Voltage spike suppressors - TVS Diode or MOV

[anishkgt]

I've been having doubts on which to go with when protect the underlying components from voltage spikes and other surprises in the line. TVS Diodes or MOVs.

Is it true that TVS diodes are better than MOVs as they last longer and MOVs tend to degrade as they suppress line voltages. Would it be a good idea to use both ?

TVS are circled in Red in the attached circuit.

[anishkgt]

here the load will be MOT (Microwave oven transformer) secondary rewired with 2AWG wire.

[T3sl4co1l]

Holy shit, eleven thousand pixels just to ask about a diode?!

Tim

[anishkgt]

and its only 600kb isn't that good. I normally put a smaller pic but they seem to be pixelated.

[T3sl4co1l]

Too bad Chrome has a bug with huge pictures.

And, why 600kB when 66kB is perfectly legible?

Anyway.

The 11 pin header has no filtering, and has I2C on it.  I2C is an on-board interface, and rapidly gets into trouble once you put it on any length of cable.  Even just 15cm of cable.

D1-D3, D10 and D11 are superfluous.

Why are T1-T2 shown in a bidirectional SCR configuration, when they are shown as TRIACs, and the nearby label says "TRIAC DRIVER"?  (If they are actually SCRs, D1 and D2 are needed.)

I wonder if R12 should be a smaller value.  Depends on load.  YMMV.

D4 and D13 being outside of fuses is.......interesting.  Especially for devices that aren't rated for mains surge...
They're also shown as unipolar diodes, but the part numbers are bidirectional TVSs.  Intentionally confusing?

The fuse is also on the wrong side of TRANSFORMER, it should be primary side.  Actually it might not matter much either way, if it's an impedance limited type transformer.  Check the datasheet for ratings.

If those are fuses at all.  One has an amp rating, neither has a designator or part number.

Tim

[IanMacdonald]

MOVs give relatively 'soft' clamping, their conduction increasing gradually with voltage. Because of this a MOV can't be rated to limit the voltage to much less than twice normal max, otherwise it would be conducting to some extent all the time.  The upside is that that because they are  basically just a dumb lump of metal oxide they can take an amount of punishment that would blow the hell out of a precision semiconductor.

Clamping diodes behave more like a zener, giving a rapidly rising current once the conduction threshold is reached. The diode is more likely to protect the circuit, but also more likely to fail s/c itself under overload. If you can tolerate the occasional s/c failure then the clamping diode is superior. If its failing s/c would do more damage than the transient would have, then that is a foot-shooting exercise, though.

[CopperCone]

You want multiples of both protecting each other.

Only reason not to do so is cost , and possibly leakage. MOVs will leak alot more (due to their fairly gradual characteristic compared to a diode)

there is alot of different types of MOV too. I thought that it was good to have a MOV to clamp the voltage at some range reasonable for the diode to act. I think they typically handle more power then a diode.

[tszaboo]

Anyway.

I have to agree. This is probably the worst schematic I saw this month. There are more problems than I can list. Starting by the explosive TVS D13, which will explode on overvoltage. This is what the datasheet said:

TVS devices are ideal for the protection of I/O Interfaces,
VCC bus and other vulnerable circuits used in Telecom,
Computer, Industrial and Consumer electronic
applications

Does it list AC applications? No.
Sometimes I think the electronics as a widespread hobby is a bad thing.

[anishkgt]

Really ! Those are Bi directional so aren't they meant for it ?


Sent from my iPhone using Tapatalk

[T3sl4co1l]

Bidirectional is certainly more correct than unidirectional, but more important is that AC mains normally fluctuates, with modest peaks (of say, double the normal peak voltage) occurring from time to time, and much larger surges (up to a few kV) occurring as a result of lightning strikes.  And the pulses carry a lot of current, like hundreds or thousands of amperes.  They might be only microseconds long, but it's still a lot of energy, much more than a diode can handle*.

*There are some TVS diodes rated for mains service, but they are huge and expensive, like $100.  Nice thing is they don't wear out over time, as MOVs tend to.

Which is the usual solution: MOVs handle orders of magnitude more energy, at the expense of a softer voltage curve (expect a peak voltage 2-3 times the nominal rated DC voltage).

If you're using 800V SCRs (from your other post, it seems I guessed correctly that they are actually SCRs), and mains is 240V, a 250VAC (actually ~360VDC) MOV will do an okay job clamping most junk in the 700-800V range, where the SCRs probably won't retrigger.

Mind that, an MOV in series with a load (i.e., in parallel with a switch) will pass input transients into the load.  This might not be desirable.  (Probably not a big deal in a spot welder.)

Tim

[tszaboo]

Tim, what I would add to the very good explanation: The name of the device. Transient voltage suppressor. It is ment for transient voltages, not surges.

[T3sl4co1l]

Mmm, sorta -- they are rated for IEC 61000-4-5 surge.  But not at mains levels (>1kV surge, 2 ohm generator), not for mains-rated diodes (100s V).

The surge rating is useful in the lower voltages, where you'd see lightning induced surge on otherwise-low voltage circuits: long wire runs for telecom, industrial and facility wiring (like 60950-1 TNV type circuits).  These circuits expect smaller surges (1kV or less), and have more series resistance (20 ohms is in a standard, somewhere?).  The surges may be longer, however (like 10/700 and 10/1000us pulses).

Automotive surges, too, except for load dump (which again, deposits far too much energy).

Tim

[anishkgt]

i guess my previous schematic was a bit messed up since i had replaced through holes with SMD and unknowingly replaced the SCR with a TRIAC.

So from what i learn here is, i would need MOV's rather than TVS diodes. My intention to use D13 was to mainly protect the underlying components and R1 was to take care of any spikes or transients that may come from the MOT. Now i am in doubt if D13 should be replaced with a MOV as well ?

[floobydust]

I think you are missing reverse diode on OPT2 pin 1,2 like you have on OPT1; I would just move D10. Never mind H11AA1 is AC input. Check LM7805 pinout pin 2 is GND.

The old thread on the Arduino spot welder, I think you could use a single TRIAC and not two SCR's as discussed there:
https://www.eevblog.com/forum/beginners/basics-ac-load-swtiching-thyristor-or-triac/msg1082209/#msg1082209

A few voices say zero-cross switching a transformer is the worst possible.
Due to dV/dT being highest and transformer magnetization.
See graph at 4 - Inductive & Transformer Inrush
Your 9VAC power transformer has phase-shift so MCU zero-cross signal is out anyhow. It would be possible to wait a few msec after, then turn on the TRIAC.
I'm not sure why you need zero-cross when the OPT1 is a zero-cross TRIAC driver. I guess to count AC cycles.

[anishkgt]

Check LM7805 pinout pin 2 is GND.

That was strange, it seems the TO252-3 package's pins were not assigned correctly and now i've corrected it.

The old thread on the Arduino spot welder, I think you could .....:

I was sticking to the same design because of the nature of SCR and  so far i did not have any problem with it.

I'm not sure why you need zero-cross when the OPT1 is a zero-cross TRIAC driver. I guess to count AC cycles.

The H11AA1 detects the zero on every half cycle which then fires the SCR's at the peak of the sine wave triggered by the MUC. So the trasformer is switched on at the peak rather than at the zero.

I know i'd need an MOV at the load so added one. Not sure what spec to go with. any would be appreciated.

[wraper]

Anyway.

I have to agree. This is probably the worst schematic I saw this month. There are more problems than I can list. Starting by the explosive TVS D13, which will explode on overvoltage. This is what the datasheet said:

TVS devices are ideal for the protection of I/O Interfaces,
VCC bus and other vulnerable circuits used in Telecom,
Computer, Industrial and Consumer electronic
applications

Does it list AC applications? No.
Sometimes I think the electronics as a widespread hobby is a bad thing.

EDIT: did not notice it was placed on the primary side  They work well for AC applications as there are bipolar versions. Not across the mains though. Put for something like 12V AC generally they are the best choice.

TVS diodes are rated for
Tim, what I would add to the very good explanation: The name of the device. Transient voltage suppressor. It is ment for transient voltages, not surges.

Take typical P6KE series TVS, 600W peak power dissipation, 100A Peak Forward Surge Current @ 8.3 ms, ½ Sine Wave, Single Phase Duty Cycle, @ 4 Pulses Per Minute Maximum. They are well suited for surges but not across the mains.

[tszaboo]

Piece of nonsense.

OK. This is from the same company. Application notes for TVS diodes.
http://m.littelfuse.com/~/media/electronics/application_notes/littelfuse_tvs_diode_overview_application_note.pdf.pdf
Please show me, which diagram has a TVS, alone, sitting directly on an AC input.

Again, there are TVS diodes, that are for AC line. Like Littlefuse AK6. Strangely, it has UL rating for 300+V, unlike the series you proposed.

[anishkgt]

I have a rigol 1054z oscilloscope and probes that came with it. Is it possible to measure the spike or transients to determine which MOV/TVS diode to go with ?

[wraper]

Please show me, which diagram has a TVS, alone, sitting directly on an AC input.

I edited the post just a few minutes after posting, even before you wrote yours. That circuit looks confusing and with low resolution combined it was not apparent that TVS is sitting on the primary side.

[David Hess]

Too bad Chrome has a bug with huge pictures.

And Firefox.

[tszaboo]

Please show me, which diagram has a TVS, alone, sitting directly on an AC input.

I edited the post just a few minutes after posting, even before you wrote yours. That circuit looks confusing and with low resolution combined it was not apparent that TVS is sitting on the primary side.

I went for a covfefe before hitting the post button.
Good to know we are on the same side.

[floobydust]

TVS is physically smaller than typical MOV which can absorb more energy, but softer clamping and they age.

I would use a MOV to protect the SCR's and opto againt HV spikes.
This is not welding all the time so a 10-20mm 275VRMS MOV. Epcos S14K275 is 151J for 2msec., clamps to ~500V typ. and max. 710V but that is a 100A lightning strike.

One protection scheme I have seen two SMB TVS (SMBJ220 back-back) used across TRIAC opto pin 4,6.
A HV spike thus turns on an SCR to shunt the HV surge for the one half-cycle. The TVS also protect the opto.
This worked well enough in an 240VAC 35A industrial controller.

MOC3041 is 400V part, MOC3063 is 600V, MOC3083 is 800V.

I would have no protection for the MCU power supply. A HV spike has to make it through the transformer, and bridge, filter caps and it's more psychological to put a MOV there.

IC2 became a 7824... should be a 7805.
I would have C7 over with C9 filter cap. (total ~2,200uF) as this gives longer hold time for the MCU during sags. 100uF after Vreg is ok. D11, D10 I can see having no purpose?

Pedal input must have ESD protection, it should not go straight into MCU. It will pick noise due to high value pullup in AVR. I would use a low value pullup, series resistor, clamp diodes and a RFI cap so it does not false-trigger.

[anishkgt]

I would use a MOV to protect the SCR's and opto againt HV spikes.
This is not welding all the time so a 10-20mm 275VRMS MOV. Epcos S14K275 is 151J for 2msec., clamps to ~500V typ. and max. 710V but that is a 100A lightning strike.

thanks that would be an added protection to the Opto. So i've added https://www.digikey.com/product-detail/en/epcos-tdk/B72214S0271K151/495-5964-1-ND/4841531 to pin 4,6 on the opto and R17 at the load terminal protecting the SCRs.

I would have C7 over with C9 filter cap. (total ~2,200uF) as this gives longer hold time for the MCU during sags. 100uF after Vreg is ok

i did not get what you meant here but i've changed the 25v 1000uf to 100uf after the Vreg but leaving 25v 1000uf before the Vreg. The pedal pin is actually set to pullup in the MCU code itself.

[anishkgt]

The 11 pin header has no filtering, and has I2C on it.  I2C is an on-board interface, and rapidly gets into trouble once you put it on any length of cable.  Even just 15cm of cable.

The connector wire is not very long, less than an inch. Would it require as a standard ? which pins would need them ?

D1-D3, D10 and D11 are superfluous.

Removed D10 and D11 as you said but left the D1 and D2 as they are needed for the SCRs.

Why are T1-T2 shown in a bidirectional SCR configuration, when they are shown as TRIACs, and the nearby label says "TRIAC DRIVER"?  (If they are actually SCRs, D1 and D2 are needed.)

had a bit of mess up when changing through hole to SMD. Those are actually SCRs.

I wonder if R12 should be a smaller value.  Depends on load.  YMMV.

The load is an MOT.

D4 and D13 being outside of fuses is.......interesting.  Especially for devices that aren't rated for mains surge...
They're also shown as unipolar diodes, but the part numbers are bidirectional TVSs.  Intentionally confusing?

They are bi-directional diodes but D4 is replaced with a varistor. Eagle cad did not have a TVS device big enough so used a diode instead.

The fuse is also on the wrong side of TRANSFORMER, it should be primary side.  Actually it might not matter much either way, if it's an impedance limited type transformer.  Check the datasheet for ratings.

The Fuse are on the primary/LIVE side and are rated 5A.