Overcurrent and MOSFET short protection

[geshka]

Hello. 

I am trying to implement high current, high voltage protection for the motor driven with single MOSFET with PWM.  Here is a partial schematic for the power board.

While overcurrent looks seemingly straightforward  - measure voltage drop on shunt resistor ( or even based on Hall sensor) , I wondering how to properly attach microprocessor to measure Vds on MOSFET?
Top right side connector will provide full voltage or the power supply  - 140-160V DC
Lower right side connector should provide average voltage on drain of MOSFET  and that has to be Vds. So the question is  - direct measurement with resistor divider or some sort of opto isolator (preferable) ?

Thanks

[ajb]

Isolating analog signals while maintaining accuracy is hard.  Optocouplers tend to be rather nonlinear, and can have wide unit-to-unit and temperature variations in their CTR.  They can be okay for something like an SMPS where the signal is maintained very close to a threshold, but they are not a great choice for isolating a signal with a large dynamic range.  There are dual-output optocouplers designed to help with those problems by using a second phototransistor to provide feedback to linearize the output, but they only help so much.

Anyway, if your MCU shares a ground connection with the power stage, then isolating the Vds sense connection alone isn't very beneficial.  A simple divider circuit, with adequate protection, will probably be fine. 

One other note: since the shunt resistor is between source and ground, you need to subtract that voltage from the drain voltage to get Vds.  The ADCs built into most MCUs cannot sample channels simultaneously, so if you digitize both Vd and Vs and then do the subtraction in software, you will get an error from the difference in sampling time. You could solve that by using an MCU with multiple ADCs that can be synchronized, an external simultaneous-sampling ADC (or multiple synchronized external ADCs), or a differential amplifier that does the subtraction in the analog domain.  On the other hand, with such a low shunt value and a relatively large range on Vd, the error is probably negligible and you can treat Vd as close enough to Vds. 

[langwadt]

ir2127

[moffy]

I am trying to implement high current, high voltage protection for the motor driven with single MOSFET with PWM.  Here is a partial schematic for the power board.

Measuring average voltage won't protect the MOSFET from an overvoltage spike which is one of the typical ways they die. The best protection against spikes is as you have a high speed diode across the motor but also a snubber across the MOSFET and good layout.

[geshka]

Isolating analog signals while maintaining accuracy is hard. 

Thanks for the answer, ajb . I am really worried working with non-isolated MCU. even logging to computer will be difficult.  For the current I am implementing Hall sensor instead of shunt  This will give me overcurrent protection.  What would be solution to protect against blown (short) MOSFET other then measure Vds?

Thanks.

[geshka]

I am trying to implement high current, high voltage protection for the motor driven with single MOSFET with PWM.  Here is a partial schematic for the power board.

Measuring average voltage won't protect the MOSFET from an overvoltage spike which is one of the typical ways they die.

Right moffy.  In this case I am trying to protect motor and the whole contraption to go with full voltage when MOSFET is shorten, so trying to sense that and shut down main power.

[moffy]

I am trying to implement high current, high voltage protection for the motor driven with single MOSFET with PWM.  Here is a partial schematic for the power board.

Measuring average voltage won't protect the MOSFET from an overvoltage spike which is one of the typical ways they die.

Right moffy.  In this case I am trying to protect motor and the whole contraption to go with full voltage when MOSFET is shorten, so trying to sense that and shut down main power.

Then current sense is all you really need, but for a shorted MOSFET you need a way to break the current e.g. a fuse and crowbar etc. The best protection though is to protect the MOSFET, a snubber and gate clamp zener are a start.

[S. Petrukhin]

I think the speed of the MCU is not enough to talk about protection, especially in case of a short circuit.
The best solution would be an autonomous hardware protection with an appropriate alarm.

[ajb]

I think the speed of the MCU is not enough to talk about protection, especially in case of a short circuit.
The best solution would be an autonomous hardware protection with an appropriate alarm.

A software solution could be fairly slow, but a lot of modern MCUs have timer units designed to control switching power stages that provide that sort of hardware control.  Typically they have one or more configurable 'fault' inputs that can be triggered from an internal analog comparator, and will stop the timer and set the PWM outputs to a defined safe state.  With multiple fault inputs per timer, you can even have one configured to restart the timer or terminate an output pulse for hysteretic or constant on/off time control modes with a second at a higher threshold for short circuit protection. 

[max_torque]

Most loss of saturation detectors only worry about a threshold rather than an analogue measurement of the actual Vds.

Most mofsets show significant rise of RDSon as current climbs out of "Normal" levels, and you can look at the SOA graph in conjunction with the RDSon characteristics to see this.  LOS is a pretty "gross" afair normally, so the thersold can be set high, higher than you would normally expect for the switching element in the worst "normal" case ie at elevate junction temperature at peak normal current

Once you have a threshold type afair with say a comparitor, then moving that digital signal across to the control system is trivial, and of course, even better you can simple use the signal locally to actually interupt the gate drive to ensure the FET is switched off.

Of course, all this has been thought about and perfected for years and rather than re-invent the wheel i'd just suggest using a gate driver with over-current protection built in. ie IR2127 already mentioned above 

[geshka]

IR2127 already mentioned above 

datasheet says it is a high side driver.  Will it work as a low side? or is there any equivalent for the low side ? 

[jkostb]

Isolation is only needed if your microcontroller has external communication interfaces. But Isolating only the current sensing does part is not sufficient. You then also need to isolate the PWM signal to gate driver. So if you need isolation then you have several options:
1) If your microcontroller directly interfaces with high voltage part, you could opt for only isolating external communication interfaces. Have a look at Analog Devices and TI. Analog devices for example have ADUM family. TI has similar products.
2) You can also opt for isolating gate driver and current sensing. For example
https://www.ti.com/power-management/gate-drivers/isolated-gate-drivers/products.html
For current sensing you can use AMC1300

Finally if you tend to sell your product it must meet requirements of low voltage directive and machine directive in europe. You need isolation, otherwise your design does not comply with the EU law.

[geshka]

Thanks jkostb.
I do have isolation from MCU on PWM. it is already done.  you suggestion to use AMC1300 looks promising, indeed. 
As per business - I am not selling this product of course  -just a hobby.
Full disclosure   I've got few similar treadmill board for DC motor and trying to fit it to my lathe(s).  It worked fine right away with constant speed provided to the motor. Then I made a step ahead and implemented  PID to compensate load on the motor. It worked  OK'ish and one of reason being just "ish" is that power board expects 20Hz of PWM signal from control board, which is too low for PID to react properly.  Now, I am driving the whole thing with 1kHz PWM applied directly to the MOSFET driver (R21) . And this where my problems started - MOSFET is blowing. It is either from transient processes or anything else.  I am not sure yet. Still investigating. Meanwhile  trying to protect MOSFET ...     

[S. Petrukhin]

The first thing that immediately comes to mind is not the excess current, but the incorrect operation of the MOSFET - it opens or closes for a long time, is in high resistance and overheats. Especially when you mentioned an increase in frequency. It may be necessary to take care of the correct and reliable control of the gate, to study what is happening there. 

[langwadt]

IR2127 already mentioned above 

datasheet says it is a high side driver.  Will it work as a low side? or is there any equivalent for the low side ?

from the first page of the datasheet:

"The floating channel can be used to drive an N-chan-
nel power MOSFET or IGBT in the high side or low
side configuration which operates up to 600 volts."

[geshka]

Thanks comrade Petrukhin     I will check rising and falling slopes of the PWM... Though,  driver seems to be good enough  - MIC4227

[S. Petrukhin]

Thanks comrade Petrukhin     I will check rising and falling slopes of the PWM... Though,  driver seems to be good enough  - MIC4227

The driver may not have enough voltage or current from the PS to implement the plan. 

[moffy]

  Now, I am driving the whole thing with 1kHz PWM applied directly to the MOSFET driver (R21) . And this where my problems started - MOSFET is blowing. It is either from transient processes or anything else.  I am not sure yet. Still investigating. Meanwhile  trying to protect MOSFET ...   

The MOSFET you have chosen has some avalanche capability, but MOSFETs die really quickly from overvoltage spikes, hence the need for snubbers close to the device and good layout, all protection devices such as the clamping diode need to be very close to the MOSFET as well as low impedance paths to the supply, place some large capacitors near the MOSFET to absorb the energy from the clamping diode. Another point, the gate resistors should be as close to the gate as possible to prevent parasitic VHF oscillations. Your values of 100R are a bit large maybe 22R each would be more appropriate.