My 5v to 7v level shifter, shifts 1.5v to 10v

[righteous]

Q1, Q3 and M1 (referring to schematic) are on the purple round board, I assume, but there are too many SOT packages for your schematic (which I guess would have 6 FETs) and no TO packages for the push pull stage. Can you tell what exactly is in the mentioned board?

Please see attached schematic.
View attachment bldc-driver.pdf

The "OUT" signal is above the threshold of the FETs, so yes, you will have shoot-through if there are FETs in half bridge config.

Yes, and I can't measure the signal while there is power on the driver pcb.

So you are actually using driver ICs ? Not the discrete one in post #1 ?

The discrete one as in post #1

 

[CataM]

Measure voltage between grounds of driver PCB and power PCB.

 

[KlausST]

Hi,

With the new wiring I don't expect GND problems anymore...but "knowing" is better than "guessing", thus I also recommend to measure it.

If I remember right you have a 4 channel scope.
With one measurement we can verify a couple of things.
I recommend:
* connect all channel cable GNDs to the copper foil of your microcontroller board.
* connect channel 1 to the GND connection of the power PCB
* connect channel 2 to the VCC of the power PCB
* connect channel 3 to the high side signal after the level shift circuit
* connect channel 4 to the low side signal after the level shift circuit (same half bridge as channel 3)

Klaus

 

[righteous]

Reverting later today, I have to step out into human society for a few hours.

 

[righteous]

Hi,

With the new wiring I don't expect GND problems anymore...but "knowing" is better than "guessing", thus I also recommend to measure it.

If I remember right you have a 4 channel scope.
With one measurement we can verify a couple of things.
I recommend:
* connect all channel cable GNDs to the copper foil of your microcontroller board.
* connect channel 1 to the GND connection of the power PCB
* connect channel 2 to the VCC of the power PCB
* connect channel 3 to the high side signal after the level shift circuit
* connect channel 4 to the low side signal after the level shift circuit (same half bridge as channel 3)

Klaus

Human society, I can not recommend it, but now I'm back in the safe comfort of my lair.

I could not make any measurements on the PCB as it short circuits. However I noticed that when all the MCU outputs were (static) low, then obviously the FET's did not short circuit, and there were correct high levels (~7.5v) coming to the driver trannies But when I start the program in the MCU, it short circuits the FET's so I could not make all the recommended measurements (again, the program works and the motor spins with all 5v i.e. MCU/PCB/no level shifters).

Here is what I could measure with no PCB connected:

ch1 (AC): GND on level shifter board
ch2 (AC): VCC on level shifter board
ch3 (DC): MCU digital output (I generated a 2KHz square with the MCU)
ch4 (DC): output of level shifter (with VCC at ~7.5v)

 

[KlausST]

Hi,

It seems like Ch1 and Ch2 are in AC mode. But in AC mode you can not check for GND bounce.

Sadly Ch3 is not "after" the level shift
and I'm not sure if Ch3 and Ch4 are high side and low side signals of the same halfbridge ..thus we can not check for deadtime.

Klaus

 

[righteous]

Hi,

It seems like Ch1 and Ch2 are in AC mode. But in AC mode you can not check for GND bounce.

Sadly Ch3 is not "after" the level shift
and I'm not sure if Ch3 and Ch4 are high side and low side signals of the same halfbridge ..thus we can not check for deadtime.

Klaus

I apologize, I forgot to mention that I checked for GND bounce, and found none, GND is same on both MCU and level shifter, so I thought it would be more interesting showing some noise rather than two complete flat lines at same level.

I only used one level shifter due to the short circuit, so ch3 is the MCU output/level shift input, and ch4 is "after" i.e. output of the level shifter, which shows inverted output and correct levels.

I'm thinking about how to measure at least one halfbridge (hi and low) without blowing the PCB up and make to much changes in the assembly code.

 

[KlausST]

Hi,

I apologize, I forgot to mention that I checked for GND bounce, and found none, GND is same on both MCU and level shifter, so I thought it would be more interesting showing some noise rather than two complete flat lines at same level.

Good to know.

I'm thinking about how to measure at least one halfbridge (hi and low) without blowing the PCB up and make to much changes in the assembly code.

* disconnect the three motor wires from the power PCB
* connect both inputs of two half bridges to GND (4 signals)
...now only one halfbridge is active.
* limit the power supply current to 300mA if possible.
* with the limited current there should be no (big) risk to destroy anything.
.. and you may be able to measure the signals.

If the voltage drops because the 300mA limit (this indicates short circuit in the halfbridge)
* connect the remaining power PCB inputs to GND.
now the power PCB is completely inactive.
Then measure the signals before and after the level shifter of one half bridge.

Klaus

 

[righteous]

* disconnect the three motor wires from the power PCB
* connect both inputs of two half bridges to GND (4 signals)
...now only one halfbridge is active.
* limit the power supply current to 300mA if possible.
* with the limited current there should be no (big) risk to destroy anything.
.. and you may be able to measure the signals.

I did like this:

1. Kept the motor connected
2. Connected A-hi and A-low to the level shifters
3. Connected the remaining half-bridges to VCC
4. Hand cranked the motor to make the sensors change

I noticed that A-hi and A-low mostly changed levels from GND to VCC, but at some motor positions when A-low was supposed to be off (GND) the level was 2.5v and A-hi was supposed to be on (GND) the level was 2.5v. So at one point both A-low and A-hi were on at 2.5v.

At this point I'm inclined to abandon the level shifters, I'm wasting too much time on it since the system is complex, I think from a time management pov I better find a driver ic.