Best Snubber for Nasty Spikes?

of

so

If this valve is juiced up by DC, then all you should need is a diode antiparallel to the valve coil. Something like a 1N4004 should be fine. Adding this diode will add a little bit of turn off delay, but it doesn't sound like that would be much of a problem.

Buy some of those encapsulated RC networks that clip into the mounting system (DIN Rail ... Clip_on_top_of_Relay ... WhatEver) that the controls are using; it looks much neater and the site technician can get/install new ones, should the need arise. Discrete components are subject to abuse, vibration, moisture e.t.c.

The values are not that important - it seems to be a "one-size-fits-all" device - shouldn't surprise me they are 47 Ohm, 0.1 uF as you say.

PS:

Diodes across relay coils are Evil:

They are *easily* blown out by one of the many spikes in an industrial environment; Especially the wimpy 1N4148 diodes that some designers are so fond of. When that happens they will fail short, and then the the wires or the controls will be damaged too! RC-Networks, MOV's or Tranzorbs (if one absolutley MUST have a diode) will survive much longer!

I prefer to put the voltage limiter across the *driver*, contacts/transistor, not the coil, because it is the *driver* and not *the coil* that needs overvoltage protection (and some site techies might forget it if it is not soldered in ;-)!!

You posted something like this in sci.electronics.basics, but I don't think you mentioned, there that this coil is driven with DC. RC networks are normally used only for AC inductive loads. For DC you can do much better with a diode in the network.

I am not sure of your supply situation, but here is the concept:

supply positive----------+----------+ | | C | | | +--+ | | | C | - C (coil) R ^ C contact | | | supply negative---||-----+--+-------+

The resistor, R limits the capacitor current when the contact closes. The resistor can be a higher value than with the AC version, because it can take the entire actuation time to precharge the cap. I would probably go with 470 ohms, 1 watt to handle the 21 watt peak power at closure.

But the diode allows the coil current to detour directly from the contact to the cap, C when the contact opens with only a volt across the contacts, preventing the arc. Then the cap charges smoothly as the coil transfers its energy to the cap, so the rate of change of voltage is limited. Since the inrush current can be kept low with a high value of resistor and the cap has to precharge only once per actuation (not every half cycle as with AC) the cap can practically be larger than with the AC version. I would look for a 1 uf class X1 cap (for across the 240 volt AC line service, about 1000 volts peak capability). The one ampere coil current will start the charge ramp at 1 volt per microsecond, which should be slow enough to radiate very little.

The diode is a 1N4004.

This kind of snubber will not slow the turn off of the hydraulic valve as much as putting a diode directly across the coil.

The comments I made about wire routing in the other group still apply.

--
John Popelish

deg

When

2 of a

affecting

work, so

every

John I think your comment about routing is important, the relays are in the same box as the logic! I'll relocate.

I was unsure that the .1uf @ 600V cap was large enough to absorb the valves coil energy. the solution will be used on AC as well as DC valves.

Also I was unaware that a snubber was available, and thought I would have to make up something.

Peter

Good points. How is a 1N4004 as a flyback eater diode? Is it fast enough? When I look up the specs they are not very specific about speed, since they are intended to be rectifiers, however when I test them they seem fast enough, my

100 meg scope shows no transient. If not a 1N4148, what is the stronger cheapest option? How about a BAW78? Do you think a coil driver really benefits from a faster clamping device?

Rocky

I think 'fast' with diodes applies to the opposite direction: how fast they go from conducting to no current again. How much this bites you depends on switching speed to the on state (off state for the diode) and the frequency.

Thomas

Correct - this is called reverse recovery. Diodes also have the equivalent forward recovery time, usually unspecified. a 1n400x has a reverse-recovery time of tens of microseconds. That just means dont use them in switchmode power supplies

Cheers Terry