UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Voyager
Voyager
330 Views
Registered: ‎06-20-2017

ZCU111 current monitoring

Jump to solution

On the ZCU111, there is a current sense resistor R447, listed on the schematic as 2 mOhms.  The voltage across R447 can be monitored via U67 (an INA226) via U23 (a PCA9544ARGYR) via either:

  • PL method:  U20 (a PCA9306DCTR level shifter) via pads AW16 and AV16 of the XCZU28DRFFVG1517.
  • PS method:  U19 (a PCA9306DCTR level shifter) via pads P29/U28 of the XCZU28DRFFVG1517.

I'd like any information about using this interface for purposes of monitoring the VCC int current.

1.  Is there an application note or whitepaper on this.

2.  Is there an example design (IPI block diagram, VHD/Verilog, C code) that is available?

 

UPDATED:  I see that the PS I2C via U19 (PCA9306DCTR) is a possilbe alternative path...looking into it on R5 via system.mss examples....any other person who has been down this path is welcome to chime in. 

Mike
0 Kudos
1 Solution

Accepted Solutions
267 Views
Registered: ‎01-22-2015

Re: ZCU111 current monitoring

Jump to solution

Hi Mike,

This seems destined to be a lonely and unanswered post (ie. a little used feature on an expensive board).

So, thought I’d offer you some company and thoughts. 

I “havn’t been down this path before”, but from looking at the ZCU111 schematics, I can confirm the connections you have described.  It seems to be an I2C thing (U67 is I2C ADC,  U23 is I2C mux, U20 is I2C level-shifter).  So, you need to make the FPGA talk I2C – and use the I2C address of U67. 

I am not familiar with a PS-side approach to I2C (although I’m sure it exists and is probably the easiest route).  However, for the PL-side approach (via pads AW16 and AV16 in bank-64), you could probably use the AXI IIC IP from Xilinx (ref document PG090).

Cheers,
Mark

View solution in original post

5 Replies
268 Views
Registered: ‎01-22-2015

Re: ZCU111 current monitoring

Jump to solution

Hi Mike,

This seems destined to be a lonely and unanswered post (ie. a little used feature on an expensive board).

So, thought I’d offer you some company and thoughts. 

I “havn’t been down this path before”, but from looking at the ZCU111 schematics, I can confirm the connections you have described.  It seems to be an I2C thing (U67 is I2C ADC,  U23 is I2C mux, U20 is I2C level-shifter).  So, you need to make the FPGA talk I2C – and use the I2C address of U67. 

I am not familiar with a PS-side approach to I2C (although I’m sure it exists and is probably the easiest route).  However, for the PL-side approach (via pads AW16 and AV16 in bank-64), you could probably use the AXI IIC IP from Xilinx (ref document PG090).

Cheers,
Mark

View solution in original post

Teacher drjohnsmith
Teacher
254 Views
Registered: ‎07-09-2009

Re: ZCU111 current monitoring

Jump to solution

The user guide for the board

https://www.xilinx.com/support/documentation/boards_and_kits/zcu111/ug1271-zcu111-eval-bd.pdf

page 88 has a bunch of details on voltage and current monitoring, 

   I seem to rember it can be read via the default app on the card. 

 

<== If this was helpful, please feel free to give Kudos, and close if it answers your question ==>
Voyager
Voyager
209 Views
Registered: ‎06-20-2017

Re: ZCU111 current monitoring

Jump to solution

Yes, markg@prosensing.com , I agree.  But this card and chip is pretty awesome.  The VCCINT on the ZCU111 is rated for 30 amps.  Depending on how many DSPs used (out of the 4272) and at what frequency they are clocked at, it is not hard to get to 30 amps.  My lab measurements are showing 28.65 Amps using just 50% (2136/4272)  of the DSPs at 600 MHz.  At 62.36% (2664/4272) of the DSPs, I am pulling 35.35 Amps, and the card/28DR can run continuously, even though I am exceeding the max current of 30 amps specified for the card.  At 68.54% (2928/4272) of the DSPs, we're pulling 38.8 amps, and the card resets after about a minute or so.  At 75.3% (3216/4272) of the DSPs I am drawing 41.2 amps, and can continue to do so for about 30 seconds before the card resets.

I am starting to suspect the ffvg1517 package cannot handle this current (based on waring in report_power, after changing the default switching activity to what I am doing), or there is a voltage or monitor generating a reset to prevent damage to the card.  In other words, based on my experiments, I think I am going to need some sort of throttling mechanism based on current and/or temperature.  However, even at 41.2 amps, the system monitor is reporting only about 63 degrees C when it resets, so I think it is related to the current.  It would be nice if a throttling infrastructure was available since I guess anybody using this chip at its apparent limits will run into a similar problem.  My switching activity is probably higher than many applications (think cryptography like distributions, with no correlation between one sample and the next), so this may be a pathological case that is not likely to be encountered in most designs.

Note:  My post previously referenced a different VCCINT from U59 on the ZCU111 that was only rated for 10 amps.  I am actually concerned about U68 which is rated for 30 Amps.  (Table 3-29 is a multi-page table not sorted by the voltage rail, and I missed the *real* VCC INT I was supposed to be interested in.)

Mike
0 Kudos
168 Views
Registered: ‎01-22-2015

Re: ZCU111 current monitoring

Jump to solution

Mike,

Wow!!!  -and thanks for sharing that with us.

Mostly, I putter around with 7-Series devices/boards.  When/if I graduate to the big leagues, I'll know who to come to for advice.

Thanks,
Mark

0 Kudos
Voyager
Voyager
142 Views
Registered: ‎06-20-2017

Re: ZCU111 current monitoring

Jump to solution

Ha!  I am sure you will figure it out, markg@prosensing.com .  It is just slower going if you don't have a team of people.

Mike
0 Kudos