Hi all
I'm currently working on a chemistry project for school, one part of which involves measuring electrode potentials and relating then to concentrations.
This will involve making accurate measurements of voltages which will be all less than 1V. I have calculated that a 10x change in concentration will cause a change in the voltage measured of just under 30mV, if you want an idea of why I need accuracy.
The voltages will be DC and not changing significantly though time.
I currently have an old fluke 87 DMM, which reads to 0.1mV on the smallest scale. However, it has not been calibrated in at least 15 years. Leaving it switched on in the mV DC scale with no leads gives a reading of 0.4mV.
I was thinking that perhaps creating a simple op-amp circuit with a gain of about 10 and using a higher scale might give a better accuracy. However, this raises the question of how well I can determine the gain, and if that measurement would introduce more inaccuracies than it prevents. To determine the gain I would either measure the R1 and R2 values in the circuit (and use 1% resistors) and calculate the ratio, or put a known standard voltage through the op-amp and measure the output (google suggests using an LED Vf). I would need a specialised low DC offset op amp. Is a typical inverting/non inverting setup adequate, or would a more complex setup be needed.
I don't really want to spend much money of my own on this, so I would rather not need to send off my meter for calibration. I have a good set of parts but no PCB making facilities so it would have to be a stripboard project.
Thanks all in advance.
I'm currently working on a chemistry project for school, one part of which involves measuring electrode potentials and relating then to concentrations.
This will involve making accurate measurements of voltages which will be all less than 1V. I have calculated that a 10x change in concentration will cause a change in the voltage measured of just under 30mV, if you want an idea of why I need accuracy.
The voltages will be DC and not changing significantly though time.
I currently have an old fluke 87 DMM, which reads to 0.1mV on the smallest scale. However, it has not been calibrated in at least 15 years. Leaving it switched on in the mV DC scale with no leads gives a reading of 0.4mV.
I was thinking that perhaps creating a simple op-amp circuit with a gain of about 10 and using a higher scale might give a better accuracy. However, this raises the question of how well I can determine the gain, and if that measurement would introduce more inaccuracies than it prevents. To determine the gain I would either measure the R1 and R2 values in the circuit (and use 1% resistors) and calculate the ratio, or put a known standard voltage through the op-amp and measure the output (google suggests using an LED Vf). I would need a specialised low DC offset op amp. Is a typical inverting/non inverting setup adequate, or would a more complex setup be needed.
I don't really want to spend much money of my own on this, so I would rather not need to send off my meter for calibration. I have a good set of parts but no PCB making facilities so it would have to be a stripboard project.
Thanks all in advance.
Connect the Voltmeter leads across a low value resistor. It should read 0.0mVdc.
Leaving them open is equivalent to measuring the voltage across 17 Teraohm resistor.
Check the Voltmeter reads the same when measuring a 1.5V battery with the leads swapped for both polarities, eg. +1.502Vdc and -1.502Vdc.
Check a pair of batteries. in series and both polarities. Now swap one battery around to read near zero net voltage. These will give an idea of linearity of the Voltmeter. eg +1.502 & +1.513 should read 3.01 or 3.02 on the 20.00Vdc scale. it might even flash those values alternately. It averages lots of readings and the average can be either above 3.015 or below 3.015.
If the reading spends 50% of it's time on each of those reading then you can estimate that the actual reading is 3.015 +-0.002Vdc. But it is an estimate and you still have to add in the basic meter tolerance. So this is a linearity test only.
adding an amplifier will introduce extra DC offset that is not easy to ensure is zero.
The Voltmeter already has the circuitry to compensate for opamp offset.
A DMM set to 2.000V will have a tolerance/accuracy.
A good one will be +-0.1% of full scale +-1digit.
A not so good one will be +-0.5% of full scale +-2digits.
Watch temperature.
I don't know how your cell will vary with temperature. Certainly the DMM will vary slightly with temperature.
Don't try to correct for temp, just accept the test room temperature. Note Ta at the beginning and at the end of your sessions. You could use this as evidence for your poor repeatability of your results. Or to show how well you set up your experiment.
Oh. Check the meter's own battery voltage. and note if you need to change the battery part way through your sessions.
There are probably a lot more you can do to show you are thinking about how the environment and test set up can affect your results.
Anybody cam measure a cell and get the same answer as anybody else.
It's the experimenter who gets a different result (on different days) and can explain why that happened, that gets the higher marks.
Leaving them open is equivalent to measuring the voltage across 17 Teraohm resistor.
Check the Voltmeter reads the same when measuring a 1.5V battery with the leads swapped for both polarities, eg. +1.502Vdc and -1.502Vdc.
Check a pair of batteries. in series and both polarities. Now swap one battery around to read near zero net voltage. These will give an idea of linearity of the Voltmeter. eg +1.502 & +1.513 should read 3.01 or 3.02 on the 20.00Vdc scale. it might even flash those values alternately. It averages lots of readings and the average can be either above 3.015 or below 3.015.
If the reading spends 50% of it's time on each of those reading then you can estimate that the actual reading is 3.015 +-0.002Vdc. But it is an estimate and you still have to add in the basic meter tolerance. So this is a linearity test only.
adding an amplifier will introduce extra DC offset that is not easy to ensure is zero.
The Voltmeter already has the circuitry to compensate for opamp offset.
A DMM set to 2.000V will have a tolerance/accuracy.
A good one will be +-0.1% of full scale +-1digit.
A not so good one will be +-0.5% of full scale +-2digits.
Watch temperature.
I don't know how your cell will vary with temperature. Certainly the DMM will vary slightly with temperature.
Don't try to correct for temp, just accept the test room temperature. Note Ta at the beginning and at the end of your sessions. You could use this as evidence for your poor repeatability of your results. Or to show how well you set up your experiment.
Oh. Check the meter's own battery voltage. and note if you need to change the battery part way through your sessions.
There are probably a lot more you can do to show you are thinking about how the environment and test set up can affect your results.
Anybody cam measure a cell and get the same answer as anybody else.
It's the experimenter who gets a different result (on different days) and can explain why that happened, that gets the higher marks.
Last edited:
The key to getting accurate measures of electrode potentials is very high input impedances. You absolutely need opamp buffers, using opamps with extremely low bias currents (generally FET opamps) and low offset voltages. Speed (slew rate, gain-bandwidth) is totally unimportant, you need tight DC parameters.
If you use 0.1% resistors to set the gain, error from the opamp will be the least of your worries. Circuit leakages will be far more likely to be error sources. Real electrometers use exotic techniques like guard rings and active shields.
If you use 0.1% resistors to set the gain, error from the opamp will be the least of your worries. Circuit leakages will be far more likely to be error sources. Real electrometers use exotic techniques like guard rings and active shields.
The offset you see on your meter is what is known as a thermal offset. (Temperature related.) If you put your meter in a refridgerator to cool off you should see it change (Do not freeze your meter, that isn't an answer. You are just cooling it a bit to see how much if any it changes.)
The others are right about load resistance and offset cancelation using the REL button. But you may also have picked up some acidic or alkaline coating between the terminals that are inducing a voltage like a battery. Clean the area between the terminals (both inside and outside) with good high percentage >70% alcohol.
For reliable readings lower than your millivolt range is set up for, you should look into chopper stabilized op amps such as the LT1052 and the ICL7650. The later has an inherent input offset of 0.0007 mV (.7 uV) max. I wouldn't try for better than 10X gain and simply count upon using battery only supply voltages. Read up on chopper stabilized circuits. Plan on using differential inputs.
A premade version of what you need is called and electrometer. Keithley is the world leader in making them. They are fairly pricey though. Might pick up an old workable model off ebay if you are lucky.
Doc
The others are right about load resistance and offset cancelation using the REL button. But you may also have picked up some acidic or alkaline coating between the terminals that are inducing a voltage like a battery. Clean the area between the terminals (both inside and outside) with good high percentage >70% alcohol.
For reliable readings lower than your millivolt range is set up for, you should look into chopper stabilized op amps such as the LT1052 and the ICL7650. The later has an inherent input offset of 0.0007 mV (.7 uV) max. I wouldn't try for better than 10X gain and simply count upon using battery only supply voltages. Read up on chopper stabilized circuits. Plan on using differential inputs.
A premade version of what you need is called and electrometer. Keithley is the world leader in making them. They are fairly pricey though. Might pick up an old workable model off ebay if you are lucky.
Doc
Last edited:
Okay, after further checking I think the ICL7650S op amp powered by 4 lithium cells per rail and using Vishay CMF series 0.1% 25ppm/C resistors will do the task for you. I'd used no higher than 10K/100K gain resistors. Use mylar or polyproplyene bipass caps and keep everything short and close to op amp built on double sided board with traces cut or etched out of ground plane.
For a cheap fairly accurate voltage reference, look at the LM-10 which is a buffer amp combined with a precision voltage referrence. Not as good as the old LM199-399, but they are mostly old stock now and getting very pricey.
If you can find a copy, fluke used to put out an excellent book on precision measurment. I have no clue where my copy might be right now. But somebody may have scanned it.
There are lots of options, such as finding an old lead compensator, but I think the best low dollar solution is the chopper stabilized op amp. Keep in mind things get very strange below a mV. Don't expect absolute stability. There is a reason that things that will inherently measure such things cost a lot.
Best of luck
Doc
For a cheap fairly accurate voltage reference, look at the LM-10 which is a buffer amp combined with a precision voltage referrence. Not as good as the old LM199-399, but they are mostly old stock now and getting very pricey.
If you can find a copy, fluke used to put out an excellent book on precision measurment. I have no clue where my copy might be right now. But somebody may have scanned it.
There are lots of options, such as finding an old lead compensator, but I think the best low dollar solution is the chopper stabilized op amp. Keep in mind things get very strange below a mV. Don't expect absolute stability. There is a reason that things that will inherently measure such things cost a lot.
Best of luck
Doc
Am going to make an alternative suggestion that you get hold of a Solartron 7075 DMM.
I use up to 3 of these for various measurements including from thermocouples and they have microvolt measurement capability with good thermal stability after a warm-up time of about an hour.
They are available here in the UK for as little as £25 on Ebay, plus postage; which may well be the same cost as you building an external circuit for your Fluke.
For you to have an idea of their resolution, have a look at the 7075specs.pdf available here:
Friedrich Messtechnik - Manuals-Downloads
They include the Guard facility for their inputs mentioned earlier and very high input impedance.
hope this assists
Mik
I use up to 3 of these for various measurements including from thermocouples and they have microvolt measurement capability with good thermal stability after a warm-up time of about an hour.
They are available here in the UK for as little as £25 on Ebay, plus postage; which may well be the same cost as you building an external circuit for your Fluke.
For you to have an idea of their resolution, have a look at the 7075specs.pdf available here:
Friedrich Messtechnik - Manuals-Downloads
They include the Guard facility for their inputs mentioned earlier and very high input impedance.
hope this assists
Mik
Rather covers the Solartron or other high end DMM.
The chopper stabilized amp solution in front of his fluke might run $12 US
Doc
A tough one. What you really need is what is known as an 'electrometer', basically it is a voltmeter with both a very high input impedance, specialized circuitry to deal with enviromental noise and it's been designed to work at low potentials (mV/uV).
A standard DMM will give you a false reading as it has a relatively low impedance (20Mohm perhaps) and is not really designed to be accurate at the uV scale.
This doesn't solve your problem though! You've got a DMM and you still need to measure your cell. I would probably build an 'electrometer' front end with some gain in it (the gain will put you in a more linear part of the typical DMM's range, perhaps only x10 needed), if you look-up the AD846 datsheet (Analog Devices | Semiconductors and Signal Processing ICs) it'll give you an idea, or trawl through their applications page: http://instrumentation.analog.com/en/chemical-analysis/segment/im.html
A standard DMM will give you a false reading as it has a relatively low impedance (20Mohm perhaps) and is not really designed to be accurate at the uV scale.
This doesn't solve your problem though! You've got a DMM and you still need to measure your cell. I would probably build an 'electrometer' front end with some gain in it (the gain will put you in a more linear part of the typical DMM's range, perhaps only x10 needed), if you look-up the AD846 datsheet (Analog Devices | Semiconductors and Signal Processing ICs) it'll give you an idea, or trawl through their applications page: http://instrumentation.analog.com/en/chemical-analysis/segment/im.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.