high voltage opamp?

[sgt_bear]

Hello

I have to measure a current in a high voltage system (150-250V DC). The measurement has to be made on the high side via a shut.  The shunt can be relatively large, anything <1kOhm will do.

Usually at low voltages, a differntial opamp would do.  All approved solutions are found are for high currents in high voltage, but i need very low current measurements. The possible currents are 0-20mA and i would like to have a 100uA accuracy.

What would be the best way to solve this problem?

[frozenfrogz]

Maybe this article is for you: http://faculty.virginia.edu/phys-electronics/doku.php?id=tips_info:low_i_high_v_meter
Also: http://www.ti.com/lit/an/sboa165a/sboa165a.pdf

[jeroen79]

Battery powered ADC on the high side that transmits the measurement optically to the rest of your equipment on the low side.

[f5r5e5d]

https://docs.broadcom.com/docs/AV02-1333EN

[ebclr]

http://www.ti.com/lit/ds/symlink/amc1306m25.pdf

[sgt_bear]

Yes, but HOW does the ADC can measure the current safely

Basically all i need is  the same function which a multimeter has, you just place the multimeter as shunt, and the current measurement will work. How do multimeters measure the current

[Paul Moir]

They float, which is to say they have no earth ground reference.  That's why they're battery powered and if they do have communications it's usually done optically. 
If you need isolation from this `150-250V, it's going to boil down to a "DC current clamp in a chip" like solution (eg LEM current sensors) or "Floating multimeter with an isolation barrier" like solutions (jereon79, frozenfrogz, etc).

[daqq]

Basically all i need is  the same function which a multimeter has, you just place the multimeter as shunt, and the current measurement will work. How do multimeters measure the current

They measure voltage on a resistor through which current is passing. The simplest solution these days would probably be to make a floating ADC with some support circuitry. You isolate the digital and power interface from the grounded side via isolators.

Some inspiration: http://www.linear.com/product/LTM2893

Mind you, this can be done with far less exotic parts

[T3sl4co1l]

Any of those current output type current sense amps can be used, given that you can spare the quiescent current, and that you cascode the output with a high voltage transistor to shift the signal down to ground level.

Tim

[David Hess]

If isolation is not required, then I would use the configuration shown in figure 1 here:

http://www.analog.com/en/analog-dialogue/articles/high-side-current-sensing-wide-dynamic-range.html

The only high voltage parts are the Rbias resistor and the output transistor.  Use a low power operational amplifier to minimize the current through Rbias.  For a wide voltage range, replace Rbias with a constant current sink.  The amplifier requires an input common mode range which extends to its positive supply.  In the past, an LM301A might have been used but a TL031 has a lower supply current and would be my first choice now.

[Zero999]

If isolation is not required, then I would use the configuration shown in figure 1 here:

http://www.analog.com/en/analog-dialogue/articles/high-side-current-sensing-wide-dynamic-range.html

The only high voltage parts are the Rbias resistor and the output transistor.  Use a low power operational amplifier to minimize the current through Rbias.  For a wide voltage range, replace Rbias with a constant current sink.  The amplifier requires an input common mode range which extends to its positive supply.  In the past, an LM301A might have been used but a TL031 has a lower supply current and would be my first choice now.

You mean this circuit? Yes, I agree.


There seems to be a reasonably wide selection of suitable MOSFETs available. I've had a quick flick through Mouser and the ZVP0545A and ZVP4525G seem to be the most cost effective.

http://www.mouser.com/ds/2/115/ZVP4525G-94131.pdf
http://www.mouser.com/ds/2/115/ZVP0545A-92870.pdf

[sgt_bear]

The AD8628 is 5V supply?

I guess if i connect the AD8628 as shown in the schematic to BUS VOLTAGE (which is 250V DC) in my case, i'm sure it will DIE very fast!

[sgt_bear]

I was planning to use the

LT6375 +-270V Common Mode Amplifier, but sadly the part is designed for higher currents. if used with low currents the precision gets wasted because it only has very low REF Resistors which consume current themself...


http://cds.linear.com/docs/en/datasheet/6375fa.pdf

Basically i'm looking for the same solution as shown on page 1, but with a lower current leak through the amplifier...

[T3sl4co1l]

Supply is zener diode regulated.

Tim

[Kleinstein]

The OP for the current sense part has a zener to provide a regulated Supply of about 5 V. So the OP will be OK.
The only part that needs to be high voltage are R_Bias and the P-channel MOSFET. If the requirements are not high a PNP transistor would work too and might have an advantage if the drop at the shunt is larger than about 2 V.

It the bus voltage changes a lot, one might want to change R_bias to a current source (e.g. a depletion mode MOSFET + resistor).
There are lower power OPs than the AD8628. One might not not need an AZ OP - it could as cheap as an MCP6001.

[technix]

You can try making a small battery operated current sense module and let it pump data down an optical or RF link. You can run RS232 over TOSLINK POF or run BASE-X Ethernet with a SFP module if you want to go optical. If you want to go RF just get ESP8266 and drop it into your local Wi-Fi network.

EDIT: Speaking of POF, you cal also use analog POF with an analog battery-operated circuit on the high side converting the current into light intensity in the POF, and pick it up using a photodiode on the low side.

[sgt_bear]

Hm... am i not understanding opams here right? If i put the in - and + to high voltage, will the opamp survive ? On the shown circuit my 250V would directly go into IN -  ?

[David Hess]

Hm... am i not understanding opams here right? If i put the in - and + to high voltage, will the opamp survive ? On the shown circuit my 250V would directly go into IN -  ?

The operational amplifier only sees the voltages between its various terminals.  The positive supply terminal follows the +250 volt input.  The negative supply terminal follows the +250 volt input minus the zener diode voltage which could be anywhere from 5 to 30 volts depending on what zener diode is selected.  The current drawn by the Rbias resistor is the sum of the current from the operational amplifer's negative supply terminal and the current through the zener diode.  The bias current should always be higher than the maximum supply current for the operational amplifier.

[technix]

Hm... am i not understanding opams here right? If i put the in - and + to high voltage, will the opamp survive ? On the shown circuit my 250V would directly go into IN -  ?

Think this way: you got a floating ground for the op amp, and you need to shift the ground reference down. The RF and optical methods I suggested simply breaks the ground, and that schematic shows a way to level shift the ground down.

[technix]

Another breaking the ground idea: there are isolated power supply modules out there, and you can use chips like ADuM1401 or ISO7241 to get an isolated SPI bus. You can power the op amp and the ADC using that isolated power supply as well as the slave end of the signal isolator, and pull data from the low-voltage host end.

[T3sl4co1l]

Related:
https://www.eevblog.com/forum/beginners/high-voltage-high-side-current-amplifier/

The circuit shows precisely what I mentioned in #9.


Tim

[sgt_bear]

The OP for the current sense part has a zener to provide a regulated Supply of about 5 V. So the OP will be OK.

But if the OP AMP runs on 5V, why do i have to take the 5V from the BUS? Can't i supply just regular 5V there?

[RES]

[technix]

The OP for the current sense part has a zener to provide a regulated Supply of about 5 V. So the OP will be OK.

But if the OP AMP runs on 5V, why do i have to take the 5V from the BUS? Can't i supply just regular 5V there?

You need to be aware of the reference point from which the measurement of 5V is taken. Voltage is relative. This is why I have been babbling about a little concept called floating ground as the reference ground of the "5V" the op amp is operating from and the reference ground of the rest of the system is very different.

[sgt_bear]

I have found the part i want to use

http://www.analog.com/media/en/technical-documentation/data-sheets/AD8479.PDF

[T3sl4co1l]

A $10 boutique part, that isn't stocked anywhere, just because you don't understand a solution that half a dozen more knowledgeable engineers have proposed?

Well, you know what they say about money and certain types of people...

Tim

[sgt_bear]

Yes i saw the part is unavailabel
Also i have read the datasheet and sadly the device also steals current 500uA

I simulated one of the shown solutions, but they all consume massive current  (like 1.5mA) through the Zener. 1.5mA is way to bad.  I need 100uA precision, so the current has to be WAY under 100uA (10uA) or so.


What solutions do good multimeters use? They have a way more extreme job, they have to measure 10uA to 10A, in  mV to 750V range,

[C]

 sgt_bear
     Think you are missing something very simple.

When you add the  measurement circuit, your high voltage system (150-250V DC) is how supplying two loads.
One that you want to measure current in.
One that supplies power to measurement circuit.

[David Hess]

Yes i saw the part is unavailabel
Also i have read the datasheet and sadly the device also steals current 500uA

Internally it duplicates the other posted solution but with worse accuracy and higher cost.

I simulated one of the shown solutions, but they all consume massive current  (like 1.5mA) through the Zener. 1.5mA is way to bad.  I need 100uA precision, so the current has to be WAY under 100uA (10uA) or so.

The supply current and output current are completely separate circuits although they could be joined and the offset produced by a fixed supply current removed.  The supply current with that high voltage is still annoying so it is advantageous to use a low power operational amplifier which is why I suggested the TL031 which only requires 200 microamps but there are modern more expensive parts which draw much less current if you can live with lower bandwidth.

What solutions do good multimeters use? They have a way more extreme job, they have to measure 10uA to 10A, in  mV to 750V range,

Multimeters measure the voltage across a switched current shunt and are floating so there is none of this high side and low side nonsense.  As far as the multimeter is concerned, it is just another low, typically 100s of milivolts, voltage measurement.  Duplicating this would be more expensive and complex; it means floating the analog to digital converter to the high voltage supply.

[Greg Robinson]

Yes i saw the part is unavailabel
Also i have read the datasheet and sadly the device also steals current 500uA

I simulated one of the shown solutions, but they all consume massive current  (like 1.5mA) through the Zener. 1.5mA is way to bad.  I need 100uA precision, so the current has to be WAY under 100uA (10uA) or so.


What solutions do good multimeters use? They have a way more extreme job, they have to measure 10uA to 10A, in  mV to 750V range,

If your zener has that much current flowing through it, the design needs to be adjusted - those example circuits are just that - not a one-size fits all solution. Typical zeners need about 300-500uA bias current to keep above the negative region of their di/dt slope. Adjust the current limiting resistor to suit the zener you have selected, and the lowest voltage your supply will ever be during normal operation.

And you do realise that those circuits only measure the current in the load right? They do not include the current drawn by the high-side current sense amplifier and it's zener-regulated floating power supply, so depending on which device you choose, you can easily get 100uA precision. Is the power supply for the load not able to spare another couple of milliamps? If so, you should probably re-think the power supply, not ditch the proper current sense solution.

By the way, your comparison to multimeters with a range of 10uA to 10A is not at all fair, as that is over multiple ranges, with different sense resistors, different burden voltages and different accuracies in each range.

Anyway, you should probably tell us what your application is, that way we can give more specific advice or understand why something may or may not be appropriate.

[David Hess]

If your zener has that much current flowing through it, the design needs to be adjusted - those example circuits are just that - not a one-size fits all solution. Typical zeners need about 300-500uA bias current to keep above the negative region of their di/dt slope. Adjust the current limiting resistor to suit the zener you have selected, and the lowest voltage your supply will ever be during normal operation.

I usually figure on at most doubling the total current so a 200uA TL031 allows a reasonable 200uA for the zener diode.  If I used a lower power operational amplifier like a 20uA OPA347 or TLV2451, then I would use a micropower shunt regulator instead of a zener diode.

Another alternative to power a micropower operational amplifier is to use a charge pump and high voltage capacitor for isolation and this would allow using a standard micropower operational amplifier with better precision like an LT1077 because the operational amplifier's positive supply can be pumped to be greater than the high voltage.

[macboy]

...

What solutions do good multimeters use? They have a way more extreme job, they have to measure 10uA to 10A, in  mV to 750V range,

Multimeters measure the voltage across a switched current shunt and are floating so there is none of this high side and low side nonsense.  As far as the multimeter is concerned, it is just another low, typically 100s of milivolts, voltage measurement.  Duplicating this would be more expensive and complex; it means floating the analog to digital converter to the high voltage supply.

Exactly. All multimeters, even the cheapest ones, have an optically-isolated output! You look at the screen with your eyes, so the multimeter can sit happily at hundreds of volts above or below ground (and your body) without issue. You don't need a common ground connection to the multimeter to view its screen 

The usual issue with high-side current measurement is that you need to shift that measurement result back down to ground/common for use by the circuitry which needs the current measurement in the first place. Unless of course that circuitry can reside on the high side as well.

Of course the OP didn't even say if the measured current is going to be used by anything other than human eyes. Maybe the best solution is a simple panel meter with a transformer-isolated floating supply? Too many assumptions.

[David Hess]

The usual issue with high-side current measurement is that you need to shift that measurement result back down to ground/common for use by the circuitry which needs the current measurement in the first place. Unless of course that circuitry can reside on the high side as well.

For current control circuits like in power supplies, my preferred solution is to move the entire error amplifier to the high side and use the level shifter to shift the control signal to the high side.  Reading back the actual current would require a second level shifter or a floating readout.

[f5r5e5d]

https://moosh.im/mooshimeter/ ??

[sgt_bear]

You mean this circuit? Yes, I agree.

I tried this circuit to simulate with LTspice and the LT1672 (2uA Supply Current).
But my simulation does nothing it only has a % counter in the bottom which never finishes

Attached my circuit and simulation.