[SUCCESS] Gigabyte Designare Z390 (Thunderbolt 3) + i7-9700K + AMD RX 580

[cylsfar]

@cylsfar
Have you tried flashing the original Thunderbolt firmware ?

No.
Why do you flash Thunderbolt? To do what?

 

[rj510]

I'm planning to use this hot air gun that I purchased in July of 2018. If you think that's not a good idea, please let me know!

View attachment 456264 View attachment 456265 View attachment 456263

I used something similar last night to remove an Macronix MX25L8006E. It required only a few seconds to loosen, and I had sharp forceps at the ready to gently lift off the chip.

 

[Covalt]

Have you tried connecting an outboard antenna to the Bluetooth port (see @trs96's post here). If I may borrow the screenshot from that post:

View attachment 455924

We can attach an outboard WiFi/BT antenna to the Bluetooth connector. We can use the antenna that comes in the Designare Z390 box, or something like this from Amazon.

It's an interesting idea, though I've placed some of my Bluetooth devices right next to the antennas in the back, so I don't think it's an issue of not picking up the signal. Unless you're thinking it'd help to separate the Bluetooth signal away from the WiFi signal?

 

[CaseySJ]

*** Possible Improvement for In-Circuit Read/Write to SPI Flash ROM Chips ***
Procedure Applies to Raspberry Pi Model 3B and Model 4
Please do not quote this post in its entirety. Post a link instead.
​

(This procedure is not necessary for Gigabyte GC-Alpine Ridge and GC-Titan Ridge because they are very easy to read/write with an external SPI ROM flasher.)

The Thunderbolt firmware chip on many motherboards and add-in-cards has been rather difficult to read and program. We've had to "wiggle" the SOIC8 clip or remove components from the motherboard or use "-F" (force) option or make many many attempts before the chip could be read or programmed successfully.

After some online research and experimentation, I believe there might be a way to significantly ease the process. It requires the use of a 47uF capacitor and a 2K Ohm resistor. Some boards may respond better with a 100uF capacitor instead.

This additional circuit complexity is not as bad as it sounds. We can use a simple half-size breadboard that is familiar to Arduino and Raspberry Pi users and wire it up as shown below. Please see attached PDF for higher resolution image.

With the 47uF capacitor and 2K Ohm resistor I was able to read the Designare Z390 quite easily. But it did require one removal/reinsertion of Pin 3 (see Note 3 below).

Caveat 23-Mar-2020:

• On the Designare Z390, this procedure may not be as repeatable as I had hoped. Buyer beware.

Parts needed:

• Half size breadboard.
• Breadboard connection wires (this set includes M/M, M/F, and F/F).
• 47uF capacitor.
• 2K Ohm resistor.
• 100uF capacitor. This is optional for Asus ThunderboltEX 3, ASRock Thunderbolt 3 AIC, and ASRock motherboards with on-board Titan Ridge.
  • I was able to read Asus ThunderboltEX 3 and two different ASRock Thunderbolt devices repeatedly with no problem using this circuit. Some of you might know that ASRock Thunderbolt motherboards are extremely difficult to read/write.

Note:

1. Capacitors and resistors often come in kits of varying farads and ohms. They may also be available from certain local electronics stores. If you have an Arduino or Raspberry Pi sensor/accessories kit, look there first.
2. Connect SOIC8 clip to the target chip before powering on the Raspberry Pi.
3. If flashrom cannot read the chip, try gently pulling out the 2K Ohm resistor from the breadboard, wait a couple of seconds, and push it back in.

Circuit:

• When clipping to a motherboard:
  • System must be powered off.
  • Power cable must be removed from the power supply unit. Physically disconnect power plug from the power supply unit.
    • Wait 15 seconds.
• Always attempt to read the chip first.
  • Read the chip at least 3 times and save each result into a different file.
  • Use shasum <filename> to make sure all files have the same checksum.
  • Type strings <filename> on any file to make sure there are some meaningful strings (i.e. make sure the file is not empty).

Potential Alternative Design:

• Although I have not yet tested this, it may be worth experimenting with a 1K Ohm or 2K Ohm resistor before Pin 7 (/HOLD) of the SOIC8 clip.

 

[CaseySJ]

For the last month or so I've been almost only booted into windows. Today I went back to my hackintosh, and it has been completely freezing / locking up after being booted for a while, requiring a reset with the case rest button. This was not happening back in early February. The first two times it was while using safari, so I tried switching to firefox, but it still happened. Watching /var/log/system.log has not had anything interesting at the times of the freezes. It does not seem related to sleeping, I was able to manually sleep the computer and it resumed fine. Twice it has frozen while trying to restart.

I had originally followed the fresh catalina guide in late January with 10.15.2, but have updated to 10.15.3 a week after that came out.

So far today I've tried updating whatevergreen and I found and did the ocquirks rev 15 update. I'm on clover 5013, and am not using the newish native nvram stuff.

I'm at a loss to figure out how to further diagnose this issue,

At the top of the Catalina Mini-Guide we've added a warning for RX 5700/XT owners. Have you seen this?

 

[rj510]

*** Possible Improvement for In-Circuit Read/Write to SPI Flash ROM Chips ***​

After some online research and experimentation, I believe there might be a way to significantly ease the process. It requires the use of a 47uF capacitor and a 2K Ohm resistor. Some boards may respond better with a 100uF capacitor instead.

This makes sense. I've done a lot of audio building (tube preamps, etc). Adding shut capacitors is commonly done on long runs of DC, such as between a power and signal chassis. The shunt caps are placed on the receiving end, the signal chassis. It keeps the voltage from sagging, allowing better current delivery too. There's no reason not to go high with a 100uF cap; the source voltages are low at 3.3V. A 100uF/10V (greater voltage rating is okay, but the size gets larger). 16V rated caps are easily found.

Placement closer to the DUT is optimal. Maybe even place the cap and resistor across the back end of the SOIC clip. The schematic looks like you're essentially tying Pins 7,8 together and running the 2K between them and pin 3 with the cap going from Pins 7,8 to ground at Pin 4. Or, put on the CHK341A chip card (a little sloppy, but functional).

Using just a GB-TR-PCIe card, I did not get consistent reads with 2k in circuit. After cutting lead at arrow (removing 2k from circuit, but ends could be touched together), I got read and writes. I did not try on a mobo mounted chip.

Spoiler

 

[decibell]

We are trying to find an easier way. In the near term I'm afraid an external flasher is the only option. We might have an answer regarding Windows-based updater in the next day or two...

cool! Happy to help sacrifice my motherboard to try out some windows-based solutions, just didn't want to go and order tons of stuff for a 1-off solution

 

[qthegaijin]

Okay all! @CaseySJ @Elias64Fr @NorthAmTransAm

Here are my results from my Z170X-DESIGNARE. About 2 hours worth of back and forth flashing, testing, reading the logs, for all of the versions posted. I am super grateful of you all taking a look at this firmware file.

ALL FIRMWARE WITHOUT SSDT:
- Nothing populates in IOreg or SYSInfo
- No mention of "THUNDERBOLT" in the log files, no errors or anything.
- USB-C works with No-Hotplug
- Only shows USB-C on cold boot.

ALL FIRMWARE WITH SSDT:
- Same as without SSDT, see above.
-

After reflashing the Original TB3 firmware, Thunderbolt would NOT work, no matter what I tried. So I ended up going back into Windows and running the Gigabyte FlashTBT app, and, after the powerdown/unplug/reboot, Thunderbolt was back to stock.

Here is what it looks like with the Antelope plugged into the built in TB3 at clover boot screen in IOREG with stock TB3 firmware fresh from windows FlashTBT. No SSDT:

After another reboot the Antelope will fully load into IOreg but like always drivers/coreaudio don't see the device.

Looks like it will be back to Titan Ridge testing for the time being. That has the potential for more successful outcomes it seems.

I may try reripping the stock firmware again and having one of the fine folks here check to see if its different after doing the windows reflash.

UPDATE:
Reinstalled the modified GC-Titan Ridge card, in doing so disables the Alpine Ridge controller. Nothing populates under RP05 when plugged in.

 

[scottkendall]

cool! Happy to help sacrifice my motherboard to try out some windows-based solutions, just didn't want to go and order tons of stuff for a 1-off solution

The solution is the same you flash the BIOS the method of implementation is what is different. Having an external programmer can save if you if the windows based solution bricks your TB during testing.

RPi is super cheap and so is the clip not sure that classifies as a Ton of stuff and people buy tools for a single job all the time. You can also use the RPi for other things not just programing chips.

 

[qthegaijin]

cool! Happy to help sacrifice my motherboard to try out some windows-based solutions, just didn't want to go and order tons of stuff for a 1-off solution

$11 Programmer that is already 3.3V: This Programmer

This just seems to work and work well once clipped on tight. Definitely use a USB extension cable with it to make it easier on yourself.