To JT : non T analog switches?

[Fred Bartoli]

A while ago I asked how analog switches could pump current in the following
condition:

Good point, it probably IS charge injection. The measured effect of
that WOULD vary with R value, as John seems to be seeing.

That's pretty obvious in this configuration. But the real situation was

How do you pump fraction of a uA through the resistor with the switch wired
as above?

The answer is because the 4053 has 3 switches connected in a T arrangement
to obtain high off isolation, the grounding switch conducting some of the
injected charges to ground.


Now I need a SPST with low injected charges (pC order) switch with *no* T
arrangement, and ultra low leakage (few pA order) and Ron under say 50R.

I'd like to not resort to the SD200 mosfets.

The Onsemi NLAS323 or NLAS4501 seems a perfect fit (typical 0.1pA to 1pA
leakage, Ron = 10R, Qinj=0.4pC), except I don't know how they are designed.

Jim, I think you said you've designed some switches of the Onsemi series.

Do you have any info on thoses?

And there's one strange point in the datasheet, which is the Ron vs
temperature.

 

[Jim Thompson]

A while ago I asked how analog switches could pump current in the following
condition:


That's pretty obvious in this configuration. But the real situation was


How do you pump fraction of a uA through the resistor with the switch wired
as above?

The answer is because the 4053 has 3 switches connected in a T arrangement
to obtain high off isolation, the grounding switch conducting some of the
injected charges to ground.


Now I need a SPST with low injected charges (pC order) switch with *no* T
arrangement, and ultra low leakage (few pA order) and Ron under say 50R.

I'd like to not resort to the SD200 mosfets.

The Onsemi NLAS323 or NLAS4501 seems a perfect fit (typical 0.1pA to 1pA
leakage, Ron = 10R, Qinj=0.4pC), except I don't know how they are designed.

I don't either, I'm not familiar with those part numbers.

Jim, I think you said you've designed some switches of the Onsemi series.

Do you have any info on thoses?

I did the LVX4051, LVX4052, LVX4053. They have conventional switches,
NOT T-arranged.

And there's one strange point in the datasheet, which is the Ron vs
temperature.

I referred to this once before...

http://www.analog-innovations.com/SED/Sims-pp11-39-SED.pdf


...Jim Thompson

 

[Winfield Hill]

Fred Bartoli wrote...

A while ago I asked how analog switches could pump current in the following
condition:


That's pretty obvious in this configuration. But the real situation was


How do you pump fraction of a uA through the resistor with the switch
wired as above?

The answer is because the 4053 has 3 switches connected in a T arrangement
to obtain high off isolation, the grounding switch conducting some of the
injected charges to ground.

Actually, the standard 4053 and its copies do not have a T arrangement.

Now I need a SPST with low injected charges (pC order) switch with *no*
T arrangement, and ultra low leakage (few pA order) and Ron under say 50R.

I'd like to not resort to the SD200 mosfets.

The Onsemi NLAS323 or NLAS4501 seems a perfect fit (typical 0.1pA to 1pA
leakage, Ron = 10R, Qinj=0.4pC), except I don't know how they are designed.

A few things on those datasheets. One, the NLAS4501 says SPDT, but it's
clearly an SPST switch! (Use an NLAS1053 or NLAS4053 for SPDT.) Second,
the NLAS323 datasheet says "Ultra Low < 5 pC Charge Injection," but hey,
that's not even low, let alone Ultra-low! Sheesh! Those are cute parts,
but go to Analog Devices for low and ultra-low charge injection, below 1pC.

 

[Joerg]

Hello Win,

A few things on those datasheets. One, the NLAS4501 says SPDT, but it's
clearly an SPST switch! (Use an NLAS1053 or NLAS4053 for SPDT.) Second,
the NLAS323 datasheet says "Ultra Low < 5 pC Charge Injection," but hey,
that's not even low, let alone Ultra-low! Sheesh! Those are cute parts,
but go to Analog Devices for low and ultra-low charge injection, below 1pC.

Or use quad diodes.

Regards, Joerg

 

[Fred Bartoli]

Hello Win,

Or use quad diodes.

Sorry Joerg, but I didn't give all the story.
pA level leakage, low charge, but also few 10s of nV level thermal recovery,
high isolation on high impedance and low noise. I'm afraid a diode bridge
won't do.

This is for a sample&hold or more precisely a block and pass switch (don't
know how to call that better) which will be ahead of my low noise
preamplifier and will allow the DUT transient recovery analysis without
suffering from the low noise preamplifier recovery. (few 10s nV accuracy
after a 0.5V transient).

But I'm now more carefully looking at some AD parts I discarded before and a
(very) few look promising.

 

[Winfield Hill]

Fred Bartoli wrote...

This is for a sample&hold or more precisely a block and pass switch
(don't know how to call that better) which will be ahead of my low noise
preamplifier and will allow the DUT transient recovery analysis without
suffering from the low noise preamplifier recovery. (few 10s nV accuracy
after a 0.5V transient).

You could design better overload capability right into the preamp.

 

[Joerg]

Hello Fred,

Sorry Joerg, but I didn't give all the story.
pA level leakage, low charge, but also few 10s of nV level thermal recovery,
high isolation on high impedance and low noise. I'm afraid a diode bridge
won't do.

That may not be a job for diodes. Although you can achieve zero charge
injection.

This is for a sample&hold or more precisely a block and pass switch (don't
know how to call that better) which will be ahead of my low noise
preamplifier and will allow the DUT transient recovery analysis without
suffering from the low noise preamplifier recovery. (few 10s nV accuracy
after a 0.5V transient).

As Win said a saturation-proof amp would help. Also, look at T/R
switches such as those used in ultrasound equipment. There we have to
stomach pulses up to 200V and the receiver has to be able to hear the
grass grow within a usec or so after that.

Regards, Joerg