TE0720, TE0790: eFUSE programming, Standalone debugging

[andreas]

Hi,

(Possibly a duplicate post?! My login session expired, and I'm wondering if the first post went through.)

We are using the TE0720 on our custom board, which was inspired by the TE0703. JTAG for standalone debugging is implemented via a mounted TE0790. We are using Vivado, Vitis, and PetaLinux (version 2023.2).

Normally, we initialize the TE0720's QSPI and eMMC after booting Linux from the microSD card.

So far, we haven't had any problems (more than 20 systems have been successfully initialized).

The problems started when we tried to program eFUSEs for Secure Boot or debug FSBL.

Both require a mounted TE0790.

I have already successfully debugged FSBL, read ILA cores, and initialized eFUSEs with our custom board and a mounted TE0790.


However, there are cases where the TE0790 and/or the TE0790 module are damaged!

The TE0720 module operates with a 3.3V power supply (like the TE0703). The TE0720's FPGA banks are supplied with 1.8V (VCCIOA, VCCIOB, VCCIOC, and VCCIOD are all set to 1.8V). The TE0720's CPLD revision is 07 (default).

The TE0790 is powered via VCCJTAG. The TE0790's switch has the positions [On, Off, On, Off].

Do you have any further information? Any hints how to prevent further damages?

--
Andreas

[Antti Lukats]

To our best knowledge, TE0790+TE0720 should not burn no matter how you connect them or what the DIP settings on TE0790 are. So it is a bit strange that you see damages.

But your dip settings doesnt seem right: you power the FT2232H with power from VCCJTAG and TE0790 IO buffer from the onboard LDO.

On-off-off-off would be better if you supply both VCC and VCCIO from your carrier.

[andreas]

My intention is to supply the TE0790 completely from our carrier.
But I use different Switch settings.
And I think our switch settings are at least not harmfull.

J2-5 is powered from TE0720 JM2-91 (VREF_JTAG, 3.3VIN)
J2-6 is open on our carrier.

S2D is Off (No connection from LDO 3V3 to 3.3V)
S2C is On  (3.3V is directly connected to VIO. No 10R from J2-6!? Is that a problem?)

I attached the DIP-Switch part of the TE0790-03 Schematic.

As mentioned earlier, the error I see is sporadic.
But when an event hits me, my hardware is damaged. MACHXO on TE0790 and on TE0720 seem to be destroyed afterwards.
In case of an error the TE0790 is still detected as USB-Device, as long as I don't switch on VIO power.

The damaged TE0720-Boards do not activate there 3.3V output power.

The last error event hit me, when I already booted a secure linux system with attached TE0790 and tried to start a FSBL debug session from Vitis. (standalone)
The running system must have died, during the JTAG-Reset-Sequence.

What else can I check and test?
My next problem is, that I don't have TE0790s left.
During the last months this error damaged all our TE0790 (5 pieces).




 

[Antti Lukats]

This is really strange.

the best option would be to FULLY supply TE0790 from the carrier switches 3.4 BOTH off and VCCJTAG supplying pins 5,6

[andreas]

I'd like to implement your suggestions in the next test.

But how do I get one or two TE0790s right now? On your Web site they are out of stock and mine are broken.