Is this op-amp uV/C that high?

I'm try to figure what the offset voltage temp drift for the Ina116P op-amp,

Ina116P datasheet:

formatting link

At the bottom left side of page 5 is a chart. Looks like there's an appreciable chance of getting an op-amp that near 20uV/C drift! I would like to build an electrometer that has extremely low input bias current and low drift, but that kind of drift seems far to much. If it's 20uV/C and a gain of 1000 then it's 20mV drift per C, correct? I'm more used to op-amps like the AD8551, which has a typical drift of

0.005uV/C.

AD8551 datasheet:

formatting link

I appreciate any help or advice. Paul

Reply to
Paul
Loading thread data ...

Hi,

I'm try to figure what the offset voltage temp drift for the Ina116P op-amp,

Ina116P datasheet:

formatting link

At the bottom left side of page 5 is a chart. Looks like there's an appreciable chance of getting an op-amp that near 20uV/C drift! I would like to build an electrometer that has extremely low input bias current and low drift, but that kind of drift seems far to much. If it's 20uV/C and a gain of 1000 then it's 20mV drift per C, correct? I'm more used to op-amps like the AD8551, which has a typical drift of

0.005uV/C.

AD8551 datasheet:

formatting link

I appreciate any help or advice. Paul

Reply to
Paul

AD8551_8552_8...

Paul, It looks like you are reading the spec sheet correctly. You'll notice that the INA116 instrument amp has a Vos of 2mV (max) to start with..... so at a gain of 1000 you have 2 volts to deal with already... (Why worry about a measly 20 mV?) With that much offset it's hard to ask the opamp to be much more temperature stable than is listed. What's wrong with the AD8551? You could always look into chopper stabilized opamps for low offset and drift.

Reply to
George Herold

Thanks for the reply George. That's very disappointing to read that it truly has that much drift. The DUT voltage is near 10uV DC, so I need that sensitivity. Also I would like to eliminate as much external bias current flowing through the DUT as possible, hence my big attraction to electrometer type op-amps with their femto level bias current. The AD8551 is a great chopper op-amp, but it has far too much bias current. Also I'm not to fond of the possibility of the chopper pulses getting past the filter.

Uggg, must I scrap my entire circuit and find another electrometer type op-amp?

Regards, Paul

Reply to
Paul

What is the DUT?

John

Reply to
John Larkin

Hi John,

Special high frequency low signal diodes used to pickup weak signals. The diodes are used as a passive method.

Would you happen to have any recommendations on a low bias current op- amp that has a max Ib of 100fA?

Paul

Reply to
Paul

Can you chop the input signal (light? RF?) and use a lock-in technique? Or occasionally auto-zero, with a relay maybe?

National has some fA rated opamos, but any fA part must be jfet or cmos, so don't expect great DC specs. As you note, choppers have charge-induced bias current and are usually noisy.

You could temperature control an opamp, or put a temperature sensor nearby and tweak a compensation factor.

John

Reply to
John Larkin

Thanks for the good advice. I just have a bit of experience in temp controlling op-amps, but it sounds worth it. I nearly have a truck load of thermistors ranging from 402 SMD to the typical large round versions. As far as I know it would take a bit of trial and error to calibrate such a thermistor offset circuit for the op-amp? Also I suppose there's no reason to add more thermal insulation to prevent the op-amp temperature from varying as much.

Regards, Paul

Reply to
Paul

s/AD8551_8552_8...

l

I'm curious though. In chart (page 5) in the datasheet seems to indicate that the drift per typical op-amp decreases with higher gains? Does anyone know how to interpret that chart in terms of gain? For example at gain of 100 they show that 19% of the parts have a drift near 10uV/C, but is that 10uV/C per gain of 1 or per gain of

100? If the gain is set to 100 and the *output* only varies by 10uV/C then that's not so bad. Somehow I doubt that's correct though, no? It's probably 10uV/C * 100. :-(

Regards, Paul

Reply to
Paul

datasheet:

formatting link

I'm curious though. In chart (page 5) in the datasheet seems to indicate that the drift per typical op-amp decreases with higher gains? Does anyone know how to interpret that chart in terms of gain? For example at gain of 100 they show that 19% of the parts have a drift near 10uV/C, but is that 10uV/C per gain of 1 or per gain of

100? If the gain is set to 100 and the *output* only varies by 10uV/C then that's not so bad. Somehow I doubt that's correct though, no? It's probably 10uV/C * 100. :-(

Regards, Paul

Reply to
Paul

--
"No reason _not_ to add", yes?  Also,:

http://www.isotemp.com/product.html
 Click to see the full signature
Reply to
John Fields

t
s
s

What about using a thermistor for the Rg resistor (gain resistor)? Of course I'd have to hope that the drift is in the correct direction, otherwise the thermistor would only add to the drift. Perhaps a thermistor and metal film combo.

Paul

Reply to
Paul

datasheet:

formatting link

Your interpretation is correct.

As you suspect, the offset is referred to the input-- that's why it's called "Input Offset Voltage" on the graph, and that's what the "RTI" means in the Specifications. At the output, you'll see G * delta-Vos. Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

Thanks for link! Similar idea, if not the same as John's suggestion. I could place a small metal place near the op-amp. Attached to the metal place is a heating element and a thermistor. A circuit maintains the temperature. Once concern is the heating element inducing a voltage on the input device / op-amp. I think if the heating element reacts slow enough then the induced voltage caused by a change in heating element current should be low enough.

Paul

Reply to
Paul

One reason not to put an electrometer amplifier in an oven is that you'll greatly increase the input bias current at normal operating temperatures (since the oven has to run somewhat hotter than the hottest possible Ta to be effective).

Of course the variation will be reduced.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

s

of

eets/AD8551_8552_8...

[snip]

"The Journey is the reward"

formatting link

eff.com

OK, I was at least hopeful.:-) Why would using a higher gain resistor, RG, narrow the chart width. The chart makes it appear as if using a high gain R could be helpful in decreasing uV/C per gain.

Regards, Paul

Reply to
Paul

"The Journey is the reward"

formatting link

eff.com

I had a possibly good idea. By placing a double switch on the input DUT, in this case a mechanical switch would be most appropriate IMO, I could flip the polarity of the DUT across the electrometer input and the change in DC voltage output divided by gain and divided by 2 equals the true DC voltage input.

So far example, say the DUT is +10uV DC, and the electrometer output is 120mV DC. The switch now reverses the polarity of the DUT such that the DUT is now -10uV DC relative to the electrometer. So now the electrometer output is say 100mV DC. Given a gain of 1000 the DUT is ((120mV - 100mV) / 2) / 1000 =3D 10uV.

Is that a conventional method?

Thanks, Paul

Reply to
Paul

datasheet:

formatting link

A portion of the offset drift is actually related to the output, and the total effect (referred to the input) is reduced by increasing the gain. The drift in mV/K or whatever at the output only goes up as gain increases, of course, there's no free lunch. Same deal with input-referred noise at lower frequencies.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

Sure, it works if the DUT is resistive. What exactly is it? Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

Hi Paul, This sounds like you're "inventing" a lock-in technique as suggested by John. Can you instead modulate the bias current to the diodes? or some other parameter? Then you look at the signal at the modulation frequency and the unwanted DC offsets go away. (Well they don't go away but they are stuck there at DC.) George

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.