a low leakage (<1pA) electrical switch?

Hi, I'm looking for a low leakage (

Hi, I'm looking for a low leakage (

DigiKey has several hundred max326 parts in stock, DIP and soic.

Right.

I agree, discrete FETs (rather than ICs with their substrate leakage, etc.) are best for the under 10pA region. WRT the 5-ohm 2n7000, I'd say even that's not quite small enough, *very* small MOSFETs are best for sub pA use. zhenzhongwang should know that low Ron is the result of a large active area, which is not compatible with very low leakage.

There are switch circuit configurations that can help. E.g., the "T" form can be used to improve the 1/Ron vs leakage-current tradeoff. A set of three sd210 lateral DMOS fets can be used (with an inverting opamp, similar to AoE fig 4.50) to achieve an Ron of about 70 ohms with under 1pA of effective leakage.

However, not meant to detract from the truth of Robert's remark.

One CMOS switch IC with good-looking datasheet leakage curves (under 10pA) is Vishay's 16-year-old DG417 series,

You were damn lucky to get anything that Maxim/Dallas advertises! That aside, at 1pA, who cares about the on resistance - even 1meg is an almost short. And any FET-based device is going to have temerature leakage doubling (per 10C) problems. Try a 2N7000 which is a small MOSFET.

Relays are wonderful. You can get a sealed DPDT latching relay, 1T ohm off, 1500 volt standoff, zero thermal EMF, tiny surface mount, isolated logic-level drive, for under $3. Most of these tiny relays stand a bit off the board, so wash nicely... no trapped flux.

John

I agree.

I see in my work email that Zhenzhong Wang also wrote a scientist here at the Institute, asking about a drawing for an STM preamp I designed some years ago for this fellow. I'm not sure which one that was right now, but most of my STM preamp designs (if they had the dual-feedback resistor feature), used NAiS Aromat TN2 relays.

What leakage is it that concerns you? Are you worried about leakage from the input to the output or from input to ground, or from switching signal to output? Also, are you concerned with leakage in the on condition or the off condition? Give us some idea of where the problem is.

If you just need isolation between input and output, I'd suggest a

Norm Strong

Relay ? Seriously.

Hill, many thanks. I'm a student of Prof. DongminChen, and use lots of boxes named RIS***, which you designed several years ago. Sometimes they work not really well.

A relay is only true contender.

As difficult 1pA was to measure I found common HEF4051 to have this kind of leakage. Stray surface leakages is what usually gets you.

Adam Seychell

Hello Win, Hello Zhenzhong,

Reed relays may be an option here but they are pretty expensive nowadays. IIRC even the regular Coto 9800 series lists 10^12 ohms insulation.

Digikey has lots of them.

Regards, Joerg

Sometimes they don't work well?

But I sometimes wonder if using a pair of reed relays instead would have some advantage over the TN2 "conventional" relay. Yes, glass reed relays have vacuum contacts, but the TN2's gold-clad silver contacts in air continue to work well, even though at 10pA it's certainly a classic case of "dry" switching.

You are right, i forgot about the DG series (been ages..).

I think only if you make it yourself..reed capsule carefully cleaned and then not contaminated by pinkies, coil shorter than capsule length and preferably not touching the glass envelope.

If you've ever been a smoker, seeing the tobacco goo that collects on even moderately high voltage surfaces is down-right scary :-(

...Jim Thompson

in

fig14, what is a wire spring relay?

martin

Hello Win,

Yes, sometimes Japanese products are a miracle in miniaturization. My preference for Reeds because of their encapsulation dates back to the

Regards, Joerg