EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: annapurna on February 03, 2017, 11:59:07 am
-
hi.
i want to sense current at the high voltage side.
i have the buck interleaved converter with 3 legs. i want to sense the voltage at 3 legs and output voltage is 126V.
but i dont find current sense amplifiers with 126v of common mode voltage. can anyone help me.
Thank you
-
Sense resistor in the positve lead and 1:50 voltage dividers to ground before and after feeding into a differential amp.
-
That solution probably results in a significant error. A HAL based sensor from Allegro is likely to perform better but it depends on how large the current is which needs to be measured.
-
That solution probably results in a significant error.
Please quantify. "Probably" is not really a statement.
This solution is proven and works, provided you have a reasonable drop over the sense resistor.
-
If an AC current transformer cannot be used, I prefer a singled ended high side current sense circuit with an operational amplifier which has an input common mode range that includes the positive supply (unless a positive bias supply is available) and a high voltage PNP or P-channel MOSFET output transistor. Today rail-to-rail input operational amplifiers are ubiquitous but in the past suitable types include the LM301A (!) and many JFET input operational amplifiers like the TL031, TL051, TL061, TL071, TL081, LF351, LF353, LF355, and LF356. Do not use the LF357; it is only stable for gains of 5 or greater.
http://www.analog.com/en/analog-dialogue/articles/high-side-current-sensing-wide-dynamic-range.html (http://www.analog.com/en/analog-dialogue/articles/high-side-current-sensing-wide-dynamic-range.html)
http://www.ti.com/product/LMV934-N-Q1/datasheet/application_and_implementation (http://www.ti.com/product/LMV934-N-Q1/datasheet/application_and_implementation)
There are other tricky ways to achieve the same thing but the above is the simplest. No dividers which increase noise and limit common mode rejection are needed. Two transistors with a cascode connected to the negative supply will provide better performance.
-
That solution probably results in a significant error.
Please quantify. "Probably" is not really a statement.
This solution is proven and works, provided you have a reasonable drop over the sense resistor.
The error between the divider resistors is multiplied by the multiplication factor of the amplifier. This is the classic problem when using dividers to measure a voltage differential.
-
The error between the divider resistors is multiplied by the multiplication factor of the amplifier. This is the classic problem when using dividers to measure a voltage differential.
That statement makes very little sense.
Yes, you can have a voltage offset between the two divider outputs due to resistance tolerances.
You can also have a division difference resulting in gain error between the two for the same reasons.
But standard resistors from the major distis (RS, Farnell, Mouser, Digikey etc.) are today E96, 1%.
I'm convinced that using an LM741 for this it can be a problem.
Really! Using modern opamps with high impedance, practically no offset and bias, it's moot.
It works.
@David Hess: nice ICs, but they are not really suited for 126 V.
-
An easier method would be to use an AMC1100 diff isolator. You do however need a power source on its secondary side. And accept its limited bandwith and gain.
-
The error between the divider resistors is multiplied by the multiplication factor of the amplifier. This is the classic problem when using dividers to measure a voltage differential.
That statement makes very little sense.
Maybe use some Google-fu? For example:
https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/04/24/difference-amplifiers-the-need-for-well-matched-resistors
But there is more out there.
@Jeroen3: that AMC1100 looks like an interesting solution! However I did not dare to look at the price yet :P
-
The error between the divider resistors is multiplied by the multiplication factor of the amplifier. This is the classic problem when using dividers to measure a voltage differential.
That statement makes very little sense.
Take an example: 126Vin, sense resistor sized for a 1V drop, dividers to knock the common mode down by say a factor of 100.
Both Rtop are 100k, both Rbot are 1k, 1%
Vdiff if your dividing resistors were ideal = +0.0099V
Divider before the sense resistor, due to component tolerance, is actually 101k and 0.99k. Divider after the sense resistor, due to component tolerance, is actually 99k and 1.01k.
126V into the sense resistor, 125V out. The output of your voltage divider is 1.2231V on the input side, 1.2624V on the output side. Vdiff = -0.0393
Now reverse your resistor dividers. The output of the dividers is 1.2725V on the input, 1.2134V on the output side. Vdiff = +0.0591
The error caused by the 1% resistor tolerance is up to 5x larger than the measurement itself.
-
That solution probably results in a significant error.
Please quantify. "Probably" is not really a statement.
This solution is proven and works, provided you have a reasonable drop over the sense resistor.
The error between the divider resistors is multiplied by the multiplication factor of the amplifier. This is the classic problem when using dividers to measure a voltage differential.
There are several problems using dividers:
1. The resistor dividers divide the common mode level but also the signal making DC and AC noise added by the following stage worse.
2. Dividing the signal means more amplification is need limiting bandwidth.
3. The ratio matching of the resistor dividers limits common mode rejection.
4. The dividers also need to be frequency compensated for good AC performance including AC common mode rejection.
With all of that said, it can work and I have done it myself although the results were poor.
-
@David Hess: nice ICs, but they are not really suited for 126 V.
Look at the example from Analog Design again. A zener shunt regulator controls the supply voltage to the operational amplifier. The output transistor is the only high voltage semiconductor.