EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: Jorge on April 21, 2016, 09:29:29 am

Title: Current Sensing
Post by: Jorge on April 21, 2016, 09:29:29 am
Hi guys, I'm new in the forum but I need some help of you.
I built a current mesuring pcb, but It's not very accurate. I've put a INA217 mesuring the drop voltage of a 1 ohm resistor with 1% precision, and the output of the amplifier is conected to my osciloscope, but the problem starts when I try mesure uA, It's starts to get to unpresise with an error of about 40%. Any ideas of what's causing this?
I mesured the voltage on the resistor with a multimeter and its acording to the Ohms Law, but on the output is not, Sometimes the voltage on the input changes just a liltle bit, but my output stays the same...
Title: Re: Current Sensing
Post by: Buriedcode on April 21, 2016, 10:27:52 am
Do we get a schematic of your setup?  Also, what power supply are you using?  The input voltage range is given on page 6 of the datasheet:

http://www.ti.com/lit/ds/symlink/ina217.pdf (http://www.ti.com/lit/ds/symlink/ina217.pdf)

Quote
INPUT VOLTAGE RANGE
VIN+ – VIN– = 0V (V+) – 4 (V+) – 3 V
Common-Mode Voltage Range
VIN+ – VIN– = 0V (V–) + 4 (V–) + 3 V

The datasheet also mentions that very low impedance sources can cause it to oscillate and recommends input filters on page 12.
Title: Re: Current Sensing
Post by: kripton2035 on April 21, 2016, 10:29:52 am
an oscilloscope is not the best way to measure voltages : a multimeter will be far more precise for that.
Title: Re: Current Sensing
Post by: dannyf on April 21, 2016, 11:03:33 am
Quote
with an error of about 40%.

It depends on the design of your circuit and what you are comparing the measurement to.
Title: Re: Current Sensing
Post by: Jorge on April 21, 2016, 11:09:12 am
I'm using a CMR0515S3C 15V dc-dc converter, and it's powered by my AimTT power supply.
I need to measure the current consumption of a small display that peaks at about 8mA, and has a rms consuption of a couple uA, so I have to measure current on a very precise way, and when I use a multimeter I cant se those peak currents.
You can see a version of the circuit, were R1 is 10ohm (G=1001), R2 is 1ohm.
Title: Re: Current Sensing
Post by: dannyf on April 21, 2016, 11:22:30 am
Quote
You can see a version of the circuit

Compare that with the datasheet and it is very easy to see what went wrong.
Title: Re: Current Sensing
Post by: Jorge on April 21, 2016, 12:00:04 pm
I tried adding ceramic caps acording to the datasheet, but the noise stays the same, so does someone have any ideas?

I'm sorry but I'm still on HighSchool, I dont have the knowledge or the experience of an engeneer, so Im still stuck with this dilemma  |O
Title: Re: Current Sensing
Post by: bitslice on April 21, 2016, 12:44:08 pm
Compare that with the datasheet and it is very easy to see what went wrong.

I'm sorry but I'm still on HighSchool, I dont have the knowledge or the experience of an engeneer, so Im still stuck with this dilemma  |O

Dannyf, on a scale of 1 to 10, imagine how much the OP would like to punch you in the face right now...

148 people looked at this thread, they didn't know either  :--
Title: Re: Current Sensing
Post by: digsys on April 21, 2016, 01:09:44 pm
You have to be careful with the INAxxx series current amps. What worries me about this one is that there is NO % linearity (or gain linearity) vs Vin charts !!
Not even a white paper etc. I made a bad mistake assuming all the INAxxx's were accurate over the entire Vin range, and designed in INA198s, made several designs.
Then I found out the hard way, that they had terrible accuracy (in some cases not even registering for the first 0-2to5%). Digging up white papers confirmed it.
I switched to the INA286, and can report excellent linearity / accuracy right down to zero. Without those particular charts, I'd be VERY sus.
Title: Re: Current Sensing
Post by: Jorge on April 21, 2016, 01:56:29 pm
This are some measurements that I've done, and you can see the max gain that I can atchive is about 900 and should be at 1001, when I droped the gain I got better results but in the lower currents (<200uA) I still had a HUGE error on the gain (up to 40%).
Title: Re: Current Sensing
Post by: Jorge on April 21, 2016, 01:59:22 pm
You have to be careful with the INAxxx series current amps. What worries me about this one is that there is NO % linearity (or gain linearity) vs Vin charts !!
Not even a white paper etc. I made a bad mistake assuming all the INAxxx's were accurate over the entire Vin range, and designed in INA198s, made several designs.
Then I found out the hard way, that they had terrible accuracy (in some cases not even registering for the first 0-2to5%). Digging up white papers confirmed it.
I switched to the INA286, and can report excellent linearity / accuracy right down to zero. Without those particular charts, I'd be VERY sus.

You were able to measure uV with that amplifier?
Title: Re: Current Sensing
Post by: digsys on April 21, 2016, 02:41:44 pm
Quote from: Jorge
  You were able to measure uV with that amplifier? 
In my case, I was working with an ADC Ref of 2.4V, on 12 bits = 500uV / bit. Using a x50 INAxxx > gives approx 10uV
I can't verify 1-10uV easily, I'd need to fire up the development / calibration jig, but it was a HECK of a lot different to the INA198 !
What I am alluding to is - IF there isn't a TRUSTED linearity chart for that part #, I'd be VERY wary of it ! There are many instances of parts with dodgy or missing charts.
Title: Re: Current Sensing
Post by: Skimask on April 21, 2016, 02:56:25 pm

I'm sorry but I'm still on HighSchool, I dont have the knowledge or the experience of an engeneer, so Im still stuck with this dilemma  |O

Dannyf, on a scale of 1 to 10, imagine how much the OP would like to punch you in the face right now...

148 people looked at this thread, they didn't know either  :--
Doesn't matter if that scale runs up to 4,692...  Facts are facts.  The O/P is trying to half ass it and expect it to work correctly.
WTF...
Title: Re: Current Sensing
Post by: suicidaleggroll on April 21, 2016, 03:12:49 pm
OP - I think what they're trying to say is you need to look at the common mode input voltage range for that part.  It runs from (V-)+3 to (V+)-3.  This means when using it with the nominal +/- 15v rails it's designed for, you have a common mode input voltage range of +/- 12v typical, +/- 11v minimum.  You're using the device with V- = Gnd, and trying to use a common mode voltage very near Gnd.  That's way outside of the valid range for that part.  Assuming you're using a Vcc of at least 9v as the spec sheet calls for, that would give you a valid common mode input range of only +3 to +6v typical (+4 to +5v min).  You can't use that part as a low side current sense amplifier, not without giving it a negative rail at least 3-4v below gnd to work off of.
Title: Re: Current Sensing
Post by: MatthewEveritt on April 21, 2016, 03:25:31 pm
It's not entirely clear, without toggling the part number,  but the DC/DC converter the OP is using is duel rail: Vdd isn't gnd, it's - 15v.
Title: Re: Current Sensing
Post by: FlyingHacker on April 21, 2016, 03:35:17 pm
The OP said he is using a CMR0515S3C, which will do a +15V and -15V output.

You have the -15V as VDD and the +15V as VCC? You measured these for voltage, and ripple as well?
Title: Re: Current Sensing
Post by: suicidaleggroll on April 21, 2016, 04:28:10 pm
It's not entirely clear, without toggling the part number,  but the DC/DC converter the OP is using is duel rail: Vdd isn't gnd, it's - 15v.

Didn't catch that from the post

OP - could you provide an actual schematic with all of your decoupling caps, labeled rail voltages, labeled resistances, etc., as well as an image of your board layout?
Title: Re: Current Sensing
Post by: Jorge on April 26, 2016, 09:34:41 am
I've already abandoned the INA217 but now I need a replacement, I was thinking on the SSM2019, what do you think?
Title: Re: Current Sensing
Post by: PeterFW on April 30, 2016, 05:16:20 am
I've already abandoned the INA217 but now I need a replacement, I was thinking on the SSM2019, what do you think?

I can tell you that i have measured reasonably well at 100µA resolution with a INA226 with a 1Ohm shunt.
Do not know if that helps...