New WD 3TB 2.5" Passport

[fzabkar]

I'm not sure why we are even arguing about this.

Maybe it's because you don't understand the importance of the concepts of "transmission line", "characteristic impedance", and "signal reflection".

Actually, I do.

It must be very satisfying to sit in front of your computer and make all kinds of theoretical pronouncements that have no basis in experience, and premature judgements that are off the mark. Maybe with a bit more presumption & arrogance, I could try it myself?

If you truly understand these concepts, then why are you defending the indefensible?

I see that you often lecture visitors on the benefits of professional services, which hopefully make use of professional tools and practices, yet here you are telling a professional audience that a bodgy hack is OK. I'm telling you how to do it properly ... with just a modicum of extra effort.

BTW, I have built plenty of jigs and cable harnesses to facilitate bench testing and repair of all kinds of equipment, so I know very well what is required.

[jono-ats]

In this thread, I've disclosed no techniques; shown no implementations; but simply warned against using lengths of single wires. And yet the undisclosed technique is a "bogey hack?"

I have to say one thing for you fzabkar -- at least you are consistent in your unfettered, condescending, and erroneous imagination.

[fzabkar]

In this thread, I've disclosed no techniques; shown no implementations; but simply warned against using lengths of single wires. And yet the undisclosed technique is a "bogey hack?"

I have to say one thing for you fzabkar -- at least you are consistent in your unfettered, condescending, and erroneous imagination.

My knowledge of communications engineering basically ends with what I learned at Uni. Nevertheless I know enough about the subject to see the problems with the method put forward in this thread. It's like extending a TV coax with two bits of wire. Sure, it may work, at least in some cases, but that doesn't make it right. I explained the problem and provided an easy solution which costs nothing, but the response was "we've been doing it this way for yonks -- we don't need some pedantic know-it-all to tell us experts what to do".

BTW, the fact that you didn't even consider that the word "bodgy" may be Australian slang says a lot about your US-centric perception of the universe. If there is arrogance or condescension in this thread, it isn't coming from me.

[jono-ats]

Wrong again.

I assumed bodgy was an Australian idiom and originally re-typed it that way. However, autocorrect got the last word and I didn't catch it.

My "US-centric perception of the universe?"

You're such a tool (sorry if you don't care for my "US-centric" slang).

[Doomer]

My knowledge of communications engineering basically ends with what I learned at Uni. Nevertheless I know enough about the subject to see the problems with the method put forward in this thread. It's like extending a TV coax with two bits of wire. Sure, it may work, at least in some cases, but that doesn't make it right. I explained the problem and provided an easy solution which costs nothing

You did not provide the solution and here is why
1. You assumed that only you were smart enough to think about it, so nobody tried. While in reality that's not the case
2. You didn't know that your solution would work, you just assumed that it would (as most of the times)
3. Based on your desire "to show them all" you assumed that your solution is the best one, which is way far from reality.

Now here are the problems with your solution
1. Soldering SATA cable to the drive's PCB requires you to connect the other end directly to a SATA controller, meaning that you would have to get inside a computer every time you need to connect a drive and that's not convenient
2. SATA cables use aluminum coated polymer tape wrapped around individual isolated pairs of wires to create proper shielding. Good luck soldering that tape to the ground (as you bravely suggested)
3. Individual wires are pretty thick and also have thick sturdy individual isolation. Soldering these wires to the PCB directly would be pretty painful task. Keeping them from popping off the PCB would be even more interesting task. You could try to glue the SATA cable to the PCB to keep it still but it's messy. You could a tape to wrap around the drive to keep the cable firmly attached to the PCB but then you wouldn't be able to take the PCB off the drive of you need to (and you do need it from time to time on WD drives)

Of course you would know all that if you actually tried all that.

[jono-ats]

Here is a photo of an adapter that I have actually used.

Small gauge wire-wrap wires are soldered to the SATA cable wires, to make it easy for connecting to PCB signal points. Wires are numbered (and color-coded) to reduce chance of error. You are welcome to copy it Fzabkar, if it will come in handy for the recoveries that you don't actually do.

Attachments

[Doomer]

Here is a solution that works for PebbleBeach

Attachments

[fzabkar]

Here is a photo of an adapter that I have actually used.

Small gauge wire-wrap wires are soldered to the SATA cable wires, to make it easy for connecting to PCB signal points. Wires are numbered (and color-coded) to reduce chance of error. You are welcome to copy it Fzabkar, if it will come in handy for the recoveries that you don't actually do. :-)

I believe I already have the patent on that technique. At least I've done it often enough. In any case I prefer your method to Doomer's.

[Doomer]

In any case I prefer your method to Doomer's.

So soldering thin wires to a SATA cable are better than soldering thin wires to a SATA connector.
That's the Australian way, got it.

[fzabkar]

In any case I prefer your method to Doomer's.

So soldering thin wires to a SATA cable are better than soldering thin wires to a SATA connector.
That's the Australian way, got it.

No, you don't get it. His solution is neat, colour coded, labelled, and he has at least attempted to understand the function of the ground wire. That said, I would have used all 3 grounds, but the central one (which separates Tx and Rx) would be the most important, AFAICT.

[jono-ats]

I still don't understand why we are arguing about this . . .

[wksk]

I'll just leave this right here....

Attachments

[Doomer]

That said, I would have used all 3 grounds, but the central one (which separates Tx and Rx) would be the most important, AFAICT.

Still waiting for you to try something before claiming that it's the best

[Doomer]

That's the Australian way, got it.

Scratch that, actually
The Australian way would be don't do anything but claim that it would be the best w/o even providing some kind of proof.
Some phony engineers are down under.

[HaQue]

"Some phony engineers are down under"

Guilty, your honour!

[fzabkar]

Here's how my USB-SATA/IDE bridge PCB does it:

My eyesight wouldn't be good enough, but ISTM that those people who solder monoliths would be able to do something similar, even if it requires an extra few mm of thin wire.

BTW the wavelength is 5cm at 6GHz and 10cm at 3GHZ, so there really isn't that much to work with, at least theoretically.

[jono-ats]

Nice soldering job, but how does it solve the problem of getting a SATA signal off of a hard drive's PCB -- especially when there are not corresponding signal paths for all of the cable leads?

[fzabkar]

The PCB has 4 capacitors. Tap into their pads. It shouldn't matter that they're on the underside of PCB because the cable length will not be a problem. I would think that the motor and HDA connectors would still be making good contact. The grounds can go anywhere.

In retrospect my first post in this thread was too obscure. The point that I had intended to make was that the length of a cable is not so much of a problem as the quality of the work. A SATA cable is a 100 ohm transmission line, as are the SATA traces on the PCB. When you splice bits of wire in between, you introduce attenuation and reflections. These result in a weaker and noisier signal.

PCB designers use a rule of thumb which states that when a PCB trace exceeds a certain length (typically 1/10th of a wavelength), the trace is to be treated as a transmission line. At 6GHz the wavelength is 5cm, so 5mm becomes their cutoff point. When I first saw the USB-SATA conversion solution, I was surprised that it worked, but I just filed it away as one of the many thing that I don't understand. I would think that in the event of instability a 6Gbps SATA link would downshift to 3Gbps and then to 1.5Gbps, so this might explain why the trick works as well as it does.

ISTM that most people don't understand the concepts involved and simply treat the job as regular point-to-point wiring.

[Doomer]

The PCB has 4 capacitors. Tap into their pads.

How would you put PCB back on the drive?

[fzabkar]

Obviously that would depend on the drive, but ISTR one post where someone was able to fit a SATA PCB to a USB drive even though there was interference with the casting. If it's not possible, then just use the test points on the top side. The test points in each pair are in close proximity to each other. It doesn't matter if the two Tx test points are not adjacent to the two Rx test points. Just keep each pair of conductors the same length as much as possible.

The point is that there are ways to get close to the optimal solution with minimal effort and cost.