RF relays

Fred Bartoli a écrit :

Looks like they've been replaced by

ge

l,

far

lay

first hit on google for low capicity is something like

We use some HF3 relays, it's an RF relay no idea what capacitances are but you can get one is latching maybe that will help ?

-Lasse

That's about the same, quoting 100 fF across the open contacts. That's less than half the coil-contact capacitance, so they must be grounding the coil to make that measurement.

The self-capacitance is of that order as well, so there's a limit to what one can expect to do by shrinking stuff.

Thanks

Phil Hobbs

They make one with a guard ring around the coil.

Thanks, I saw that. They said that the ones without the guard ring had lower capacitance, though. Grounding or bootstrapping the coil isn't too difficult, so that's a win.

I'll probably either stick with the G6KU ones I have, or go with the new Panasonic ones Fred found. I'll get a few of them to test.

Cheers

Phil Hobbs

I have some Cherry company reed-relay keyboards (circular magnet in key plunger.) Still working just fine and they are almost 40 years old and have been used quite a lot. I also have a large tray of reed relays for replacement there. Never used them, though I thought I might have when I bought them back then.

Reeds were used in the IBM Electronic Model 85 (and similar) typewriters, as well. I know, because I turned mine into a printer and scoped out their use before using transistors to bypass them for control by software. Probably used in many keyboards I'm not aware of, too.

Jon

Why are you range switching at such a critical place with such hard-to-work-with components?

There's one other possibility: if the amplifiers aren't too expensive, just use two, one with 50G, and the other with 1G, and switch at the input node. That way, the capacitance of the relay is to ground (really, pseudoground).

A third possibility, putting the two resistors in series and shorting the

message

(I'm selling

improved TIA as

wrinkle: I have

otherwise the Johnson

and dominate the

far are the

coil and

relay coil, or

guys

There are more modest RF relays, still not all that small.

See:

Maybe

Finding much of anything for this is a bit challenging. The search engines do not cooperate.

?-)

Hobbs

and

as

interesting

mode,

through

kHz.

ones

capacitance

the

relay? I

G6K

pF.

ballpark.

That's

grounding

Hmmm. Ya know, it may not be practical to do this "on board". When resistances exceed 1Gohm the board conductances may swamp things. You have been around this block many times before though, i would like to know what board material you use if this is all on board. Just some things i saw when i took the covers of a Keithley 619 that had problems before we sent it to the cal/repair lab. It was a bit strange from a operator or a programmers POV (i had both) but otherwise a very nice instrument.

?-)

voltage

well,

the

far

relay

Wow, that is a real nice part in a lot of ways. The isolation resistance specification while typical of the class is quite lower than real typical values but may not serve Phil Hobbs needs well.

?-)

voltage

well,

the

far

relay

Fine. You made your point, and i agree. The relays i saw in a mere 1 kW HF transmitter were much larger (about 5/8 diameter by 3 inches spdt). It is possible that such relays may be useful for Phil Hobbs needs. The questions involved are shunt conductance and capacitance.

?-)

At 5 KW & up they don't even try to use RF relays at a lot of sites. They use multi pole power contactors typically used for 240 or three phase power at similar levels. They are usually mounted in the center of a large sheet of 1/4" or thicker Phenolic to both reduce capacitance & arcing. :)

Some use solid copper busboys & servo motors, like the old VOA site in Bethany Ohio. For those applications there were no off the shelf relays on the market. :(

Phil does come up with some fairly uniqe needs, but he seems to find answers for them. :)

That switch arrangement is more or less what I'm doing, but with a single amplifier. The effectiveness of the bootstrap depends critically on keeping the stray capacitance to ground very low, ideally below 30 fF, so the whole first stage has to go inside a bootstrapped shield. That's much easier to do with one amp than with two.

I've used positive feedback occasionally in linear circuits, but it never seems to make the final cut, because it tends to be noisy and flaky. If the feedback is close enough to exact cancellation that you get a useful signal improvement at one frequency, Murphy dictates that it oscillates at another. Plus that only works for capacitance--if the FB resistance is too low, you've already taken the SNR hit.

Also, this is the front end for a big expensive semiconductor tester, so reliability and blameless performance are key.

Cheers

Phil Hobbs

It's for inside a semiconductor fab, so the environmental conditions are benign except for EMI. (Switching motor controllers are a bear.)

We'll probably do it on some nice Rogers material, but more for the lower dielectric constant than for the low leakage. Flux will be more of a worry, I expect. I might put vias under the larger components in the front end, so that the solvent can get underneath.

Cheers

Phil Hobbs

(I was travelling yesterday, so I answered a few posts via Google Groups on my phone, but it doesn't seem to have worked.)

SNR. This is the front end of an expensive tool, and there's a sharp tradeoff between the available coupling capacitance and the spatial resolution. Once the amplifier noise and the Johnson noise of the resistor get mixed into the signal, there[s no getting them out again, so I have to switch at the front end.

Cheers

Phil Hobbs

Conductive vias? That may not be so cool here. How about slits or at least drill-outs?

Maybe not PTHs, true. But getting the flux out from under SOT23s and

Any wisdom on getting flux out from under small parts? How about 4-oz copper on the top layer, or something like that?

Cheers

Phil Hobbs

Either that, or us thru-hole parts. Since volume is low and price not a concern you could also have someone bend down the pins some more and hand-solder. Like a lift kit for a SOT23 :-)