What would be the best method to measure current?

You mean to measure current within your circuit so as to feed that measurement to a controller? How much current are you talking about?

 

Ever heard of the Amploc devices?

 

You mean to measure current within your circuit so as to feed that measurement to a controller? How much current are you talking about?

Exactly to measure the output rms from the power modules.
Peak phase current would be around 350A.

 

Ever heard of the Amploc devices?

I just had a look and the sensors looks great but I wanted to know if Using Shunts are better in terms of availability and performance, Since I can easily source shunts and related parts pretty easily here, Or What about ACS772 Hall effect sensors?
Or using this hall ic is good?

 

Exactly to measure the output rms from the power modules.
Peak phase current would be around 350A.

To get the exact rms linear output you'd need either a very complicated circuit or a more or less expensive chip. I've used the combination of said chip with amploc sensors and they worked like a charm.

 

I've used the ACS772-200B and it works well. The -400B will give you <1A resolution (5mV/A, so +/-2v centred on 2.5v full scale) and ~0.5% total error, so realistically that's within 1-2A. There's little point in using more than a 14bit ADC as the LSBs are just noise. The RC filter on the output is important to get right, to manage noise without impacting loop stability. The trickiest bit at 400A is finding a good way to mount it. RMS involves a bit of calculation (area under the curve, so summing samples - needs a fairly fast processor) or you feed the output of the ACS772 to a RMS chip as suggested above.

I've not used the Amploc sensors but they are essentially the same idea. They don't appear to do a 400A version, but the 300A might work, depends what its overload capacity is. It may be easier to use by mounting it around the stud that the phase cable is attached to.

 

I've used the ACS772-200B and it works well. The -400B will give you <1A resolution (5mV/A, so +/-2v centred on 2.5v full scale) and ~0.5% total error, so realistically that's within 1-2A. There's little point in using more than a 14bit ADC as the LSBs are just noise. The RC filter on the output is important to get right, to manage noise without impacting loop stability. The trickiest bit at 400A is finding a good way to mount it. RMS involves a bit of calculation (area under the curve, so summing samples - needs a fairly fast processor) or you feed the output of the ACS772 to a RMS chip as suggested above.

I've not used the Amploc sensors but they are essentially the same idea. They don't appear to do a 400A version, but the 300A might work, depends what its overload capacity is. It may be easier to use by mounting it around the stud that the phase cable is attached to.

I have selected some readily available Current sensors.
https://www.mouser.in/datasheet/2/397/L32PXXXS05FS-267585.pdf

 

I have selected some readily available Current sensors.
https://www.mouser.in/datasheet/2/397/L32PXXXS05FS-267585.pdf

Interesting devices & relatively new to market...

 

I think Going With SMD is more better So I have Came Up with some more designs

These are Completed Modules for the To-247 Fets
Below Is another possibility for Mosfets which are SMD Based But Heat Dissipation is my biggest concern, Could Someone give feedback on how can I effectively dissipate heat.

 

Below Is another possibility for Mosfets which are SMD Based But Heat Dissipation is my biggest concern, Could Someone give feedback on how can I effectively dissipate heat.

Personally I don't think you can do this with SMD parts, the thermal management is much harder with dpack2 or similar devices because you need lots of copper both sides stitched together with vias and then have to bond that to the heatsink in some thermally more effective way:

 

 

Personally I don't think you can do this with SMD parts, the thermal management is much harder with dpack2 or similar devices because you need lots of copper both sides stitched together with vias and then have to bond that to the heatsink in some thermally more effective way:

For that reason, I am going with the Through Hole Devices only. Heat Displacement will be hard to achieve at this stage,
Right now I am focussing on Logic Board,
I would love to hear feedback on what features should be there in the controller

 

Did you move the gate drivers off the bridge module and back on the control board? If so, bad move....

 

Did you move the gate drivers of the bridge module and back on the control board? If so, bad move....

I put the Gate Drives on the logic board itself for 2 reasons,
1)Mosfet Board will be smaller now and more compact
2)Debugging will be easy

the distance between the gate driver and MOSFET isn't too large since I have put the gate driver near to the side of the logic board and it will connect with the down power card directly with berg strip.

 

I put the Gate Drives on the logic board itself for 2 reasons,
1)Mosfet Board will be smaller now and more compact
2)Debugging will be easy

the distance between the gate driver and MOSFET isn't too large since I have put the gate driver near to the side of the logic board and it will connect with the down power card directly with berg strip.

But your gate drives are fast rise-time, high-current; by putting them even a few cm away on a header will slow your turn on/off times and will increase losses, plus you're introducing all sorts of parasitic capacitance & inductance, which will have effects we can only guess at. Suggest you read, and then re-read several times, all the application notes here.

Are you still using the IR2110 gate driver? Have you verified it'll drive 4 gates fast enough? Its only rated at 2.5A...

 

But your gate drives are fast rise-time, high-current; by putting them even a few cm away on a header will slow your turn on/off times and will increase losses, plus you're introducing all sorts of parasitic capacitance & inductance, which will have effects we can only guess at. Suggest you read, and then re-read several times, all the application notes here.

Are you still using the IR2110 gate driver? Have you verified it'll drive 4 gates fast enough? It's only rated at 2.5A...

I am still trying to figure out the best possible scenario for the driver board and logic board , And the Finalised Gate Driver is IRS2186 its 4A/4A Device Now

 

And the Finalised Gate Driver is IRS2186 its 4A/4A Device Now

OK, but still may not be sufficient, need to run the numbers... but I think you're marginal at 15v/50degC, you may need two gate drivers per phase to be sure. What's your proposed PWM switching frequency and the track widths for the gate drives?

 

OK, but still may not be sufficient, need to run the numbers... but I think you're marginal at 15v/50degC, you may need two gate drivers per phase to be sure. What's your proposed PWM switching frequency and the track widths for the gate drives?

T I am sticking with 50 Khz frequency over here but it may come down to 20 Khz as well that depends upon the Motor developers but this system need to be flexible and easy to adapt as per change. Could you share the Mosfet gate driver calculations?

 

Charge Q = Current I x time t, therefore I = Q/t

gate drive Ig = gate charge/switch-on time = Qg/(t(on) + tr)=215nC/(107nS) = 2A peak, per MOSFET

Gate resistance = Vgs/Ig = 15/2 = 7.5ohm, of which 2ohm is internal to MOSFET so external gate resistor = 5.5ohm.

Per MOSFET switching loss = 0.5 *Io * Vds * (tr + tf) * fsw = 0.5 * 87.5 * 60 * ( 86 +77)*1e-9 * 50000 = 21W

Reduce gate current to 0.7A per MOSFET, t = Qg/Ig = 215/0.7A = 307nS, & assume change to rise/fall is proportional then tr + tf = (86 + 77) * 307/107 = 467nS (good to 1st approximation)

Per MOSFET switching loss = 0.5 *Io * Vds * (tr + tf) * fsw = 0.5 * 87.5 * 60 * (467)*1e-9 * 50000 = 61W

Editted to correct formula & math...

 

Charge Q = Current I x time t, therefore I = Q/t

gate drive Ig = gate charge/switch-on time = Qg/(t(on) + tr)=215nC/(107nS) = 2A peak, per MOSFET

Gate resistance = Vgs/Ig = 15/2 = 7.5ohm, of which 2ohm is internal to MOSFET so external gate resistor = 5.5ohm.

Per MOSFET switching loss = 0.5 *Io * Vds * (tr + t(on) + tf +t(off)) * fsw = 0.5 * 87.5 * 60 * (21 + 86 +77 + 100)*1e-9 * 50000 = 37W

Reduce gate current to 0.7A per MOSFET, t = Qg/Ig = 215/0.7A = 307nS, & assume on time = off time

Per MOSFET switching loss = 0.5 *Io * Vds * (tr + t(on) + tf +t(off)) * fsw = 0.5 * 87.5 * 60 * (2 * 307)*1e-9 * 50000 = 80W

80W of heat???