Author Topic: Beware of Aksa ProSlim SATA cables! (Scope screenshots)  (Read 7039 times)

0 Members and 1 Guest are viewing this topic.

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« on: September 02, 2018, 02:13:36 pm »
The cables in question are these:
Aksa ProSlim http://www.akasa.com.tw/search.php?seed=AK-CBSA05-50BK#
My particular part number: AK-CBSA05-50BL
Similar ones: AK-CBSA05-50BK , AK-CBSA05-30 , AK-CBSA05-15
Photo of what is inside the cables: https://www.eevblog.com/forum/chat/beware-of-aksa-proslim-sata-cables!-(scope-screenshots)/msg1798562/#msg1798562

So i have been doing a NAS server build. Its quite a compact one totaling about 12 liters in volume but packing 4 hotswappable drive bays, full i5 quad core x86 PC and cooled by a single fan using a 3D printed airflow guide and a slightly modified heatpipe CPU cooler.





Because space is at a premium here i bought SATA cables that have particularly compact connectors on the ends because i need to do the bend quickly and 90 degree ones would get in the way due to the ports being to close on the backplane. I already desoldered the molex power connectors from the backplane due to clearance to the large CPU heatsink. And to keep things neat and tidy i bound the cables together and attached them down with a screw. Looks pretty tidy if you ask me. These fancy low profile cables cost more than the standard ones too.





Everything works nicely and im messing around with unRAID to run it all, but i start to notice some odd behavior after a while. At one point i thought i corrupted my installation so i put a fresh factory image on it and it seamed to be fine for a while. However once i added a extra parity disk to the RAID array things started getting weird again. During the array rebuild i kept getting errors, eventually they became really reproducible errors. The drives seamed to be disappearing all of a sudden and needed a power cycle before the PC would detect them again (reboot was not enough, the drives themselves needed a power cycle).

With some testing it seamed to be that this only happened reproducably when i was simultaneously writing and reading to all 4 drives in the array. I could trash the commonly failing drive as much as i wanted and it would never throw errors. Yet rebuilding 2 drives in the array where 2 are reading and 2 are writing at max speed seamed to make 1 or 2 drives fail within minutes. Then i moved the drives outside of the drive cage and used different cables and it ran fine for an hour before i gave up and called it stable. Hmm....

At this point im starting to look at my nice and tidy SATA cabling job from another perspective. What if im causing crosstalk by routing the cables so close together? I google some images of cut up SATA cables and sure enough the TX and RX pair in the cable are shielded pretty nicely, surely enough to keep such a crosstalk problem from happening. Well i do have a fast scope and differential probes so lets actually measure the crosstalk. Here is the test jig i put together where i soldered down a SATA connector and attached a active differential probe to one of the pairs.





The other end of the cable was going into the 3rd drive bay with a drive preset in it so that the end is terminated. And when the PC is off we get a nice low noise floor as our baseline.





Okay now we actually let the PC boot up so that the SATA lines start moving and... Oh DEAR LORD!!! When the drives are working and communicating there is 250 mV peak to peak of noise!!! :o :o :o

I have tried to disconnect the other end of the cable in case the drive was sending something back or this was noise picked up in the hotswap backplane, nope still same thing. Its only when i started pulling out drives that the noise started going down with every drive removed. This is really looking like a horrible case of crosstalk.





Okay now maybe i am picking up some common mode noise or something. These diff probes do have pretty amazing common mode rejection but they are not perfect after all. So i grab some standard SATA cables i have laying around and see what it does. I tried plugging the other end into a powered up drive, into the motherboard, into the backplane, ran it parallel to other sata cables to try and get crosstalk. In all cases it looked like the flowing photo. I was barely seeing any noise at all.




CONCLUSION:
Its pretty clear that these fancy Aksa brand cables have NO SHILEINDG AT ALL. Infact i am guessing this is simply a FFC cable wrapped in some extra insulation as i have noticed these cables are also significantly more flexible than normal SATA cables.

Do NOT buy these cables.
« Last Edit: September 04, 2018, 03:52:07 pm by Berni »
 
The following users thanked this post: tom66

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5632
  • Country: au
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #1 on: September 04, 2018, 05:06:08 am »
Could you not just use standard right-angled SATA cables?

 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #2 on: September 04, 2018, 05:43:20 am »
Could you not just use standard right-angled SATA cables?


I did try some that i had laying around but the ports on the back plane are so close together that they interfere with each other. You can probably get it in with enough brute force but from my experience SATA connectors are pretty fragile to twisting forces. Regular non angled SATA connectors still fit if you give the cable a tight radius bend and they don't put sideways force on the connector.

The whole thing is mostly my fault for putting such a massive CPU heatsink in there (This case is really not designed for it). But i do get excellent CPU temperatures with the fan running nice and slow to stay quiet despite being a fairly high performance desktop chip.
« Last Edit: September 04, 2018, 05:46:00 am by Berni »
 

Offline stevelup

  • Regular Contributor
  • *
  • Posts: 184
  • Country: gb
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #3 on: September 04, 2018, 07:06:08 am »
I suspect if you repute them side by side instead of tightly bundled into a stack, this problem may well go away. Won’t look as pretty though...
 

Offline Halcyon

  • Global Moderator
  • *****
  • Posts: 5632
  • Country: au
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #4 on: September 04, 2018, 07:17:31 am »
I suspect if you repute them side by side instead of tightly bundled into a stack, this problem may well go away. Won’t look as pretty though...

There should be no reason why you couldn't bunch them together (with decent cable). This is quite commons with SAS/SATA cables inside servers and workstations, of course they are made of decent quality cable (which are usually quite stiff).
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8240
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #5 on: September 04, 2018, 08:12:52 am »
Try adding shielding foil yourself and see if that makes a difference.

This also reminds me that coiling SATA cables used to be a thing in PC building --- I wonder how much that affects the signal integrity.
 

Offline Zucca

  • Supporter
  • ****
  • Posts: 4298
  • Country: it
  • EE meid in Itali
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #6 on: September 04, 2018, 08:18:47 am »
Since you will trash the  Aksa brand cables anyway time for a surgery....  >:D
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 
The following users thanked this post: Mr. Scram

Offline timgiles

  • Regular Contributor
  • *
  • Posts: 236
  • Country: se
  • Programmer, DB architect
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #7 on: September 04, 2018, 08:49:56 am »
I always thought akasa made decent quality cables. Bit shocked you are having these problems. Its right that servers, especially those prebuilt, have their cables bunched up and tied together.

I might have a couple of spare cables if you want some?
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #8 on: September 04, 2018, 08:59:41 am »
Try adding shielding foil yourself and see if that makes a difference.

Proper SATA cable have individual shield for each pair. Anyway OP confirmed that particular cables are garbage, there's no point to invest more time.
 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #9 on: September 04, 2018, 10:37:17 am »
I do plan to cut up one of these to see what is inside. For a moment i thought about wrapping them in some copper tape, but have decided that its better to just replace them entirely with a proper shielded cable. When using the proper cables i could measure pretty much no crosstalk so that's how it should be. This is a server after all.

There are enough random SATA cables laying around here to get it connected back up, but i have decided to do it properly and ordered a bunch so that all the cables are identical in shape and color (I know it doesn't matter but i want it to look neat).

I also thought of Akasa as a known good brand for computer parts, but from now on i will be more careful.
 

Offline Mr. Scram

  • Super Contributor
  • ***
  • Posts: 9810
  • Country: 00
  • Display aficionado
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #10 on: September 04, 2018, 11:45:09 am »
Since you will trash the  Aksa brand cables anyway time for a surgery....  >:D
Don't turn it on, take it apart!
 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #11 on: September 04, 2018, 03:50:40 pm »
Here is the photo then.

It turns out they have simply used oldschool 1.27mm pitch ribbon cable as was used to connect floppy drives back in the day. For comparison i have included such ribbon cable in the photo, i think its a perfect match. :--

So yes these are going into the trash where they belong.

EDIT: Actually i think i could probably protocol decode the SATA traffic if i left my scope hooked up and only connected 1 sata drive in that bundled pair. The SATA interface is AC coupled and you only need to look at one pair since one is RX and other is TX. I had plenty of signal size above the noise floor of my scope and i can probably correct the cables crappy response using the scopes de-embeding feature.

« Last Edit: September 04, 2018, 04:05:52 pm by Berni »
 
The following users thanked this post: Zucca, ogden, The Soulman, Mr. Scram, ttelectronic

Offline Monkeh

  • Super Contributor
  • ***
  • Posts: 7990
  • Country: gb
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #12 on: September 04, 2018, 04:07:42 pm »
I always thought akasa made decent quality cables.

Akasa don't make decent quality anything.
 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #13 on: September 04, 2018, 04:21:21 pm »
Also i have taken apart the connector to see how that works.

Unsurprisingly its again the exact same thing as on ribbon cables.
 

Offline filssavi

  • Frequent Contributor
  • **
  • Posts: 433
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #14 on: September 04, 2018, 04:34:16 pm »
I’m really surprised...

Not at the existence of such shitty cable, but that they worked at all, SATA cables are supposed to be parallel micro-coax lines
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #15 on: September 04, 2018, 07:10:25 pm »
Also i have taken apart the connector to see how that works.

I do not see way this crap can pass SATA interoperability tests meaning it does not qualify for "SATA cable" name. In my opinion it's scam, they know what (kind of crap) they are doing.
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 6278
  • Country: ca
  • Non-expert
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #16 on: September 04, 2018, 07:41:40 pm »
If anyone else was curious, here is a proper SATA cable, foil shield and internal ground wires.
Profile -> Modify profile -> Look and Layout ->  Don't show users' signatures
 
The following users thanked this post: Zucca, timgiles

Offline tom66

  • Super Contributor
  • ***
  • Posts: 6680
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #17 on: September 04, 2018, 07:48:01 pm »
What a load of crap!
Bad form from Akasa.
 

Offline Kevman

  • Regular Contributor
  • *
  • Posts: 146
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #18 on: September 04, 2018, 07:58:17 pm »
The picture doesn't make it clear, but SATA is actually double Twinax, one for each direction.  :-+

https://en.wikipedia.org/wiki/Twinaxial_cabling#/media/File:SATA3-TwinAxCable.jpg

Silverstone has ridiculously thin SATA cables here:

https://www.silverstonetek.com/product.php?pid=445

I haven't used them, but Silverstone has a good reputation and would not cheap out like that. The reviews are good, too.

I wonder Akasa literally just took ribbon cable and ran it through a PVC extruder setup for normal SATA?
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #19 on: September 04, 2018, 08:56:06 pm »
Silverstone has ridiculously thin SATA cables here:

https://www.silverstonetek.com/product.php?pid=445

It's for occasions where you need translucent side cover for your PC, not to mention finger-spreader :D
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8240
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #20 on: September 05, 2018, 12:58:13 am »
Try adding shielding foil yourself and see if that makes a difference.

Proper SATA cable have individual shield for each pair. Anyway OP confirmed that particular cables are garbage, there's no point to invest more time.
He shows the problem is crosstalk between cables, so I suggested a fix for that.

I’m really surprised...

Not at the existence of such shitty cable, but that they worked at all, SATA cables are supposed to be parallel micro-coax lines
That reminds me of a saying "in theory, there is no difference between theory and practice. In practice, there is."

In practice, SATA transmitters/receivers have a lot of pre-emphasis/equalisation and link training to compensate for bad cabling and noise, and don't forget that the signals run on a PCB aren't going to be shielded either. This isn't microwave voodoo territory.
 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #21 on: September 05, 2018, 05:46:27 am »
If anyone else was curious, here is a proper SATA cable, foil shield and internal ground wires.

Yep this is exactly what i was expecting to see in my SATA cable. individually shielded diff pairs. Boy was i in for a surprise


The picture doesn't make it clear, but SATA is actually double Twinax, one for each direction.  :-+
https://en.wikipedia.org/wiki/Twinaxial_cabling#/media/File:SATA3-TwinAxCable.jpg
Silverstone has ridiculously thin SATA cables here:
https://www.silverstonetek.com/product.php?pid=445
I haven't used them, but Silverstone has a good reputation and would not cheap out like that. The reviews are good, too.
I wonder Akasa literally just took ribbon cable and ran it through a PVC extruder setup for normal SATA?

I think they indeed simply ran normal ribbon cable trough a cable extruder to give it the blue outer layer (The dimensions look identical and insulation material feels similar to cut). From what i see the outer layer serves no purpose since the connectors grab normal ribbon cable just the same. This means the only reason they put the outer blue layer on the cable is to hide the fact that its a ribbon cable inside and make it look more like a SATA cable instead.

Those silverstone ones look nice. It seams they simply skipped the outer molding step of the cable and just gave you the 2 diff pairs bare. You can even see the shielding on them so nothing wrong with those. I could see them being useful in space constrained builds.


I’m really surprised...

Not at the existence of such shitty cable, but that they worked at all, SATA cables are supposed to be parallel micro-coax lines
That reminds me of a saying "in theory, there is no difference between theory and practice. In practice, there is."

In practice, SATA transmitters/receivers have a lot of pre-emphasis/equalisation and link training to compensate for bad cabling and noise, and don't forget that the signals run on a PCB aren't going to be shielded either. This isn't microwave voodoo territory.

Yes SATA is designed to be reasonably robust over a long cable due to those eSATA external drives (Never seen one in person so far). But this is not an excuse to use crappy cables, sure it works, but does it work reliably?

I ran high speed diff signaling over improper cables during prototyping of some stuff (like LVDS, MIPI, HDMI) and on short runs of cable they can survive surprisingly bad cables as long as the pairs are run close, but will often result in wonky operation if you fiddle with it. Like it might become glitchy if you move the cables around, attach probes or even just touch the outer insulation on the cable.

As for PCB traces they do sort of have shielding, but its only from one side. The common way of running really high speed things over a PCB is to use a 4+ layer PCB and have the internal layer be a ground plane, this places the ground really close to the pair (usually ~0.1mm) and this gives it good coupling to it in order to keep the electromagnetic field trapped down in the pair and provides a good close ground return path. Additionally the proximity of the ground plane helps determine the characteristic impedance of the line(for SATA its 100 Ohm). It also helps the impedance math to work out in a way that lets you place the diff pairs really close together and keep the right impedance, this reduces the crosssectional area between the two traces so its more difficult for it to radiate out energy and makes it more difficult for outside RF energy to create a differential signal on the line. There is a lot behind all this high speed signal integrity stuff.

I focused on the crosstalk because obviously 250mV of noise is NOT okay (keep in mind signals on SATA are under 1V). I could go characterize the high frequency characteristics of the cable and do eye diagram measurements and such but this is a frigin ribbon cable. We all know this stuff was never designed for high speed (Have a look at what poor speeds IDE hard drives can muster trough these cables despite using a wide parallel bus).
 

Offline analogo

  • Regular Contributor
  • *
  • Posts: 85
  • Country: at
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #22 on: September 05, 2018, 06:22:09 am »
So i have been doing a NAS server build. Its quite a compact one totaling about 12 liters in volume but packing 4 hotswappable drive bays, full i5 quad core x86 PC and cooled by a single fan using a 3D printed airflow guide and a slightly modified heatpipe CPU cooler.

By the way, which case are you using?
 

Online BerniTopic starter

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #23 on: September 05, 2018, 06:41:36 am »

By the way, which case are you using?

It is a CFI A-7879 V2:
https://www.mini-itx.com/store/~CFI-A7879

Quite well made case with its fairly thick steel construction, an average person stand on top of any of the sides without it bending in. That is apart from the front plastic face that flows typical plastic standards as most computer cases, but it does look pretty "servery" with its design. And do make sure you get the V2 version of the case because the V1 has some things riveted down and that makes building in it more annoying as it can't be completely disassembled for easy access.

The CPU cooler is a Scythe Katana 3:
http://www.scythe-eu.com/en/products/cpu-cooler/katana-3-cpu-cooler.html

You are supposed to use one of those low profile coolers in this case because the drive bays interfere with the space above the CPU. But that was not good enough for me so i bought this cooler due to its sideways offset fin stack. I bolted the whole fin stack between two wood boards to get a good grab on it and used a big wrench to bend the heatpipes back so that the fin stack is standing vertical rather then angled as it is on the original cooler. All the space inside the body of that black 3D print is the CPU cooler fin stack.

« Last Edit: September 05, 2018, 06:45:11 am by Berni »
 
The following users thanked this post: analogo

Offline filssavi

  • Frequent Contributor
  • **
  • Posts: 433
Re: Beware of Aksa ProSlim SATA cables! (Scope screenshots)
« Reply #24 on: September 05, 2018, 07:21:49 am »
That reminds me of a saying "in theory, there is no difference between theory and practice. In practice, there is."

In practice, SATA transmitters/receivers have a lot of pre-emphasis/equalisation and link training to compensate for bad cabling and noise, and don't forget that the signals run on a PCB aren't going to be shielded either. This isn't microwave voodoo territory.

Pre-emphasis and equalisation surely help but they are not a substitute for a correct impedance matching, after all they can’t do miracles,

What I would guess is happening here is that the link is negotiated down and working at lower speed

It is not a problem of shielding but of correct impedance, on pcb the signals have the correct impedance so that is not a problem

Probably that link is already on the brink of not working due to the cable alone, the extra stress induced by the parallel wires is what pushes it over the edge
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf