Microohmmeter using LT3080

[aneesh]

I am designing a microohm meter to measure resistance from 1mE to 1kE.
I tried designing it using opamp current source but it doesnt give precise voltage with cheap opamps..
Then i tried with Lt3080 which works fine but i want a way to control the current with my microcontroller.. Without using fix resistor..  Please advise..
Also i tried using PWM with RC filter it doesnt give me low as 1mA current

[Kleinstein]

There is not much sens in using the Lt3080 for this purpose. The typical choice would be to send a low frequency (e.g. 5 Hz) AC signal (usually kind of square wave with some bandwidth limiting) to the DUT and than measure voltage and current. The logical choice is an OP circuit - even more precise OPs are cheaper and better than the LT3080 and they alow for both polarities. The high power driving part may not need to be stable if the current and voltage are measured.

[aneesh]

hi,

LT3080 has internal percise opamp and a driver as well... saves lot of external components...
and the normal practise is to send a stable DC current through the resistor and measure the voltage across it..
It works fine for me but to generate a stable voltage using  microcontroller becomes difficult..

[fcb]

I've powered a PIC micro from a precision shunt reference before and used the PWM to generate a precise voltage -worked a treat.

[aneesh]

HI,

can you share the schematic or the rc values ... how my rc stages did you use

[fcb]

HI,

can you share the schematic or the rc values ... how my rc stages did you use

Just two RC low pass networks in series driven by a PIC micro pin outputing PWM, from memory something like 220nF and 10K and then a low-drift opamp buffering the output.

[sarepairman2]

whats wrong with using the lt3080? you can get a nice high output current with them. a microohm meter needs a beefy pulse. the ltl308x variants can go as high as 3Amps... idk about slew rate and stability for capacitance but it should not matter.

now, i wonder what the drift specs are.. but in pulse mode with low duty cycle you might have found a perfect solution. the only problem is at this point you have 2 voltage references, in the LT3080 and in your control circuit. this may or may not be good enough for your application, i suspect that this will be a fairly solid solution from something that is put together from only a few parts with minimal design effort though.

use a EXTERNAL digital to analog converter with reference to control the lt3080 in a precision application.... i don't see why you would use a god damn pwm from a pic to do this (so ghetto  ikea solution! don't do this for nice test equipment ). its great if you want a interview prop for some kind of bullshit cheap ass company only
(p.s. you wont want to work there for long)

if you want to be l33t figure out how to use 2 of them so you can send a negative and positive pulse, then find the difference in the measurement (offset error/noise elimination). like, use a ADC + multiplexer to measure the exact voltage of each lt3080, set pos/neg rails appropriately, then send pulses through your load using mercury wetted relays, then tune your sampling correctly.

if you build it well this will be a absolute beast

[Marco]

a microohm meter needs a beefy pulse.

You don't pulse, you throw an AC current at it and determine the voltage with a digital or analog lock-in amplifier.

[sarepairman2]

no, you use a trapezoidal bipolar pulse and carefully sample.

you will never get high digits with a lock in,

would you really trust a lock in amplifier for high resolution ohms? not to mention the reactance effects.

lock in is if there is no other way or if you wanna measure something in circuit (i.e. lock in + voltage drop to measure resistance of something without preturbing it)
you can make a somewhat useful power meter using a lock in system, but for a benchtop meter used to actually determine "standard grade" measurement on wires and such you want a trapezolidal  pulse.

[sarepairman2]

[Marco]

no, you use a trapezoidal bipolar pulse and carefully sample.

So a lock-in amplifier with blanking/dead-time and slew rate limiting for the excitation?

you will never get high digits with a lock in

He didn't say he wanted it, and he wouldn't get it without using a high INL ADC either (well, not with very meaningful results any way).

You have a rather limited definition of lock-in amplifier, a sine based digital lock-in amplifier is very standard and can get you to the INL limits of the ADC (and give you complex impedance for free). Using a subset of samples with a lock-in amplifier while it didn't cross my mind has crossed the minds of others, who didn't stop calling it a lock-in amplifier.

[sarepairman2]

you never see that used in commercial micro ohm meters. only LCR meters.

you also probobly dont want to use a super low current in a micro ohm meter due to contract resistance issues. search user "robrenz" posts for more details. hence the need for high current pulse.

again, lock in for resistance makes alot more sense for spying on things in circuit or maybe RTD's that are solder bonded etc..

if your going high current then the tempco is enough of a problem that it does not make sense to use a traditional lock in amplifier, why would you use a high current sin wave rather then just a trapezoidal pulse sampled in the middle away from any impedance mismatch?

[Kleinstein]

Modern Lock-In amplifiers can actually get a rather high resolution / number of digits. If they really want, they could give you 8 digits with not problems with quantization errors and even the INL can be surprisingly good, e.g. on par with the best DMMs in AC mode. This is however with a digital kind of Lock-in, not the classical old style. For just milli-lohms this is however overkill.
But digital Lock-in technique is an real option - though at lower cost level, down to the point of using µC the internal ADC.

The difference is not really big: its a low frequency generator and a good resolution reasonable speed ADC to sample the data. It's a question of how you look at the data and what waveform one use, whether you get ohms only or also LCR functionality with maybe limited range.

Anyway the LT3080 is relative accurate for an voltage regulator, but rather poor compared to a precision OP. Also the LT3080 is one polarity only So no real use for it, except in the power supply where is belongs.

[sarepairman2]

how do you know its poor? do you have a tempco stat on it? he is looking to set it digitally. he can trim with a dac

also CONTACT RESISTANCE PEOPLE

and he also wrote micro and not mili

[Kleinstein]

The LT3080 datasheet has a typical offset curve. This shows something like a 5µV/K tempco and offset limits of about +-5 V are given. So overall performance is about at the level of an LM358. One also has the power stage in the same case - so the chip might actually get hot.

So you can use the chip, but a simple transistor could do much of the same as well.

There are actually low noise, low drift amplifiers build with an LM358 - but only as one of many parts.
It's a little like asking on how to build a car using a nail.

[sarepairman2]

well as a one shot thing (like a uOhm meter) it should not really matter

otherwise you need like a 3 amp buffer and a precision op amp, which would be better if you can make the entire system work around 1 vref.

[sarepairman2]

i still think a circuit that just uses a single lt3080 would be fairly impressive and work well. that drift is not much and its all on one chip.. bullet proof

and if you do use PWM or DAC to power it, be sure to buffer the output with a voltage follower op-amp that has some muscle, if you need more current a MCU PWM pin or DAC can provide.