Can dirty power supply cause bad sectors

[bigal.nz]

That pretty much sums up the question : Can a bad PSU cause bad sectors on hard disk, or do hard disks have sufficent circuit protection to stop this?

I ask because a friend has replace his hard disk twice in as many weeks and it keeps developing bad sectors. He swears black and blue that he has not knocked it or otherwise mishandled it.

TIA

-Al
PS: 4.v on 5v rail and 12.46v on 12v rail into HDD in question, both fall within +/- %5.

[N.C.]

Hi,

1. you can check PSU quality only with oscilloscope.
Forget the DMM.

2. yes, can cause!
Especially the voltage noise, unstability.
But most common case the PCB or preamp burnout.
Can cause SA corruption as well.

3. in case of your friend, i would replace the cable and connector wich goes to this hdd, if have anything between the PSU and HDD.

4. This is not an option for you, because i have only this unit for protecting hdds.

Janos

[bigal.nz]

Hi,

1. you can check PSU quality only with oscilloscope.
Forget the DMM.

2. yes, can cause!
Especially the voltage noise, unstability.
But most common case the PCB or preamp burnout.
Can cause SA corruption as well.

3. in case of your friend, i would replace the cable and connector wich goes to this hdd, if have anything between the PSU and HDD.

4. This is not an option for you, because i have only this unit for protecting hdds.

Janos

Thanks for the reply.

I just dont quite understand what you are refering to in point 4? Is it something to do with a special unit that you put between Molex connector and HDD?

Cheers

-Al
PS: I have a scope so I will hook it up and see what sort of noise is there.

[N.C.]

Hi,

1. you can check PSU quality only with oscilloscope.
Forget the DMM.

2. yes, can cause!
Especially the voltage noise, unstability.
But most common case the PCB or preamp burnout.
Can cause SA corruption as well.

3. in case of your friend, i would replace the cable and connector wich goes to this hdd, if have anything between the PSU and HDD.

4. This is not an option for you, because i have only this unit for protecting hdds.

Janos

Thanks for the reply.

I just dont quite understand what you are refering to in point 4? Is it something to do with a special unit that you put between Molex connector and HDD?

Cheers

-Al
PS: I have a scope so I will hook it up and see what sort of noise is there.

I have one protection unit wich is good for protecting the hdd from all the harm things can come from the PSU.
I have designed it, and i sell these in Hungary.
If you want one, you can PM me.

Janos

[bigal.nz]

Hi,

1. you can check PSU quality only with oscilloscope.
Forget the DMM.

Janos

Hi Janos,

I have attached two images of the 12v and 5v rails by back probing the molex with the scope.

What do you think? Looks bad? Settings for both pics was :

5mV/div (but 10x probe - so 0.005 x 10 = 0.05v/div)
.5ms

I should connect a spare Enermax I have here and compare....

Cheers

-Al

Attachments

12v

5v

[N.C.]

um, this is not a good measure methode.
A capture scope can be better, because the idle computer not makes current spikes...
Anyway, this looks noisy anyway.

Janos

[scratchy]

You need to look at single traces of two or three cycles.

Then look at the trace, it should look something like a square waveform.

[bigal.nz]

You need to look at single traces of two or three cycles.

Then look at the trace, it should look something like a square waveform.

Ok well I am keen to try again with a better technique, but might need some more information.

1) Is it ok to use not the molex that goes into the HDD but another molex on the same wire chain?
2) When you say lookat it over 2/3 cycles I assume I needto zoom in on the time quite a bit?
3) GND to GND of molex?

Thanks

Al

[bigal.nz]

um, this is not a good measure methode.
A capture scope can be better, because the idle computer not makes current spikes...
Anyway, this looks noisy anyway.

Janos

So your saying that I am looking for intermittent spikes, (in addition to a dirty waveform).

Wonder if I could do ith using triggering?

-AL

[N.C.]

um, this is not a good measure methode.
A capture scope can be better, because the idle computer not makes current spikes...
Anyway, this looks noisy anyway.

Janos

So your saying that I am looking for intermittent spikes, (in addition to a dirty waveform).

Wonder if I could do ith using triggering?

-AL

Yes, you can catch it using trigger.

Set up trigger level in 5V line to 5.4V (+400 mV) than 4.8V (-200 mV).

When measuring, do something on the PC.
Like boot up to windows, and start some programs.
It would be interesting if you try to write a DVD when measuring.

Let me know what was the result.

Janos

[bigal.nz]

You need to look at single traces of two or three cycles.

Then look at the trace, it should look something like a square waveform.

Ok I have had another crack at it. Some pics are time/div zoomed to 5us. Some are 1ms. (check file names)

Both are heavily zoomed on v/div to 5mv (but thats a 10x probe so its really only 0.05v/div).

This was not with a load (or hdd) attached to the molex.

The old PSU picture certainly has some spikes in it (they are a bit faint in the photo).

I assume these spikes are the bad things, though in reality the biggest spike was 4 divisions, so only 0.02v....

Thoughts?

Cheers

-Al

Attachments

5v rail
New PSU
Time/Div = 1ms
Volt/Div = 0.05
This was as close as I could get to making it display anything near a squareish wave?

5v rail
New PSU
Time/Div = 1us
Volt/Div = 0.05

5v rail
Old PSU
Time/Div = 1ms
Volt/Div = 0.05

5v rail
Old PSU
Time/Div = 1us
Volt/Div = 0.05
Those spikes go all the way to the top of the screen - just.