In my "current sensing" stable I have about 7 of the INAxxx chips , but the INA170 slipped between the cracks, it has the big advantage over the ZXCT10xx in that you can <accurately> add an offset current , very helpful in single supply measurement, where things get a bit muddy near zero current.
Good for a recent design I have with a battery operated motor drive, the currents go from several amps while running , down to milliamps during charging and standby operation.
As it uses PV charging it's important to wring out every last mA during "shade" operation.
PS thanks for all the links in the blog , I've downloaded several datasheets as a result.
PPS I've read Part 2 of the blog just now, good work!
There is a pertinent article by Mark Peffley on the acuracy of the resistor shunt in current measurement in Design News "Obtain Accurate Current Measurement". Although Mr Peffley is talking about shunts used with DMMs, his comments apply equally well to any shunt discussed in my blog.
Nice summary of the technolgy out there, looking forward to the next post. I've used the ZXCT10xx before and they are pretty good, and they work well using the stray resistance of wiring or PCB traces.
For low budget motor overload detection, I've also used BCV61 and BCV62 (current mirrors at ~ 20c each) in conjunction with copper trace sensing , not terribly accurate , but the positive TC of the copper trace produces the helpful effect of reducing the trip current as the temperature increases.
You can also use the BCV61 as a current amplifier , when using say a UC3845 or similar current mode PWM, and utilise tens of mV drop on a PCB trace , rather than a hot expensive current sense resistor.
I've become quite fond of the hall effect sensors recently, for example the ACS712 style in SO8 package fits nicely on a PCB, gives bidirectional DC and AC to 10's of kHz, and is galvanic isolated, accuracy is a bit ordinary though. Once you get past 10A, current shunts get large, and "ground" voltage varies significantly around the PCB or system, and noise is typically higher. I've found some good closed loop hall sensors at 0.5% accuracy for ~ $30, they are bulky ,but in reality, smaller than an external shunt, and comparative pricing. Calibrating anything above 10A is challenging, but a 10A closed loop hall effect can use 10 turns of wire at 1A as a much easier calibration method.
Yes it is easy enough to do a high side current measurement. I intended this blog to cover that. Obviously I didn't do a good enough job. You will need to insert a resistor in the high side current flow and then measure the voltage difference across it using a difference amplifier (or even op-amps configured as a difference amplifier) or a device specifically designed for this ("current monitor"). There is often a problem when the supply voltage of the difference amplifier is less than the value of the voltage on the higher side of the shunt. If that is the case the configuration you choose it must be able to tolerate this. There are many low cost dedicated devices that will do this. The TI INA170 is a recent one that have designed with, but there must be dozens more. If you are working with voltages less than 20V the Diodes Inc/Zetex ZXCT1008/9 looks like an interesting device.
Check out the Microchip and Diodes Inc reference in the blog. They actually will take you to app notes on the subject.
If you supply is going all the way down to 0 and it is powering to sensor circuit, then I am not sure if I know any technique other than some kind of external supply for the sensor circuit.
Are there any technicques that can be used in such a situation and will you be covering them?
Part 2 will cover other techniques like CTs and Hall effect devices.
Hi Aubrey. Good stuff as usual and I look forward to part 2.
I have always wondered if it is possible to measure the high-side current of a variable power supply (say a lab PSU putting out 0-30 volts). I have seen many designs that put a resistor in the low (ie ground) leg and measure the current on that, but I reckon ground ought to be ground and anyway this can give problems with voltage sensing. But if your supply is putting out close to 0V you get other problems.... Are there any technicques that can be used in such a situation and will you be covering them?