Author Topic: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs  (Read 1814 times)

0 Members and 1 Guest are viewing this topic.

Offline aussie_employerTopic starter

  • Newbie
  • Posts: 4
  • Country: au
Looking to find someone familiar with or happy to quickly independently resolve the following...

  • Kicad 5.1
  • SPI bus annoyances (long bus, 50cm, 6+ devices)
  • Routing 10BaseT over non-standard connections (physical push-pull capable connections; see image attachment)

A few different projects immediately, plus sporadic ongoing work. Would be great if you have practical experience preparing hardware for small volume production (assembly and cabling concerns, safety, testing, reliability, etc.) Please send an email through the forum (https://www.eevblog.com/forum/emailuser/?sa=email;msg=2509890) to express interest.
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3893
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #1 on: June 26, 2019, 06:44:56 am »
Ethernet should not be routed accross these connectors. Neither should SPI.  Different connectors should be picked.
 

Offline aussie_employerTopic starter

  • Newbie
  • Posts: 4
  • Country: au
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #2 on: June 26, 2019, 06:48:07 am »
Thanks, SPI not using these. Our info is 10BaseT can take a lot more punishment than faster ethernet. Welcome to recommend a different push-pull connector.
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3893
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #3 on: June 26, 2019, 07:44:08 am »
At least step the connector size down to 3.81mm pitch (3.5mm pitch are not as robust imho). But for such very small signal stuff as SPI and data comms, I'd suggest picking a crimp style connector, where wiring error is minimized by the  end-user/customer/installer.

I know even from my experience, that ETH runs fine even through wirenuts and such, but a small fine pitch connector, such a JST-XH* or any similar will be a) cheaper b) better suited for the task.

*Note: most crimp style connectors may be used only with stranded wire, ETH is prevalently just hard wires.

If an ETH cable needs to be terminated to your board, why not to use a standard RJ45?

I would leave these huge pluggable screw terminals for  power stuff only (ie, really not SPI).

But I do not know a lot of details about your application, so my suggestion may become void. I would gladly try to help, but unfortunately my tribe does not speak KiCAD.
 
The following users thanked this post: aussie_employer

Offline aussie_employerTopic starter

  • Newbie
  • Posts: 4
  • Country: au
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #4 on: June 26, 2019, 08:03:53 am »
Thanks for sharing your thoughts. The push-pull only (zero manual lock/release allowed) requirement means both standard RJ45 and JST-XH are no good. The current thinking is that, in the unlikely event of slight misalignment (significant physical motion guides are present), some force translation may occur through the connector, so a more industrial connector may help assure long term reliability. Again, this is only for short run 10BaseT ethernet, not for SPI.
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3893
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #5 on: June 26, 2019, 10:30:58 am »
So the PCB shall act as a slot/card, that is pushed/slided into an encosure?

In such case, I'd opt for completely different connectors, these green terminal block I can't even tell how you would mount that (the plug). Any drawings/renders available?

I'd  use a either connectors specifically designed for hotswap/edge plugging and design the backplane such, that it will provide correct terminations (i.e. RJ45 for ETH).

For example, PCI-Ex style connectors shall be pretty cheap and usable fort this purpose.
 

Offline wilhe_jo

  • Regular Contributor
  • *
  • Posts: 174
  • Country: at
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #6 on: June 26, 2019, 03:46:26 pm »
Might be happy to resolve this but email-through-forum seems to be broken...

I'll try to PM you.

regards
 
The following users thanked this post: aussie_employer

Offline aussie_employerTopic starter

  • Newbie
  • Posts: 4
  • Country: au
Re: A few KiCad 5.1 design / rework (2-layer / ATMega328P) jobs
« Reply #7 on: June 27, 2019, 07:17:33 am »
Replied to PM.

The issue is less the connector and more the circuitry enabling it. Usually the RJ45 box has magnetics inside which allow the 10BaseT link layer to function. We need to replicate that manually on PCB owing to the use of an alternate connector and haven't yet had success. Anyone with experience here would be well placed to take my money.  :-DD
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf