MEMS relays still suck

Compared to....?

Am I missing some sort of in-joke?

bob prohaska

Real relays.

I don't think it's funny. MEMS relays have promised to be wonderful for decades, but still aren't. A lot of startups have died.

MEMS oscillators and sensors and gyros and microphones and micromirrors are great.

There are a zillion academic papers on MEMS varicaps, but I haven't seen any commercial products.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

A great idea. Lets use solis state semiconductor tech to make something mechanical, that can break.

I think the idea of 'moving' semiconductors was always fascinating.

Dreamt once that I emptied a bag of parts on my bench and watched as they automatically reassembled themselves into their last physical iteration.

It was extremely depressing, at the time . . . .

RL

Their web site blather is victim to the big-number fallacy. "If every ceiling fan used our switches, 17 coal-fired plants could be taken off the grid". Yeah, right!

I just perused one of their RF switch datasheets. You can't use them for hot switching, so you'd need a switch to protect the switch. Also, the ON resistance isn't very impressive, in the

There's more to worry about, but I've already lost interest.

Oh well.

Jeroen Belleman

When I look at the datasheet I see a combination of size, bandwidth and switching speed that seems very hard to approach with a traditional electromechanical quartet of relays.

Admittedly, I'm old and so is my knowldege. The last relay I handled that had a prayer of carrying 18 GHz was a mercury-wetted reed device about an inch long that could _maybe_ have been built into a coax placed inside a solenoid to switch it. Thats a single channel, ten times bigger than the example cited which has four channels. Slower switching, too.

It just occurred to me that a multiplicative mixer has somewhat the same "multiply by zero or one" attribute of a relay. For signal handling that might do a similar job. No experience with them, however.

What are your requirements?

bob prohaska

We use a lot of relays, over 100K of one Fujitsu part so far. Relays have a combination of on resistance and capacitance that no semiconductor can match. They are kind of big and are tricky in manufacturing.

I'd love to have a tiny MEMS relay with similar specs. They seem to be unreliable. The classic failure mode has been contacts welding closed. That Menlo relay costs $100 and has a hot-switch max rating of 0.5 volts. I assume that more will fry the contacts.

Here are some Fujitsu DPDT relays:

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

MEMS makes some very cool and reliable parts. Just not relay contacts.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

There should be specifications on on the current you can afford to have running through the contacts at the time of of opening, and the amount of current that could flow through the contact during closure, not to mention the capacitance across the contacts at make or break, and the series inductance.

It isn't voltage that fries a contact, but current.

It sounds more as if the gap is very small, and 0.5V across an open contact might create enough electrostatic attraction to close the contact, or prevent the contacts from coming apart.

Well, you are the MEMS expert.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

MEMs grew out of lithography, primarily 2D patterns transferred by chemical etching. Electromagnetics are intrinsically 3D, the only example I'm aware of is write heads for disk drives.

The combo of 3D and size needed for what you want is awkward; too vertical and large for chemical etching, too small and intricate for traditional mechanical assembly. It looks like the mechanical watchmaking industry is starting to use some MEMs components, maybe progress will come from that direction eventually. Not soon, nor cheap.

Might that be a crossbar switch?

Thanks for writing,

bob prohaska

Capacitance is interesting. One can measure kilofarads to attofarads with simple equipment.

But I doubt their fF numbers in a packaged part.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It's a sort of general-purpose tester. The front panel has a bunch of different connectors, D25s and such, and the back has connectors for test equipment. It will pretty much connect anything to anything.

We maybe should have used latching relays. It's borderline for testing thermocouples, from relay coil heating. A big fan on the top cover helps.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

The patterns are transferred as two dimensional images, projected photon-by-photon onto a layer of resist by an optical syste, or electron by electron by an electro-optical system such as an electron-beam microfabricator.

The surviving resist protects part of the surface from being etched away, which adds the third dimension to the structure.

There are techniques which let you etch deep narrow channels.

Printed circuit transformer windings seem to exploit them to lay down thick - tens of microns - layers of copper with only a coupe of microns of spacing between parallel tracks.

The copper can be as thick as the printed circuit substrate material (which can be pretty thin in multi-layer boards).

Printed circuit transformer windings have been around for some twenty years. Maybe you haven't been talking to the right people.

John Larkin being sarcastic. I do seem to know more about reed relays that he does, but that doesn't make me any kind of expert. I've never used a MEMS relay, or even read a data sheet with an application in mind.

You do claim that reeds are good for billions of operations. I've found them to be unreliable.

They have lots of other bad points.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I have mentioned that mercury-wetted reeds are claimed to be good for about 100 million operations, while regular dry reeds are only claimed to be good for about 10 million.

There are lots of ways of using them that lead to shorter lifetimes.

They were invented to improve telephone exchanges, and they worked reliably enough in that application until we found a better way of doing the job.

Somebody like you who doesn't seem to be able to remember crucial details is unlikely to have been able to get the best out of them.

Slow, bulky, thermocouple voltages - the alloys in the magnetic reeds do generate a thermoelectric potential when coupled to copper at a different temperature and the activating coil dissipates heat when the reed is held closed.

Design is all about exploiting the good points and minimising the bad one.

I can certainly understand the appeal of relays in that application. Thanks for the explanation!

bob prohaska