until a few days ago when I was coming to the conclusion that I should not do that, I was pushing my tubes to the 30 to 35 mA area. I suppose the 25 to 28 mA area would be better?

I have three other amp builds - two marshall 50 watt style and one AB165'ish (or at least used to be) build. Beside that I have a 100 Marshall plexi re-issue and a JTM45 reissue. Oh yea, and a Fender Hot Rod Deluxe also. None of them (though I do not remember the plexi's readings) put out 515VDC on the plates of the tubes, or 510 on the screens. SO these are the reasons I arrived to the original post. In general I am worried about such high readings.

Thank you, everyone, if I can learn one thing from posting then I am happy. I have definitly learned and drew better conclusions from your comments!!

[J M Fahey]

Blanket statements should have little disclaimer labels attached

A choke is "better than a resistor" because for any given impedance it has less DC resistance than an equivalent resistor, so it has lower voltage drop.

Since in this particular case we do need significant voltage drop, the opposite is truth .

[Mick Bailey]

I've also been using Shuguang EL34s in high voltage amps without any problems. My previous high-voltage tubes were late-90s Svetlana stock that's now too expensive to use. Shuguang tubes have thin envelopes compared with TADs, but in my post-mortem on both tubes I was unable to discern any difference in the spacers, electrode assembly, finish or construction. The TADs weigh a little more (I think about 6g heavier) due to the thicker envelope. TADS are less microphonic than factory-branded Shuguangs, possibly down to the glass.

As a sidestep the JJ KT77 spec sheet rates it for high plate and screen voltage, though I have not personally used these.

I agree with other observations that a white tube is a mechanical failure of the vacuum seal and is not usually symptomatic of high voltages. A tube that fails due to voltage issues will often short, arc internally or if you're lucky the screen will go open.

For me, I use a MOSFET voltage reducer to tame high amp voltages. Just a few components that are hard-wired onto the MOSFET pins and bolted to the chassis. It's dead easy, super-reliable and easy to change the voltage if needed. I reduce the overall voltage rather than regulate the screens. See R.G.'s 'MOSFET follies' for details. Here's how it looks (not my picture). There's no free ride - the MOSFET still produces as much heat as a resistor or Zener. Just cheaper, easier and neat.

[Pedro Vecino]

Thanks, Mick. Two questions: What is the reference of the mosfet? Does it need an insulating mica or can it be screwed directly onto the chassis?

[dstrat]

Have a look at the bottom of the page.

http://www.geofex.com/Article_Folder...osfetfolly.htm

Although not sure have never tried one, but it should be isolated from the chassis and large enough heat sink to keep the smoke in.

[Gtr0]

Blanket statements should have little disclaimer labels attached

A choke is "better than a resistor" because for any given impedance it has less DC resistance than an equivalent resistor, so it has lower voltage drop.

Since in this particular case we do need significant voltage drop, the opposite is truth .

And right after rectifier output is the right place? or in exact replacement of the choke? (between B+1 and B+2) - Thanks

[dstrat]

that would be the place.
you asked "I suppose the 25 to 28 mA area would be better?"
Yes it would , Tubes dont so much care what the voltage is , its the current thru the tube that makes it hot.
with that kind of plate volts 60-65% bias would not be a bad idea.

[Zozobra]

If you're after a tight metal/high gain sound then I'd try out the mosfet regulated screen supply as was mentioned earlier. I'm going to try this in my next high power build. It's a pretty simple circuit that you can squeeze on a bit of tag board quite easily.
Check this out: Amplified music : Guitar valve amp service

regulated screens deets are towards the bottom of the page.

[Gregg]

Although not sure have never tried one, but it should be isolated from the chassis and large enough heat sink to keep the smoke in.

Unless the metal back of your MOSFET is other than the drain then you can bolt it to directly to the chassis which will be both your ground connection and heatsink.

[J M Fahey]

And right after rectifier output is the right place? or in exact replacement of the choke? (between B+1 and B+2) - Thanks

In original choke place.
I am dropping only screen voltage, which to boot is a way lighter and colder duty (just a few mA) than reducing the *full* power supply voltage, including power tubes, which means *a lot* more power dissipation.

Just what was asked for in the original question; itīs assumed that plates by themselves are quite more robust and can take higher voltage without big trouble.

Of course, if you want to reduce full supply voltage, be my guest

EDIT:

Tubes dont so much care what the voltage is , its the current thru the tube that makes it hot.

Well, both do, to be more precise their product
W=V*A

EDIT:

Unless the metal back of your MOSFET is other than the drain then you can bolt it to directly to the chassis which will be both your ground connection and heatsink.

Please look at the picture shown above, that drain is hovering at around +400/500V above ground; I would very carefully insulate it from chassis/heatsink.

[Chuck H]

A note about dropping the screens with a resistor instead of the choke. This will drop voltage on the PI and preamp tubes as well and THAT will likely change the tone of the amp more than the difference on the screens. The following HV rail resistor feeding the PI and preamp in the SLO schematic is a 10k/2W. You should be able to reduce this value to bring the downstream voltage up again. Just measure voltage on the downstream side of that 10k resistor before changing the choke to a resistor so that you have a target voltage.

[Mick Bailey]

Thanks, Mick. Two questions: What is the reference of the mosfet? Does it need an insulating mica or can it be screwed directly onto the chassis?

The last batch I had were STW9NK90Z - 8A/900v/TO247. I've stuck with that rating for quite a few years, though various different types, depending on what's cheapest or easily available. I've never had an amp back with a failed circuit and it's robust and reliable - fit and forget. I just use mica and heatsink grease. The rating is way over what's needed and generally a 75W device such as the IRF830 is fine (4.5A/500v). I like the larger packages because the greater surface area means the device runs much cooler on a steel chassis as a heatsink and there's no problem in finding space. The legs are also more robust and the spacing is ideal for mounting the other components.

[J M Fahey]

A note about dropping the screens with a resistor instead of the choke. This will drop voltage on the PI and preamp tubes as well and THAT will likely change the tone of the amp more than the difference on the screens.

Well. *that* is the point.
If original preamp expected to be fed from around 400V and in this particular amp itīs fed from around 500V , in both cases from the screen supplying node, taming that extra high voltage to the expected one is the reasonable thing to do.

The following HV rail resistor feeding the PI and preamp in the SLO schematic is a 10k/2W. You should be able to reduce this value to bring the downstream voltage up again.

Why?
You are now feeding that 10k resistor the proper voltage, why would you want to overvolt the preamp?

Just measure voltage on the downstream side of that 10k resistor before changing the choke to a resistor so that you have a target voltage.

I suggest you use the original SLO voltages as "targets".
Why correct a wrong one (in this case a too high one) at the screens and then "uncorrect" it?

[Gregg]

Please look at the picture shown above, that drain is hovering at around +400/500V above ground; I would very carefully insulate it from chassis/heatsink.

It looks like it's not used as suggested in R.G.'s article (to CT or to bridge negative) but in series with the B+ supply.

[J M Fahey]

The last batch I had were STW9NK90Z - 8A/900v/TO247. I've stuck with that rating for quite a few years, though various different types, depending on what's cheapest or easily available. I've never had an amp back with a failed circuit and it's robust and reliable - fit and forget. I just use mica and heatsink grease. The rating is way over what's needed and generally a 75W device such as the IRF830 is fine (4.5A/500v). I like the larger packages because the greater surface area means the device runs much cooler on a steel chassis as a heatsink and there's no problem in finding space. The legs are also more robust and the spacing is ideal for mounting the other components.

It may be fine, but in general the datasheet rated dissipation is the most misleading rating of them all.
None of your fault of course, but "creative" Datasheet writing.

In the Real World no semiconductor will ever be able to dissipate that much, and will sweat a lot to even approach *half* that, if at all.

Practical example, and straight from IRF830 datasheet:

Maximum Power Dissipation TC = 25 °C PD 74 W

Looks good, huh? 74W!!!!!

read again: to be able to dissipate that, the *case* , not the heatsink, needs to be held at steady 25C .... good luck with that.

In the beginning I corrected my mental image to transistor, this one or any other one (as in *all* others), being rigidly mounted to, say, a 1 cubic meter block of Aluminum held at steady 25C , only guarantee for it to *stay* there at least for some time.

Hardly a realistic heatsink but whatīs implied by the rating.

Only later I noticed an even bigger flaw in that rating: holding that "infinite heatsink" (their words, not mine) at steady 25C is not enough, since we are ignoring the thermal resistance between case and heatsink .... conveniently ignored in most modern datasheets (it was present in old ones, such as old 2N3055 datasheets, those which came, ugh!! , *printed*!!!! and using, forgive my French: ink and paper !!!
Ok Mom, Iīll wash my mouth with Laundry soap now
Maybe that spoiled me because to this day I use that value to design.

I am talking Rcs here:


Searching a lot (since itīs not even mentioned any more) I found "typical" minimum values of 0.3 to 0.5 C/W , no mica, greased case to sink junction.

Assuming 0.5C/W as a safe value (since datasheet shows none), 75W through that means a 35C thermal drop.

Which also means that if case is to be held at 25C, and itīs mounted to a "best heatsink in the World" 2.5 Ton aluminum cube, said cube still needs to be held 35C below that, or minus 10C ... 10C below ice temperature.

Donīt know you but I find that dissipation rating "slightly" unrealistic

Ok, ok, what is realistic then?

Just a practical example, using same IRF820.

Suppose itīs in an SS environment (nearby HOT tubes will definitely worsen the situation) , inside a cabinet, inside temperature may reach 40C , maximum design junction temperature is 120C (considering absolute maximum is 150C and we want a little safety margin) and we use a reasonable 2C/W heatsink such as this one:

then final junction to air thermal resistance is: 1.69 + 0.5 + 2 = 4.19 C/W
(the 1.69 Rjc value is "shown but hidden" in the datasheet, as its inverse: "linear derating" parameter )

So maximum *actual* dissipation is (120-40)/4.19=about 20W .... a far cry from big bold lettering "74 W"

And that in a very conservative design: mild ambient temperature, no mica, large heat sink, etc.

For poor MosFets inside a HOT Tube amp chassis, bolted to not-too-thick steel sheet and of course using mica or Silpad, using a larger TO247 one such as suggested by Mick Bailey is the way to go.

And in any case I would search (by Finger-O-Meter at least) for the relatively coolest spot on that chassis to mount that MosFet.