Bought a TS-453 Pro recently from a Ebay seller, however it has the same problem(no beep, no HDMI output and nothing) as everyone here. Since the seller provides a full refund without having to send it back(Canada), I decided to take it apart and see if a simple SPI reflash will fix the issue, as few successful stories. Unfortunately as I suspected, it does not do anything for majority of the people. After reflashing the BIOS and powering it up couple times, the image read back is exactly the same. That means either the CPU failed to fetch/execute any code or the BIOS does not run long enough to write back some data to the flash. By the way, the pin header for the SPI is exactly the same as the Dediprog header, which is commonly used in the PC development.
I also captured the pin 1 CLK on the LPC header(LPC_CN1) to see if it's the same as C2000 Atom LPC defect. Although the clock capture looks different than this capture, https://youtu.be/Y8_emfoR_MI?t=788
, I felt it is probably the same issue and just shows up as a different failure pattern. I also noticed the low of the LPC clock is quite high, probably not normal. Voltage measured with a multimeter will show ~2.4V if you don't have a scope. I also probed the Fintek F71869AD pin 38(PCICLK) and it seems that's the same clk as LPC header, 25 MHz. The pin39(CLKIN) is clocked at the 24 MHz(but looks clean/good), so some clocks are still good. Plus there's no clock(low) to the SPI (actually the clk is only transmitted when there're transactions), pin 1(Vcc), 3(CS), 5(SO) are high. That explains why the system failed to boot and unchanged SPI image after power up.
After doing some Google search, I saw this issue discussed below matches exactly the same as my system, but the workaround doesn't work well for me [Updated status below]. Tried lower the resistor to 100 ohm and the low is still around 1V(with High reduced to ~2V), but this might be the solution for some lucky ones.
https://community.intel.com/t5/Processo ... rue#M46864
we have measured that the LPC CLK signal is clearly degraded, and most often that the negative cycle transistor is broken, meaning that the clock signal is swinging between 1.6 and 3.3V, not 0V and 3.3V. We have tried to put a 120 ohm resistor between the degraded clock signal and 0V, and that seems to solve the problem, the computer boots again.
Continuously trying in the evening, I actually got it to boot with the resistor back to 120 Ohm (fine tune might be needed for a specific system). It seems it has a higher chance to boot if the processor is heated up hours before it is ready to boot (without fan and leave it on even if it hung). Most the time it hung at various places on the 1st boot(post code B2 or during decompress/booting to kernel). After power cycle and if it booted successfully, it seems running fine afterward, until next cold boot.
Still trying to boot after an hour wait the next day, looks like I might have to warm it up a lot longer before it was able to boot last night. After waiting over 2.5 hours, it is finally showing signs of booting again (hung randomly) and it seems 100 Ohm works better on my system.
This was captured when it was repeatable booting last night (with 120 Ohm). That big dip and more stable in the clock might be what's making it boot.
This is when it was cold (not booting) this morning with nothing changed in the setup
When more than 5 hours of warm up still failed to boot (with 120 Ohm), it boots immediately after I reduced the resistor to 100 Ohm.
Final thought, it seems the 100 Ohm might be the solution for my system. When it failed in the earlier test, it's probably because I forgot to put back the memory after testing various conditions. It boots right away consistently(except once) after cool down with that 100 Ohm in place between pin 1 (LPC CLK) and ground (pin 8 or 10). From chatting with my friend, the rework might cause clock buffer degradation even faster with this strong pull down to the clock. Even though it works now and the boot is repeatable, not sure how long can it last?
Also this LPC clk is fed into the Fintek F71869AD, which acts as the EC for the onboard power management, as well as the FSC (fan speed controller). Once the clock went wild, the CPU/EC lost the communication, the system will either hang, reboot or log fan errors(if lucky) as some have seen. When that happened, power cycle the system without G3 (mechanical off) does not fix the problem for me and the system continue to fail to boot until removing the power cord. It's very easy to reproduce this problem. Once the system boots up to QTS, those bad things happened immediately as soon as I disconnect the resistor.
Besides replacing the CPU (which is a riskier/costly option), an alternative clock from LPCCLK1 or other clock source might be a solution as well. The problem for that is it will be a challenge to find/cut the trace from the LPCCLK0 and hook it up to an alternative clk source without a board file.
You do not have the required permissions to view the files attached to this post.