MOV's AND flyback diodes?

[Gaz]

Is it okay to use MOV's across the each half-winding of the primary on an output transformer AND diodes from plate to ground as well? ...Or is that gonna make the magic smoke come out of my amp? Is there such a thing as too safe?

[Enzo]

Offhand I don't know why not. None of those things is active in a circuit until the conditions they are there to prevent occur.

[R.G.]

Enzo beat me to it. Yup, fine to use both as long as the MOVs (ACKKK! TVS is lower capacitance!!) as long the MOV/TVS never trips on maximum normal voltage peaks - approximately 2x B+ or a bit more.

[Gaz]

Enzo beat me to it. Yup, fine to use both as long as the MOVs (ACKKK! TVS is lower capacitance!!) as long the MOV/TVS never trips on maximum normal voltage peaks - approximately 2x B+ or a bit more.

Thanks. Do the MOV's only need to be rated at some margin above B+ if there is one across each half winding (opposed to a single MOV/TVS across the entire primary?

[R.G.]

Just thinking about the issues of driving an output transformer a bit, you can put one MOV across the whole thing, but two may be both better clamping for transients as well as being easier to find.

Here's what is happening. Some transient causes a voltage spike by making the output tubes try to interrupt the current flowing in the transformer primary. The spike carries energy proportional to half the inductance times the current squared. The energy in the main transformer is largely damped by the secondary loading, but the energy in the leakage inductances on the primary are by defiinition the inductive energy NOT coupled to the secondary windings, so that's the one which will (mostly) be kicked back into arcing the output tubes.

The two half-primaries are related like a see-saw around B+. When one output tube pulls one half primary down, the other (off) side rises above B+ an equal amount, by transformer action. The maximum voltage in normal operation across one half-primary is B+ minus the "saturation voltage" of one output tube. Normal output tubes can't "saturate" to less than maybe 50V across them, so the maximum voltage they can make across a half-primary in normal operation is B+ minus about 50V. And the maximum voltage the off side tube sees in normal operation is B+ plus (B+ minus 50), so two times B+ minus one tube "saturation voltlage".

On spikes, this gets different. The spike energy probably comes mostly from the leakage inductances, so the "on" side tube will see its half of the primary spike negative, while the "off" side tube sees a spike positive above the end of the transformer half primary, or at most two times B+ minus a saturation voltage plus the spike voltage.

Protection diodes act differently on the "high side" versus the "low side of this". On the low side, the diodes simply turn on and clamp the inductive spike to ground. On the high side, they either reverse breakover/zener to clamp the leakage inductance spike, or offer no protection at all and the tube arcs. The leakages are not coupled to each other by definition, so transformer action provides little or no protection to the high side.

A MOV directly at the transformer clamps the energy relative to the transformer itself. It is OK to clamp the whole transformer with one MOV rated at more than 2x B+, which is the biggest normal operation voltage. In my thinking, it's better to clamp each half-primary separately with more than 1x B+, so the uncoupled leakage spikes are diverted back into B+ or just eaten in heating the MOV.

And as I get reminded periodically, MOVs have higher capacitance than TVS "transzorb" devices, which do much the same thing, so TVS might be better for this than MOVs. They do the same function - clamping voltage spikes to a specific level.

[trobbins]

The capacitance of a MOV or a transzorb is very device selection dependent, so imho it is not correct to say that MOV's have relatively higher (or lower) capacitance. I think the higher capacitance view results when a large MOV device datasheet is looked at (eg. 20mm diameter style disk), compared to a more applicable 7mm diameter style disk.

Tranzorbs have a very sharp clamp voltage, where pretty much all the transient energy available to be absorbed at the start of clamping will dissipate in to the transzorb.

MOVs have a very soft clamp voltage, where transient energy available to be absorbed at the start of clamping is split between the MOV, and further increasing the voltage across the capacitance of the primary winding.

[Gaz]

Thanks, R.G. the explanation is clarifying. So is it worth the effort of using both methods?

[Gaz]

The capacitance of a MOV or a transzorb is very device selection dependent, so imho it is not correct to say that MOV's have relatively higher (or lower) capacitance. I think the higher capacitance view results when a large MOV device datasheet is looked at (eg. 20mm diameter style disk), compared to a more applicable 7mm diameter style disk.

Tranzorbs have a very sharp clamp voltage, where pretty much all the transient energy available to be absorbed at the start of clamping will dissipate in to the transzorb.

MOVs have a very soft clamp voltage, where transient energy available to be absorbed at the start of clamping is split between the MOV, and further increasing the voltage across the capacitance of the primary winding.

Hey trobbins, I've actually asked you about this very subject years ago on AX84 I'm still a bit confused about selecting the MOV for each half-winding, and I really don't want to bugger it up. I've also waded my way through your article on output transformer protection, which is great, but admittedly over my head in sections.

In a Plexi type amp with a B+ of 450vdc, does something like this look appropriate?

https://www.mouser.com/datasheet/2/3...C1-1141750.pdf
ERZV10D681

The V1 mA is 612V to 748V. B+ x1.5 is somewhere in there. I'm really not sure about the other specs though ???

Thanks.

[loudthud]

I don't disagree with what R.G. said above, but ... you can't divorce the transformer from it's load, a speaker, unless you are talking about driving an OT with no load connected (not a bad conversation to have). Most spikes come from the speaker, the OT just passes them through to the tubes. Leakage inductance is small, too small to worry about IMHO.

Check the files I posted in this thread: https://music-electronics-forum.com/...ad.php?t=35493

You can see inductive spikes flying off the scope screen. What I didn't mention in that thread is that the amp is solid state. The output MOSFETs (operated common Source) are arranged to allow spikes beyond the power supply rails and they are clamped by an MOV. No transformer was in the signal path.

I would be interested if someone (smarter than me) would take the equivalent circuit of a speaker, and calculate the apparent values as seen from the primary side of an OT. The values must be huge, but maybe not as large as the primary inductance.

[trobbins]

Voltage rating seems ok - the 612V minimum 1mADC rating is nicely above your expected B+ given additional margin from tube saturation, but noting mains AC tolerance. At the other end, the 1120Vpk clamping at 5A, when added to B+, is also not really high (ie. < 2kV, and so should be fine for an OPT designed to work at 400-500V B+).

The impulse withstand ratings aren't really targeted at repetitive pulses of short period, but do indicate robust performance for one-off pulses (eg. >1000 fault events of 5A surge lasting a few millisecond).

The 170pF shunt capacitance of the MOV across each half-primary shouldn't be noticeable unless you have a hi-fi amp with finely tuned feedback. That is consistent with a half primary winding shunt capacitance of at least 400pF. The winding shunt capacitance will obviously depend on the transformer - I've only measured 15W hi-fi styles with well-interleaved sections that had circa 600pF per half - I should try and measure a few run-of-the-mill types of various power levels. If MOV capacitance was a concern, then using two ERZV07D331 in series presents a much lower capacitance.

When it comes to maintenance testing a MOV, a megger at 500Vdc or 1kVdc should be ok to use (perhaps with a series resistor) - my handheld digital megger has a 1.4mA short circuit limit.

Imho it is too fuzzy to try and design a MOV energy rating that suits a particular application or the range of fault mechanisms that could arise. That probably needs some testing like Loudthud's 6L6 PP output stage testing.

Ciao, Tim

[Helmholtz]

I would be interested if someone (smarter than me) would take the equivalent circuit of a speaker, and calculate the apparent values as seen from the primary side of an OT. The values must be huge, but maybe not as large as the primary inductance.

Effective speaker series inductance varies with frequency. I measured around 0.7mH@1khz and 0.4mH@10kHz with 8 Ohm speakers. With an OT having an impedance ratio of 4k/8= 500, speaker inductance will be reflected to the primary multiplied by 500, resulting in something around 300mH.
Primary inductance is much higher, typically >20H.

[Gaz]

Voltage rating seems ok - the 612V minimum 1mADC rating is nicely above your expected B+ given additional margin from tube saturation, but noting mains AC tolerance. At the other end, the 1120Vpk clamping at 5A, when added to B+, is also not really high (ie. < 2kV, and so should be fine for an OPT designed to work at 400-500V B+).

The impulse withstand ratings aren't really targeted at repetitive pulses of short period, but do indicate robust performance for one-off pulses (eg. >1000 fault events of 5A surge lasting a few millisecond).

The 170pF shunt capacitance of the MOV across each half-primary shouldn't be noticeable unless you have a hi-fi amp with finely tuned feedback. That is consistent with a half primary winding shunt capacitance of at least 400pF. The winding shunt capacitance will obviously depend on the transformer - I've only measured 15W hi-fi styles with well-interleaved sections that had circa 600pF per half - I should try and measure a few run-of-the-mill types of various power levels. If MOV capacitance was a concern, then using two ERZV07D331 in series presents a much lower capacitance.

When it comes to maintenance testing a MOV, a megger at 500Vdc or 1kVdc should be ok to use (perhaps with a series resistor) - my handheld digital megger has a 1.4mA short circuit limit.

Imho it is too fuzzy to try and design a MOV energy rating that suits a particular application or the range of fault mechanisms that could arise. That probably needs some testing like Loudthud's 6L6 PP output stage testing.

Ciao, Tim

Thanks for walking me through the specs. I am still a little confused by the "maximum allowable voltage" specs - not sure what those are really.

And to clarify, the 1mADC rating is the window in which the MOV starts conducting, and the clamp voltage is when it becomes a dead short? Would there any benefit to increasing the 1maDC rating in this example as long as the clamp voltage is does not exceed ~2kV? It's unclear to me how much higher the B+ will go under normal conditions. You made it sound like the margin I chose was ok.

Is it smarter to to use two in series for power ratings too, or is the rated power ok for this application?

Thanks for the clarification!

[trobbins]

No probs.

The 'maximum allowable voltage' ratings in a datasheet relate to an application's typical maximum operating level. For example, if the MOV was used across an AC mains supply then it recommends that supply be no more than 420Vrms (eg. perhaps a 400V 3-phase supply with the MOV across phase-to-phase terminals, and allowing 5% tolerance on 400Vac). Similarly if you had a 560Vdc max power supply, where the supply was limited to 560V on the dial. That then gives a 10% margin if 560V ever gets to 612Vdc, where it is possible that some MOV's of that model could start to conduct 1mA (ie. they would be dissipating 0.612W, which is above their 0.4W 'rated power' rating, and so the part would start to thermally walk/run away).

In an output transformer winding protection application, the MOV shouldn't get close to a continuous 560Vdc level. That would be the situation for your amp, even if one output stage tube was driven on full-time (eg. a coupling cap failed), where the amp wouldn't remain in that state for long anyway.

The 1mADC rating is just a nice generic MOV datasheet level that luckily is easy to comprehend in a valve amp. If a MOV across a half winding was conducting 1mA in your amp, it would be loading that winding with a resistance somewhere between 612kohm and 748kohm. The MOV will be conducting a titch of current if the winding has say a 400Vpk sine wave across it - the voltage versus current curve in the datasheet indicates that the MOV current would be a distorted sine'ish current with a peak of about 1uA (ie. equivalent to having a 450 megohm resistor across the winding).

Certainly you could choose a MOV with a higher or lower voltage rating, depending on which region you were confident about (eg. if your OPT had a rated 3kVac insulation rating then that might entice you to use a higher voltage MOV). Few commercial OPTs provide an insulation rating, or even a manufacturing test voltage that may have been applied. I've only seen a rating on vintage Partridge hi-fi OPTs. It is likely just speculation what an OPT will survive. I test vintage OPT's to 1kVdc insulation resistance before I use them - which to me should not stress old types of insulation, and gives some confidence that the OPT should not have an intrinsic fault - I have a tester that could apply up to 12kVdc, but applying a level above say 1.5-2kV could lead to internal arcing and I may then not want to use that OPT in an amp

I would suggest that a part in the hand is worth 10 at Mouser - selecting a part is likely to come down to what you've got, or what you can get. MOV voltage ratings that are aligned to mains AC voltages will be cheap and plentiful. I have lots of VE13 0421K, which is essentially the MOV you are looking at, and lots of 350Vdc/1ma 7mmD MOVs, and so just use one or t'other of them for each restored amp - whether it is 1, or 2 or 3 in series doesn't really matter to me (the only time I would get concerned and use the lowest capacitance arrangement I could conveniently come up with is with a vintage Williamson). At the moment it is a bit wishy-washy to think that a higher power rated MOV will survive better in a thrashed guitar amp - that would take some interesting testing to quantify, and may just be peculiar to your amp/OPT/speaker combination anyway.

[R.G.]

The capacitance of a MOV or a transzorb is very device selection dependent, so imho it is not correct to say that MOV's have relatively higher (or lower) capacitance.

That's always been my position. I included the note about MOV capacitance, somewhat wearily, because every time I touched on the issue of MOVs (which was what I used in my ampwork) for protection, I promptly got a reply post telling me that TVS devices had lower capacitance. I should go look up the previous times I've posted on this topic, I guess.

I did use MOVs, the capacitance wasn't an issue that I could tell, and so I think they're OK. Other people may have issues with MOV capacitance, so the TVS note is there for those people. Everyone's mileage may vary.

EVERYTHING is device selection dependent, all mileages may vary, and I think I need to put together a more inclusive disclaimer.

[R.G.]

I don't disagree with what R.G. said above, but ... you can't divorce the transformer from it's load, a speaker, unless you are talking about driving an OT with no load connected (not a bad conversation to have). Most spikes come from the speaker, the OT just passes them through to the tubes. Leakage inductance is small, too small to worry about IMHO.

Good reasoning; but the issues in a transformer divorce, like a human divorce, get complicated.

Every wire coil has a certain self-inductance. Put a ferromagnetic core in it and the inductance goes up. However since ferromagnetic materials are only 3000 to 10000 times as good a magnetic "conductor" than non-ferromagnetic materials, some of the flux from the coil is not pulled into the core. It leaks around the core.

Put two coils on the same core, and they are coupled by the mutual shared flux in the core area they share. If the core is, for instance, a one-foot-diameter ring of iron, and the coils are on opposite sides, a (relative) lot of flux from the coil used as a "primary" doesn't make it all the way around the ring to the coil being used as a secondary. It's a high leakage situation.

Any load on the coil being used as a secondary causes a current flow in that coil, which causes a reverse EMF and a resulting opposing flux, offsetting some flux from the primary. The primary lets enough current flow to bring the core flux back up to the previous level, and the excess power flow through the core, through the secondary and out to the load.

So there is current, and current-caused flux on both the primary and secondary sides. And there is leakage - literally magnetic flux that does not couple across through the core between windings. And there is leakage - literally, uncoupled power flow - from every coil (every turn of wire, actually) to every other coil.

Electronically, this looks like an inductor between the voltage/current source and the primary coil(s) and between the load(s) and the secondary coil. Those leakage inductors store and release energy by V= L di/dt and E = (1/2)*L* I². For this discussion, all of those leakage inductances can cause a spike voltage locally by V=L di/dt and by that voltage being transformed to the other coils by transformer action.

So a load on a secondary is only divorced from the inner workings of the transformer by the local leakage inductance it faces in trying to get back through the coupled part of the flux caused/sensed by its local winding. Not having a load is a divorce, but having leakage is a partial separation.

The local leakage inductances act just like series inductances you might (for some very odd reasons, I guess) put in series with the coil, only they're hidden inside the transformer where you can't get at them. So they do V=L di/dt to puncture the wire insulation on local windings, and by transformer action cause V = (Ns/Np)*L di/dt on other windings, the Ns and Np being the winding ratio to the other coils individually.

So a secondary, load-side inductor spike can be transmitted to primary side windings. So can a primary side spike, and this sees the reflected secondary loading, including its invisible leakage inductor in series. A secondary-side voltage spike is seen at the secondary, across the secondary load, and transformed to the primary.

Getting back to the hypothetical iron-ring transformer and high leakage, the only thing that changes in a normal transformer is the size of the leakage inductances. In fact, getting the leakage inductances down is a huge amount of the funny considerations in winding and interleaving coils in OTs.

I would be interested if someone (smarter than me) would take the equivalent circuit of a speaker, and calculate the apparent values as seen from the primary side of an OT. The values must be huge, but maybe not as large as the primary inductance.

It does get tricky when you start adding in all the things that might be connected as loading across any winding. A speaker has the inductance of the voice coil, which is in the 5 to 20mH range IIRC, for most speakers. Then you can have arrays of speakers series/paralleled, and various capacitances, and even crossover networks.

There is the additional imponderable that what punctures insulation is voltage, not current. the voltage makes the puncture as a chain of insulation molecules give up and let current through, and there's a puncture. How big is the puncture? The first one is probably a few molecules wide, and didn't necessarily have a lot of energy expended to widen the hole's diameter. But with a hole through the insulation, the next one is easier, especially if the charred byproducts of the first puncture are less conductive than the original insulation. Next spike, the hole gets bigger and easier to punch through. Some day, some time, the arc vaporizes some metal to coat the hole, and the NEXT time the main currents and voltages come through and have a party.

What I make out of this all is that there are too many cases. I get back to the idea that if protection is what you're after, figure out what the protection values are, then protect there. We're pretty sure that primary windings can handle 2x B+; they have to do that in normal operation. There are specs on the withstanding voltage of the insulation on magnet wire, but for most OTs we don't know what brand/kind the manufacturer used.

So I just punt. Puncturing insulation with overvoltages is an issue, as witness that repair shops get amps in to repair that have shorted primaries. Some of those are overheating, some are insulation puncture. The actual voltages and energies are tricky to figure. My approach is to say "it has to stand 2x B+" and to pick something that keeps it from going much outside that.