Trenz Electronic GmbH Support Forum

Trenz Electronic Products => Trenz Electronic FPGA Modules => Topic started by: rkbluecubed on January 05, 2022, 05:27:04 PM

Title: Marginal voltage on MR when using TE0790 with TEBT0808 and TE0803.
Post by: rkbluecubed on January 05, 2022, 05:27:04 PM
I believe I have finally solved an intermittent reboot problem that has been vexing me for a few months. I am using the following products:

The intermittent reboots appear to stem from a marginal voltage on the master reset (MR) net.

On the TE0803, MR is connected to U41 (TI TPS3106K33DBVR) where it is internally pulled up to VDD (LP_DCDC, 3.3V).

On the TE0790, the "G" pin on the CPLD is used to drive MR.  The TE0790 TRM states that pin G has an internal pull-up in the CPLD.  This pull-up is to VIO, which needs to be 1.8V for this system config due to direct connection of the UART to the PS 1.8V bank.  This pull-up is fighting against the 3.3V pull-up in U41 on the TE0803.

The minimum input high (V_IH,min) of the TPS3106K33DBVR is 0.7*Vdd.  Vdd is connected to 3.3V, so the input needs to be kept above 2.31 V to avoid resets.  On the units I have hear, MR tends to float around 1.9 V.

Does this analysis and conclusion seem correct?

My temporary workaround is to sever the connection to the TE0790 CPLD.  The downside is that the reset button on the TE0790 is no longer usable.

It would be great if the CPLD firmware could be revised to disable the pull-up on pin G, but I'm not sure how that would impact other use cases for this product.

Title: Re: Marginal voltage on MR when using TE0790 with TEBT0808 and TE0803.
Post by: Antti Lukats on January 11, 2022, 02:04:14 PM
Thus may really be a problem with TEBT0808, good catch!

on your custom board you can hopefully fix it.

br
Antti Lukaks
Title: Re: Marginal voltage on MR when using TE0790 with TEBT0808 and TE0803.
Post by: rkbluecubed on January 11, 2022, 05:22:11 PM
Yes, that was my conclusion as well.  I have corrected the issue on my own designs.

I suspect that the most common use case for the TEBT0808 is normally just used for short/quick SOM verification using JTAG or the like.  This design deficiency probably isn't too noticeable for that use case.

The main problem I can see this causing is when customers use TEBT0808 as a reference design or starting point for their own carrier boards.  I took this path because it was simpler and more approachable than the TEBF0808, which had tons of features I don't need or want.  That's how I got into this mess.

I will also note that the TEBF0808 is not impacted by this issue because there is an intermediate CPLD with 3.3V IO.