Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

High side current sense Amplifer

Status
Not open for further replies.

berger.h

Member level 5
Joined
Sep 30, 2016
Messages
93
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
2,355
I need measure current on high side,
max voltage 50V
max current 30A
shunt 3mOhm
Gain min 20 max 50
Bandwith about 100kHz

I tried ina168 with Gain 10 (25k resistor), but this amp. have input offset +-2mV and in real was PROBLEM on low current.
For current under 500mA was output unstable for example for 400mA current, 1,5mV on shunt was on output senseless 600mV, for 600mA 68mV, for 700mA 10,6mV. 600mV for low current .
The problem is high voltage (600mV) for low current (under 400mA)

I thing I need amplifier with lower input offset.
 

Have you looked at the purpose-made current sense
chips from Maxim et al?

If you have architectural freedom you might come out
ahead by doind a high side current -compare- rather
than measuring; autozero comparators abound, super-
low-offset instrumentation amps are fewer and, maybe
critically, slower - no chance of following a single
power pulse with fidelity, at 100kHz chop or above.
 

Sensing small voltage 'at a distance' (many volts of common mode) is fundamentally challenging.

First, are you sure you laid out the design well and followed all the notes in the datasheet?

Second I see TI recommending INA282 as a higher performance alternative (it has a built-in voltage amplifier so it does more of the work for you).

I also have AD8210 in my notes as a relatively high performance high side shunt monitor similar to the INA282.


Finally there are a couple other alternatives. Hall effect sensors are a good one though the cost to performance ratio is probably higher but they're easier to use because they're isolated by nature. See Allegro's IC based hall effect sensors, or this style of larger and more expensive (but more accurate) closed loop style:
https://www.digikey.com/product-detail/en/lem-usa-inc/LESR-50-NP/398-1200-ND/9488904


Finally a solution that's guaranteed to work is to create an isolated circuit and reference it to your high side shunt. There are now more and more single chip solutions that provide signal isolation and a small amount of power such as the below. This would let you power an amplifier and ADC for example:
https://www.analog.com/en/products/adum5411.html
 

Hi,

I have the INA286 on two homemade ammeters fpr power supplies and while this device is easy to "plug in" and works correctly down to 0mA, I believe due to datasheet stated offset (mainly, as well as my layout a little presumably), both ammeters read 0.004A at 0mA input (the ammeters are on different, not 100% identical, boards so I think both having that ugly 4mA is "suspicious") - my only regret with the circuit was not knowing more at the time and adding some subcircuit to subtract that value by offsetting the output somehow to read 0mA at 0mA.
 

INA28x have bandwith only 10kHz or low ,
AD841x for example AD8417 with G=60 is good choice but if is powered GND +5V is mimimum output voltage about 55mV , i,e. for 3mOhm shunt this solution not measure curren under 350mA.
Solution is any negative voltage on GND pin 2 , for example VSS +4V, Pin2 -500mV, Ref1 and Ref2 on GND, and output will be works from zero voltage/current of course with full input offset error projected to output.
 

From what I've seen of power supplies it's rare to have
more than a 100:1 current range, and you should be sure
you are not "challenging yourself for no benefit". Current
mode control works within a servo loop (voltage) and
current reading errors can be taken out to some extent
by outer feedback. As long as transimpedance (net) has
some monotonicity and doesn't lose its gain altogether,
or get buried in offset (losing gain on the other side of
zero).

An approach could be to inject an offset current from
the low side to ensure that power-leg I=0 results in a
sensed value that's "right side up" (never mind that it
carries a fixed error) across the whole Vio distribution.
If the application really is a CMC power supply the added
"error term" can be taken out by the outer loop no prob.
Although you might check PSRR to ensure that the adder
current sink is not a HF "injector"....

Easier by far than procuring an amplifier with
Vio=0.0000 +/- 0.0000 and (less obviously, but critical)
this holds across the common mode input range, to
0.0000 (referred to the hgih side rail). Many rail-rail op
amps see gain loss and/or Vio increase as you approach
either rail. You might look at "beyond rail" amplifier
offerings, although many of these are super low power
and turtle-on-'ludes slow.
 

Hi,

INA168 has specified typical offset voltage of +/- 0.2mV, even if +/-2mV is specifed for extreme temperatures I expect the real values close to the typical values. I rather expect a design error, either schematic or PCB layout.

How did you measure the 2mV offset?

Especially your "low current" problem also seems to be a design problem.

--> show us your complete measurement circuit with all the voltages and part values, your PCB layout and your test conditions.

Also, please mind that "bandwidth" relates to sine waveform. For any other waveform you need (much) higher bandwidth.

Every current sense amplifier manufacturer has selection tools to choose the best device for your application. Did you use them?

Your description gives no specifications about accuracy and precision.
Most designs without negative supply voltage suffer from a problem outputting voltage close to zero. Thus one may add current on the input side and subtract it on the output side or subtract it on the digital side when an ADC is involved.
Stable offset usually is not that big problem, "drift" is more complicated, because it is not that predictable.

Klaus
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top