Electronics SATA Question/Challenge

Anyone in the Reading area want to let me stick my scope accross their Tx pair ? Make a small hole in a cable, if the scope can pick it up without damaging anything, i think we might be able to design something.

(SATA isn't something i've bought into, been a SCSI j'aid, so don't have a rig i could whack it on)

 

lol well all this is baffleing since none of us know what way is right so we need to experiment some idea so gimme intructions simple lol, and i shal test it

 

the way I read the specs you have to keep the cable impeadance to 50 Ohms +/- 5% and the termination impeadance has to be 40Ohms +/- 5%. If you can keep to this no data degredation should happen. My main consern is that With LVDS you dont have to have both your signaling cables at 0V at logical 0 you just need them to be the same.

In my eyes if you use a very high impeadance/ very fast OP AMP (SATA data signals are in the 1.2Ghz range) Then it is possible by connecting Tx- to the -ve terminal and Tx+ to the +ve terminal.

However if anyone wants to try this out be prepared to have your data hosed.

edit: I am workign on a proposal to try. What ever the answer will be it wont be that cheap. (might be about £20-£30) and will require a microcontroller to filter out the sync signals that are sent down the cable every so often. (i.e. we might end up trying to intercept the commands going down the cable to work out activity)

 

It's perfectly doable to make something like this under Linux by cat'ing /proc/yourharddiskhere and checking for the results AFAIK... There are even more civilised tools then going after the kernels virtual fs (read, /proc, for the less Linux-knowledgeable), several tools provide the current transfer rate to the drive, I guess you could run awk/sed on the output of that and define when it's, say, half a MB per second (just to weed out any useless stuff and only get /REAL/ data transfer), and then if that happens use the parport tool to manually set one of the data pins on your parrallell port... Once you've got that, LED bliss is only a transistor away ;-)

 

I think my method sounds a whole hell of a lot better than this complicated nonsense.

 

I'll agree with that now.

1ghz is pretty fast for an op-amp to work at. I'm sure you could find something on the drive or the board as activity light. Since it is running at 1.2 ghz, I'm sure it dosen't run straight to the northbridge, so there has to be a controller chip, and it probably is in a package you can try to put a wire on one of the leads. If you have good hands and a verrrrry fine iron, or a hot air station (you can ghetto make one from a $13 radioshack desoldering iron and a fish tank air pump if it intersts you)

 

Only thing stopping you doing that directly in NT is the kernel (it actually has security). So you still make a device driver for it, say make the LPT port lights add up as the access increases is somethign i made a while ago (but i messed up my code a little and somehow slowed the HDD down).

 

hey lynx have u got anyware with your idea, or has anyone else tried and succeded in doing this?

 

The signaling is LVDS which causes all sorts of problems. Its possible but not realy viable. It is eayser to tap off the controller chip. Though that could be a problem if its a southbridge.


edit: I have decide to give this whole problem my undivided attention. What chipset do you have? I am going to find a solution even if it requires designign an LVDS system

 

I came across something on the WD web page that could help. It said that SATA drives can engage an activity LED via pin-11 on the Hard Drive power connector. http://www.wdc.com/en/library/sata/2579-001097.pdf

 

Ok, Looks like it works for WD HDDs anyway. I hooked a LED up to pin11 on the power cable and it lights up. It is not that bright, but it seems to correspond to disk access. Also, according to the "Serial ATA attachment specification" http://www.sata-io.org/docs/serialata10a.ZIP (Page 45-46) Pin 11 is reserved. Looks like this could work with other HDDs

 

I'm crash reading sata specs as I type. In particular I'd be interested if I could get a signal when a HDD is connected at all so I could have ready and active lights.

There's a market for someone to design a modified power connect here.

 

Damm - all these SATA power connectors seem to be moulded in a single piece. I'm hunting a few sites for a more 'accessible 'connector but it looks a bit thin as yet.

 

You could try this cable. It looks like the one I am using. http://www.performance-pcs.com/cata...d=384&osCsid=7791170a24237afbf05fb979f809f207

Any ideas on how to make the LED brighter.

 

Aaah yes, the wonders of SATA cables! I had looked into cobbling together a simple pass-thru backpanel for a raid array, connected to a separate controler card. Well, i found lots of info on the pin-out from Molex (www.molex.com) and realized that the pins themselves were so friggin small that i wouldn't be able to solder them myself. On top of that, scrweing with the tracelengths will seriously corrupt the transmission. The only one that's pretty much safe to mess with is the power. the pins are grouped into sets of 3, isolated by longer ground pins inbetween. That's part of the hot-swap setup, i.e. the ground pins contact first, then the power.

Hmm, i thought i had found some more detailed info somewhere else, but looks like all i found was the SATA spec sheet mentioned above, with the "pin 11 reserved" bit.

I wonder of there's anywhere else that might have it...

 

Just connect an optoisolator between the led and the hd. Not only will it protect the hard drive if the led pulls too much current, but it also should be able to sink more current.

Nice find on this pin 11 thing!

 

Good news! Looks like pin-11 will also work on maxtor drives. This is what they say, " Pin 11 on Maxtor DiamondMax 10 or MaXLine III SATA hard drives becomes an output, which can be connected to LED circuits to indicate activity. Pin 11 activity LED output can sink up to 13mA of current in the active state, which is sufficient to allow connecting this output directly to the cathode of an LED "

 

if you need a way to make your own cables, check out mouser.com and look for SATA backplane connectors. those should let you isolate pin11 and still connect power to the rest of the drive. you might have to order the whole thing and jsut chop off the power section though...

 

I haven't given up on this yet:
http://www.cpuser.com.tw/SATA-Backplane.htm
seems the similar to
http://www.yesway.com.tw/product.asp?xsel=09

Now I just need to find somewhere who supplies these and get a price (for four in my case )

 

Alas I can't find the parts anywhere except those two links above (No idea of the cost but not sure I could shift the spare 96 on the order

However I've just found a SATA mobile ract at GBP8 here, so I'm gong to buy 1 and see If I can take it apart and use the backplane.

(Ok maybe I'll buy all four and see if I get lucky )



This project is really one for some company with proper manufacturing capability to design (ok hardly any design needed) a SATA power cable with led header. I might email Thermaltake or some one to suggest.

Hell, you could have a flashing lighty-uppy power connector if you had to