How to save two Mitsubishi M2896 drive (spindle motor smoked!)

Brent Hilpert hilpert at
Thu Mar 19 16:43:16 CDT 2015

See schematic:

On 2015-Mar-19, at 11:36 AM, Enrico Lazzerini wrote:
> Well the verifications taken here:
> - Observe (C358)pin7 while alternating the 0/+5 test voltage on the 3rd
> connector pin, pin 7 should change accordingly to 0V & 9V. => NO it remain
> statically at 7,76V
> - Similarly (LB1620)pin11 should be changing. => NO it remain at 3,44V

> - The hall sensors may still be suspect. They are supplied by being stacked
> in series with two resistors, one each in the ground leg side and +9V supply
> side. First thing to do would be to measure the V on the two supply pins of
> each hall sensor (the pins closer to the motor center).  The supply voltage
> should divided between the 3 sensors and 2 resistors. 
> => The chain is :
> - from the +5vdc to R109 = 6,76V
> - then from R109 to the 1st hall sensor = 5,17V
> - then from the 1st hall sensor to the 2nd hall sensor = 3,62V
> - then from the 3rd hall sensor / R110 to the gnd = 2,2v
> So at each hall sensor we have (potential difference):
> 1st = 1.59V
> 2nd = 1.55v
> 3rd = 1.42v

This may be a concern. We don't know just what hall-effect sensors are being used but I found a datasheet for functionally similar 4-pin sensors.
The specs may of course be different, but the supply voltage range in that datasheet is 2.2-12V with 3.25-5mA current.
From calculations, operating at 9V, I was wondering if the voltage drop across the resistors left what may be too low a voltage on each sensor.

Recall we concluded the regulator voltage would be 9V because that fit inbetween the limits of the 9142 and 1620 specs. However we now know the 9142 GND is raised by 3.5V, lowering the supply to that IC.

Ii may be the regulator voltage is supposed to be 12V rather than 9. This would give:
	- 12V for the 1620, within spec,
	- 12-3.5=8.5V for the 9142, within spec,
	- and raise the voltage across each hall sensor.

> The halls sensors feed the LB1620 with a differential drive from each sensor
> to avoid having to make a fixed GND reference, which allowed them to be
> stacked in series for the supply. In theory you would be able to observe a
> change on one of the drive pair of each sensor with the scope as you
> manually rotate the rotor but the change may be too small to observe within
> the bias from the stacked supply. Properly one would use two scope channels
> in A-B difference mode to observe the drive pair.
> => confirm that rotating the motor my voltmeter change a bit its measured
> value

How much (voltage change)?

> - Instead, or in addition, you might try watching the (LB1620)pins1,2,3
> outputs while spinning the rotor manually to see if they change.
> => confirm that rotating the motor my voltmeter change a bit its measured
> value
> Could at this point the problem to be on c358c or the transistor q101?

The 358 could be a concern, or the diode at location R117, or Q101.
R117 looks like it was supposed to be a pull-down resistor but was installed with a diode as a reverse-bias clamp instead, for who knows what reason.

You could try removing Q101, then with the limiting resistor feeding the 3rd connector pin, watch what happens to (LM358)pin7 while toggling the control voltage into the limiting resistor. But also watch what happens at (LM358)pin5 to ensure it changes. The concern is that the R117 diode, if damaged/shorted, could be clamping the input voltage to the op amp comparator.
Expected operation of the 358 half with pin 7 open (Q101 removed):
	pin 5	pin 6	pin 7
	------	------	------
	0V		1.25V	~ 0V
	>1.25V	1.25V	~ 9V

Also with Q101 removed, you could try connecting (LC9142)pin11 to pin 1 (see 9142 datasheet). If the motor still doesn't run it may be time to consider changing the voltage regulator to 12V.

More information about the cctalk mailing list