Testing SATA HDD Controllers on the Motherboard

[jdituro]

What is the best external software or plugin device to evaluate the heath of HDD contollers (SATA and IDE) on the motherboards.

[hddguy]

something that supports SMART...

[jdituro]

Not necessarily. I not all motherboards comply, or have chipsets that support S.M.A.R.T. I am looking for some external software tool or diagnostic device that I can attach and run to evlauate the condition of the SATA controllers

[Vulcan]

I note that you're talking about motherboard controllers and not disks - motherboards have minimal interaction with the SMART specification, but before getting into that:

What amount of confidence are you hoping to get from your test? Does it have to be 100%?

What level of detail are you hoping to test e.g. just basic electrical voltages? Or rise&fall times (IDE) / eye patterns (SATA) as well? Or detailed protocol tests? Or something else?

What limits do you have on testing time & device cost?

What have you already considered & eliminated (if anything)?

What can you tell us about the context e.g. Is this PC motherboard testing for consumers, or are you designing new motherboard PCBs from scratch, or ...?

Without answers to those questions to help beter understand your requirements & expectations, then IMHO the options are too wide...

[jdituro]

I note that you're talking about motherboard controllers and not disks - motherboards have minimal interaction with the SMART specification, but before getting into that:

What amount of confidence are you hoping to get from your test? Does it have to be 100%?

What level of detail are you hoping to test e.g. just basic electrical voltages? Or rise&fall times (IDE) / eye patterns (SATA) as well? Or detailed protocol tests? Or something else?

What limits do you have on testing time & device cost?

What have you already considered & eliminated (if anything)?

What can you tell us about the context e.g. Is this PC motherboard testing for consumers, or are you designing new motherboard PCBs from scratch, or ...?

Without answers to those questions to help beter understand your requirements & expectations, then IMHO the options are too wide...

Thanks for the quick response.
1. What amount of confidence are you hoping to get from your test? Does it have to be 100%?
I am IT repair Technician with a busy repair shop. I have been involved in PC repair for many years. Back in the day, I used PCI card driven diagnostics and also Hardware "burn in" software to stress the video RAM and HDD on new systems. Regarding confidence level; there would be two levels. The first for a quick evaluation and the second for a more in depth analysis.
2. What level of detail are you hoping to test e.g. just basic electrical voltages? Or rise&fall times (IDE) / eye patterns (SATA) as well? Or detailed protocol tests? Or something else?
The quick and answer would be the cost of equipment/software determining the level of sophistication. I am not a PC/HDD manufacturer but repair PC’s and laptops and am looking for some level of diagnosis and confidence that a SATA controller is functioning correctly, given the correct windows driver and stability of the OS.

[drc]

I am not a PC/HDD manufacturer but repair PC’s and laptops and am looking for some level of diagnosis and confidence that a SATA controller is functioning correctly, given the correct windows driver and stability of the OS.

Why would it not be? What percentage of the time (aside from known issues which are later corrected with software or firmware updates) would you estimate that PC problems are caused by a malfunctioning SATA controller?

[Vulcan]

@drc:

Agreed - we now know that the OP is fixing PC systems, and it's not a frequent type of hardware failure on the (admittedly non-PC) systems I work with.

@jdituro:

I didn't see answers to some of my questions, and I couldn't map the answer about "the cost of equipment/software determining the level of sophistication" to my question - I was hoping for a much more specific answer. So I'm going to reply based on my overall interpretation of what you said.

Several of the bits of equipment I use will likely be outside the cost/time limits of what you have now explained. I suspect that within the limits of a PC repair operation, a fairly effective way to test typical SATA controllers would be based around running disk tests, to a known-good disk. Most SATA controller hardware fault types which I can think of, and which you could detect in a reasonable time, would affect disk I/O typically causing either timeouts or interface CRC errors. Peronally I wouldn't use Windows for doing that testing, unless all other options (MHDD, HDAT2 etc.) had to be ruled-out.

Additional tests (e.g. loopback) are possible with at least some SATA controllers, but require specific programming for that chipset and in some cases need extra documentation from the chipset manufacturer to get the necessary details.

As drc has said, this doesn't seem like a common area for faults. What SATA controller faults have you seen, which you're trying to test for?

[jdituro]

I agree with you both that Motherboard SATA controller errors are rare, how if I run typical read and write tests on the HDD how can I determine if the failure is local to the controller on the HDD itself or that of the controller on the motherboard. I recently had a siduation were I found a delayed write error during an HDD evaluation and assumed it was the HDD. After replacing the HDD, OS etc. came to get the same errors with the new HDD. Rather than wasting time guessing can you reccomend a method that is more precise.

[Vulcan]

if I run typical read and write tests on the HDD how can I determine if the failure is local to the controller on the HDD itself or that of the controller on the motherboard

That's why I said an HDD used for testing should be known-good* - then the cause of any errors is highly unlikely to be the HDD. Also the specific type of failure will sometimes make the suspect component (more) obvious (though not in this case).

I recently had a siduation were I found a delayed write error during an HDD evaluation and assumed it was the HDD. After replacing the HDD, OS etc. came to get the same errors with the new HDD. Rather than wasting time guessing can you reccomend a method that is more precise.

As well as the point above, that's also why I said I wouldn't be using Windows for the testing, as you are affected by its limited error messages, and hidden driver behaviour (e.g. retries could hide marginal issues).

In your example, you should be looking in the Windows system event log, to see which specific driver(s) logged errors, and what that error was, which then led to the displayed messaged about delayed write error. Also, if you had seen hardware errors logged before that delayed write error, then you could have started with the hypothesis that the OS was unlikely to cause hardware error messages, and started considering other causes first, rather than reinstalling the OS... (Personally I've never seen a software cause for those specific Windows errors, but I rarely use Windows these days, so I don't claim it can never have an software cause - just that I've never seen that.)

I'm usually running *nix, where I can easily see nice, verbose kernel driver messages, which typically give a better clue that what I have seen in Windows error messages, in my experience. However it really can be difficult to identify the true cause of an interface problem, from any error messages alone. Properly interpreting error messages from any tool, and knowing what cannot be assumed or believed, can be an art.

There are also DOS-based disk I/O tools that I mentioned before (e.g. MHDD, HDAT2 etc.), although you may need to temporarily switch the SATA controller into IDE / legacy / compatibility mode (whichever it is called in each BIOS) to use that type of tool. You will likely get less ambiguous error messages from such tools (no drivers involved), for some types of problem.

If you had used a known-good* disk for testing, you wouldn't have started by guessing the disk was faulty. Or you could have moved the customer's disk into a test system which you run in your workshop, to see if the problem moves with the disk - in your case it would not have done that, and so again, you would be narrowing-down the cause.

(*Of course known-good disks can become faulty, but that's why you use a lab system to run tests on items like those, to confirm that there are no detectable problems, before then using them on customer systems.)

There are limitations & caveats to these types of techniques, and without lots more time to explain things, nor having the faulty hardware here to find out exactly what was wrong with it (additional test equipment might have helped, but is costly), I can just give some suggested approaches for you to consider.

[jdituro]

Thanks for your responses. I appreciate your responses and I currently employ the techniques you suggested. However it does not answer the question I am posing. Is there an external device or software that can determine whether a SATA controller error is local to the HHD or the SATA controller on the motherboard, without resorting to trial and error with known good equipment?

Using known good equipment is a flawed method because the tester is making assumptions that are not necessarily correct. Also it is time consuming, and in my my business time is money.
From your responses in this case, can I assume the such a device or software does not exist, except for the proprietary equipment and software used by the PC hardware manufacturers?

Thanks again for your time

[Vulcan]

I currently employ the techniques you suggested.

In that case, I don't see how you made the mis-diagnosis which you described

However it does not answer the question I am posing.

That's because IMHO your question is too poorly defined, to allow others to help you identify what equipment best suits your technical & commerical expectations (but you probably haven't seen the range of available test equipment to see what I mean). As I explained initially, in my experience it depends on which specific types of SATA controller test you want (different faults/problems need different types of test, and potentially need different equipment), but I didn't see a clear answer to that part of my question.

As always, a simple "pass" from a PC-Check type of diagnostic, does not mean that no fault exists. I work in the area of diagnosing the very-difficult-to-find faults, and that needs a range of different test equipment, but I can't understand your expectations well enough to be sure they are realistic, and which equipment (if any) will meet them. Hopefully someone else will understand your needs better than me. Good luck in your search.

[guru]

As we don't have much time. And the answer is not so easy without spending time ;o)

Try

http://www.elpeus.com/index.php?main_pa ... ts_id=1619

They have SAS Loopback Adapter Module. Give them a call and pick their brains, maybe they can find a SATA version.

I use Loopback when testing FC cards.

Good luck

[Vulcan]

@guru,

Thanks for adding your comments.

I use Loopback when testing FC cards.

Me too - indeed that's related to one of my patents. As you may know, not all test patterns are equally likely to expose problems, and correct interpretation of the LESB is critical.

Unfortunately loopbacks on any interface can help identify some, but not all, types of faults (mutter, mutter, old Infineon FC transceivers, mutter.. ) and I can't understand exactly what the OP's expectations are. The Finisar equipment I have used is good but costs mega $$$, hence my (unanswered) questions to the OP about his cost limitations. He would also need suitable software to enter loopbck mode on a specific SATA controller, and then send / receive / compare the data etc.

Anyway, hopefully your suggestions will help the OP more than mine

[guru]

Oh....I hope I don't need to pay you for your patent ;o) That would be mega face palm

[Vulcan]

That would be mega face palm

I could come and sit next to you, to do that double face palm, if needed (But no, I don't have an ST:TNG costume!)

[drc]

I can't understand exactly what the OP's expectations are

I think he was just wondering if a device existed that could be plugged into a SATA port on any mobo, left to sit for an hour, and then would give a readout of GOOD/BAD

[JWCC]

In my experience it's almost never the motherboard but the cable. Smaller wires give better signal quality (less diameter=less coupling) but they also break much easier!

Transferring one very big file and then verifying the checksum will tell you if the interface is functionally stable but it won't tell you anything about analog (RF/packets) or mechanical issues. You can be working just fine and then barely cross some line (say, a cable is cheap and starts to come loose because of plastic deforming) and suddenly you notice the problem a few weeks or months later.

[nikaragva]

I have problem with GA M68MT-S2P motherboard.
Sometimes windows freeze. In event viewer stay error for disk. Something wrong with driver or controller.
Then I check HDD with hdd sentinenel. Surface pass ok, but selft-test can't start:
http://www.dodaj.rs/?k/pl/3QCQqvyf/untitled.png

I check other cable, connect other HDD to this motherboar, always same error. HDD on other motherboards work ok, and self test pass ok.

Then I remove nVidia storage driver, and install microsoft default driver for sata device, and then self test can start and pass ok.

Problem with freezing windows is hard to reproduce, every 2,3 days. Sometimes more then one per day. There is no rule.

I want to return motherboard to seller, or service, but I want somehow find way to always reproduce that error, because they turn on pc, everythng is ok, and return back to me with same board:|

Maybe problem is in driver, but I was install newest driver from gigabyte site.

Now I regret that I bought motherboard with nVidia chipset for IDE.