current moniter

[electros6]

Hi guys ,
          I want to measure the current flow in a 150V dc rail in the high side , I am not able to find a chip above 100V. Somebody please help

[Rerouter]

that is bloody high.  you can make one, by using a P-fet and an op amp that is run about 10V below the supply rail (zener supply or something similar)

you measure your output current across the 100K resistor in this image, you may need to adjust the gain to suit. and the shunt resistance,

equally i would reccomend strapping back to back diodes across the op amp inputs in this case, and having a zener also across the 100K limiting maximum voltage.

[dannyf]

Flow the ground to the 100v rail sufficient and then use a level shifter to reference the output to the true ground.

[Jay_Diddy_B]

Hi,

Here is another example of the technique proposed by Rerouter from the LTC6101 datasheet:



This can be extended using multiple MOSFETs:



I have built this circuit, and it works quite nicely:



Regards,


Jay_Diddy_B

[electros6]

Thanks Jay

[electros6]

Hi jay
     can you please tell me how this is working

[dannyf]

The zener floats the current sensor's ground to within the spec of voltage supply with regards to the real supply voltage. In this case, it is 500v-62v. The zener is isolated via Vgs of M2. So the real potential on the current sensor is 500v-62v+Vgs(M2).

M1 is the "level shifter" that translates the output with regards to the true ground.

[fcb]

As we don't know how much current you are measuring or the required accuracy it is difficult to pin down the most appropriate answer.

It's quite possible that you just need a 10:1 (20:1 would give you 7.5v if you haven't got more than 15v available) potential divider either side of a shunt, and then an instrumentation amplifer looking at the difference between the two.

Or you might require lots of current and choose an allegro current shunt (ACS714 or bigger).

[dannyf]

Why is Vout = 49.9Vsense?

Because of the negative feedback, The opamp's inverting and non-inverting inputs must be at the same potential. That means the current going through Rin is Vsense / Rin. As the opamp's input current is very small, that current has go flow through the mosfets and eventially onto Rout -> Vout = 49.9*Vsense.

This is where the use of mosfets is helpful. The opamp also needs to have high common mode voltage range.

[Jay_Diddy_B]

Hi,

The LTC6101 is not a standard op-amp. It is a current sense amplifier designed to have its inputs at or close to the positive rail.

Link: http://www.linear.com/product/LTC6101

I have attached the LTspice model for my circuit so that you can se how it works.



Jay_Diddy_B

[nctnico]

Allegro has several hall based current sensors which are easy to use.

[digsys]

My favourite IC for picking of "small" signals on HV rails is the LT1990. Pretty much nothing needs to be added.
It's a true diff op-amp, with 250V+/- Inputs, and optionally a x10 link setting, and very nice accuracy figures. It's magic :-)

[Rerouter]

Wow. did not even know such a gem existed...

[nctnico]

My favourite IC for picking of "small" signals on HV rails is the LT1990. Pretty much nothing needs to be added.
It's a true diff op-amp, with 250V+/- Inputs, and optionally a x10 link setting, and very nice accuracy figures. It's magic :-)

How about creepage distance? 

[Rerouter]

how about conformal coating?

[digsys]

  How about creepage distance?     

OK That depends on what actual voltages you expect to up to. Keep your pads 1-4, HV on 2+3 as skinny as possible,
don't route tracks between 1-4 and 5-8 side any closer to 2+3. Have a spare TEST IC PAD or two on the PCB permanent,
and run a 250V / 500V megger test before assembly. I've used Kapton tape under the IC, ie a long strip right down the
length of the PCB as I run 22X in a row (in my BM modules). If you DO expect to hit + and - 250V on adjacent ICs,
then, do the lot, HV coating and micro grooves / routes. The max I design to is 200VDC between any pins, and rarely
ever need to go to any extra lengths (besides skinny pads etc). A megger is your best friend !!!

Edit: I've been making BM pcbs for several years using this IC and have never had a flash-over yet ! (Touch wood?)

Edit2: Given the alternate definition of wood, is the expression "Touch wood" appropriate any more ??

[Psi]

The stacking of multiple fets is just to handle the high voltage correct?
You could use less fets if they were rated higher?

Gussing it's more cost effective using multiple cheap/common mains-voltage fets vs using a custom high voltage one

[Jay_Diddy_B]

Psi asked:

The stacking of multiple fets is just to handle the high voltage correct?
You could use less fets if they were rated higher?

Guessing it's more cost effective using multiple cheap/common mains-voltage fets vs using a custom high voltage one

I assume that you are talking about the LT6101 circuit?



The challenge is that the circuit requires P Ch. MOSFETs. I  used 3x Diodes Inc ZVP054 in the design for a total of 3x 450V for a 700V application. This gave a safety margin of about 2:1. The SOT223 package was chosen for creepage distance. There are relatively few H.V. P. Channel devices  compared to a large number of N. Channel MOSFETs.


A single MOSFET will work in the requested application of 150V. I just wanted to show how the technique can be extended to really high voltages. The circuit can be extended or reduced depending on the voltage rating required.

Because it use P. Channel devices it is easier to understand if you turn the circuit upside down. M4, M5 and M6 are configured as source followers. The voltage on the negative supply pin of the LT6101 is equal to the Zener voltage minus the gate threshold voltage of M4. M5 and M6 are also source followers. Their Vds is determined by the resistor string R7, R8 and R9.

MOSFETs M1, M2 and M3 are operated in common gate. The source current equals the drain current. The MOSFET inside the LT6101 sets the current for the whole string. Voltage sharing is forced by the divider R7, R8 and R9.

The voltage gain is R3/R2. As shown the gain is 50x. The output voltage is 50x the voltage on the sense resistor. Combined with the 0.1 resistor the total gain is 5V/A.

Jay_Diddy_B

[dannyf]

This is where the use of mosfets is helpful.

To put it in perspective: the benefits of using a mosfet diminishes as the current going through Rin goes down.

So the simplest solution here is to use an opamp with high common mode voltage range and level shift via a pnp to ground: you can easily reduce the impact of base current to below 1% (and 0.2% likely). And that are plenty of pnp with high Vceo.

[Feanor]

Texas intruments INA149 common mode goes +/- 275 V.

Datasheet -

http://www.ti.com/lit/ds/sbos579b/sbos579b.pdf

[nctnico]

Repeat: creepage distance.

[Feanor]

digsys's post below is spot on regarding how to avoid creepage problems.

If you want a number then 1mm per 100V is a very rough rule of thumb for anything below 1000V. (Veroboard used for prototyping has roughly 0.5mm spacing between traces and this will withstand 1000V when it is new and clean)

You will need to test your PCB at the highest voltage (or double that for some confidence) that you expect during normal service before populating it (test it for several hours).

You can use 1mm wide slots to increase creepage distances to 10mm per 100V if the PCB may be subject to dirt/dust/moisture.

If flashover (or component failure) could result in electric shock creepage distance is cold comfort. You should consider an earth trace around all circuits above ELV (extra low voltage), or change your design.

[digsys]

If flashover (or component failure) could result in electric shock creepage distance is cold comfort. You should consider an earth trace around all circuits above ELV (extra low voltage), or change your design. 

In all my BM and HV Isense systems, I have the control logic Isolated as well, both power and all signals. Usually 500v-1KV,
so IF flashover does occur, it's an extra higher capability safety. In fact, as the BM modules live in the Battery compartment, I have ONE
more final isolation layer, 2KV+. I embed this against the battery enclosure wall, so nothing can move or have access.

I also doubt that they could sell that IC if it DIDN'T meet the specs they advertise. Assembly precautions is up to the user.

[Jay_Diddy_B]

I also doubt that they could sell that IC if it DIDN'T meet the specs they advertise. Assembly precautions is up to the user.

An example of this from the Fairchild 4N26 optoisolator:

 

The isolation voltage is only specified for 1 minute. This doesn't mean that it o.k. to use this for 5.3kV ac applications.

In fact the UL file (E90700) for this part includes:



They can rate the voltage at whatever the manufacturer chooses. They are not constrained by regulatory requirements.

The user of the parts is constrained by the regulatory requirements that they need to meet.

If the user of the part may determine, if in their application, the insulation system fails and it doesn't create a shock hazard, fire hazard etc. They may determine that this o.k.

This would be true of low energy power supply. It would not be true in a high power battery management system.

This best method would be to use a LEM module:



http://www.lem.com/hq/en/component/option,com_catalog/task,displayserie/serie,LA%2025-200%20-P/output_type,/

Think of some of the pictures of destroyed DMMs .....

Jay_Diddy_B

[peter.mitchell]

for all the effort and cost, why not isolated supply (like this) to opamp + micro then beam it back by IR or bluetooth or Opto?