HDD GURU FORUMS

Old Tech wrote:I have also traced the entire circuit of a hard drive and I have no difficulty understanding what is going on in the circuit. Back in the old days of computers, in the late 1970's and '80s, I used to replace heads in the field and re-align them using a special disk that had a track dedicated to a signal that could be seen on an oscilliscope. You aligned the heads by watching the amplitude and vertical evenness of the signal.

I'm an electrical engineer who has plenty of experience as a component level tech. In fact I worked on those same drives (eg Control Data, Ampex, Fujitsu) way back in the 1980s. I well remember those cat's eyes patterns. I wrote my own head alignment program (30 words of machine code) for a Data General minicomputer. This allowed me to use regular software packs instead of expensive CE packs and disc exercisers. Those were fun times.

This was the sort of gear I worked on:

Recovering a ComputerVision CGOS200X file system:
http://www.hddoracle.com/viewtopic.php? ... 203&p=5729

BTW, I would love to see your reverse engineered circuit diagram. Information like that is very hard to come by.

Old Tech wrote:There was a suggestion that you add a free-wheeling diode to the relay on the board. If the power inputs to the relay coil are isolated from the rest of the circuits on the board, there would be no need for such a diode at 12 volts. There are MOVs used on such boards anyway to absorb transient spikes.

Diodes should always be used to protect the switching transistor. Even at 12V, very high back-emf voltages are induced when the relay's coil current collapses. Snubber diodes are just standard practice for any good engineer.

Most of WD's VSCs are documented within the scripts packaged with TREX. This factory tool is all over the Internet, including the HDD Guru files area. That would be a good start.

The only feature of the PC3000 "dongle" that may need to be duplicated is power control and load current monitoring. The former could be achieved with USB controlled relays or similar, while current monitoring may not be necessary if one were to use a current limited supply.

BTW, here is an "IDE Grabber" that you could build:

Reverse Engineering HDD Software using IDE Port Sniffer:
http://www.hddoracle.com/viewtopic.php? ... 611&p=1613

Under certain circumstances you could use a drive's SMART logging feature to sniff and record its own ATA traffic, including VSCs.

fzabkar wrote:Diodes should always be used to protect the switching transistor. Even at 12V, very high back-emf voltages are induced when the relay's coil current collapses. Snubber diodes are just standard practice for any good engineer.

Thanks for reply.

I have no argument with you in the use of snubber, or freewheeling diodes. On the board in question, they seemed to have been omitted and someone suggested adding them to the board. I would reason that someone designing such a board would have a reason for not including them.

Perhaps the relay is driven by a chip that has protection built in. Perhaps the diode is located in a location that is not obvious. I don't know. I did suggest that a common design on modern boards is to employ MOVs to cut off spikes and there are sometimes hundreds of capacitors used all over the board to absorb spikes.

I don't know the reason for omitting the diode, I was only suggesting to the board owner, if the board is working fine, I'd be hesitant to add a snubber diode without fully understanding the operation of the circuits involved. It's not worth blowing a board based on a bad assumption.

fzabkar wrote:Most of WD's VSCs are documented within the scripts packaged with TREX. This factory tool is all over the Internet, including the HDD Guru files area. That would be a good start....
BTW, here is an "IDE Grabber" that you could build:.

Thanks for all the info fzabkar. This is actually a case of data acquisition and storage. I only scanned your link but they send the output to a printer port and it's not clear what they are doing with it. I did notice a picture of what appears to be the trace of a multichannel osciliscope, and they can get pricey.

I was thinking along the lines of capturing a good slice of data and written a program to analyze it. You could single step it into the storage medium to determine which instruction produced which output then you could stream it to gather enough data to get the bigger picture.

Modern microprocessors are very sophisticated but there's nothing to stop anyone using the older TTL logic to build a very simple capture unit like the one at your link.

fzabkar wrote:

Old Tech wrote:I'm an electrical engineer who has plenty of experience as a component level tech. In fact I worked on those same drives (eg Control Data, Ampex, Fujitsu) way back in the 1980s. I well remember those cat's eyes patterns. I wrote my own head alignment program (30 words of machine code) for a Data General minicomputer. This allowed me to use regular software packs instead of expensive CE packs and disc exercisers. Those were fun times.

....BTW, I would love to see your reverse engineered circuit diagram. Information like that is very hard to come by.

I worked on what were called minicomputers like the Datapoint 1100 series through the 5500 series. The lower end units had nothing more than 4 x 8" floppy drives ganged together for a whopping 1 meg of storage. RAM was 4 x 4K boards for a total of 16K. That escalated fairly quickly to 64K then 128K.

I imagine you'll remember the stacked packs of hard drive disks that were 18" in diameter with 5 meg per platter.

I also remember writing realtime programs in Basic to move the heads out to read a certain cylinder on disk.

My circuit diagram is in a major mess. I made no attempt to clean it up. The hardest part was tracing the memory circuitry and trying to identify which chips and components were used.

If I come across the notes I'll try to clean the drawings up and maybe post them.

Old Tech wrote:This is actually a case of data acquisition and storage. I only scanned your link but they send the output to a printer port and it's not clear what they are doing with it. I did notice a picture of what appears to be the trace of a multichannel osciliscope, and they can get pricey.

I was thinking along the lines of capturing a good slice of data and written a program to analyze it.

It has already been done for you. The Grabber includes software that captures the IDE traffic.

http://nazyura.hardw.net/Grabb8b.zip
http://nazyura.hardw.net/Grabb16b.zip

The ZIPs include logs of the taskfile registers.

Old Tech wrote:Perhaps the relay is driven by a chip that has protection built in. Perhaps the diode is located in a location that is not obvious. I don't know.

ISTM that the relay is controlled by 5V logic via a simple NPN transistor. The only reason that I can see for omitting the diode is a possible manufacturing error. Some relays have diodes built-in, but the datasheet would suggest that the board does not employ one of these. Perhaps the design allows for both relay types, in which case it may be that the factory messed up the BOM.

fzabkar wrote:ISTM that the relay is controlled by 5V logic via a simple NPN transistor.

I was researching this question but I have been under the weather the past few days and it's giving me a headache.

Suppose you use a 2N2222. It has a forward current of 1 amp and a reverse breakdown voltage of 40 volts minimum.

The relay is a Rayes RS-12 with a nominal operating current of 12.5 milliamps and a coil voltage of 12 volts.

The question is this...can the induced voltage spike be high enough to exceed the 40 volt reverse breakdown voltage of the 2N2222 transistor? I was trying to find the actual back-EMF generated by a 12 volt coil with no diode across it to bleed off the current produced. I could not find a reference for the inductance of the coil.

fzabkar wrote:It has already been done for you. The Grabber includes software that captures the IDE traffic.

Thanks, fzabkar, that's good to know. I have not been well the past few days and have not done much. Feeling better now.

Old Tech wrote:

fzabkar wrote:ISTM that the relay is controlled by 5V logic via a simple NPN transistor.

I was researching this question but I have been under the weather the past few days and it's giving me a headache.

Suppose you use a 2N2222. It has a forward current of 1 amp and a reverse breakdown voltage of 40 volts minimum.

The relay is a Rayes RS-12 with a nominal operating current of 12.5 milliamps and a coil voltage of 12 volts.

The question is this...can the induced voltage spike be high enough to exceed the 40 volt reverse breakdown voltage of the 2N2222 transistor?

From Wikipedia ...

https://upload.wikimedia.org/wikipedia/commons/7/76/FlybackExample.GIF

Old Tech wrote:

fzabkar wrote:It has already been done for you. The Grabber includes software that captures the IDE traffic.

Thanks, fzabkar, that's good to know. I have not been well the past few days and have not done much. Feeling better now.

I'm having problems as well.

Just FYI, here are the taskfile registers (mainly VSCs) for a Samsung HDD whose maintenance cyclinders (MC) are being dumped (1X7 = command register):

Code:

IDEGRABB Log File 2004 Krasnodar (C)NazYura Email: NazYura@rambler.ru
---------------------------------------------------------------------
1X1=22  1X6=A0  1X2=80  1X3=00  1X4=00  1X5=00  1X7=C0
1X1=00  1X6=A0  1X2=01  1X3=01  1X4=00  1X5=00  1X7=EC
1X1=00  1X6=A0  1X2=03  1X3=01  1X4=00  1X5=00  1X7=EC
1X1=00  1X6=A0  1X2=03  1X3=01  1X4=00  1X5=00  1X7=EC
1X1=00  1X6=A0  1X2=01  1X3=06  1X4=00  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=04  1X3=07  1X4=00  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=02  1X3=01  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=02  1X3=03  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=04  1X3=05  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=01  1X3=0A  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=08  1X3=0C  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=08  1X3=13  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=08  1X3=1E  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=01  1X3=4E  1X4=01  1X5=00  1X7=80
1X1=00  1X6=A0  1X2=10  1X3=4F  1X4=01  1X5=00  1X7=80
etc etc etc

The previous dump was generated by the 8-bit version of NazYura's IDE Grabber.

@Old Tech , maybe you should check as well our brand new and on-going thread about MAXTOR Firmware research here : - http://www.hddoracle.com/viewtopic.php?f=59&t=1735

If we can extract vendor specific commands from the IDE Grabber logs then we can use those same commands on free MHDD scripting to retrieve drive firmware modules and resources.

Of course that it wouldn't help on your specific case as you don't want to copy translator tables from one drive to the other but it's a start

This could possible be used to unlock ATA passwords on MAXTOR by scripting and to copy modules that are not unique from a drive to another as long as the drive is still in working condition and don't have to be "loaded" from an outside file/code.

fzabkar wrote:

Old Tech wrote:https://upload.wikimedia.org/wikipedia/commons/7/76/FlybackExample.GIF

This does not answer my question. They have made an assumption that without the diode, using a 24 volt supply, that the back EMF will rise to 240 volts. I wanted to know what the actual back EMF would be with a 12 volt coil.

Besides, the current directions in the diagrams are all wrong. In the field of electronics and in the electrical field, current is deemed to flow from negative to positive. It's only in engineering schools they persist with the incorrect notion of conventional flow from positive to negative.

I had to suffer through that when I returned to the uni to study electrical engineering. Having been immersed in electronics theory based on negative to positive flow, I held my nose and did it. I found in most cases I could simply ignore conventional flow and use my old negative to positive flow to work out problems.

Perhaps a word on that might help people confused by diode markings. On some diodes the diode symbol is used to indicate polarity and the arrow points in the direction of conventional current flow, from positive to negative. Any one thinking in terms of negative to positive electron flow must think of current flowing against the arrow. Same with BJT transistors.

Many diodes simply have a bar on one end which indicated the cathode end of the diode.