Help:Digitally controlled current source

[Roff]

Busy today. I might be able to look at it tonight. Post a schematic. See for instructions.

 

[abicash]

Hello

Whenever i try to attach an image,the small pop-up window starts uploading the image and goes blank after uploading it ,with no way to close the window.The attachment too is not visible in the message.
I don't understand whether its a bug with my browsers (firefox and opera in ubuntu) or the site itself

 

[abicash]

I can use a single ended supply for ICL7650S, right?
(V+=+8v , V- =0v or GND)

With the low i/p impedance circuit i don't get any voltage at pin 6 (o/p) .

(the same problem with file upload)


**broken link removed**

 

[abicash]

Ok now i changed the resistors on the ICL7650 to 4K7.
With these values i do get a linear differential voltage across the load resistor..but i have a huge offset of around 60mV at pin6 of the opamp

The resistors are all 1% MFR .I also tried varying the gain on the inverting terminal by replacing R3 by a multiturn preset.But that doesnt help.
As i see ,ICL7650 quotes almost zero offset voltage.
Is something wrong????

Please advise

Thanks and regards

 

[Roff]

I'm on a road trip. I drove 670 miles yesterday, and will do the same today, so I don't have much time to spend on this right now.
Why did you choose 8V for the second power supply? this won't work. You need to use the 12V supply here also. The input common mode range of the ICL7650 will not support the use of the 8V supply.
You need to swap the input connections for R1 and R2. As drawn, the output will have to swing negative, and you don't have a negative supply.
Also, keep in mind that the output can never get all the way to zero without a negative supply. I would probably create a low negative supply for the output amp. An LM337L running off the -12V supply could create -1.2V, which would be sufficient. Mind the minimum load current requirement of the LM337L.

 

[Roff]

Why aren't you using the ICL7650 for the current source?

 

[Roff]

Here are the changes I recommended.

 

[abicash]

Hi Ron i hope your long road trip ended well.Or you still on it?
Well i used OP07 for the current sink since the resolution for current is 250uA which is within limits of OP07 and its cheap.
the 7650,i read, has max supply of 8v.And you say that 12v is the correct voltage there.I dont understand!I have tried swapping inputs and i can get the change.I will try some more tomorrow.

 

[abicash]

Hi Ron

I did some changes.
I added a -1v supply on the pin 4 of 7650.And 12v as recommended on pin 7.
With this i got negative saturation at the o/p pin 6.So i changed the gain of the inverting terminal as in figure.With this i can adjust the o/p voltage to zero at no i/p differential.
But now i can get wonderful voltage variation with the current flow.
I hope what i did is correct

**broken link removed**

 

[Roff]

Resistor tolerance causes a differential gain error, and an offset error due to the commom mode input voltage. The offset with 1% resistors and your 12V common mode input can be as much as ≈ ±240mV. It should not saturate your op amp.
You can correct for the common mode offset with the "preset" pot you added, but it will not correct for gain error.
If you want to learn something from this, I would encourage you to calculate the common mode and gain errors. If you understand op amps, you can do this yourself.

 

[abicash]

Ron please tell me what needs to be done.I will learn better this way..This is just a small part of the project and i never thought it would be so tough.
I hope you wont mind
Thanks and regards

 

[Roff]

Ron please tell me what needs to be done.I will learn better this way.

Yeah, right.
I don't know what needs to be done. I don't have the project requirements, I don't know what your budget is. I am giving you suggestions. You have to do some work to figure out the details.

This is just a small part of the project and i never thought it would be so tough.
I hope you wont mind
Thanks and regards

Below is a block diagram of the idea I mentioned earlier. If Rload is low (you said <5Ω, IIRC), you shouldn't need a buffer amplifier, and you definitely won't need an output level shifter.
Another idea would be to use the current sink you have now, and use a current source as a level shifter, as in the attachment.

 

[abicash]

Hi Ron thanks for getting back..Well..That was in frenzy that i begged for the irrational help.This was since i can find a drift in the o/p of the diff.amp over time and i really dont want YOU to do my project.I could not figure out why that is.Maybe i will struggle today
i am basically a digital person.More into microcontrollers.

 

[abicash]

Hi

I will brief my project here.

I have to pass a current value initially for a very small time (10ms) so that there are no temperature dependent coeff. changes in the resistor load.This drop across the resistor is saved as ref.cal. value.After this,the same current ,is passed for a longer duration.The resistance change and subsequently the voltage change is noted (above the ref.cal. value).This voltage change will be in micro volts.This has to be amplified and read in a 12 bit ADC.

With my existing circuit i am proposing some more amplifier stages (my ADC is 3.3v max and the max drop across the R will be 5v) to make an auto-ranging ckt.

I am finding drift in the o/p of my 7650 over time.Maybe this is due to Differential gain error.I will look into it.

Thanks for the support since i have been demanding at times **broken link removed**

 

[abicash]

Also one more thing...

Since i have to measure a voltage which is 'above' few tens or hundreds of millivolts , does it really matter if i don't use a precision low offset op-amp?
Since even though the actual voltage is in micro volts ,its above the i/p offset voltage range of most op-amps...

 

[Roff]

I suspect that the component most susceptible to drift is the 1 ohm current sense resistor. What type of resistor are you using there?
Pots are also suspect in my mind, especially if used as a rheostat, as in your differential amplifier.
If this were my project, I would probably use the current source version that I posted. I would avoid the differential amplifier, unless it were monolithic, like AD8276.
I would also make the actual measurement as soon as possible after the calibration procedure, which should minimize drift problems.
I think you are correct about offset in the output amplifier (which I would avoid). It should not be a problem if you measure immediately after calibration.
If you are trained as an engineer, and understand op amps, you should be able to make error budget calculations, such as where your gain and offset errors occur, how they are affected by resistor tolerances and amplifier offsets, etc.

 

[abicash]

Reading your post
..just checking what you meant by i being correct about offset in output amplifier..Does it mean that the actual measured voltage in microvolts(above some millivolts) is above the offset voltage so any opamp would do?(although i will use a precise opamp)..Is that correct?
Meanwhile will check the link u posted
thanks for your time

 

[Roff]

Reading your post
..just checking what you meant by i being correct about offset in output amplifier..Does it mean that the actual measured voltage in microvolts(above some millivolts) is above the offset voltage so any opamp would do?(although i will use a precise opamp)..Is that correct?
Meanwhile will check the link u posted
thanks for your time

If I understand your calibration procedure, when you subtract the cal value, that will (I think) also subtract all offset (but not gain) errors in the measurement chain. Offsets will still show up in the initial current magnitude measurement, so if that's important, you will need to use low-offset amplifiers. For example, 50mA will give you 50mV across the 1Ω sense resistor. 1mV of offset in the input amplifier will cause 1mA of error. You can't compensate for that unless you are using a high-precision, low drift sense resistor.

 

[abicash]

Hello Ron
i have said earlier that for the current magnitude measurement i am using OP07 since the accuracy required is below the offset stated by OP07.I will dig some p-channel mosfets and test the source version today.
The sense resistor i am using is not precise for now.But WILL sure use a precise one actually.I am getting a board painted for this today since on a general purpose board i have lots of wires and links which must be adding to noise and offsets.
You have been very helpful..

 

[Roff]

Hello Ron
i have said earlier that for the current magnitude measurement i am using OP07 since the accuracy required is below the offset stated by OP07.I will dig some p-channel mosfets and test the source version today.
The sense resistor i am using is not precise for now.But WILL sure use a precise one actually.I am getting a board painted for this today since on a general purpose board i have lots of wires and links which must be adding to noise and offsets.
You have been very helpful..

In my current sink schematic, I added a frequency compensation network consisting of resistors and capacitors. I determined their values empirically, by simulations. You may need similar compensation in the current source circuits to prevent ringing or oscillation.