Constant current source using voltage regulator and an n-type MOSFET
MJHanagan
Posts: 189
I need to measure the resistance aging of several samples over a long period of time (ranging from a few hours to several months). I am currently using the Propeller to automate the thermal cycling of these samples and now I want to add the ability to automatically measure their resistance values periodically (ranging from once to several times each day).
I figure with a good steady 10 mA current source and a 16-bit differential input ADC I should be able to easily detect aging of at least 1% in the resistance. So now I am looking for a circuit that will produce a nice steady and repeatable 10 mA of current that I can send to each sample via some DPDT relays. I came across what looks to me to be a very simple two-device circuit to generate a 5 mA source. It uses a voltage regulator (e.g. MAX6126) and an n-type MOSFET (see Figure 2 on page 15). The figure shows an n-type bipolar transistor but the text suggests using a MOSFET for better accuracy (I assume this is because the base current is much less than a transistor??). This is where I am stuck I have no idea how to spec out which MOSFET would be suitable.
The DUTs (samples) will generally range from about 10 to 200 Ω. I would like the Vref to the ADC to be 4.096 V to make the math simple so the maximum voltage at ADC (+) needs to be less than 4.096 V. I assume as long as I keep Vin less than 4.096 V that should protect the ADC if any of the samples break (high resistance). Assuming the MOSFET has a negligible Vce(??) and Rset for 10 mA is about 200 Ω then the maximum resistance for the DUT would be 200 Ω. The most important aspect of the current source is repeatability and stability over time and temperature (<0.1% would be nice over weeks/months and as much as ±15°C). I will probably also put and ADC across Rest and to measure the current to correct for any drift and help maximize the accuracy and repeatability of the measurement.
Can any of you MOSFET savvy out there help with selecting a suitable device?
Circuit:
Maxim MAX6126 spec sheet: Maxim MAX6126 Voltage Reference.pdf
I figure with a good steady 10 mA current source and a 16-bit differential input ADC I should be able to easily detect aging of at least 1% in the resistance. So now I am looking for a circuit that will produce a nice steady and repeatable 10 mA of current that I can send to each sample via some DPDT relays. I came across what looks to me to be a very simple two-device circuit to generate a 5 mA source. It uses a voltage regulator (e.g. MAX6126) and an n-type MOSFET (see Figure 2 on page 15). The figure shows an n-type bipolar transistor but the text suggests using a MOSFET for better accuracy (I assume this is because the base current is much less than a transistor??). This is where I am stuck I have no idea how to spec out which MOSFET would be suitable.
The DUTs (samples) will generally range from about 10 to 200 Ω. I would like the Vref to the ADC to be 4.096 V to make the math simple so the maximum voltage at ADC (+) needs to be less than 4.096 V. I assume as long as I keep Vin less than 4.096 V that should protect the ADC if any of the samples break (high resistance). Assuming the MOSFET has a negligible Vce(??) and Rset for 10 mA is about 200 Ω then the maximum resistance for the DUT would be 200 Ω. The most important aspect of the current source is repeatability and stability over time and temperature (<0.1% would be nice over weeks/months and as much as ±15°C). I will probably also put and ADC across Rest and to measure the current to correct for any drift and help maximize the accuracy and repeatability of the measurement.
Can any of you MOSFET savvy out there help with selecting a suitable device?
Circuit:
Maxim MAX6126 spec sheet: Maxim MAX6126 Voltage Reference.pdf
954 x 487 - 7K
Comments
In your case the resistor value would be 120 ohms.
-Phil
A simple approach :
On clean DC source of 4V and a precision resistor Rref in series with the DUT.
Use one channel of the ADC to measure the voltage over Rref, divide by Rref and you hav the actual current value.
A precision Rref is about a few USD for a 0.1% and 10 USD for a 0.01%.
/Stefan
If this is an aging test, you need something better than a LM317, and a precision reference is the right part - get the highest spec you can afford.
Almost ANY Logic level mosfet will do, just choose a package with some thermal margin, and voltage margin
SOT223/DPAK for example ?
I see that MAX6126 even has a GNDS spin, with 300mV max compliance, so you could test relative to ground if the load values will drop < 300mV.
If you also measure the drop across the 250 ohm sense, then you will get a second opinion on actual current sense value.
** I forgot to check the GNDS current, check that is low and stable, if you want to try low side readings.
5 mA is not much. It might be possible to float the DUT in the feedback path of an op-amp. Like this:
The current is set by Vref and Rref, and the DUT Rx is the only unknown. Feed the output voltage into an ADC. If the ADC has a differential or psuedo-differential input, then one input can be Vref and the other input Vo, so the measurement is simply Vdiff = Vref * Rx/Rref. You can use one Vref for all the op-amps.
I'm with Fisherman on this one. Here is a circuit I use for RTDs.
The ADS1115 is a 16 bit converter with PGA. It converts the two differential channels to find the voltage across the DUT and across the reference precision resistor. The other resistors provide a current, but the absolute value does not matter. All the currents and reference voltages cancel out in the calculation. All that is left is the ratio of the unknown resistor to the reference resistor.
Do you know what the limitations are?
Maybe even be lower than Vsence? This way you could use a "Depletion Mode MOSFET".
Depending on the limitations it would be nice if the DUT was between the MOSFET source and the shunt.
The advantage would be the AtoD would only need to measure the voltage on the high side of the DUT and just subtract the 2.5V.
Gee Tracy you beat me to it.
Duane J
Thanks for your inputs!