TVS diodes versus MOVs / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

TVS diodes versus MOVs

1 2 next

Move Thread

LAN_403

sehrgut

Tue Oct 02 2007, 04:14PM

Registered Member #1017 Joined: Fri Sept 21 2007, 09:08PM
Location:
Posts: 24

This was brought up briefly some time ago on the pupman list, but was quickly redirected to talking about Terry filters. Does anyone have experience using TVS diodes in flyback and ignition driver snubber circuits? MOVs I'm out of, and TVSs I've plenty of . . . I could just burn some up, but I'd rather not burn up driver semiconductors as well, just to try them out! *grin*

Steve Ward

Tue Oct 02 2007, 07:07PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

TVSs are just avalanche (break down) devices. They are similar to MOVs, but they dont have an energy rating, but rather a power rating (this is good or bad depending on your desired use). For clamping inductive kick in an ignition coil or flyback driver, then a TVS is the proper device. If you are clamping rare events (like power surge, or other unexpected transients) the MOVs are great... but they degrade as they soak up more energy.

Really, you should put a small capacitance across your MOSFET switch so that some of the leakage inductance's stored energy is transfered to the capacitor, which will clamp the voltage spike. The TVS should only be used as a back-up, in case the capacitor is not sized properly to snub the voltage transient. Otherwise, you will just be wasting power in the TVS making it hot (while a capacitor will have considerably lower power losses). Using an oscilloscope you can easily observe the effect of adding a capacitor. Be sure to connect the cap very close to the mosfet terminals (D-S). As a rough guess, for a flyback driver you would want around 1-10nF, and for an ignition coil maybe up to 100nF (ignition coils have far larger inductance).

Sulaiman

Tue Oct 02 2007, 07:08PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I NEVER use MOVs for clamping/snubbers as they are for OCCASIONAL clamping of transients.
I do use a lot of 1.5KE350s though (I got 1000 of them off eBay!)
They work GRET for ignition coil clamps.

There are only two things critical;
- peak clamping voltage, e.g. for 1.5KE350: 350V @ 1mA, 482V @ 4A, 618V @ 16A
- power dissipation, e.g. 1.5KE350: 5W MAXIMUM, I'd say 1 or 2 W in practice

and one important;
- capacitance, e.g. 1.5KE350 C=430 pF

So for my Ignition Coil flyback I use
10 x 1.5KE350 in parallel mainly due to power/heating when there's no load on the output
and a UF5406 diode in series to prevent power wastage/heating due to charging/discharging the TVSs each cycle.

Now my ignition coil driver is VERY reliable.

P.S. you can put the diode/TVS clamp across the transistor, or for slightly higher output voltage, across the coil.

P.P.S. I just saw Steve's post above and I disagree - I prefer not to use a capacitor.

sehrgut

Wed Oct 03 2007, 04:48AM

Registered Member #1017 Joined: Fri Sept 21 2007, 09:08PM
Location:
Posts: 24

Thanks to both of you for your help! First, I'm going BJT on this one . . . I mentioned somewhere else I was feeling rather Luddite! *grin* No modern MOSFETs, IGBTs, or the like for me! I've actually got a number of TIPL753A NPNs (designed for large SMPSs and other inductive loads - 1KV blocking!) This is for the old standard 3055 circuit . . . with a real transistor instead.

Of course, I imagine once my scope gets here (I'm having it shipped out, since I didn't want to trust it to checked baggage) I'll be able to try it out, but I'm curious as to your difference on the idea of a snubber cap. Besides the fact that many snubbers (i.e. the standard Zobel network) include them, Steve's reasoning makes sense. Why exactly don't you care for them, Sulaiman?

Oh, and while we're on the topic, there seems to be a lot of disagreement as to what kind of transients you have to be prepared for. My electrical theory being quite a distant memory, it still seems to me that inductive kick should be limited to 2*Vcc. However, given the ratings of devices routinely destroyed by it, it's obviously more. What exactly does define its voltage range?

Sulaiman

Wed Oct 03 2007, 08:46AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

For a simple ignition coil driver like mine (555 and mosfet) that has no feedback/regulation
once I set an output power using a potentiometer, that power will be processed continuously.

Many experiments (playing) involve the coil temporarily not having a load
so I sized my TVSs to handle full power (only 30W in this particular case)
which means I didn't need a capacitor snubber.

Also, unlike ferrite cored TV/Monitor flybacks, ignition coils have relatively high leakage inductance
(I guestimate about 10% of the primary inductance)
The value of snubber capacitor required to catch that much leakage inductance energy is quite large.
Not that it matters as for me the entire primary energy needs to be "snubbed" in the case of no-load.
Adding a capacitor would do no harm - it would reduce the initial dV/dT which would be good.

As for the voltage of the primary flyback spike / 'inductive kick' here are some limits
1) Output voltage x (Npri/Nsec) plus whatever leakage inductance 'spike'
2) Ipri x SQRT(Lpri/Cpri)
3) Vclamp
4) Vceo or Vcex

Basically you start with energy stored in the magnetic field (E= 0.5 x Lpri x Ipri^2 )
when the transistor turns off the energy has to go somewhere.

Case 1 is ideal, all the energy goes to the output, the primary 'sees' (output voltage) x (transformer ratio)
Case 2 would be for a primary capacitance large enough to 'snub' all of the energy - what a waste
Case 3 is like my TVSs absorbing all the energy when the output doesn't - another waste
Case 4 is when the transistor exceeds it's voltage rating and avalanches - usually not good.

Case 2 gives the most important information;
with little or no added capacitance the voltage will rise until something stops it.

There is no fundamental reason why the flyback spike should self-limit to 2 x Vcc
that would be bad as the output could only then rise to 2 x Vcc x (Nsec/Npri)

Hope that makes sense.

P.S. For a bjt driver, given that Ton is usually larger than Toff
and the primary current starts from zero and rises almost linearly,
quick/hard turn-on of the transistor is unneccesary
quick turn-off of the transistor is required for good performance
SO, supply the base current for the main switching transistor via a resistor
and use an npn transistor to short the base to ground for turn-off.
This allows a small drive current to be used (555 etc.) due to the gain of the second transistor
it precisely defines the maximum base current for the main transistor
the main transistor turns off quickly
the secondary transistor doesn't dissipate much power / get hot
as it has no current during the ON time and negligible voltage during the OFF time.
The resistor will of course get hot and needs to be sized accordingly.

If you can make transformer drive for the base of your switching transistor work then that would be better
but there are many hidden 'challenges' with transformer base-drive.

RickR

Wed Oct 03 2007, 03:36PM

Registered Member #93 Joined: Thu Feb 09 2006, 04:11PM
Location: Houston, Texas
Posts: 64

Sulaiman,

In your earliest post on this thread, I saw that you use 10 paralleled 1.5KE350s to clamp transients. This brings up a question that has often bothered me. I've seen others doing similar things to increase the power handling capabilities of TVSs. However, it was my impression that what happens is, like paralleling diodes to handle higher currents, only one of the TVSs will conduct. That is to say, whichever of the TVSs conducts first will handle all (or a very big share) of the current and the others won't conduct at all.

Am I perhaps missing something? Does what I'm asking make sense?

Thanks,

Rick

Sulaiman

Wed Oct 03 2007, 07:55PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

In my post of 02 Oct I wrote
peak clamping voltage, e.g. for 1.5KE350: 350V @ 1mA, 482V @ 4A, 618V @ 16A
which contains the answer.

A TVS is basically a zener diode and it's voltage increases with current
somewhat like a zener with series (non-linear) resistance
so as one TVS takes a little more current the voltage increases causing more current to flow in the other TVSs
so the internal resistance of each TVS causes current sharing.
It will not be perfect as each TVS will be slightly different and some will get a little hotter than the others
but the sharing occurs anyway.

Although it's not a practice that I like,
I have seen several designs where ordinary rectifiers are paralleled for increased current rating
relying on the same principle that real rectifiers can be considered as an 'ideal' diode
(which wouldn't share current due to a negative temperature coefficient)
in series with resistance - which does force current-sharing.

I have not measured the current-sharing but the continued operation of my inverter is good enough for me.

Steve Ward

Thu Oct 04 2007, 12:48AM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

The size of the snubbing cap is dependent on the blocking voltage of the switch. Since the energy storage goes up with V^2 the cap can be pretty small if you can stand 400 or 800V. As far as i know, television flyback transformer drivers use a "snubbing" cap, but someone correct me if thats not true.

Using TVS alone is probably just fine for stuff under a few hundred watts of power, but id like to see you try that trick with a 1kW++ converter

Shaun

Thu Oct 04 2007, 02:54AM

Registered Member #690 Joined: Tue May 08 2007, 03:47AM
Location: New Jersey, USA
Posts: 616

Sorry about the thread-jacking, but does it matter if you use bipolar or unipolar TVS diodes? I have the bipolars installed in my DRSSTC half-bridge, and was wondering if they will cause problems.

sehrgut

Thu Oct 04 2007, 03:32AM

Registered Member #1017 Joined: Fri Sept 21 2007, 09:08PM
Location:
Posts: 24

It shouldn't. A bidirectional TVS is an open circuit both directions until it avalanches. A unidirectional TVS is, for all other intents, a diode, and must be installed the right way 'round to avoid shorting all your drive current through it (and with some designs, directly shorting the power rails).

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob