search

opulo-inc / feeder

Source for LumenPnP Feeders
GNU General Public License v3.0
76 stars 30 forks source link

Reconsideration of the whole power subsystem. #34

Open willyvmm opened 1 year ago

willyvmm commented 1 year ago

Version Number

All

Bugfix or Enhancement

Both

Description

Reconsideration of the whole power subsystem.

Suggested Solution

I did a quick design review, and I've found some (possible) issues. Some of them have already been pointed out, some are still hidden.

  1. Why MC34063 ??? there are absolutely no benefits to using this chip. Instead, we have a lot of possible issues. So in general all I can tell about this chip is: It is cheap(not exactly) and it works. Nothing else. - NEW
  2. No Fuse. #31
  3. No UVLO (Under Voltage Lock-Out) Protection. - NEW
  4. No Softstart on Input side (aka inrush current). #11
  5. No Sofstart on internal power. (Actually, it's derived from the lack of the softstart on the Input side, and it's uncontrollable) - NEW
  6. Missing capacitor on CMU Reset Pin (Delayed reset is a VERY DESIRABLE feature. In such a HotPlug design I would call it a must-have feature) Partially mentioned: #25 #16

My recommendations are:

  1. Use modern chip: Fx. TPS62932/TPS62933/TPS62933x - Overall BOM cost may be even lower or at least comparable to the current. (smaller and cheaper inductor, synchronous converter instead of the external diode, smaller output cap, built-in OVC protection, etc., etc.)
  2. NIS4461 by OnSemi https://www.onsemi.com/pdf/datasheet/nis4461-d.pdf (e-Fuse with dv/dt control for same price as quality resettable PCT fuse)
  3. Fixed by 1 and 2
  4. Fixed by 2
  5. Fixed by 1
  6. Add an extra 100n Cap on the NRST pin, and remove the pullup. This is also the recommended circuit mentioned in ST Datasheets.

@sphawes I can elaborate a bit more about all of these features if you're interested.

As I'm planning to build a set of feeders for myself, I will implement these anyway. My plan is to do that later this summer. Unfortunately, KiCad projects are not easily mergeable ... BUT if I get a message that You're interested in these changes I Can calculate all components, draw schematics and update PCB ASAP then create a pull request.

Best Willy.

30350n commented 1 year ago

Looking at the NIS4461, this is something that really should be done at the source ideally, right? Adding one of these per feeder is quite expensive and probably not worth it, given that the only benefit would be per feeder fuse funcionality. From what I can tell, regarding inrush current, softstart etc. a single device on the source of the power bus would be totally sufficient.

What about just adding a small fuse board between the motherboard and the IDC feeder bus as a fix for now? Ideally this would later be added to a new mobo revision, but this might be an easier and more cost effective fix for the people having inrush issues than shipping new power supplies to them (which might or might not solve the problem reliably).