Jump to content
EwinRacing Flash Series Gaming Chairs
Tech Inferno Fan

US$189 AKiTiO Thunder2 PCIe Box (16Gbps-TB2)

Recommended Posts

Wow. I came into conclusion after playing with a multimeter that the 4pin yellow/black cable was compulsory. Would you like to describe (maybe with a picture) how you did it?

I will post my picture after work! sorry for the late-ish reply.

  • Thumbs Up 2

Share this post


Link to post
Share on other sites
goalque: Is the 280X stable under Mac OS X though?

Well, I have focused on the Win8.1 instability - haven’t much tested it with the AKiTiO under OS X yet, but tomorrow I can run Valley benchmarking for an hour for example, let’s see how it goes.

Share this post


Link to post
Share on other sites

I soldered wires right to the PCB so I could connect a PSU right to the PCIE power lines going into GPU.

I have been unable to tell a difference, certainly no night& day change.

Share this post


Link to post
Share on other sites

Hi all!

first of all thanks for the posts and the articles in this forum, they have been EXTREMELY helpful so far!

I am trying my first experience with an eGpu.

I've followed the guide here

My setup is:

- Macbook Pro 2011 13", i7, HD3000

- OS Yosemite

- Windows 8.1

- Akitio Thunder2

- Asus GTX660Ti

- EVGA 500W PSU

- pCIe X16 riser with molex

- Start option (jumper) for the PSU to start as soon as switched from the back switch

- Thunderbolt cable provided with Akitio

- DVI cable to external Benq monitor

I am using only the EVGA PSU and powering the Akitio through the molex

I have installed all the hardware following the guide.

Nothing connected to the Macbook so far.

I have made a clean installation of Windows 8.1 from the DVD using the superdrive and Boot Camp. Boot Camp did the hard drive partition and started the installation.

I have installed the Boot Camp tools/drivers.

I have then switched OFF my macbook

Connected the Thunderbolt to it and powered the EVGA PSU: The green light of the Akitio is ON, the GTX660Ti fans start to run, low speed.

I switch ON the MacBook pressing the Option button, when the macbook is ON, the blue led on the Akitio switches ON as well

Select my BOOTCAMP partition to start

The external monitor stays OFF, but also the internal switches OFF. The computer seems frozen, doesn't go anywhere and it doesn't respond to anything.

I have to switch it OFF by holding the power button for 5 seconds

A picture of the hardware setup

post-33016-14494999371031_thumb.jpg

I try then to leave the Thunderbolt disconnected and connect it only when the Windows OS is going to start

Windows starts but the GPU is not recognized.

Same happens if I power the EVGA PSU only when Windows is going to start with the Thunderbolt already connected.

I've looked on the internet and found someone doing this:

starting the computer with the EVGA PSU ON and the Thunderbolt connected, but the 4 wires (2 yellow and 2 black) cable on the Akitio connecting the Thunderbolt board with the pCIe board disconnected.

A picture of the cable:

post-33016-14494999373383_thumb.jpg

When starting the computer with the cable disconnected the fans of the GTX660Ti run at full speed.

In a very rapid way, I click on the BOOTCAMP icon to start Windows and at the same time connect the 4 wires yellow/black cable back where it should be. When the cable is connected, the fans slow down to normal speed.

Magically Windows starts and my GPU is recognized, I can install the drivers, the external monitor works.

post-33016-1449499937408_thumb.jpg

Even after installing the drivers etc, everything written before doesn't change, the only way it works is by the "4 wires cable" trick on startup, otherwise it either freezes or starts without GTX660Ti.

I tried also to continue the project by installing the Optimus procedure HERE, to see if it was going to do any good and resolving the above.

I get the EFI NVIDIA icon on startup, all good.

post-33016-14494999371795_thumb.jpg

I click on the EFI Boot to start, I get the console

post-33016-14494999372677_thumb.jpg

No matter if I do the procedure ("ls" etc etc) to declare the EFI partition, every time I try to start from EFI Boot it opens the console

I try again booting from Windows, conditions are still the same as above: it starts if nothing connected, but doesn't recognize the GTX660Ti, it works if you do the "4 wires cable" trick and it doesn't start at all if you try starting with everything powered and connected.

By the way, if I start my computer in MacOS, on "About" I can see the Akitio connected, but not the GTX660Ti.

I tried also the guide to make it work on MacOS, but it didn't help. So I would focus on Windows, don't need the GTX660Ti on MacOS anyway.

This is all I've done so far (sorry for the long description... but it could help avoiding many posts to explain)

Any help guys to make it work in a decent way and maybe also eventually get Optimus to work?

Cheers

Alex

Share this post


Link to post
Share on other sites
What happens if you boot from the "EFI Boot" Partition?

Hi Dschijn,

if I try to boot from "EFI Boot" it goes straight into the Grub 2 console (as in picture). I tried the procedure on the guide to declare the EFI partition on GRUB, but it still just goes on the console instead of starting Windows...

Share this post


Link to post
Share on other sites

The good news first. OS X ran Valley benchmark test (full screen mode) fluently that one hour. Displays turned off, so I went to energy saver settings and prevented sleeping. Energy meter showed ~180-190W. The test was done by a Mid 2014 Mac Book Pro (Iris Pro only) + Asus R9 280X + AKiTiO with a powered riser + 450W ATX PSU, Samsung 28” UHD monitor via DP 1.2, using 3840x2160@60Hz resolution.

The internal retina display is also accelerated when you locate the menu bar to the external screen. But now the bad news…. When I started to play a YouTube video on Safari, it seemed all good but suddenly the video and mouse got sluggish and finally froze totally.

Restarted, OS X reported a graphic problem. Then switched to NA211TB, continued the YouTube video, opened a Valley benchmark window too, no sign of crash or sluggishness. It’s a rock stable.

Remember that Nvidia gives you better OpenGL performance under OS X, but the OpenCL is a different story - even with mid-range AMDs.

So far my methods have been non-warranty voiding… J11 looks interesting :D

Share this post


Link to post
Share on other sites

uff…

Hard to tell, because imo you did it right. You had the chance to install the Nvidia driver in Windows 8.1, right?

Can you try:

- to fix the GRUB problem? Step 6: http://forum.techinferno.com/implementation-guides/8558-%5Bguide%5D-macbook-enabling-optimus-internal-lcd-mode.html#post110757

- when you "boot" into Windows with the eGPU and all the screens are black, can you see anything happen on the MacBook screen during the first seconds? Like the Windows spinning circle? Sometime I boot into Windows and the external display is "grey'ish" (not crashed). So apply the following steps without any visable interface: I can press any button 1x, enter my password and can go into windows.

Share this post


Link to post
Share on other sites

Thanks for testing! If you want to play a youtube video while using your eGPU, can't you just use your internal display accelerated by the internal graphics? Oh, and do you think my internal display would be accelerated, even though I've got dedicated 6750M graphics?

Share this post


Link to post
Share on other sites
Thanks for testing! If you want to play a youtube video while using your eGPU, can't you just use your internal display accelerated by the internal graphics? Oh, and do you think my internal display would be accelerated, even though I've got dedicated 6750M graphics?

Maybe the videos are more stable in the internal screen, not sure. One user has said that an older MBP showed greenish YouTube videos on Safari via NVidia, but I didn’t notice that with my MBP, Nvidia or AMD.

I found a screenshot for you:

post-28870-14494999375077_thumb.jpg

Regarding questions about NA211TB and the Late 2011 MacBook Pro with 6750M, you can contact directly Netstor.

Share this post


Link to post
Share on other sites

I must have hit all the bad combinations... I am hoping with some help here I can get it sorted, even without Optimus at least that it works without haing to pug a connector every time...

Tried again that... no success

this is the configuration I have on GRUB when I start the system and I select "EFI Boot"

post-33016-14494999376597_thumb.jpg

And this is how I configure the grub.cfg inside the EFI partition

post-33016-14494999377336_thumb.jpg

- when you "boot" into Windows with the eGPU and all the screens are black, can you see anything happen on the MacBook screen during the first seconds? Like the Windows spinning circle? Sometime I boot into Windows and the external display is "grey'ish" (not crashed). So apply the following steps without any visable interface: I can press any button 1x, enter my password and can go into windows.

Well, I think the computer literally "freezes". Nothing is shown on the monitor(s), the internal one just switches OFF and nothing happens. When I start Windows without the Thunderbolt connected or when I do the trick with the "yellow/black" cable, the windows logo appears and I didn't put any password so it goes straight on the desktop without pressing any button.

My doubt here I think is that there must be a problem with the power is the "yellow/black" does the trick... I am tempted to buy a new pCIe X16 riser with molex, but I am almost sure it won't change anything. But probably someone here has some suggestions on this as well?

As said before, I would be still happy if I could use it at least with an external monitor... otherwise I sell everything and buy a desktop :D

Share this post


Link to post
Share on other sites

Have you tried with the barrel connector in the Akitio?

If you don't want to use it's power supply, you can rig it to the EVGA PSU.

2011 shouldn't be so much trouble. I have frequently found that connecting TB once on Windows desktop then moving display cable at least works to get drivers loaded.

It might be worth double-checking your OS X install fixes. If you had it working perfectly in OS X and Windows still bad you would know it was drivers.

If bad in both it could just be hardware.

Share this post


Link to post
Share on other sites
Have you tried with the barrel connector in the Akitio?

Yes, tried it without the molex on the riser and even with the molex connected to the riser. Both cases same result: when trying to start windows it freezes.

In this case I didn't try the unplug-plug yellow/black cable as it could be messy I think being the barrel on that board

If you don't want to use it's power supply, you can rig it to the EVGA PSU.

Good idea, I tried this as well just now. Bad news is it didn't fix the issue :( It still freezes

This is the cable I made

post-33016-14494999377699_thumb.jpg

2011 shouldn't be so much trouble. I have frequently found that connecting TB once on Windows desktop then moving display cable at least works to get drivers loaded.

You are right, that worked already: the first time I tried the trick (I start windows with the yellow/black cable removed on the Akitio and I connect it back just right when I click on the icon to start it) the GTX660Ti was recognized and I was able to install the drivers. No every time I do hat unplug-plug trick everything works perfectly. I got the driver installed, I even played a game a little bit to try, did a 3D Mark... Only by doing that trick though, otherwise... freezes.

And on top of that, the Optimus also doesn't work (Grub always presents the console instead of starting Windows as it should)

It might be worth double-checking your OS X install fixes. If you had it working perfectly in OS X and Windows still bad you would know it was drivers.

The OS X fixes didn't work... they are still there (driver CUDA installed and the 3 files modified to support it), but Yosemite doesn't see the card.

The difference with Windows though is that I don't need to do any trick, I can leave the Akitio powered and plugged on startup and OSX starts without any problem.

THough it doesn't see the GTX660Ti on OSX, it does see the Akitio correctly, here is from the system info

post-33016-14494999378469_thumb.jpg

- - - Updated - - -

And if that could add some interest on this, reading back in the old posts I found someone else having the same problem! Here it is

http://forum.techinferno.com/enclosures-adapters/7205-us$200-akitio-thunder2-pcie-box-16gbps-tb2-51.html

Didn't say how they sorted it out though...

  • Thumbs Up 1

Share this post


Link to post
Share on other sites

i got na asus g75vx with thunderbolt port watt kind egpu i need to buy for my pc ,sorry i im really a noob at this , ty in advance.

Share this post


Link to post
Share on other sites
MSI HD 7970 is loud as a jet engine, but looks like it’s more stable than 270X and 280X (both Asus). I picked up some interesting 3DMark11 power consumption numbers, they may help to find the common factor for AKiTiO instability with some GPUs. Note that the small 4-pin cable was completely removed during the tests (value columns 2-5).

GPU Whole system - one PSU - 4-pin cable in place - Idle

TB card (4-pin) + x16 slot - max

TB card (4-pin) + x16 slot, physics TB card (4-pin) + x16 slot - Idle

DC jack (idle / tests 1-4 and 6 / physics) Whole system - one PSU - 4-pin cable in place - 3DMark11 highest peak
MSI HD 7970 (stable so far) 22W 36W 13W 12.4W 7.5W / 8.8W - 8.9W / 7.7W - 7.8W 231W (test 2)
Asus R9 280X (not stable) 25W 36W 22W, 34W … crash 12.1W 9.1W / 10W - 10.9W / 9.5W - 10.4W 218W (test 2)
Asus R9 270X (not stable) 25W 51W 17.6W - 18.3W 15.2W 7.8W / 8.6W - 8.7W / 8.1W - 8.2W 151W (test 2)
EVGA GTX 780 OC (stable) 21W 67W 18W-19W 15W 7.4W/ 8.8W / 7.4W 265W (test 2)


If we take a look at the second column, the specified 75W (PCIe 1.1) is not close at all when we subtract the TB card power consumption. That’s the maximum value, metered from the TB card 4-pin socket alone, feeding power via x4 connector to the x16 slot as well. I used a riser, but it was not powered. The DC jack (molex-to-barrel adapter) gives power to the PCIe board alone, and it stays quite constant. I believe 3DMark11 gives realistic power consumption numbers. FurMark is for torturing the GPU and does not describe the real use so well.

Interestingly, isolating the PCIe board powering for another PSU resulted more stable 3DMark11 physics runs for R9 280X and R9 270X. I don’t know why. Maybe a single PSU with multiple 12V rail gives the same result?
  • Thumbs Up 2

Share this post


Link to post
Share on other sites

@pacheco: AKiTiO with a GPU of your choice should be fine with an external display. Internal display will be difficult.

Share this post


Link to post
Share on other sites

@Ramses82: Have you tried to type “quit” or “exit” (at least twice) when you are stuck at GRUB console? For me that worked and continues booting if I get the console screen (maybe because of the external USB SSD).

Earlier than 2013 MBPs with Win8.1 Boot Camp installation are not the easiest to get eGPU working. You may get some help from @entzoe, he has the same MBP:

http://forum.techinferno.com/implementation-guides/8488-%5Bguide%5D-2011-13-macbook-pro-gtx-970%4010gbps-tb1-akitio-thunder2-win8-1-a.html

OS X should be easy, see the OS X installation part from here (the same instructions work for the new web driver which you can download from here). Track down the reason from the system log if the card is not recognized.

Share this post


Link to post
Share on other sites
@Ramses82: Have you tried to type “quit” or “exit” (at least twice) when you are stuck at GRUB console? For me that worked and continues booting if I get the console screen (maybe because of the external USB SSD).

Good call!!! "quit" didn't work, but "exit" did. Same as before, I have to leave the yelloe/black Akitio cable unplugged and plug it back just when I press "ENTER" after exiting.

Tried then to run windows normally, selected nvidia for all the applications by default and the hardware acceleration is working on the built-in screen! Does it mean we've got it working? :D

I hope it is going to work, I have ordered a single double pole single throw switch, I will put together the 2 yellow and the 2 black wires and use the switch to do the "plug" trick before I ruin the contactor... in theory the 2 yellow and the 2 black wires are the same in parallel, they both measure 12V when powered and I tried when unpowered to check for continuity between the 2 yellow and between the 2 black and the multimeter says there is continuity... fingers crossed.

Here is the configuration on the nvidia settings

post-33016-1449499938285_thumb.jpg

Earlier than 2013 MBPs with Win8.1 Boot Camp installation are not the easiest to get eGPU working. You may get some help from @entzoe, he has the same MBP:

http://forum.techinferno.com/implementation-guides/8488-%5Bguide%5D-2011-13-macbook-pro-gtx-970%4010gbps-tb1-akitio-thunder2-win8-1-a.html

Il'll try that as well, let see how it goes otherwise I'll post there... at the moment it seems the switch is the only solution but who knows...

OS X should be easy, see the OS X installation part from here (the same instructions work for the new web driver which you can download from here). Track down the reason from the system log if the card is not recognized.

I tried this as well, good shot about the pkg... I didn't install the right driver before, that for sure!

I can get Yosemite to see there is an Nvidia card now, no need to do yello/black cable pull on macos, as before.

post-33016-14494999383162_thumb.jpg

BUt still from the Nvidia settings the option to select that card is greyed out... I can't chose it. Like it really doesn't see the card.

I tried to connect the monitor on the DVI and HDMI, but it didn't change the situation.

Share this post


Link to post
Share on other sites
Good call!!! "quit" didn't work, but "exit" did. Same as before, I have to leave the yelloe/black Akitio cable unplugged and plug it back just when I press "ENTER" after exiting.

Tried then to run windows normally, selected nvidia for all the applications by default and the hardware acceleration is working on the built-in screen! Does it mean we've got it working? :D

I hope it is going to work, I have ordered a single double pole single throw switch, I will put together the 2 yellow and the 2 black wires and use the switch to do the "plug" trick before I ruin the contactor... in theory the 2 yellow and the 2 black wires are the same in parallel, they both measure 12V when powered and I tried when unpowered to check for continuity between the 2 yellow and between the 2 black and the multimeter says there is continuity... fingers crossed.

Great! :) If the alt/option boot + “exit” aren’t enough to do the trick, the manual delay is necessary as you described. 3DMark11 test will confirm the internal screen acceleration. The delay mechanism may be possible to do electrically, with some sort of adjustable timing but that goes beyond my knowledge.

The purpose of the 4-pin cable is to share +12V current between the Thunderbolt controller and PCIe board. The direction of the power flow can be either from the powered riser or from the DC jack / J4. The PCIe board needs only ~9W-10W:

http://forum.techinferno.com/diy-e-gpu-projects/8844-targeting-akitio-thunder2-pcie-boxs-stability-issues-3.html#post121721

Be very careful, avoid ground loops, double check the correct voltage and polarity. Some tips:

- The 4-pin wires can be released by pushing the side wings with a sharp tool

- You can modify a floppy drive plug with a knife. The best would be a plug that AKiTiO uses, but I haven’t found similar.

BUt still from the Nvidia settings the option to select that card is greyed out... I can't chose it. Like it really doesn't see the card.

I tried to connect the monitor on the DVI and HDMI, but it didn't change the situation.

Some kexts are not loaded.

1) Check that you are running OS X 10.10.2 (14C109)

2) Check that “grep IOPCITunnelCompatible /var/log/system.log” does not output anything. If it does, you haven’t correctly modified the kexts. I am not certain if your GPU is supported, but examining the system log will often help.

Share this post


Link to post
Share on other sites
Great! :) If the alt/option boot + “exit” aren’t enough to do the trick, the manual delay is necessary as you described. 3DMark11 test will confirm the internal screen acceleration. The delay mechanism may be possible to do electrically, with some sort of adjustable timing but that goes beyond my knowledge.

That worked perfectly, I tried to do 3dMark with an external 1080 monitor and got just above 10000, done it with the internal and got just above 9000 (the internal is just 1200x800 so it's easier to handle)

The purpose of the 4-pin cable is to share +12V current between the Thunderbolt controller and PCIe board. The direction of the power flow can be either from the powered riser or from the DC jack / J4. The PCIe board needs only ~9W-10W:

http://forum.techinferno.com/diy-e-gpu-projects/8844-targeting-akitio-thunder2-pcie-boxs-stability-issues-3.html#post121721

Be very careful, avoid ground loops, double check the correct voltage and polarity. Some tips:

- The 4-pin wires can be released by pushing the side wings with a sharp tool

- You can modify a floppy drive plug with a knife. The best would be a plug that AKiTiO uses, but I haven’t found similar.

I'll try to put a switch on that, I would be fine with switching it every time I turn on the computer, as soon as the trick works and it doesn't compromise the hardware.

Another thing that I was thinking about doing is taking another +12V and GND from a molex of the PSU and provide it straight to that pin as well on top of the already powered riser... that should bypass any diode or other circuits that may restrict the current capacity down to that part of the board... I'll see what I can do, if just the switch trick works I'll stick only to tha

Some kexts are not loaded.

1) Check that you are running OS X 10.10.2 (14C109)

2) Check that “grep IOPCITunnelCompatible /var/log/system.log” does not output anything. If it does, you haven’t correctly modified the kexts. I am not certain if your GPU is supported, but examining the system log will often help.

Thanks for this, it must be it... I am quite happy at the end that the thing works at least on Win, I do play games only on that at the end... but I'll give it another go when I've completed the Windows configuration ;)

Thanks!!!!

  • Thumbs Up 1

Share this post


Link to post
Share on other sites

Im not too experienced in electronic but what about replacing existing capacitors on the akitio board for larger ones ? I see two 20uF. is this enough ?

Share this post


Link to post
Share on other sites

Great news! We can now get this enclosure for a cheaper US$189 price directly from AKiTiO-US. Opening post has been changed to reflect this.

On Thursday, 12 February 2015, 11:43, Richard Wright <richard@akitio.com> wrote:

Hi Nando,

We have received and shipped quite a few of our Thunder2 PCIe Boxes to members of your forum. In appreciation of that, we have decided to lower the price to your group. The new price is $189 plus shipping. Thank you for helping us to spread the word!

Best regards,

Richard

RICHARD WRIGHT

Vice President

Sales & Marketing

AKITIO

Tel: (657) 216-2222 / Fax: (408) 657-4686

2951 Saturn St. Suite B, Brea, CA 92821 - USA

  • Thumbs Up 7

Share this post


Link to post
Share on other sites

INFO: passing 75W slot power without a PCIe riser

 

I've been working with a user who get PCIe signal instability when using a PCIe riser, but gets power instability without it. So how can 75W power be routed directly to the PCIe board of the Thunder2, eliminating the need for a Gen2 signal degrading riser? The following process done successfully by goalque, resulting in a fully stable AMD eGPU implementation. Details at http://forum.techinferno.com/enclosures-adapters/7205-us%24189-akitio-thunder2-pcie-box-16gbps-tb2-101.html#post124478 .

 

Two other positive side effect of doing this are:

1. the yellow/black J4 cable rated at 22AWG and the Thunderbolt board itself would no longer be overdriven past it's specced 25W.

2. Users who have or are choosing a powered PCIe riser instead of above soldering could also benefit from the cellophane tape isolating of the first 3 pins on both sides as shown in (1) below or even easier by doing the same on the PCIe riser itself as shown in the spoiler. That then separating their two different 12V inputs (12V_ATX and 12V_filtered) to increase stability.

Spoiler

fx8ILnm.jpg

 

How is power routed to the PCIe slot?

 

Some continuity tests with a PCIe slot schematic referring to slot pins as Ax, By where done with reference to the J4 12V-Yellow and GND-Black wires, finding:

* 12V (B1, B2, B3, A2, A3) are NOT wired directly to 12V-yellow. They go via the Thunderbolt board providing filtered_12V input specced at 25W (max)

* GND (A4, B4, B7) are wired to the BLACK so there is a common ground

 

How can we bypass the 25W (max) input board limit and pass 75W to the slot directly?

 

We also know from http://forum.techinferno.com/enclosures-adapters/8848-%24299-owc-mercury-helios-16gbps-tb2-discussion.html#post121898, using a nearly identical internal board, that the input DC barrel plug and lines to J4/J6 have been specced to accept up to 120W input power -> 75W fits easily within spec.

So then if you have a clean input 12V from say a ATX PSU via the DC barrel jack, the simplest solution to provide dedicated 75W to the slot while continuing to providing the filtered_12V input to the TB board is as follows, shown pictorially below:

 

CEXhXwO.jpg

 

1. disconnect the x4 board and use cellophane tape to isolate the filtered_12V pins A2, A3, B1, B2, B3. Perform continuity tests to ensure this was done correctly.

 

2. 12V_direct: solder two 18AWG yellow wires from the J6 (yellow pads) to A1 and B2. 18AWG is what your PCIe power leads use. These are wired to pins A2, A3 and B1 by the slot. J6 mirrors J4, which in turn is attached to the DC barrel connector and is a more convenient solder point than J4 since it can be more easily reversed without obvious tampering. A1 and B2 were chosen as the most convenient solder points to attach to.
 

Is there anything I can do to increase power stability?

 

Yes, solder in the missing C3 (100UF/16V) and C4 (10uF/25V) capacitors as is provided by the higher-rated but nearly identical 120W OWC Mercury Helios product to provide some additional smoothing/filtering of the 12V_ATX input power to the slot. See http://forum.techinferno.com/enclosures-adapters/8848-%24299-owc-mercury-helios-16gbps-tb2-discussion.html#post121898
 

  • Thumbs Up 5

Share this post


Link to post
Share on other sites

@Tech Inferno Fan: the graphics cards I have tested don’t utilise the specified 75W as shown in this table. It wasn’t just a single test, they are approximately same numbers after tens of tests. The interesting detail with a R9 280X was that the PCIe board power consumption numbers switched precisely 9.5W or 10.4W during 3DMark11 physics with apparently constant time period, no other value between those. With stable cards, such as EVGA GTX 780 OC, it was constant 7.4W the whole test period. The same 7.4W at idle when the GPU is not stressed. When R9 270X and HD 7970 passed the test, there was only 0.1W variation.

I am using a 3DMark11 physics test as a stability indicator, because it’s a 100% crash for AMDs with those two common powering methods: a back-powered riser, molex-to-barrel adapter or combination of these two. I tested also with two separate 6-pin to 4-pin molex cables from the PSU. All the test were done without soldering the AKiTiO board.

I have found keys to pass the physics twice in a row by isolating the PCIe board powering, so that the TB card and PCIe board power is not shared with that 4-pin yellow/black cable.

When I played with a multimeter a bit more, I found something interesting when testing the TB card alone. As you said, there is no direct connection from the 4-pin TB card socket +12V pin to the x4 slot (B1, B2, B3, A2, A3), but I can repeatedly get a short beep sound by switching red/black probes, indicating maybe a capacitor between those test points. When I did a diode test with the same test points, the multimeter showed “OL” (very high resistance) when the black proble was placed at x4 12V pin and the red probe was placed at 4-pin socket 12V. However, when I did the diode test by switching probes, the multimeter showed continuously 0.511 V (I guess the display number is the forward voltage drop). Does this mean the power is meant to flow only one direction and blocking from the other direction? Is there a risk that with back powering the TB card we may destroy the diodes on the board? (I remember one EE warned about it).

However, it seems to be possible with a back-powered riser to provide power firstly for the GPU, then for the TB card, and finally for the PCIe board via 4-pin cable. The other way is from the DC jack or J4 (where it splits for the PCIe board) -> TB card -> x4 slot -> x16 slot.

I have a possibility to test with either a back-powered riser (as most of the users here) or non-back-powered riser such as this:

8X PCIe 16X extension cable 19cm power

It’s expensive but was very stable with the Sonnet SEL. And the yellow wire is very thick. Some risers that I have got from eBay seemed good externally (two thick yellow/black wires + capacitor) but when I cut the wire, the inner part was ridiculously thin. Sometimes you cannot even trust to AWG markings.

One more thing, with the TB card attached to the x4 slot of the PCIe board, the multimeter behaved differently when testing the PCIe board reverse points (marked as red lines in your picture). The 4-pin cable was removed. And what was the difference? Well, I got those short beep sounds again (continuity test) when the TB card was attached, but when not attached, it was mute. The same result with a diode test, but this time no continuous value as with TB card 4-pin <-> x4. I am happy to continue with the aid of EE experience. Already did the electric tape test with the back-powered riser / non-back-powered riser + molex-to-barrel adapter, but unfortunately the same 3DMark11 physics crash with the R9 280X. Should I try that soldering? I have a plan to use pins at the end of the wire for x16 slot reverse board points.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.