Are SATA cables available with gold-plated contacts ?

I once wired up a server box with 128 SATA cables while watching the bit error rates. Many of them were bad. Worst was a Maplin cable which had so much crosstalk the receiver could lock onto the signal leaked from the transmitter despite the other end waving loose in the air.

They're also quite susceptible to bending - if there's any sideways force on the socket it can cause the contacts not to mate properly. SATA isn't a very good standard.

I wouldn't bother, I'd just replace the cables. I bought my selection from Farnell where there are various brands like 3M, but not sure who I would recommend. Startech stuff is usually good, or you could look for genuine Dell/HP/Lenovo/Supermicro etc server parts on ebay I suppose.

Theo

I would have thought that all reputable SATA cables would have gold contacts. But you have in fact described the problem. Even gold contacts can acquire dust, which physically prevents contact. It's not corrosion.

I used to deal with a particular digital video effects machine which had a large PCB plugged in on two edge connectors. The powerful fans to keep the thing cool drew in lots of dust. You could either leave the sponge filters over the fans, and wash them every few months, or take out the filters and clean off these edge connectors (and much else) every few months.

You have the same choice: you either improve the dust filtering, if any, or clean it out regularly. If the computer sits on the floor, just raising it a few inches will probably reduce the amount of dust pulled in.

For computer electricals, we try not to mix metal. Gold goes with gold (glides). Tin goes with tin (bites).

The SATA connectors on the motherboard are gold plated 10u (telecom plating is 50u). This means when we buy SATA cables (red or black ones, all 6Gbit capable), the connectors on those are also gold. Gold is a precious metal. 10u is thin enough, there can be "pin holes" in the finish, but this does not necessarily mean ruination. It should still work.

There are two retainers on SATA. The original retainer method is "compression". The body of the connector is shaped, when molded, such that it "presses on" its mate. This helps prevent it from walking out, thermally.

A second method was added as an afterthought. SATA cables have a "jaw" and a release surface you press on to stop the "jaw" from biting its mate. Unfortunately, on WD hard drives, the "jaw" has no where to bite, as WD drives, the WD Corporation invented their own flavour of SATA connector, and it is an interference fit which is not the same as the rest of the industry, The cable with "jaws" on both ends, it bites the motherboard (and holds well), but if there is a WD brand hard drive, only the "compression" feature holds the cable. It still works, but is annoying (it "feels wrong" while you work on it).

The signaling on SATA, is high speed low amplitude differential, where D+ and D- are carried on two coax cables. The dielectric might be polystyrene. When the cable comes out of the cable machine, it was straight at one time. The coax conductors, the plastic insulation was a constant diameter. The impedances matched.

At the motherboard factory, idiots take elastic band, and they fold up the SATA cable below its allowed bend radius. If you "kink" a SATA cable, it pinches the polystyrene-like material and changes the dielectric impedance effect. The two signals now don't go through identical environments. At the receiving end (after some loss through the cable), there is a measurable BER (bit error rate) due to the imperfect transmission environment. There is a counter in SMART on the disk drive, and it counts transmission errors and the counter does not reset. You can check the counter, and for any reasonable computer install, the count could quite well be zero. But a damaged cable could leave a permanent count of billions of errors, until the SATA cable is swapped out. You record the current value on the error counter, record the value tomorrow, take the difference, to determine whether any errors accumulated that day.

There are error counters at both ends of the cable, for the signals in either direction (7 contacts, three GND, two pairs of signals, one pair for transmit, one pair for receive). The SMART interface makes it easy for users to find the error count in one direction. I'm not sure where the motherboard count is logged.

Nobody pays attention to the *power* on hard drives!

The ideal case on computers, is two looms with 15 pin SATA power come out of the PSU. The user carefully runs this power cable past the drives. (One loom maybe, for the optical drive, the other loom for a couple hard drives). Because the cable comes right from the PSU, the voltage for the motor is an ideal 12.0VDC with little loss.

Older drives get in trouble, if the received supply voltage is

If you use around four power extension cables during your build, and the HDD is on the end of this chain of connectors, the voltage can be low enough to be around 11.0V .

So far I described an ideal situation, where with Hot Plug enabled, the OS recovers after a short outage.

But another kind of failure on low voltage, is the CPU on the HDD controller board "crashes". Not even a warm PC reboot recovers the HDD. This is because, while the IDE ribbon drives have a very nice RESET signal on the cable, there is no RESET on the seven connection SATA data cable. You have to power cycle the PC, to get the HDD CPU to reboot itself.

Summary: While you could have a badly kinked data cable, it's just as likely your power cabling needs sorting so that the drive remains sane and properly powered.

The jaw equipped cables, the one disadvantage, is on an Asrock first gen SATA board, you can pull the SATA motherboard data connector right out of the motherboard, with a jaw cable :-) a lot of later designs are swaged and the connector cannot leave (ripped solder connections) quite as easily.

SATA cable ends come in straight, left-angle, right-angle. On home computers, usually you buy a straight on one end, angled on the other. I have a few cables which are left-angle, and the cable tail points upwards when it is plugged into a right-side-up HDD. Most of the motherboard box cables are right-angle, and they point downwards when the angled bit is used on the HDD end. The motherboard end may be good enough that the straight end will work on that side. Think carefully about what kind of cable you want -- the staff at the shop will point out "the cable everybody buys", as well as cables that might be needed for best fit on obscure Dells (the upside-down one would work there).

On my Test Machine, the cables have well surpassed their service life spec. I plug them in. They still work. The retention isn't great, but, I'm not complaining. There is no particular reason to be suspicious of them. As long as the data cable is not pinch-kinked, the power chain is not too long, both HDD and SSD SATA III, they both work fine.

For a 24TB Helium drive, do not use a five wire power cable. Only use a four wire cable. You can extend the PSU loom with a Molex to SATA six inch extender for example, which uses four wires. The reason for this requirement, is the 3.3V rail wire, if left connected, it tells the Helium drive to "not spin up". Your brand new 24TB drive then looks kinda dead, and it's just the spin control signal taking liberties with your good times. When no 3.3V wire is on the cable, the expensive 24TB will spin up. Such drives can take 20 seconds to spin up and self test, so be patient.

Paul

Once a cable is plugged into a SATA connector, it is pretty enclosed, so dust ought not be a problem in that case. (it could be an issue using a port that has sat unused for years).

Yup a proper contact cleaner may help. However I would also ditch the exiting SATA lead and try a new one to eliminate the cable from the equation.

Not as a general rule - but prolonged higher temperatures tend to lower the overall life expectancy of IT kit.

It's a mini tower PC that sits on my 'desk' (ex-Boots fire door, rescued from a skip, reduced in width to 70cm and fitted between the two walls of my 3rd bedroom, sitting on battens), and with the side panel removed to keep it cool. Also has an external desk fan blowing once the room temp exceeds 24C to keep it from getting too hot. QED dust is a problem.

Actually I pulled out the two DDR3 memory sticks and cleaned the contacts with methylated spirits, and carefully cleaned the M/B sockets then re-insterted them and the problem has not re-occurred. Hopefully that was the issue and its fixed.

I only have two blue ones, currently in use. They have locking clips as well. The Novatech M/B upgrade package came with a red sata cable, but that has no retainer clips and every so often I had to give it a wiggle on the WD HD to avoid the current issue.

See my reply above, re removing and reseating the DDR3 sticks.

The Western Digital cable (covered version), I don't know if they still make it, but it relies on a drive having "dual power". If the drive has a Molex on the corner, plus the SATA 15p position, then the strengthened WD cable can be used for the data portion. When it is one-piece like that, the blocks on either end, help.

There are two versions of the cable, and on modern drives, the one that leaves the 15p exposed would be used. But that would also make the scheme a little less stable.

One solution for computers in dusty environments, is to turn the PC into a "convection cooled" unit. Zalman used to make a couple very expensive cases, where heat pipes were used to conduct heat from GPU and CPU, into the outer skin of the PC, and then convection air over the fins would cool the machine. But there could still be local heating with things like VCore, that the casing heatpipe method does not address. While in theory it could work up to about 400W, practically speaking half that value would be a better application of it.

Computer cases, some come with dust filters, but then the fans have to work harder to get airflow through the filter. And the filter has to be cleaned every three months (and sooner or later, you are going to forget to do that).

There are also industrial enclosures with fins, but you'd have to examine the designs fairly carefully, as sometimes they forget to cool certain things. As an example, one Asus motherboard, a small eight pin regulator chip (a switcher of some sort), the surface temperature of the chip was 100C. As detected by an IR camera. You can fix that with a little Arctic epoxy thermal component (permanent!) and glue a RAMsink to the top of the tiny chip to cool it.

Paul

Thanks Paul

I think removing the DDR3 sticks and cleaning their contacts and the sockets with methylated spirits seems to have done the trick. Maybe this was the problem all along, mind you it is now a lot cooler than when the problem first started occuring. We have had three heatwaves this Summer in the UK and that might have been part of the problem.

Andrew

The DIMMs have relatively high insertion force.

It is gold-on-gold, so there is no oxide to cut through.

While you could foul a DIMM slot if you put your mind to it, there wouldn't normally be any chemicals present laying on top of the gold.

One way that motherboards get fouled, is in the washing machine. They wash off the solder flux after the soldering process. This used to be done with trichloroethlyene ("trike"), as if in a dry cleaning plant. Today, the pollutants are water soluble on purpose, so a chlorinated hydrocarbon is no longer required. They could use water, or water and a bit of some alcohol.

Well, when you put a board in the washing machine, you have to install "plugs" on the open connectors. For example, an RJ45 has a plug inserted with a rubber bung to seal.

I got one motherboard, the board with the Marvel test feature (can test Ethernet impedances), and it was showing open on one pin when I tested the feature. By inserting the Ethernet cable and removing it five times (gold-on-gold connector), the eighth pin started making contact and the Eth would work properly.

And that happens, because the "dirty" wash water got to the pins and dried onto the surface.

The DIMM sockets should also be capped during the washing cycle.

But you would have had a problem during the build, if a washing error had happened like that.

At least, you know that if it was a wash problem, water or water+alcohol(isopropyl) could cut through a dirty wash water problem.

Paul

Indeed. You should check your CPU fans and remove any dust bunnies.

I thought that "no clean" flux was almost universally used now, so no washing is required.

John

That was my P4C800e Deluxe motherboard. Which is quite a few years old (and has passed away). That is the one that had the washing dirt.

I've not seen that sort of problem on any board after that point in time, so I don't know what they're using today.

The companies do things, that they would swear in the past they would never do. You can't use a trend, to predict how they will do something. For example, at one time, they would not use screws on anything (because of the assembly labour that would cost them). But they are putting Wifi on retail motherboards now, and those are screw mounted. It's hard to say whether a robot drives the screws, or a thousand ladies sitting at tables do that :-) They used to make 3.5 million motherboards per month, in some of those companies.

There have been some products, where double IR reflow might have been used, followed by hand soldering of leaded components. Activities you would assume are prohibitively expensive.

Paul

When a company I worked with was screwing transformers together they had overhead suspended air screwdrivers with magnetic tips - put a screw in the driver, a nut underneath and whoosh. In and tight to the correct torque in about 2 seconds.

Not necessarily. A UK contract manufacturer (Newbury Electronics) uses robotic soldering irons.

They also have what's effectively like a pinpoint wave soldering machine, enough to solder one pin on a through hole connector but avoid the next: