Low bias current opamps

Impressive.

You can make it even better--Pease wrote an ap note on nulling the LPC661's Ib by varying / bootstrapping the supplies.

Jeroen a écrit :

When I did my 200pV/rtHz preamplifer with some huge Interfet IF3602 I was very pleasantly surprised the IG was under one pA at 20mA Id, for it to be spec'ed at 0.5nA max in off state which is generally more favorable due to the lack of impact ionization. (Ok, I run them a real low VDS).

How did you measure that?

Yes to minimizing test time, leakage in the DUT board, handler, etc. plus guard-band. The parts also tend to get charged sliding on the handler, so these kinds of measurements are really tough in production.

It only has 1000v ESD. I assume the protection structure is minimal.

for low bias : how about the LMP7721 ? "the industry's lowest guaranteed input bias current precision amplifier. The ultra low input bias current is 3 fA, with a guaranteed limit of ±20 fA at 25°C and ±900 fA at 85°C."

the classic is the LMC6001

also I guess you may have seen ,"what's all this teflon stuff" by Bob Pease.

Jure Z.

I made it an integrator with a 10pF feedback capacitance. The output drifts slowly and linearly at about 17uV/s. It drifted a mere volt since yesterday afternoon! There is a plot of the voltage vs. time curve at . The steeper part near the start is where I put something slightly radioactive nearby.

Jeroen Belleman

Yes, I looked at that one before I discovered the LPC661. I decided to give it shot, with the result as described. An LMP7721 is ten times the price, supposedly because of the cost of testing it for that low Ib.

Yes, of course.

Jeroen Belleman

I think that chip is what is used as the front end in current Keithley electrometers.

Note that here it seems to provide yet another kind of spec: a "typically less than" spec of 40fA.

So "max" is 4pA, "typical" is 2fA, and "typically less than" is 40fA. ;)

Did you take note of their comment, "the leakage of the capacitor and socket must be taken into account?"

Jon

Yes, the input node is in the air. In fact, the whole package is sitting on top of the bypass caps. It's just a quick lash-up to see what could be done. The capacitor is a 1206 NP0 ceramic. If it had been significantly leaky, I would have seen a hint of an exponential curve. Over the time scale I have been taking data (~20 hours), there is no sign of that yet.

Jeroen Belleman

Jeroen Belleman schrieb:

Hello,

but what about the 10pF feedback capacitance and its leakage current? Are you sure that the output drift is caused by the bias current only and is not influenced by the leakage current of the capacitor?

Bye

Interesting that the data sheet shows the part in "hermetic package" to be more than 10x worse leakage (but no hermetic packages are shown, just DIP and SOIC).

If that were it, the curve would always be concave away from ground, (it would go as [1-exp(-t/RC)] if the resistance were linear), whereas Jeroen's crosses zero and is a very creditable straight line, suggesting that it's ordinary leakage.

Cheers

Phil Hobbs

(exponential the wrong way up, but you know what I mean--cut me some slack, we just had a very unseasonable snowstorm to go with our hurricane)

Cheers

Phil Hobbs

Got power back?

Yes, thanks--we were only out for a day at home with the hurricane. Internet and phone (Optimum) took a week. Cell service at the lab was out for 9 days and is still spotty, but the lab power and cable stayed on, more or less. (The UPS didn't shut down the server, anyway.)

Power didn't go out due to this snowstorm--not like Hallowe'en last year, we had 2 feet of snow dumped on us, with all the leaves still on the trees. We were out for a day then too, but friends of ours were out for 10 days.

Of course the same friends are still out from Monday the 29th. They have boatloads of money--I wonder why they don't get a natural gas standby generator. You don't have to stand in line to get fuel for that.

Cheers

Phil Hobbs

If the leakage were voltage-dependent, yes. Is there any mechanism by which there might be leakage that is fixed rather than dependent on applied voltage across it? (I can't think of one. So I think I agree with you. But the question has to be asked. What happens if the capacitor is left in free air unattached? I'm thinking about the outside environment in which all this is contained. And again, I can't think of any reason to imagine different from what you said.)

Jon

What page are you on?

The bold face limits are over temperature, so the military ("M") part has a higher leakage limit over temp. [100pA versus 4pA.]

What I find confusing is the Ib limit on datasheet page 2. The industrial temp range devices will have no leakage higher than 4pA over temperature. But the military part can leak as much as 20pA at 25 deg C.

Upper right graph on PDF page 5 (numbered page 4). "Input Bias Current vs. Temperature".

Strange- maybe that limit is actually tested? I see datasheets from as far back as 2001 that are exactly the same regarding the above-mentioned two points.

Best regards, Spehro Pefhany