@4wd:
Switchmode PSUs really require a load in order to regulate correctly - therefore you'll need to plug it into a HDD before you can accurately measure the voltage.
Hmm... Good point. I wonder what would be required for a dummy loaded test rig?-Edvard
Normally a HDD
Come on, you should be able to find an old HDD. Hard rubbish collection when people throw out their old 286's with 20MB HDDs. Visit your local tip. Visit the local PC shop. Tell your girlfriend that size does matter and she'll need to buy a bigger HDD
We are talking about a HDD.
A normal HDD doesn't run on AC, (and one diode is not sufficient to fully rectify it anyway), and a HDD doesn't need a 0.6V voltage drop, which is only relevant for generic silicon diodes - it could be from approx. 0.2-0.7V depending on diode type, (discounting zener diodes).
Right. I noticed there appeared to be two diodes there, which is not very efficient nor effective, but I have seen setups like that whose duty was to rectify one last time to eliminate as much PS line noise as possible. I have also done this for wall-wart powered audio gear, as those things tend to be noisier than they need to.
Actually, the only way to eliminate line noise is to use capacitance or inductance. You'll probably find that the diodes were there for protection against reverse power connection. Feeding DC, (or AC), into a full-wave bridge rectifier will always produce + and - on the correct terminals allowing anything connected to work. You've then got a voltage drop across two diodes to take into account.
But personally, I wouldn't bother replacing the diode. Remove it, power it up on a real ATX PSU, get your data off and then toss it if you don't want to live with it like that.
I agree. But wouldn't you want to bridge the diode pads after removing?
Aaahh, I now see where there is confusion.
Using just a single diode there is basically two methods of guarding against reverse power connection.
Method 1: You're referring to probably the most common form: a diode connected in
series with the +ve supply rail as per this diagram:
Pros: Cheap, simple.
Cons: Voltage drop has to be taken into account, requires a diode rated to handle
full load current.
The main problem with this with regard to HDDs is the voltage drop. HDDs require a regulated +5V that varies by no more than 5%, +4.95V - +5.05V, same specs as an ATX PSU is designed to supply. Already this blows using method 1 out of the water because of the voltage drop across a series connected diode, (generally silicon power diode, so drop = 0.6V), 5.0 - 0.6 = 4.4V
Yes, you can bridge the diode in this case. You gain 0.6V and lose your reverse power connection protection.
Method 2:
Pros: Cheap, simple, no voltage drop, diode doesn't need to handle full load current so can be lower rated, (ie. cheaper).
Cons: Relies on a properly designed PSU.
This is the method HDDs use, a reversed biased diode
across the supply. There is no problem as long as the power is connected correctly, the diode being reverse biased will not conduct. Also, you do not get a voltage drop because the diode isn't in the circuit, (so to speak), thus the full +5V is available to the HDD.
However, if you reverse the power connection the diode conducts causing a short circuit. This is supposed to either:
(a) cause the PSU to see a fault condition and shut down, (eg. a good ATX PSU), and remain shut down until it's removed, or
(b) cause the fuse to blow in the case, for example, of a battery powered circuit.
If you bridge the diode in this case it won't work at all
period. You've just shorted the power supply output and doesn't matter which way it's connected.
As you noted, there are two diodes, one across the +5V supply and one across the +12V supply. However, only one appears to be toast.
You'll note that the one on the +12V rail is bigger, thus blowing the idea of a two diode full-wave rectifier - the diodes would need to be rated the same, thus a similar size.
You can also see that the cathode of the diode, (end with the bar), is connected to the +12V lines on the connector - ie. reverse biased.
So does that mean just... removing it... like with tweezers? And not doing anything else except plug the HD? I must be misunderstanding something, am I not?
As you can see in my simple diagram above, (Method 2), removing the diode will do nothing more than remove the short circuit that is now there - you can put a pair of cutters through it if you want to use brute force as long as the short is removed.
However, please note that all the above is assuming the component is really a diode.Read my previous
post regarding using a multimeter to check.
Basically, I can see that the left pad in the picture goes to +5V on the SATA power connector. If the right pad shows zero resistance to ground then the odds are extremely good that it's a diode. Remove the component before testing.
This is also why I've said to use a
REAL ATX PSU - they are designed to shut down if there is a short circuit. Normally they don't even start up if there is one - the fan twitches and then it shuts down.
You've lost nothing by just removing the diode, (other than a few minutes), as the HDD isn't working at the moment. The worst that can happen is you lose the remaining
magic smoke in the rest of the HDD and still have a non-functioning HDD
EDIT: After having another look at the picture again, I can make out the band indicating the cathode end, (left side). So I'm 99.99999999999999999999%, (round it up if you like), positive that it's just the protection diode that has gone shorted.
So my amended directions would be:
1. Remove it.
2. Measure the resistance between the two pads it came off. If it reads zero then the HDD is not worth the trouble.
NOTE: There will be some resistance just hopefully not a dead short.
3. Connect it to a
REAL ATX PSU as found in a desktop PC.
DO NOT connect the data cable - just the power cable.
4. Turn it on. If it spins up let it run for 30 minutes. If it doesn't spin up, turn it off immediately and forget about it.
5. Assuming it spun up, turn it off. Connect the data cable, turn it back on and see if it's identified at the POST screen. If it is, you might be lucky and get your data off of it, (assuming the power spike didn't scramble everything).
