what is the current state of the art in DC solid state relays/transmission gates

[SArepairman]

So, while mechanical relays can be used, what is the current state of the art for solid state relays for low level DC signals?

when looking for parts to meet this criteria is it better to look for solid state relays or analog multiplexers?

[SArepairman]

I am particularly interested in the 1SK6-0001 used as a switch chip in the 34401A multimeter.

It seems that agilent made a custom IC to act as a chip that changes the range for the amplifier gain. Are chips of this caliber available commercially now?

I ask this because I feel like relays are dinosaur technology...

[Marco]

You'll want the relays any way for high voltage stand off (high voltage MOSFETs don't generally have low on-resistance in small packages).

You will want analogue muxes for everything else though, solid state relays are designed to switch high currents. I have no idea how special the HP/Agilent ones are but the input impedance of the amplifier section is probably greater than a GOhm (JFET) so even a couple ohm temperature shift in the analogue mux isn't going to matter.

PS. I notice that according to Dave the 34401A is a piss poor designed multimeter No PTC on the input path for protection, only hard clamping with a varistor (I assume they do this because of the atrocious temperature coefficient of a PTC).

[sync]

A lot of older bench multimeter use discrete ultra-low-leakage JFETs for switching low voltages. Relays for the higher voltages.

[SArepairman]

I don't understand by what you mean when you say temp-co does not matter.

As I understand it substitutes in a different resistance to change the gain.
Are you saying it actually switches in a different amplifier and does not touch the feedback loop?

SO they have one amplifier set to 10, one to 100, one to 1000, and they just switch them in?

[Frant]

So, while mechanical relays can be used, what is the current state of the art for solid state relays for low level DC signals?

when looking for parts to meet this criteria is it better to look for solid state relays or analog multiplexers?

CMOS analog switches/multiplexers and solid state relays (SSR) are widely used today, but what you should use in a particular case depends on the requirements. Mechanical relays, including reed relays for low level signals, can be used to keep you "on the safe side" when isolation, leakage current and series resistance are considered. A leading manufacturer of reed relays is Coto Technology. For many applications, however, a good solid state alternative exists, but you should be careful when choosing one. For a good overview of the current state of technology you can visit the website of Analog Devices, a leading manufacturer of CMOS analog switches and multiplexers. A leading manufacturer of SSRs is probably Omron with the G3VM series for low level signals (see G3VM-21PR11, G3VM-61LR). The most important parameters of such components are: isolation voltage (for SSRs only, CMOS switches are not isolated), output voltage, load current, output resistance (on state), leakage current (off state).

[Marco]

I don't understand by what you mean when you say temp-co does not matter.

The switch resistance together with the series resistance before the amplifier input forms a divider, the series resistance is around 100K (this is accurate both for the unattenuated path and for the attenuated path) and the amplifier input resistance is >1 GOhm. Lets say the switch resistance is 100Ohm and it has a 10000ppm tempco.

Lets ignore the tempco of the other resistors for a moment, so with a 1 degree shift we get a (1e9/(1e5+100+1e9)-1e9/(1e5+101+1e9))*1e6 ~= 0.001 ppm shift in the amplifier input signal.

They have one attenuated path (9.9M+100K attenuator) and one unattenuated path to the amplifier which they switch between with both relays and the mux, the amplifier does have a divider in it's feedback path which also uses part of the mux to change amplification. The same argument as above applies why tempco of the switch resistance is irrelevant.

PS. I should probably point out that the mux IC seems to provide voltage clamping to protect the amplifier (so with 1000 volt input after the primary voltage clamp the ~100K resistance before the mux limits current and the mux then clamps the voltage).

[SArepairman]

So, I was doing research on AD parts.

I found the best bi-polar parts have ~1 ohm resistance and a leakage current of 200-400pA with a temperature coefficient of 10000 PPM.

[sync]

That is too much leakage for a bench DMM input stage. They have usually over 100G ohm input impedance (>10G specified). Look at the 2N4117A for example. Typical 0.2pA leakage.

[pinkysbrein]

It's a little high and jfet multiplexers are probably better but but if you are orders of magnitude better than the input protection it doesn't get you much. The mux can see 1000v on it's inputs if it doesn't clamp it ... the clamping circuit will have more than 0.2 pA leakage.

PS. hmmm, guess I am wrong ... Analog has one remaining jfet mux (mux08) and they say over/undervoltage is okay as long as current is limited. I guess JFETs can survive avalanche and forward biased gates with low currents.