# Why are my reed relays releasing 10x faster than spec?

I have Magnecraft W171DIP-25 reed relays (datasheet). They are DPST relays with a built-in coil suppressor (according to datasheet). Their listed response time is 1 ms, and, following the logic of this question (Do reed relays have shorter opening than closing time?), I would expect the release time to be at least 1 ms.

I measured the release time with an oscilloscope: 0.1ms.

I'm not complaining, but why is this? I did it in two distinct (identical) setups: same result.

I don't have a 5V supply available on the board, so I just ran them off of a 8V supply and stuck a 100 Ohm resistor in from of the relay (the coil resistance is 200 ohm like on the Mouser site, NOT 500 ohm as in the datasheet). I used a NPN transistor to switch the relay.

The specified switching time is a maximum value, not a typical value.

Your relay may indeed perform better than the specification, but don't count on this for your design. Some other relays of the same type may not perform quite as well, especially under extreme conditions (e.g, when unusually hot or cold, when switching larger voltages or currents, when switching rapidly, after ten years of heavy use, etc).

• A note: as I am a hobbyist, my design only uses 2 relays, and will not be replicated. So I really don't care if a few relays take a bit long. – dpdt Apr 20 '16 at 18:22

As others have mentioned, the operating time specification is a maximum, and no indication of typical is given. The relay will release faster if you allow the magnetic field to collapse more quickly- to do that requires a lot of voltage to appear across the coil.

Often users will put a diode directly across the relay coil to prevent the voltage from rising much above the supply voltage. This is the easiest on the switching transistor and the slowest.

Note also that putting a resistor in series with the coil as you did will actually speed up the close operation as it begins to approximate a constant current source, so it reduces the effect of coil inductance. The time for the current to rise to 63% of final $\tau$ = L/R so the higher R is, the shorter the time constant will be (for a fixed final current). It will also reduce the release time if the diode is across the resistor + coil rather than just across the coil. Of course, the magnetic field is only part of the operation- even if it appeared and disappeared instantly, the contacts would take some time to move.

If you didn't put a diode in there, you should, to protect the transistor.

• Might be interesting to note that, before high speed pulse generators became available, a mercury-wetted reed relay was a common approach for providing fast-rising/falling pulses to assess the rise time of an instrument. They can achieve subnanosecond make or break times, so 1 ms seems wildly pessimistic. – Oleksandr R. Apr 20 '16 at 20:42
• @OleksandrR. The mercury-wetted reeds took hundreds of usec to operate, but the make/break itself was very fast. Cp. propagation delay vs. rise time. – Spehro Pefhany Apr 20 '16 at 20:48
• @OleksandrR. I don't have mercury-wetted reed relay, just a normal reed relay. – dpdt Apr 21 '16 at 23:47
• I don't care about the close time, just the open time. Does the fact that I put a resistor in help with the open time? Also note that those relays have built-in suppressor diodes. – dpdt Apr 21 '16 at 23:49
• No it doesn't help at all with the open time if the diode is built-in. To speed up open time you would use a relay without an internal diode and use something like a zener in series with a diode to allow the coil voltage to go up to 50 or 60V (near your transistor rating). – Spehro Pefhany Apr 22 '16 at 0:06

1 msec is the guaranteed operating time. There is no minimum listed, so there's no reason to worry about it.

What you measured was the time necessary to open a tiny gap between the relay contacts. But a longer time is needed to get at least 90 % of the nominal contact distance and the full specified insulation voltage. You tested only one part and not a large sample from different batches of manufacturing.