1
\$\begingroup\$

I've seen questions about selecting TVS diodes before (for example, How to choose the correct TVS diode and how to size it?; there are others, but they aren't about TVSs for protecting relay coils). However, the answers there are lacking easy-to-understand guidelines. The best information I've found is from https://www.eevblog.com/forum/projects/how-to-choose-tvs-diodes-to-dispel-flyback-surges-from-relay-coils/:

just ensure the peak current rating of the TVS exceeds the current drawn by the relay/contactor

However, even that leaves some open questions, which are not helped by this blog which asserts that higher is not better... although the example they give seems to assume that peak pulse power is directly related to maximum clamping voltage, which does not seem to be the case in practice?

For the sake of clarity, I'm going to use a specific example, however I am looking for answers that can be generally applied.

Say I have a G5LE-1A4 DC12. Coil voltage is 12V nominal, ~20V maximum, with a draw of 33 mA (0.4 W). My power supply (IRM-10-12) claims ±0.3 V (±2.5%) which, IIUC, means I shouldn't see more than 12.3 V on the relay coils when they're energized.

Question 1: What is an appropriate voltage for a TVS? The previously-cited forum thread advises choosing a breakdown voltage at least double the nominal coil voltage, but IIUC this guarantees that the coil will be exposed to an over-voltage condition (24 V > 20 V)? It seems that a diode with Vclamp-max < 20 V would be a better choice. Such diodes exist with common Vbr-min of 12.4 V or 13.3 V.

Question 2: What should I look for in peak power? (Alternatively, how do I whittle down the options once I've picked a breakdown/clamping voltage?)

Continuing with the concrete example, let's say I'm ignoring the 'double voltage' guideline and looking at something that isn't going to over-volt the coil, i.e. breakdown closer to nominal and clamping as low as possible. Let's say I'm also 'lazy' and want a bi-directional diode. P6KE13CA, SA12CA and 5KP12CA all look like viable candidates. One, however, has a much higher peak power (3 kW vs. 500 W and 600 W) compared to the others. Still, if I'm understanding the preceding advice correctly, even the smaller number exceeds what I need by orders of magnitude. (Is that bad? The previously cited blog seems to think so, but as noted, its reasoning seems suspect. Moreover, I'm not even finding TVS diodes with peak pulse power less than 400 W.)

Besides "obvious" criteria such as "does this fit on my PCB", or the even more prosaic "does anyone have this in stock", what criteria can be used to narrow the set of potential parts?

In summary, how do I go from a relay's coil properties to selecting a specific TVS diode?

\$\endgroup\$
2
  • 1
    \$\begingroup\$ The TVS is more to protect what is connected to the coil, rather than the coil itself. Higher voltage TVSs will kill the field faster, but expose the connected elements to that higher voltage. Therefore your decision doesn't really involve the coil, but comes down to how much back-EMF your circuit can take along with how fast you need the relay to de-activate. \$\endgroup\$ Feb 1, 2023 at 18:09
  • \$\begingroup\$ Let us continue this discussion in chat. \$\endgroup\$
    – Matthew
    Feb 1, 2023 at 20:46

1 Answer 1

3
\$\begingroup\$

What is an appropriate voltage for a TVS?

This depends on what the rest of the circuit can withstand, and how it is constructed.

Using a battery, switch, and coil, place the clamping device in parallel across either the coil or switch. Due to magnetic induction, opening the switch will cause the current in the coil (an inductor) to collapse, creating a (theoretically infinite) opposite voltage to manifest. Energy is not free, nor can it be created nor destroyed - so as the voltage reverses and increases, the available current goes down, keeping the overall power constant.

It is the current which creates the magnetism, so the lower you can get that quickly, the faster the relay will mechanically open. To get it open quickly, implies that it needs to reach a high voltage.

Of course, if you're not concerned with the relay opening speed at all, then a garden-variety 1N400x diode would work fine instead (well, except for really big relays.) This will clamp the reverse voltage to likely a volt or so, and also keep the current inside the coil for the longest time, perhaps a few tens or hundreds of milliseconds for a small relay.

What should I look for in peak power?

If your relay is 30 mA at 12 V, then that's 12*0.03 = 0.36W. Charge that relay fully and dump it into a TVS, and the most work it can do to the TVS is 0.36W. If 400W is the smallest device you can find, then that'll work fine; it'll just be capable of far more "juice" than you're giving it.

How do I go from a relay's coil properties to selecting a specific TVS diode?

  • Do you need high-speed opening? If yes, pick TVS (or resistor-capacitor (RC) snubber.) If no, pick a standard diode.
  • If using a TVS, pick the power rating to be at least 2x the expected coil power (i.e. a monster 48V/8.5A coil "holds" 408W, so the 800W variety should be chosen at least.) Keep in mind, the TVS has a (thermal) time limit for this power dissipation. The power dissipation will be non-linear - it will peak and drop quickly. Small relays won't matter for a much-beefier TVS, but it can be important for larger relays. You may find that the larger, 1.5kW or 3kW variety more appropriate for larger relays. Incidentally, the failure mode for TVS's is to short - consider what happens to the rest of the circuit in this case; perhaps consider a fuse.
  • If a 1W relay is to open quickly, and you've picked a 400 V, 400 W TVS, does the rest of the design support such high voltages on the affected traces/pads/components? Each time the relay opens, there will be a spike of hundreds of volts - care must be taken to ensure that traces are far enough away (clearance, creepage distance) from others that it doesn't arc-over when some dust or humidity gets onto the board. And if this is any appreciable distance away, those traces will act as antennae during this time, radiating an electromagnetic "blip" which could cause further issues. Can unknowing fingers accidentally touch any of this?
  • If a semiconductor is switching this (such as a BJT, MOSFET, or IGBT), then it's voltage rating must exceed the clamping voltage by a significant margin. For a 400 V clamp, pick a 600 V rated device at least. This is partly because a TVS marketed as "400 V" will actually allow a significantly higher voltage (like 575 V) across it for a very short time. The resistance of a TVS is dynamic - it changes with voltage. So under the breakdown voltage rating it may behave like a 1MΩ resistor, and beyond breakdown it may behave more like a 100Ω resistor, thus allowing some "overshoot."
  • Another possibility is powering the relay from an "H-bridge". This is four semiconductor switches in an "H" pattern controlling whether each leg is connected to positive or negative. These are usually microcontroller-driven. Such an H-bridge could theoretically power the (small) relay as normal, and when it is time to turn off, reverse it's polarity for a very short and specific amount of time, in order to neutralize the stored current very quickly, perhaps faster than any other way. If timed right, it could also open with nearly zero current remaining in the inductor. The down-side is this completely thrashes the power supply, and could lead to other issues such as droop on the power rails, regulation instability, even creation of EMI from loop current pulses.
\$\endgroup\$
6
  • \$\begingroup\$ "Do you need high-speed opening?" In fact, I assuredly don't care. Except that slow opening increases the likelihood and/or severity of arcing, which certainly is a concern for my application. More generally, my impression is that a TVS is recommended for power switching. Might be worth noting that? \$\endgroup\$
    – Matthew
    Feb 1, 2023 at 19:24
  • \$\begingroup\$ BTW, as noted in my comment in reply to evildemonic... I think I understand the voltage ratings for most of the components in my circuit... but what about the power supply itself? Say I had the most trivially possible circuit, a PSU directly wired to the relay coils with nothing else (and switching the PSU on the AC side); what then? \$\endgroup\$
    – Matthew
    Feb 1, 2023 at 19:31
  • 2
    \$\begingroup\$ @Matthew in that situation you'de almost certainly not need any "protection" at all. Switching the PSU on the AC side will not result in an abrupt event on the DC/relay side - you'd have a "gentle" decrease in voltage across and current through the relay as the PSU's capacitors discharge through it, and little to no back-EMF from the relay to worry about. \$\endgroup\$
    – brhans
    Feb 1, 2023 at 20:26
  • \$\begingroup\$ Is the arcing AC or DC? If AC, consider a snubber to mitigate the arcing. If DC, consider an "arc-quench" type relay, which is nothing more than a typical one with some magnets added near the contacts. The magnetic flux helps quench the (DC) arc faster. \$\endgroup\$
    – rdtsc
    Feb 1, 2023 at 22:03
  • \$\begingroup\$ AC, but I don't know the amperage. Maybe I just don't understand snubbers, but can you build one that will work across a wide variety of possible current (0.1 A - 20 A), and that won't upset downstream things that expect full voltage "right away"? (I'm also not sure I have physical space for snubbers...) \$\endgroup\$
    – Matthew
    Feb 2, 2023 at 0:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.