RD54 Stopped Spinning

[Jon Elson]

On 04/05/2015 02:55 AM, Robert Jarratt wrote:
>
>> -----Original Message-----
>> From: cctech [mailto:cctech-bounces at classiccmp.org] On Behalf Of Jon Elson
>> Sent: 04 April 2015 22:18
>> To: General at classiccmp.org; Discussion at classiccmp.org:On-Topic Posts
>> Subject: Re: RD54 Stopped Spinning
>>
>> On 04/04/2015 04:06 PM, Robert Jarratt wrote:
>>> So when I fired it up again with the scope, the disk started to spin.
>>> This was a stroke of luck as I was able to probe what I believe were
>>> the outputs of the hall sensors. All three oscillated, but one of them
>>> showed significant sideways wobble on the scope, which suggests to me
>>> that the signal is not absolutely regular. It does seem to suggest
>>> that one of the sensors is not working well. However, would this be
>>> enough to explain why it stopped dead in its tracks when it was
>>> working before?
>> Yes, the logic is generally that you decode 6 legal states of the Hall
> sensors to
>> decide which of the 3 motor terminals to drive high and low.  if the Hall
> sensors
>> give a signal that has all 3 high, or all 3 low, the decode logic will
> fail, generally
>> leaving all transistors off.  If it gets in that state again, a vigorous
> twist of the
>> drive around the spindle axis might shift the rotor to a position where it
> will
>> start up again.
>>> Changing one of these sensors is probably going to be beyond me :-(
>>> Thanks Rob
>> I'd get the data off it quickly and retire the drive.  You may only have a
> few
>> minutes run time before the sensors go more flaky and the drive shuts off.
> But,
>> now that you know the secret, you ought to be able to recover data, if
> that is
>> the plan.
>>
>> Anyway, it seems you have completely diagnosed the problem.
>>
>> Jon
> I checked again this morning, because something was bothering me. I realised
> that the pins I was observing with the scope were the D pins on the FETs and
> not the hall effect sensor outputs. When I went back to measure the sensor
> output properly the outputs looked fine and perfectly stable. The outputs
> from the Z8 also looked fine. The outputs of the inverters into the FETs
> looked stable, although not square (pics here: http://1drv.ms/1a39cwz here:
> http://1drv.ms/1a39kfs and here: http://1drv.ms/1a39pju). It is only the
> output of one of the FETs that looks a bit unstable, but that could be a
> triggering problem because the signal is a bit irregular (pic here:
> http://1drv.ms/1a39ymO), the other two FETs have similarly shaped outputs
> but do not wobble in the same way. I wonder if replacing that FET might do
> the trick? Or is the inverter the problem as the inputs are perfectly
> square?
>
>
Do the gate signals not have the wobble?  (Wobble could be 
the PWM of the speed control logic, so could be normal.)
If so, then the FET may be bad.  In fact, the FET may be 
entirely OPEN, and the waveform you see is coming from the 
other two driven phases.

Kind of strange that one gate signal is a smooth ramp, the 
others have some wave on the top.
But, the drive probably has a current sense resistor in the 
common source leg of the low-side transistors, so that is 
feeding some of the current waveform back to the gates.  So, 
it may well be the one with the linear ramp on the top 
indicates the drain is open.

Jon