Looking for a bilateral voltage/current sensor

[leonerd]

I'm considering making myself some test equipment for measuring/classifying LEDs, transistors, other such small devices. Something like a curve tracer or SMU. Trying to stick within quite a small footprint and budget though, so I'm probably limiting it to a few hundred mA and maybe 12V at most.

I'm slightly stuck for what to use for a decent voltage/current sensor. The rather excellent INA219 serves me just fine if I want a single-channel device; one that could source a (programmable) up to 12V and monitor the actual voltage and current. This would do nicely for LEDs. The key points of this sensor are that it has an onboard 12bit ADC which gives me resolution down to 4mV / 100µA (presuming 0.1R shunt).

However, I'd quite like a dual channel device, one that's capable of classifying transistors as well. This would have two source-measure channels sharing a common "ground"; so to be able to handle PNPs/P-channel FETs both channels will need to source and measure a negative voltage too. The INA219 is no good at this - while it can cope with differential input in either direction, the common-mode must not be negative with respect to its own ground.

So I'm looking for chip ideas. Ideally something similar to the INA219 (I²C accessed with onboard ADC) would be perfect, though at a push I might consider a separate dedicated ADC chip alongside some more traditional op-amp-like solution for current sensing.

[mazurov]

INA226 will measure current in both directions, for voltage you can build absolute value circuit, something like this -> http://www.linear.com/solutions/1567.

[leonerd]

That doesn't look useful though. Whlie it says it can do current bidirectionally, it still can't handle a negative common-mode voltage. That's the main feature here that seems to be lacking - an ability for the bus voltage to be negative, below GND.

I suppose at a push I could try shifting the INA sensor so its ground was in fact at negative supply, but then the quiescent "zero" output is then at half of rail-to-rail supply, so zero drift errors become more of an issue. I think I'd prefer a dual-rail solution to this to get full four-quadrant ability.

[danadak]

If you are using a UP, interestingly the Cypress PSOC UP, its A/D will common mode
outside its supply rails by 100 mV. That coupled with max resolution of 20 bits yields
a pretty good low or high side current measurement. And you can mux it to do both.
Note its A/D is diff in so it will reject CM due to layout if utilized properly.


Regards, Dana.

[leonerd]

Again I don't see how that helps. I have both +12V and -12V supplies available, I will have a programmable output voltage that might be anywhere between these limits, and I want to measure the voltage and current flow anywhere within this envelope. -100mV isn't enough; it needs -12V

[danadak]

The +/- 100 mV is to handle a current shunt to measure current and do it bidirectionally.

But the A/D, which measures V in either I or V measure cases, has a basic
range of Vss to Vdd, Vdd being 5 or 3.3. If you want higher V then an R divider
would be used to scale the V to the A/D's CM range. Point being you could measure
KV or MV if you wanted with it and appropriate divider.

Regards, Dana.

[leonerd]

On further thought, I wonder if I could do this a bit differently.

My original plan was to have two independent four-quadrant channels available that can both source/sink current at either positive or negative voltage, with respect to a shared common. That would then allow either NPNs/N-channel FETs by using positive voltages on both channels, or PNPs/P-channel FETs by using negative on both. Two-terminal devices (primarily in my case LEDs) are easily done on just one channel.

But if all I need to do is transistors, a simpler arrangement might be had. Instead of two independent channels, this could be arranged as a single-channel device with a voltage sensor across it and current sensor at both ends, with an additional connection that can source/sink at some voltage inbetween the two. This additional connection can then supply base current or gate voltage.

It now simplifies my problem somewhat, as now everything will always be positive with respect to the "lower" of the main channel terminals. The main channel is a double-ended current sensor with a shunt resistor at both ends (for which sensor chips like the INA226 will do well), having a differential voltage meter across it. In addition, the third terminal needs bidirectional high-side current monitoring, and needs its voltage measuring. Ideally I'd still like to be able to do this one differentially, by being able to either measure voltage above the lower terminal (for N-type devices), or voltage below the upper one (for P-type).

This feels like a more workable setup that reduces requirements on trying to find dual-rail meter chips.