Artix-7 not talking to JTAG HS3

[iain.mcclatchie]

We are trying to bring up an Artix-7 35T on a custom board.

We are using a HS3 JTAG cable from Digilent (JTAG-HS3 Rev. A).  We use this cable successfully to talk with Zynq-7000 FPGAs on two other custom boards.  Vivado recognizes the cable when we try to connect to the Artix-7, but doesn't see the FPGA.

We have probed the board.  All voltages (1.0V core, 1.8V VDDO, 2.5V VDDO, 1.8 V AUX) look good.  We can see the TCK, TMS, and TDI signals from the cable on traces connected to those pins on the FPGA.  The signals are 1.8V, which is the reference voltage going to the cable and the VDDO for bank 0.  Signal integrity looks excellent, but we've tried many JTAG frequencies to see if that might be a problem without avail.

The FPGA appears to initialize correctly, and in JTAG mode releases INIT_B which is then pulled high to 1.8V via a pullup.  In SPI mode, INIT_B is pulled down and stays down because we have not yet programmed the SPI flash.

We probed the SPI flash config clock, and verified that the FPGA is sending it a 2.7 MHz clock.  We think the FPGA is not broken, and we have verified identical behavior on several boards.

We have run out of ideas for what to check next.  Any help would be appreciated.

[drjohnsmith]

A wild guess

signal integrity on the jtag lines ?

wrong type of flash fitted,

[bpatil]

Hi @iain.mcclatchie

Try reducing JTAG CCLK. 

I will suggest to try with 2MHz.

EDIT: From snapshots: I can see when you tried configuration of Zynq with JTAG CLK = 5MHz; configuration succeeded. When you tried configuration of Artix 7 with JTAG CLK = 15Mhz; configuration failed

Regards,

Bhushan

-------------------------------------------------------------------------------------------------------------------------------------------------
Kindly note- Please mark the Answer as "Accept as solution" if information provided is helpful.
Give Kudos to a post which you think is helpful and reply oriented.
-------------------------------------------------------------------------------------------------------------------------------------------------

[iain.mcclatchie]

We got back to this board and found something very curious.  I thought I'd post for the benefit of anyone else beating their head against this particular wall.

The Artix-7 FPGA is in fact replying to the JTAG HS3 cable.  We didn't see this before because:

• We are using an AC coupled probe on our scope.
• The TCK, TMS, and TDI signals are low when inactive.  To see them, we use a positive voltage trigger (e.g. 1 volt).
• The TDO signal is pulled high by the FPGA when inactive.  We needed to use a negative voltage trigger (e.g. -1 volt) to catch these signals.  We had not done this before and so we missed the TDO outputs.

So it turns out that the problem isn't with the board at all.  It's the JTAG HS3 cable.

• JTAG HS3 cable works just fine with our Zynq board (at 3.3V).  This is the main thing that was confusing us before, since we figured if the cable works for one FPGA, it should work for all of them.
• The exact same cable (not just another JTAG HS3) does not work with the Artix-7 (at 1.8V).
• We have an older Xilinx Platform Cable USB (DLC9G) that works fine with the Artix-7, using the same Artix-7 board and the same PC running the same Vivado.  The problem really seems to follow the JTAG HS3.

Tomorrow we'll test a different JTAG HS3 cable to see if the problem is specific to just that one cable, or to that cable design.  I will update.

[iain.mcclatchie]

The different JTAG HS3 cable sees the Artix-7 just fine.

So, it appears this is a simple case of an interesting hardware failure.  Apparently the JTAG HS3 cable can fail for 1.8 volt operation (not seeing the FPGA at all), but work reliably for 3.3 volt operation.

We think we must have fried a component in the 1.8 volt path that isn't on the 3.3 volt path.  On close inspection, there appears to be some distortion of the plastic covering one of the components.  We must have burned it somehow.

[drjohnsmith]

chips are powered by smoke,

   when the smoke escapes, they stop working,

its quiet easy to put those things wrong if you don't have a cage around the pins on the board.

I'd always suggest a 10 or so ohm resistor on the vcc of the JTAG connector, 

   its only used as a reference on the cable, so only low current needed.