switch selection

A little heavy for the currents described, but:

These are great little relays:

They switch in under a millisecond, and no semiconductor will match them for Roff/Ron ratio (ballpark 1e14 maybe), or capacitance for that matter.

We used over 4000 so far, and I don't think one has failed. The latching versions are especially fun.

John

Yes they are.

Mmm... at lower frequencies, yeah, but about 100MHz or so wouldn't you really prefer something like

Well, you can do a lot with a DPDT switch, like two T-switches in succession, if you really want isolation. And Hittite has a very weird definition of DC... "blocking capacitors are required."

I doubt this one will switch 500 volts, either.

I'd guess the Fujitsu relays would be fine at 100 MHz.

John

Certainly good points.

They're certainly not bad... but we have a several projects where the standard isolation from any path to any other path from typically DC-18GHz is 60dB, which is not exactly trivial, so the more isolation we can get out of the switch itself, the better. Power is limited to more like 30dBm into 50 ohms, though, so orders of magnitude below 500V.

Someone should definitely take them to task for that nonsense.

---Joel

I've seen a few manufacturers lately who use "LF" instead of "DC" to indicate the lower frequency limit of their parts, like RF switches and mmics. That's more honest.

We used one Hittite part, an 8pst switch, that absolutely screwed up below about 80 MHz. The huge lettering at the top of the datasheet said (envision huge letters) DC. When I complained about it, they told me that I hadn't examined the S-param graphs closely enough.

Hey, it still does:

John

Thanks for your help guys. This FTR-B3SB4.5Z thing seems to be very attractive except height (have

I just checked "mechanicals". The height limits are:

--- Check this out:

You'll need to get/wind a coil for it, but at 17AT you should have no problem fitting it onto the 5mm side of the board.

What supply voltage do you have available to drive the coil?

And, BTW, can you rout a slot in the board and slip the relay into it to get a few more mm of height that way?

JF

standard

We've used jillions of them in thermocouple circuits, millivolts and nanoamps, and they work fine.

If you can live with the 27 ohms + 400 volts, that's a better choice.

John

Winding a coil.... Grrrrr. I'd rather not get into it. I would probably get yelled at by people who deal with CMs

It's a very good idea. BUT... The board is very tight as it is and cutting out 10x7mm hole doesn't make me feel very good... I'll look into it, though.

One more comment. If I have zero cutout tolerance, I will have 0.5mm of leads on each side to solder to. The relay has 0.2mm width tolerance. This leaves me with 0.3mm. The leads have +...-0.3mm tolerance. This leaves me with... nothing. :o(

Hmm... about -3dB return loss at 100MHz for a "non-reflective switch."

Apparently they went to a different class than we did in school...

Nah, they figure if they deal with half of the reflection, you can handle the rest.

Cheers,

Phil Hobbs

Turns out there are capacitors in series with the termination resistors. That screwed me up big time... unterminated signals would overshoot and blow through the "off" fets into the output. I had a conference call with a rep and the chip designer, where he reluctantly revealed the caps, and blamed me for not inferring their existance by reading the s-param graphs carefully. I asked him the value of the caps, and he refused to tell me: "that's proprietary". As if I couldn't calculate or measure them in 50 seconds or so.

Grrrrrr.

And the THS3062's are still fatally broken, and the datasheet still doesn't say so.

Somebody should start an ICBugs.com web site. I could donate a bunch of items.

John

I bet you have a bunch of half-silvered mirrors, Phil. :-)