Cheapest way to microvolt measurements

[stcoso]

What would be the cheapest way to measure in the microvolt DC range (let's say up to 1mv with 1uv resolution and 1% error) ?
A used bench top multimeter? A very low offset amplifier with a multimeter or a high resolution ADC?

[Conrad Hoffman]

Lots of time but no money, build an amplifier. Lots of money but no time, buy a good benchtop meter.

[stcoso]

I guess it's the first...

[David Hess]

The cheapest way is common voltmeter with 1 millivolt accuracy combined with a x1000 low offset drift and low noise amplifier, which would typically be precision bipolar or chopper stabilized.

[bob91343]

I think my HP 3456A comes pretty close.  Dealing with noise and thermal difficulties is at the heart of the problem.

[MiDi]

What would be the cheapest way to measure in the microvolt DC range (let's say up to 1mv with 1uv resolution and 1% error) ?
A used bench top multimeter? A very low offset amplifier with a multimeter or a high resolution ADC?

I read 1% error as 10uV absolute.
If the maximum impedance to be measured is not more than a couple of k\$\Omega\$, than a discrete built instrumentation amplifier with modern zero drift opamps and a gain of 100 could be used.
100x would require a 3 1/2 digit MM with 100mV range only.
A modern zero drift opamp is e.g. ADA4522-4 (Quad) with 6.5uV max offset.
At this level thermal EMF comes into play and has to be dealt with.

[ScoobyDoo]

Hello stcoso

Have a look to QA351 from Quantum Asylum
https://quantasylum.com/collections/frontpage/products/qa351-uv-dc-voltmeter

Herzlichen Grüßen/Meilleures salutations/Best regards/Un saludo
ScoobyDoo

[guenthert]

Hello stcoso

Have a look to QA351 from Quantum Asylum
https://quantasylum.com/collections/frontpage/products/qa351-uv-dc-voltmeter

Herzlichen Grüßen/Meilleures salutations/Best regards/Un saludo
ScoobyDoo

I like how they managed to avoid specifying any error (uncertainty in measurement). 
It's cute and I can see the appeal, but you might get an old used nanovoltmeter for that price (which might or might not meet its original specifications).

It's all pretty moot until the op tells us, what is to be measured.  At least source impedance and error budget need to be known.

[dietert1]

The only disadvantage i can see is the BNC inputs. Maybe one can add copper binding posts. Otherwise it seems to be state of the art, but no idea how isolated the interface really is.

Regards, Dieter

[stcoso]

Hello stcoso

Have a look to QA351 from Quantum Asylum
https://quantasylum.com/collections/frontpage/products/qa351-uv-dc-voltmeter

Herzlichen Grüßen/Meilleures salutations/Best regards/Un saludo
ScoobyDoo

I like how they managed to avoid specifying any error (uncertainty in measurement). 
It's cute and I can see the appeal, but you might get an old used nanovoltmeter for that price (which might or might not meet its original specifications).

It's all pretty moot until the op tells us, what is to be measured.  At least source impedance and error budget need to be known.

Source impedance would be in the MOhm range...

[Kleinstein]

The noise specs (example curve) don't look that great.  I would prefer something like a used HP3478 DMM (5.5 digits with a 30 mV range).

The source impedance can be an important factor. The usual AZ OP are good for source impedance more like < 1 M. Especially the low noise one don't like high impedance. Also some DMMs don't work that well with high impedance.
So the best amplifier depends on the source impedance and also how much filtering can be added.

[stcoso]

As much filtering as necessary... Bandwidth is not a problem here

[tszaboo]

Cheapest?
An INA128 with a 0.1% 50 Ohm resistor powered by two disposable 9V battery feeding to a chinese 3.5 digit multimeter. You might also need a trimpot to null out the offset.

[KT88]

What is your signal source?
In case you want to measure thermocouples, there are some ICs like the AD8495 that offer integrated cold juction compenstion in addition to the amplification...

Cheers
Andreas

[stcoso]

I need to measure various source. A low pass filtered photodiode signal (but this is 50Ohm) and also electrochemical potentials on very sensitive (and high impedance) electrodes.

By the way, I'm evaluating to buy an hp 34401a (at a very good price)

[stcoso]

In any case... An instrumentation amplifier plus a decent multimeter should do the trick from what I've understood here, right?

[KT88]

The 34401a is an excellent workhorse - at a decent price it's a no-brainer...
However it won't help you with the kinds of measurements you mentioned. The source impedance and the way how to deal with it requires specialized equipment. You could get electrometers, SMUs or even dedicated potentiostats for that kind of measurements - but they cost a fortune even used...
But you could build dedicated circuits to adapt your signal source to the 34401a and get nice 6 1/2 digit readings.
To get started with that I would recommend to read the *Low Level Measurements Handbook - 7th Edition* from Keithley/Tek first.
Suggestions on how to design the AFEs could be found here in the forum....

Cheers

Andreas

[RandallMcRee]

I need to measure various source. A low pass filtered photodiode signal (but this is 50Ohm) and also electrochemical potentials on very sensitive (and high impedance) electrodes.

By the way, I'm evaluating to buy an hp 34401a (at a very good price)

As KT88 mentioned, neither the 34401A nor the Quant thingie are going to handle high impedance sources with any accuracy. You can build a decent front-end that will transform your high impedance to something that a normal meter can handle. AD has various app notes, etc. on measuring such things. E.g.
https://www.analog.com/media/en/technical-documentation/application-notes/AN-1563.pdf
https://www.analog.com/media/en/technical-documentation/white-papers/ECG-EEG-EMG_FINAL.pdf

It's always good to understand current state-of-the-art before diving in to unknown waters.

[Kleinstein]

A photo-diode directly connected to 50 Ohms is a solution for relatively large, fast pulses, but not that much for a low level DC signal. One can usually increase the resistor to get more like some 100 mV without to much extra DC error. Than a more normal DMM is OK.
For a 50 Ohms source or a thermocouple  (often similar impedance) the amplifiers bias is not so critical an most AZ OPs are OK to build an amplifier.

Electrochemical cell voltages rarely need to be measured to µV resolution, as there usually is quite some temperature effect. However here one may need high impedance.  So the suitable amplifier would be more like a good non chopped OP (JFET or CMOS input depending on whether noise or bias is more important). If a chopper, than more like a low bias type like max4238 or LTC2054, not so much the low noise ones.

The two examples are quite different in source impedance and thus essentially need different amplifiers / front ends. The usual DMMs are more like good for something in between, so not really good for either case, but maybe still acceptable.

[stcoso]

A photo-diode directly connected to 50 Ohms is a solution for relatively large, fast pulses, but not that much for a low level DC signal. One can usually increase the resistor to get more like some 100 mV without to much extra DC error. Than a more normal DMM is OK.
For a 50 Ohms source or a thermocouple  (often similar impedance) the amplifiers bias is not so critical an most AZ OPs are OK to build an amplifier.

Electrochemical cell voltages rarely need to be measured to µV resolution, as there usually is quite some temperature effect. However here one may need high impedance.  So the suitable amplifier would be more like a good non chopped OP (JFET or CMOS input depending on whether noise or bias is more important). If a chopper, than more like a low bias type like max4238 or LTC2054, not so much the low noise ones.

The two examples are quite different in source impedance and thus essentially need different amplifiers / front ends. The usual DMMs are more like good for something in between, so not really good for either case, but maybe still acceptable.

Well...i guess that this arguments are really far apart. 


The photodiode is 50Ohm terminated because i'm dealing with very fast pulses (sub 2ns edges with duty cycle in the 1-10% range). The photodiode is an hamamatsu s5973-2 with a really small photosensitive area. So, even at high light flux (kW/m^2), with a 50Ohm termination i get small signals (20mV or so at full power with the photodiode really close to the source). I need to record the light waveform (i'm using a Tek 784c) and I need a mean of this value over time (here's why i use a LPF). With the same equipment i'd like to measure a continous light source with flux in the same range of the mean of the pulses that i mentioned before. And the mean is in the 1 mv range.

At the moment i'm using two different photodiodes... the 50Ohm terminated s5973 for the waveform recording and a much bigger (and slower) photodiode with its TIA to record fluxes.

I should probabily separate the waveform recording and the flux measurements...



In the electrochemical cell that i'm measuring i have an high impedance electrode and I certainly do not need 1uV resolution....

[tszaboo]

Well...i guess that this arguments are really far apart. 

No, you are committing the sin of X-Y problem.
https://xyproblem.info
The circuit you are looking for is called Transimpedance Amplifier. Look it up. Your original questing is very misleading, and we were wasting everyone's time trying to answer it.

[2N3055]

First thing you had to check whether Hamamatsu has any application notes for what you need.

At link
https://www.hamamatsu.com/resources/pdf/ssd/si_pd_circuit_e.pdf
there is, among other stuff, two circuits for exactly your diode. They also give general layout advice. That should be more than enough to get you started.

As far as other circuit, it's like Kleinstein said..

[stcoso]

Well, first and foremost if you feel that I may wasting your time you could simply ignore and skip ahead.

I know what a TIA is (it's just in the post above yours) and I definitely do not need that.

[stcoso]

First thing you had to check whether Hamamatsu has any application notes for what you need.

At link
https://www.hamamatsu.com/resources/pdf/ssd/si_pd_circuit_e.pdf
there is, among other stuff, two circuits for exactly your diode. They also give general layout advice. That should be more than enough to get you started.

As far as other circuit, it's like Kleinstein said..

I'm looking at signals (with an oscilloscope) with very fast edges. From what I know (and I may be wrong in this) there's no way to build a transimpedance amplifier that fast. In the application note that you linked they talk about 100MHz BW max which really is not enough.



Here's why the 50 Ohm.

[maat]

In the application note that you linked they talk about 100MHz BW max which really is not enough.

What is fast enough?

[2N3055]

First thing you had to check whether Hamamatsu has any application notes for what you need.

At link
https://www.hamamatsu.com/resources/pdf/ssd/si_pd_circuit_e.pdf
there is, among other stuff, two circuits for exactly your diode. They also give general layout advice. That should be more than enough to get you started.

As far as other circuit, it's like Kleinstein said..

I'm looking at signals (with an oscilloscope) with very fast edges. From what I know (and I may be wrong in this) there's no way to build a transimpedance amplifier that fast. In the application note that you linked they talk about 100MHz BW max which really is not enough.



Here's why the 50 Ohm.

One of the amps is not TIA there. It's basically what you doing, a diode terminated with resistor and amplified.
By using amp, you can also play with optimal diode termination resistor, to maximize voltage and not lose much speed.
By using faster amp and lot of care doing layout, you can get to few hundred MHz bandwidth with 10x amplification without resorting to anything esoteric...That should be enough.

[dietert1]

Some years ago we got an old HP 81519A Optical Receiver 400 MHz. Not really a precision device but turned out to be very useful. We successfully adapted it to a contemporary SMA type fiber interface. The service manual shows an example circuit. There was an intro in the HP Journal January 1985, p27.

Of course, if the photo diode is external, a 50 Ohm interface is natural.

Regards, Dieter

[MegaVolt]

I did not quite understand the required frequency. But for example AD8015 ... more than 100 MHz.