[In progress] SSDT for Thunderbolt 3 Hotplug

[jiffyslot]

"...the MOTU 8A. It works when connected to my OWC Thunderbolt 2 dock (no problems with the on-board Titane Ridge of the Z390 Designare board), but it’s not a good solution. I had tried this on my iMac (where I could compare it, because Thunderbolt 1 was just working, too) and it’s the same here: Always slight humming and sometimes other noises when it’s connected to the Thunderbolt 2 dock..."

Yeah, that's weird. There shouldn't be any noises at all. Frustrating stuff, no doubt. It would be interesting to probe the ground on the Hack and the ground on the dock and/or the MOTU with an oscilloscope and see what kind of this noise is. The old, "Is it a shield or an antenna?"

BTW, what's your roundtrip latency using the OWC dock compared vs your previous working MOTU system?

… which is clearly wrong, because Thunderbolt 2 devices are working just fine on the Z390 Designare, and Thunderbolt 1 devices connected to Thunderbolt 2 docks are working as well.

Perhaps only clearly wrong if you are not using a UAD Apollo DAW interface that relies on a direct "legacy" Thunderbolt 2 connection. You're lucky in that you can fall back on USB 3 and Ethernet on the MOTU 8A. I really wish I had a 10GBe/Dante connection on the UAD...

 

[kgp]

Just considered something that I can't read or find in the SSDT: on my Z370, it is required that the GC Alpine Ridge card goes into the last PCIe slot: Slot 4. Were you approving the SSDT knowing that? Also Legacy connection and only port 1 is working. Weird because both ports work in Windows. Thanking you, j.

This small guideline should help in properly adapting the TB-SSDT to any motherboard and slot population.

Always verify correctness of SSDT-implementation by means of IOREG and "PCI" implementation in Apple's system report. If ACPI path and ACPI replacements have been properly adapted within the SSDT, the IOREG and PCI result should be the following:

Good luck,

KGP

 

[JimSalabim]

Yeah, that's weird. There shouldn't be any noises at all. Frustrating stuff, no doubt. It would be interesting to probe the ground on the Hack and the ground on the dock and/or the MOTU with an oscilloscope and see what kind of this noise is. The old, "Is it a shield or an antenna?"

BTW, what's your roundtrip latency using the OWC dock compared vs your previous working MOTU system?

I could indeed try check this with a scope, I’ve got a small digital one. But I think I would need advice how to read the results.

I will check the roundtrip latency on Sunday when I’m back home from holiday. I know that I had measured 35 samples at 44.1K on the old system when the MOTU was connected directly to the iMac.
Could it be that the MOTU is just daisy-chained with the OWC dock? At least it doesn’t matter which one of the two Thunderbolt connectors of the dock is connected to the computer and which one to the MOTU.

 

[jiffyslot]

I could indeed try check this with a scope, I’ve got a small digital one. But I think I would need advice how to read the results.

I will check the roundtrip latency on Sunday when I’m back home from holiday. I know that I had measured 35 samples at 44.1K on the old system when the MOTU was connected directly to the iMac.
Could it be that the MOTU is just daisy-chained with the OWC dock? At least it doesn’t matter which one of the two Thunderbolt connectors of the dock is connected to the computer and which one to the MOTU.

35! Wow. At 44.1/24 that may be better than the Antelope or RME stuff.

Just look for general hum at multiples of 60Hz and whatever speed your CPU/GPU or the TB controller chip might be running. Really, it doesn't mattter too much, but you may notice a pattern that if recorded would sound like "WeeWeeWeeWeeWeeWeeWeeWeeWeeWeeWeeWeeWee!..."

See if you can still measure it when the MOTU is off. Voodoo stuff. EMI/RF interference is the groundloop of the 21st century for sure. -And it might not leak into your recordings, but test it anyway. Normalize it, etc. PM me if this gets way OT and I can give you some scenarioss and fixes.

PS: If you can repeat that awesome RT latency (or not noticeablly different) via USB 3/C etc. or even the ethernet/AVB, I'd skip Thunderbolt and never let that MOTU 8A out of your sight. -Unless you want to trade for my Apollo!

 

[JimSalabim]

OK. Now this is really strange … Now just that I have ordered a used Alpine Ridge, I thought I’d try the MOTU 8A directly at my motherboard’s built in Thunderbolt again before the Alpine Ridge arrives, just to prove that it really doesn’t work (Z390 Designare with built in Titan Ridge). But well … it does work. I didn’t work when I first tried it a month ago and it was clear to me that it would not work, because I knew that the motherboard and Titan Ridge would not support Thunderbolt 1 devices. But it’s working quite fine. The OWC Thunderbolt 2 dock is completely disconnected, it just works (just with the usual Apple Thunderbolt adapter). No idea why. It’s definitely a Thunderbolt 1 interface (the specs say Thunderbolt 2 and 3 supported, but that doesn’t make it a Thunderbolt 2 or 3 interface, right?)

The latency offset measured in Logic Pro’s I/O plugin is 20 samples at 44,1kHz and 36 samples at 96kHz. With the OWC Thunderbolt dock in between I measured 36 samples at 44kHz yesterday and 51 at 96kHz.
When connected via USB3 I have 68 samples at 44kHz and 126 at 96kHz.

So I think I’ll have to sell the Alpine Ridge again then as soon as it arrives :-D

 

[JimSalabim]

Oops … After rebooting it doesn't work anymore without the Thunderbolt dock. I didn’t reboot before when I disconnected the dock and connected only the MOTU 8A instead.

EDIT: Working again. ErP was accidentally enabled last time … Perhaps it had got to do with this?

 

[jiffyslot]

Dang. When you measure the latency, would you be so kind as to use millisecond values? Thanks!

 

[JimSalabim]

Latency offset of MOTU 8A connected directly to Titan Ridge:
0.5–0.8ms
About the same when connected via Thunderbolt dock.

MOTU 8A over USB3:
0.7–1.3ms

But this is not the round trip latency, it’s just the recording delay (which can be compensated for in the DAW settings). MOTU says the round trip latency is 1.6 ms at 96kHz
over Thunderbolt and 1.9 ms over USB.

 

[Ferz2019]

@LeleTuratti, could you help me with ssdt ? I read posts before in this topic and find a post #404 in page 41. I Used F7 bios as said in post #404 and download ssdt from @kgp from post #1 as said in post #404 (DOWNLOAD SSDT-Z370-THB3HP). then I download MaciASL and change RP05 to RP21 as said in post#404 (because GB Gaming 7 used RP21), saved it and copied two aml files to EFI/CLOVER/ACPI/patched.

My changed to RP21 by MaciASL AML Files in attachment in this post. After reboot in clover I cleared nvram using f11 key and i have DSB0 in ioreg, instead of PSPX (see screen in attachments) and Thunderbolt devices in PCI tab in sys info, but hot swap doesn't works for me. What I'm doing wrong?

P.S. of course, when I didn't copy any aml files in patched folder, I see PSPX and nothing in PCI tab in system information.

Moreover if with the inserted SSDT you do not see UPSB instead of PXSX and the respective DSB0, DSB1, DSB2, DSB3, DSB4 on your IOReg, it means that the SSDT has not been edited/loaded correctly and the Thunderbolt is not implemented.

 

[LeleTuratti]

@LeleTuratti, could you help me with ssdt ? I read posts before in this topic and find a post #404 in page 41. I Used F7 bios as said in post #404 and download ssdt from @kgp from post #1 as said in post #404 (DOWNLOAD SSDT-Z370-THB3HP). then I download MaciASL and change RP05 to RP21 as said in post#404 (because GB Gaming 7 used RP21), saved it and copied two aml files to EFI/CLOVER/ACPI/patched.

My changed to RP21 by MaciASL AML Files in attachment in this post. After reboot in clover I cleared nvram using f11 key and i have DSB0 in ioreg, instead of PSPX (see screen in attachments) and Thunderbolt devices in PCI tab in sys info, but hot swap doesn't works for me. What I'm doing wrong?

P.S. of course, when I didn't copy any aml files in patched folder, I see PSPX and nothing in PCI tab in system information.

If you remove the SSDT, your IOREG show the original paths.
PXSX is the original path that needs to be renamed > The SSDT renames PXSX to UPSB.

Your IOREG is correct, the tunderbolt is implemented correctly.
If the hotplug does not work, check that your BIOS is configured correctly.
Do some tests, disable the ERP in the BIOS, disconnect and reconnect the thunderbolt device, etc