tube rect/ss rect circuit

[mikepukmel]

Is this tube rect/ss rect switching circuit legit? It looks like switching to ss just puts the ss rectifiers in parallel with the tube. The tube is still energized and rectifying.


And a 50uf (even single) first filter cap seems too high, with 20% tolerance, couldn't this cap be right at the 60uf max for this tube? What happens if the cap is so, or right on the tolerance that the tube has trouble charging it?

From the GZ34 data sheet, its output is something like 30 or 40 v less than a ss rect, so won't there be a potential drop from the diode output to the GZ34 output?

And won't all of the voltages on the schematic be wrong when the switch is flipped one way? I mean, the GZ34 is efficient but not as efficient as the ss rectifiers.


jtm45plus_schematic_v4.pdf

[mikepukmel]

If the transformer is 330 - 0 - 330, how can the PS output higher than 330*1.414 = 466? The schematic has 482v on the output tube plates.

[Enzo]

Well, the transformer in your schematic says 690vCT, or 345-0-345. So 345 x 1.414 is 487.

When you turn on the diodes, yes, the tube is still there, but the diodes do all the work. There is no particular point in disconnecting the tube. Current seeks the path of least resistance.

Yes, the voltage on the B+ line will be lower with the tube. The two rectifier choices will give you different amplifier dynamics.

Write this on your wall where you can always see it: It is just a guitar amp.

There is nothing precision or critical here. a 50uf cap on a tube with a 60uf "maximum" will be happy as a clam. Don;t worry about tolerances. If a tube has a 60uf max, it doesn;t blow up at 61uf. Note the switch to turn on the diodes also adds in another filter cap in parallel.

When you switch rectifiers, the voltages won't be "wrong", they will just be different. It is just a guitar amp.

Read the notes lower left, all the test voltages on the drawing are taken in the diode mode. In tube rectifier mode, the readings will be lower. Also note it says all voltages are +/-10%. Your 482 B+? 10% of that is 48v. it could be well over 500v. They are aware of that,and it really won;t be a problem.

[Chuck H]

The transformer in the schematic is indicated as 345-0-345.

The parallel diodes work fine for this application because current takes the path of least resistance. Which would be the SS diode in the case of paralleling them with the tube diode.

[pdf64]

On the face of it, a single pair of 1kV 1N4007 seem rather underrated for this application.
If the HT winding is 690Vac @ 150mA, it will be somewhat higher at the lower current draw of idle, and likely at least the 708Vac needed to breach the limit.
Even at the full load 150mA, the mains only need rise by ~2.5% (to ~123Vac) for the voltage across the diodes to exceed 1kV.

[J M Fahey]

On the face of it, a single pair of 1kV 1N4007 seem rather underrated for this application.
If the HT winding is 690Vac @ 150mA, it will be somewhat higher at the lower current draw of idle, and likely at least the 708Vac needed to breach the limit.
Even at the full load 150mA, the mains only need rise by ~2.5% (to ~123Vac) for the voltage across the diodes to exceed 1kV.

Fully agree.
FWIW Fender traditionally used 3x1N4007 in series on each branch and many others followed suite.
I would use| at least 2 in series on each branch.
A production/real world 4007 might be lossy above 1000V but stand it and only punch through between 1200 and 1500V, but trusting undocumented values is at best poor Engineering.

And may even end up in problems such as "oh, 2014 made amps are unreliable" (because 2014 supplier *just* met rated specs) or similar problems, which don´t show up at Factory but raqise their ugly head in the field.

[Enzo]

But the winding is center tapped, the CT being grounded. So we start with only 345vAC at the rectifier, not 700v. SO even if this somehow went up 100v, we'd still only have 445vAC, and that means peaks in reverse of 630v, well within the 1kv PIV of the diode. So I think the argument that 1N4007 is underrated is based on a faulty assumption.


And the old Fenders like the AB763 had three diodes in series, but they were not 1N4007s.

[J M Fahey]

This 345VRMS winding charges main capacitor to its peak, 345*1.4142=488VDC

Then, while the capacitor still holds those 488VDC (that´s its job) , phase inverts and winding applies the opposite peak to capacitor anode: minus 488V ... while cathode is still at plus 488V .
So 60 times a second that diode must stand 488+488=976V .
And that with stable or low mains voltage, peaks ere common.
An accident waiting to happen.

We don´t see catastrophic failures more often *only* because, in order to have low rejection rates, diode manufacturer "aims" at a somewhat higher average value, say 1200PIV (but we don´t know that for sure) so, say, 99% production, which does have a spread, meets or exceeds official rated value.

But some at least will barely meet spec, will work perfectly on most places, buy if so tightly used now and then will blow.

Two options:
1) use selected/tested parts. Sometimes schematics show some part , generally a critical one such as a power transistor, low noise input transistor or IC, with the label "use only factory selected part" ... that´s why.
I routinely used +/-42V rails with 2N3055 which are officially rated up to 70V , since 1972 ... but I bought the full boxes or at least a few stryrofoam trays straight from the official Motorola/ON/RCA/ST distributor (think Mouser quality) and tested all of them, sticking small labels to each one.
Only used those standing at least 90V and many reached up to 112V , go figure.
But that´s expensive in time wasted, so for a busy factory, the best option is to:
2) put more diodes in series, or more transistors in parallel.
For a large buyer parts are cheap and speed of production is paramount.

Maybe Marshall used better than ordinary diodes, why not? , they certainly have the buying power for that, but I guess a random Tech out in the field will use something pulled from his own parts cabinets, random brand and batch.

If I happened to get one of these amps with a blown diode, I woukd replace both with a couple series ones each, cheap insurance.

[mikepukmel]

Thanks everyone. Re-reading all your posts.

(doing this now ... Write this on your wall where you can always see it: It is just a guitar amp.)

[nevetslab]

Concerning the Capacitance tolerance....where +/- 20% is typical of most power supply electrolytic caps we see in use, I can't recall the last time I've seen a part on the high side of the nominal value! Manufacturers use the Tolerance figure to justify installing less foil in the part, so it still meets the tolerance spec, and saves them money, getting higher yield of the rolls of foil used in fabrication.

[Enzo]

True, caps are much tighter to tolerance these days. Those -20/+80 days are decades ago.

[mikepukmel]

Re "The transformer in the schematic is indicated as 345-0-345."
Ahhh geez, half of 690 is 345 not 320. (And I didn't even have one beer when I wrote that ).


By your suggestions if the solid state/tube thing was done, it would be good to at least double up the solid state diodes.

I do have a fear of toasting tubes, especially the rectifier. It sounds like the switch should be thrown with the amp off?


Re "And the old Fenders like the AB763 had three diodes in series, but they were not 1N4007s. " yeah, always wondered why they did 3x, must have been tough to manufacture good hv diodes back in the day?

Do any of you guys ever get a request to install something like this, or build something like this, I mean "Tube to SS" switch? I've seen the solid state modules some folks sell, that drops into the octal socket, but that's now full time, there's no switching between the two.

And I always got the impression from talking to my brother that people like the natural compression of the tube rectifier amps.

[mikepukmel]

Concerning the Capacitance tolerance....where +/- 20% is typical of most power supply electrolytic caps we see in use, I can't recall the last time I've seen a part on the high side of the nominal value! Manufacturers use the Tolerance figure to justify installing less foil in the part, so it still meets the tolerance spec, and saves them money, getting higher yield of the rolls of foil used in fabrication.

I looked for ages to find a discussion on how close you could go to the design max, for the first capacitors off the tube rectifier. I know there must be a couple on this site, just couldn't find it. OK understood: the old days, who knows what actual capacitance those old electrolytics had, but now we have better tolerances so we can go closer.

I thought (probably wrong) that some places in the circuit are more forgiving for parts over/under rated max/min. If you did go over, for the first capacitor on a tube rect, say a Ruby or JJ GZ34 (not an new old stock tube) what would happen?

[nosaj]

I looked for ages to find a discussion on how close you could go to the design max, for the first capacitors off the tube rectifier. I know there must be a couple on this site, just couldn't find it. OK understood: the old days, who knows what actual capacitance those old electrolytics had, but now we have better tolerances so we can go closer.

I thought (probably wrong) that some places in the circuit are more forgiving for parts over/under rated max/min. If you did go over, for the first capacitor on a tube rect, say a Ruby or JJ GZ34 (not an new old stock tube) what would happen?

Pulling too much current through the tube shorting it out I believe. The tube data should tell you the max capacitance for the first cap after the rectifier. I found this book on Tube rectifiers that may help you.Pete Millet's site is one of the best.

nosaj
http://www.tubebooks.org/Books/Schure_Rectifiers.pdf

[R.G.]

I looked for ages to find a discussion on how close you could go to the design max, for the first capacitors off the tube rectifier. I know there must be a couple on this site, just couldn't find it. OK understood: the old days, who knows what actual capacitance those old electrolytics had, but now we have better tolerances so we can go closer.

I thought (probably wrong) that some places in the circuit are more forgiving for parts over/under rated max/min. If you did go over, for the first capacitor on a tube rect, say a Ruby or JJ GZ34 (not an new old stock tube) what would happen?

What's actually going on here is that vacuum cathodes can only put out a certain amount of current without being damaged. The damage is not one of those things where if you step over this line, the world ends things, but is an accumlation of poorer and poorer electron emission.

The circuits view of what happens when a sine wave powers a rectifier/filter forces the rectifier to work in bursts of current right at the peaks of the sine wave. The cap gets 100% of the charge it's going to dole out as DC for the next AC half cycle in the short time near the peak before the changing sine wave voltage holds the diode forward biased, so the current in the charging peak is many times the DC output current in the cap.

The bigger the cap, the shorter the time the diode can be on, and so the higher the peak current is as a multiple of the DC going out of the cap. So for a given DC average current (which the rectifier's current integrates out to) the bigger the cap, the higher the peak current. Vacuum rectifier max specs on first filter caps amount to an educated guess about what the peak currents would be in a "typical" rectifier setup, and what peaks will give a long-enough life in such setups.

There are several ways to cheat. You can use solid state diodes and resistors to parallel the vacuum rectifier to slip some current around the vacuum rectifier, taking some of the peak load off it. You can fake the forward voltage of the vacuum rectifier with SS diodes and resistors, which many people report works very well indeed. Or you can take that "FIRST filter cap" thing to heart, and install a resistor or inductor after a low value first filter cap into a second filter cap that then acts as a "first filter cap" to the whole circuit but siphons a much more constant stream of current out of the real first filter cap, taking some of the peak load off the rectifiers at the expense of a little lost B+ voltage to the using circuit.

Sigh. So many circuits, so little time.