reed relays

I've used several hundred Selfein and Pickering SPST reeds, switching every few seconds for many years. None ever failed.

Jeroen Belleman

The main reason I haven’t used reeds is poor performance at low signal levels.

The ones I’ve looked at had nylon cases (outside the glass envelope), so their leakage would be poor, and used different metals for the two contacts, potentially leading to thermocouple offsets. So I went with normal dry contact DIP relays, which had neither problem.

They also suck power, iirc, on account of a lousy magnetic circuit.

Cheers

Phil Hobbs

There are special ones with very low guaranteed thermo-EMF, very good for low signal levels. They are expensive but you can't get comparable performance from ANYTHING else.

Just look, e.g., inside venerable HP/Agilent/Keysight 3458A. Or any piece of metrological (or near) grade test and measurement equipment. They are full of such reed relays.

Coto 3501-05-511, e.g., has guaranteed < 0.5uV thermal-EMF, Dry Only with minimum insulation resistance of 1 TOhm. Electrostatically shielded. Comes with 5V (-05-) or 12 V (-12V-) coil. 350 Ohm / 2 kOhm coil for 5V / 12V versions.

Pickering also has their series 100 and series 101 relays of the similar performance, the latter specially designed for direct drive from 74HC[T]. Coto is significantly better -= they give <0.5uV thermo-EMF right in their specs while Pickering only MENTIONS "3uV or less" in the overall description WITHOUT making any promises in actual specs.

Yes, bad thermals. And they twang when they close, generate hundreds of microvolts of complex ringing for many milliseconds.

Some have to be separated from their neighbors or they may not release.

I did a data acquisition system with reeds and the twanging forced the scan rate low.

I did a many-channel liquid helium level measurement system with reeds to mux the sensors, with cold switching even [1], and failure rates were high.

Thermocouple off-sets can be a problem.

If you want good performance at low signal levels, mercury-wetted reeds can be attractive (if your boards are orietented in a way that lets you mount the reeds with 15 degrees of the vertical). Contact resistance when on is low and consistent, and the liquid mercury stops the contacts from bouncing and damps any twang.

I put them into the Cambridge Instruments 10.5 electron beam microfabricator (as a range switching device) and they worked better than the dry reeds they replaced.

And they are good for about 100 millions cycles, where dry-feed relays start getting more resistive after about 10 million operations. This wasn't a problem for the EBMF 10.5, which was mostly used for writing lithography masks, which tended to be of the same size.

John Larkin has characterised them as unreliable, but reed relays were invented by the telephony business as a more reliable relay than the original Strowger switch.

Am 11.08.26 um 02:26 schrieb john larkin:

I still have two or three trays of mercury-wetted reed relays. I bought them as surplus when I still was a student. They'll go to the problem waste on next opportunity.

A bit smaller than these, but same style & make:

No contact bounce ever, contacts just like soldered. Mounting orientation must be observed; if you shake them, you feel the liquid mercury. They would freeze however at liquid helium temp.

------------------

I had a project at Verigy wafer testers (used to be HP b4 HP was smashed to smithereens by their CEO). I had a full 4096 channel mixed-signal wafer tester (not completely populated) of my own and on some boards there were huge arrays of Teledyne RF170 reed relays. Never heard of complaints; but then my job was integrating a TDR etc and not getting lost in the 10 Meg lines of our own source code and such.

Gerhard

I inherited equipment that used relays in an area with strong stray magnetic fields. Those relays were like little thermometers, with a gas-filled glass bubble at one end that could be heated with a tiny lamp filament inside. The mercury would then be pushed away until it contacted little wire stubs fused into the capillary.

Awful things, *very* slow, unreliable, fragile.

I've got rid of them since. I should have kept a picture.

Jeroen Belleman