Author Topic: STM32F417 - any reason why a min PCLK2 speed is required for ethernet to work?  (Read 2841 times)

0 Members and 1 Guest are viewing this topic.

Offline peter-h

  • Frequent Contributor
  • **
  • Posts: 757
  • Country: gb
  • Doing electronics since the 1960s...
I looked for my original thread there (https://community.st.com/) but can't find it, can't find (there or via google) the above text, and there appears to be no way to look at my own posts in my profile there.

EDIT: the info posted came from ST by email.
« Last Edit: June 08, 2021, 10:53:36 am by peter-h »
Z80 Z180 Z280 Z8 S8 8031 8051 H8/300 H8/500 90S1200 32F417
 

Offline wek

  • Contributor
  • Posts: 9
  • Country: sk
Probably related: https://community.st.com/s/question/0D50X0000BABrWMSQ1/stm32f767-no-ethernet-when-apb2clkdivider-rcchclkdiv8-or-greater

JW

(That thread has more than 10 messages and one has to click on blue "More answers" at the bottom to show them. If there's no "More answers", reload. ST insists of using Salesforce instead of proper forum software.)
« Last Edit: June 09, 2021, 02:32:35 pm by wek »
 

Offline peter-h

  • Frequent Contributor
  • **
  • Posts: 757
  • Country: gb
  • Doing electronics since the 1960s...
Yes - same issue exactly.

Interesting that DIV8 or DIV16 is reported as breaking it. I found that DIV16 definitely broke it, but DIV8 possibly didn't; may have been right on the edge. So clearly DIV4 or DIV2 are the only sensibly available values.

That ST guy, Adam Berlinger, is the one who sent me the email.

Yes, their forum software is crap, but this is not unusual; U-BLOX use a similar useless one e.g.
https://portal.u-blox.com/s/question/0D52p0000AT0rV9CQJ/how-to-detect-waasegnos-is-actually-being-used-for-the-fix

« Last Edit: June 10, 2021, 09:07:08 am by peter-h »
Z80 Z180 Z280 Z8 S8 8031 8051 H8/300 H8/500 90S1200 32F417
 

Offline ajb

  • Super Contributor
  • ***
  • Posts: 2070
  • Country: us
Interesting that DIV8 or DIV16 is reported as breaking it. I found that DIV16 definitely broke it, but DIV8 possibly didn't; may have been right on the edge. So clearly DIV4 or DIV2 are the only sensibly available values.

If the issue really is that the MII/RMII switch setting needs time to be completed before the MAC is reset then no, DIV2 and DIV4 aren't the only available values, they're just the only values where you don't have to be aware of the timing issue between the two clock domains.  That's still a really annoying bug, and maybe they could have solved it with some better design decisions in how the MAC integrates with the system, but it's not that hard to work around it if you need to use the other DIV ratios.
 

Offline peter-h

  • Frequent Contributor
  • **
  • Posts: 757
  • Country: gb
  • Doing electronics since the 1960s...
The above suggested fix works perfectly. ETH now running with DIV16 on PCLK2

Z80 Z180 Z280 Z8 S8 8031 8051 H8/300 H8/500 90S1200 32F417
 
The following users thanked this post: thm_w


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf