Normally this should be done properly by whatever software is used to access the HDD, but the homebrew drivers used for this differ wildly, so not all of them work the same way, as originally intended. Some simply cause brief pressing of the reset/standby button to be completely ignored (eg: HDLD_svr), while others correctly cause a proper termination of all ongoing file access followed by powering off the console (eg: HDLoader).
Originally Posted by alexo
Such faith in the infallibility of these implementations is almost touching.
Um, no. The mechanism that parks the head is mechanical, not electrical.
Basically, a spring. When power is applied, the voice coil works against the resistance of the spring. When power is lost, the spring moves the head away. Another method is using the rotational momentum of the drive to move the head away. In any case, the head does not go down on the data portion of the platter when power is cut off.
But with my experiences in tow I am unable to share that faith, regardless of your eloquence.
I never claimed to make a detailed step-by-step diagnosis. I merely claimed to have seen what happened and the result thereof. And that is enough.
With all due respect, but I don't think that you have the equipment to accurately diagnose the actual cause of the crash.
Yes, which is the kind of reason why I too said that the first incident could be blamed on those spikes. But it is even so an example of how an external power event can cause damage to the disc. And surely you too must be familiar with the fact that simply cutting off current to any circuitry containing inductors can produce voltage spikes...
A spike can forcefully drive the head into the platter, which is what probably happened happened in the first case.
So simply turning off the main power breaker while the equipment is running might in some circumstances cause harmful voltage spikes.
Never-the-less a power loss was the only external event that occurred, and the result was disk damage. Whether this was caused by a direct or indirect effect of the power loss is of course impossible to say with certainty, but the fact is that it occurred, and nothing you say will change that. Nor will it change the fact that similar events have happened, and continue to happen, for others than myself too.
I don't know what exactly happened in the second case but the cause should have been something other than a power loss.
It doesn't really matter one way or another whether the connection between cause and effect is direct or indirect. What matters is that this kind of power-loss can cause damage, and that the danger of such damage is higher if it occurs during active data access. So regardless of what protection modern drives contain it still makes good sense to continue with the same old protective measures as with older stuff, of making sure that all data access is terminated before we turn the hardware off.
A lot of it is engineering talk, of course, but I do not think that any honest engineer would claim 100% protection efficiency. And when that claim is made I do regard it as sales talk and nothing else.
Nobody claimed total protection against head crashes. I said that they eliminated *one* possible cause of a head crash -- power loss. And that's engineering talk, not sales.
I don't care how many articles you find nor where you found them. Stating a claim over and over again does not make it true when the experience of many users is that the claim is overstated.
I am not alone in having lost HDD drives due to power loss incidents, it has happened to lots of people, and most likely to many in these forums too.
I agree that the improved protection against head crashing in modern drives is a very good thing, but it does not give the total protection against damage in power loss that you say.
For one thing, it should be obvious to anyone that during the act of writing data to the HDD surface any kind of interruption at all will prove fatal to the integrity of the data, which will then be crashed even if the surface itself is undamaged. And this kind of power-loss damage to data is something that no mechanical systems can protect against.
In any case, I don't think either one of us is going to budge on this, and I don't think we are contributing anything constructive to this thread. So perhaps it is best if we just 'agree to disagree' on this topic and drop it for now. OK ?
Best regards: dlanor