LS,

So there I was, happily chugging along on my trusty T42p with a Samsung HM160HC (a single-platter 2.5" 160GB PATA drive from 2007). When I had to reach for my daughter who was doing something unhealthy on the other side of the table I happened to rest my hand on the right wrist rest on that T42p, above the drive. I put a bit of pressure on it, not much, but enough for the screen to change into a rainbow of vertical lines. Aha, I thought, another T42p bites the dust due to the GPU getting unstuck from the board, time to bring out the Aoyue 968A+ again. Well, no, it turned out to be the drive which had stopped working. Not due to physical pressure, mind you, the amount of pressure I put on the wrist rest was hardly enough to flex it a little. When I restarted the machine the drive was no longer detected. After removing the drive, I noticed that the plastic layer between the aluminium cage and the drive PCB was damaged (torn, pieces missing). In the T42p the drive is mounted upside-down, with the PCB facing upward. The most likely thing to have happened is that the cage (which is grounded) made contact with some part of the PCB, causing the drive to fail. As stated, the drive is not recognized. It does not try to spin up, nor does it show any sign of life.

The PCB does not show any sign of damage. I measured the fuse at the incoming 5V line, it is undamaged. There seem to be 2 TVS diodes on the board, one close to the motor controller (a TLS2502B), one next to the incoming 5V line. Neither of these has shorted, but I measure only 18 ohm over the one next to the TLS2502. This is not due to the TVS diode, I unsoldered it to check. Not having any experience with these drives I'd like to hear from y'all what to look for next. I can post some board pictures if that is deemed helpful. The blue 1 ohm resistor next to the TLS2502 measures 1 ohm so that is cleared as well.

While this is a puny 160GB PATA drive from 2007 with no important data on it, I like to try to revive it for use in those older machines. I like those T42p machines, but they only like PATA drives (without resorting to hacks like PATA/SATA converters). I also prefer to repair instead of replace, which is one of the reasons for using those older machines in the first place.

So, anyone have any pointers? A datasheet for the TLS2502 might come in handy, to trace the possible location of more fuses on the board. It is a rather uncrowded board but that does not make it easier to find unmarked fuses in the forest of equally-unmarked resonators, resistors and capacitors.

TIA//Frank

For good measure I took some pictures of the PCB:

https://click.unternet.org/photos/album ... 4676e/list

Download the originals for a closer (10 MP) look. Not all of them, or my poor server-under-the-stairs will melt down and/or your patience will run out...

Could you measure the +5V, Vcore, and Vio voltages?

_________________
A backup a day keeps DR away.

They seem quite normal:

+5 filtered: 5.02V
Vcore: 1.34V
Vio: 3.32V

Judging from what I read in the M40S service manual (a different drive, sure, but some things are bound to be similar due to component constraints) I expect Vcore to be around 1.2V, Vio 3.3V and 5V just what it says.

Any fuses on this board that I could have missed? Do 2.5" drives typically provide the pre-amp with a negative voltage? I don't see any circuitry capable of generating such here so I guess not?

I think Vcore is OK. IIRC, some WD drives use 1.3V.

As for the preamp voltage, some preamps are powered from +3.3V only. I would measure the voltages at the HDA contacts on the PCB.

Are you certain that the drive doesn't attempt to spin? Can you hear any faint noises? Could it possibly be a stiction fault?

_________________
A backup a day keeps DR away.

Not a sound to be heard. Even stranger, nothing on the console - assuming this drive has RX/TX on the 'bottom two option pins' (where 'bottom' means 'bottom of PCB' and 'option pins' means the set of 4 pins next to the 44-pin PATA connector. I tried both 38400 and 57600 bps, in all combinations of the 4 pins (just in case). Zilch.

As the drive was running when the happening occurred, it'd be strange if this were a stiction problem...

I'm trying to trace Vcore to the processor, maybe something in between gave way? Or maybe the processor is just dead, but then I'd expect at least some magic smoke, or at least some heat somewhere.

Well, I fixed it, at first... The component you identied with 'inductor?' in the picture you marked as Vcore_current_sense.jpg seems to be some type of fuse/overcurrent-protection device. It was faulty. I replaced it with an smd fuse from another drive board and suddenly the console started responding:



So I put the drive back together and thought to test it one more time...

at 03:00 at night...

Don't do that. I happily connected the four test leads from the USB-powered level converter which I use to read the console and power the drive...

...and connected them the wrong way: power on serial port and the other way around. Way to go. Now it is dead, again, but this time I'm fairly sure there is no light at the end of the tunnel. I guess I blew up the controller by zapping those 3.3V lines with 5V.

Not a happy camper right now, having managed to grab defeat from the claws of victory in the last moment.

I'm reasonably certain that the device is actually an inductor. Its function would be to filter the supply. In fact yours is about the 5th time I have seen problems with Samsung's inductors, and in several different models. In those cases the problem was not with the inductor itself, but in the solder joints. The problem was resolved by simply reflowing the solder at the inductor.

Congratulations on locating the fault. Hopefully your experience will help others.

_________________
A backup a day keeps DR away.

First, thanks for your help and may it help others indeed.

The part I swapped for the fuse was definitely faulty in some way. It has three connections, two on each end and one in the middle. The middle connection on all these devices is connected to the ground plane. In a device which has not failed the resistance between the end and middle connections is high ('infinite' as in 'does not show on DMM'). On the one I removed it was only 18 ohms.

In my first post I mentioned that '...I measure only 18 ohm over the one next to the TLS2502...'. This device was the cause of that low resistance. It pulled down Vcore to close to 0V on its cold side. This is what makes me think it is an overcurrent protection device for the processor. There are two more of these on the board, one for 5V and one for 3.3V

Now to the current drive. While it seems dead, as stated, there is something rather peculiar: it echoes anything sent over the console. This could be due to a simple short circuit between TX and RX but I measure a very high resistance between these lines when the device is unpowered.

Maybe, just maybe, the firmware got scrambled by me applying 5V to those lines? It is a rather long shot, but it *is* odd for a dead device to echo data over its serial connection. I put it in the 'work for when the snow is 2m high' box for now...

Many thanks for the feedback. Could the component be a low pass LC filter, perhaps with ESD protection diodes?

The following are multi-channel examples:

http://www.nxp.com/documents/data_sheet/IP3337CX18.pdf
http://www.nxp.com/documents/data_sheet/IP3048CX5.pdf



As for the echo, are you certain that you are not observing "local echo"? That is, are you certain that it is not your terminal program that is responsible? Do you still see an echo if you disconnect the Tx/Rx wires?

_________________
A backup a day keeps DR away.

Local echo was disabled when testing, so that can not have been the source of this feedback. Oh well, after a long search full of hope I'll probably find the cause to be a bottle caught in a curtain string, flapping in the breeze... (classical reference...).

More later on those devices...