Favorite current sense ICs

Looking at testing a complex chip with 11 different supplies,
for latchup. I want to monitor them all. I have found some
chips that have built in limit comparators (which would be
handy for simplicity of monitoring and quenching the latchup
events), a wide range of bandwidths and gains, likewise the
common mode range (some nice ones, specwise, have an
unfortunate requirement that one end of the sense resistor
must be the same supply as runs the chip). I have a couple
of low voltage negative supplies to deal with, 9 low voltage
positive supplies (1.8 to 3.3V, which takes the options -way-
down in this niche of high voltage focused power supply
parts).

I've got my eye on INA302, LMP8645, LT1999, AD8216,
MAX9643 now (having looked at a dozen others) but none
is entirely satisfactory.

If you were going to try and monitor supplies from -1V to
+4V common mode, wanted a gain of 10 or better (V/V)
(though I am liking a current mode output with gain, and
a select gain resistor off chip too), and >500kHz BW, what
other would you suggest? I expect to have +12V, +5V
available but see no need for higher VCM and Vsupply,
which is what a lot of the parts aim to support.

Hi,

I've got my eye on INA302, LMP8645, LT1999, AD8216,
MAX9643 now (having looked at a dozen others) but none
is entirely satisfactory.

Click to expand...

--> use the interactive selection guides at the manufacturer's internet sites.

wanted a gain of 10 or better (V/V)

Click to expand...

You say "current" measurement...therefore a max voltage drop, max shunt resistance, desired output voltage, measurement current range... are more usefull values.
****

I wonder why you need BW > 500kHz..since you talk about power supplies with installed capacitors.

You say the measured power supply voltages are -1...+4V ... therefore many 5V supplied RR Opamps will do (some need extra resistors to lift the negative voltages to the desired common mode range).

Are you interested in measuring negative currents, too?
Do you need isolation?

What is your (voltage) measurement device? Seems like an ADC...

****
Some calculations:
Let's say the power supply voltage is 1.5V....and you want to install a shunt...
* then the voltage drop should be low not to introduce voltage problems
* the impedance needs to be low, not to introduce ringing or other stability problems
If the desired voltage drop across the shunt should be less than 1% of the voltage, this means 1.5V x 1% = 15mV.
Now you taked about a gain of 10V/V.... thus the output voltage is just 150mV...is this a useful value?

You say BW > 500kHz....
If you use an ADC ... you need a sampling rate of more than 1MHz for each channel to comply with nyquist...
Now you have 11 channels ..... are they muxed? Are they sampled synchronously?

Did you consider to use current transducers like LEM..
Or hall sensor equipped ICs?

Klaus

I am looking to be able to observe the evolution of
latchup currents, which really happens in sub-uS
timescales (but will be smeared by the supply filters
to some extent). I do not need isolation. Sense
resistors will be picked to impose a minimal drop
under normal conditions (a few % of full scale,
ADC-input-range-referred, just so we can see
that the value is sane, not noise).

I have 2 CT-1s in a prototype jig but this has
some drawbacks such as requiring the 'scope to
be in the loop, no easy logical combining of
detects for counting or quenching, and I do
need 11 supplies monitored meaning another
9 of these expensive little suckers.

Have used the LEM sensors for DC measurement
but their active amplifier ones are too slow and
the passive ones, too high-current. I'll look at the
Hall angle.

We need to record supply current per frame via
the 1MSPS ADC but also be able to look at the
waveforms anecdotally on the 'scope with some
time domain fidelity. Not concerned with Nyquist,
this is not frequency domain stuff, just strip chart
time domain looking for a supply current step that
sustains, increment count, quench, and look for
more. Bandwidth here is a proxy for step response
fidelity / lag. I've seen one of the parts I like, showing
pretty significant peaking (their way to get BW) while
others are better damped.

150mV is 1/6 full scale on the ADC input. So, this
is probably more than I'd pick offhand (more like
50mV baseline to leave room for the big end, on
the chance that there are "micro-latch" as well
as chip-scale latchup events.

It's been my experience that CTs show fast current
slew even into filtered supply pins - provided that
the supply source is "stiff". This is what we want -
but less bulky, costly and needy (like, I do not have
and will not afford a 12-input 'scope with math /
logic triggering and AUX output, if even such a
beast existed).