EEVblog #1262 - Designing a Flex PCB + uSupply Update

[EEVblog]

Designing a flexible PCB adapter board for the uSupply project.
Discussion of rigid flex PCB design, layout, and manufacturing.
Plus a rant on ST Micro parts and USB Power Delivery.

[jazz]

Nice to see the progress on the uSupply.

I know you're planning to use a custom LCD on it, but the renderings made me wonder. Did you choose the custom LCD dimensions to be similar to some easily obtainable colour dot matrix LCD, so if you should ever want to do such an LCD upgrade, you could reuse the case without much trouble? Or did you select the LCD dimensions more or less at random, without anything like that in mind?

Also, does David2 get paid by the hour to work on the uSupply, or does he actually have a stake in it and gets a percentage of the revenues?

[sakujo7]

I'm very curious as to why this mega-library is so big. Anyone got a link to it? Found very little on google.

Edit: Found the part of the video mentioning closed source and undocumented registers, wtf? Why even have a (partial) register listing if the only realistic way of getting it working is a bloated binary blob?

[FrankBuss]

I wonder if there could be a problem for the 8 MHz crystal with the long traces on the flex PCB. Usually the app notes say to place it as close to the IC as possible.

[sakujo7]

Had a quick look at the USB PD specification... it's 657 pages! Pretty crazy when the most common use case is just switching from 5V to 9V. 

The latest revision supports 3.3V to 21V programmable voltage and current limits... You could probably make a "uSupply lite" that's just a controller with no regulation...

[langwadt]

I wonder if there could be a problem for the 8 MHz crystal with the long traces on the flex PCB. Usually the app notes say to place it as close to the IC as possible.

yes, should probably have had the crystal and some decoupling caps on the flex

[Nominal Animal]

This is probably an utterly silly question, but why not put the adapter PCB between the board and the existing components?
There aren't that many, so unsoldering them, soldering the board, then adding the components on top, should be doable?

[Unixon]

Dave, There's a better way to solve this with a rigid patch than attaching flex PCB, but only if boards aren't assembled yet.

What you basically do is design a new version, then cut out a minimal piece of project and move it to a patch board. Existing pads covered by a patch PCB go to the bottom layer of the patch thus making an LGA-type package, replacement circuit goes to the top layer. In worst case a 4-layer patch PCB may be necessary to do all connections, but in this case you  can probably get away with a 2-layer patch board. However, it may be difficult to align a patch board perfectly with fine pitch bottom pads.

[vanarebane]

Add that bypass cap on the FPC

[Fran_PSR]

As always an excellent video, very didactic and entertaining. Now I have a question. I see that the film has some pads to connect the two-sided tracks. Is there a risk that those pads make contact with any component that is on the pcb?
Thank you

[EEVblog]

As always an excellent video, very didactic and entertaining. Now I have a question. I see that the film has some pads to connect the two-sided tracks. Is there a risk that those pads make contact with any component that is on the pcb?

I mentioned that in the video. One of the reasons to tent the vias.

[EEVblog]

Dave, There's a better way to solve this with a rigid patch than attaching flex PCB, but only if boards aren't assembled yet.

They are assembled and that's the whole point.

[Unixon]

They are assembled and that's the whole point.

Well, that's unfortunate. At least you may try to flip this chip-on-flex on its top and glue with epoxy back to the same spot to save height.

BTW, why do you need a proper USB PD and why the hell these libraries so huge?

[Paul Bryson]

I had a similar problem and came to a similar solution.  I also performed a few experiments to try out different techniques.

You can see my results here:
http://www.brysonics.com/we-dont-need-no-stinking-bodges/

[German_EE]

OK Dave, thank you for another exercise in PCB design, useful and also entertaining.

HOWEVER...............................................

I can live with most of your catch phrases, but if you use the 'Polly put the kettle on' phrase to describe a piece of plastic once more I will jump up and down on your head until the squishy bits come out.

        

Thank you.

[fmiz]

[...]
I can live with most of your catch phrases, but if you use the 'Polly put the kettle on' phrase to describe a piece of plastic once more I will jump up and down on your head until the squishy bits come out.

I thought that he was being serious and he was saying the name of a specific material, then a generic one (polyamide), and I just gave up trying to understand what he was saying because of how fast he can say 'Polly put the kettle on'! You gotta be shittin me, bloody catch phrases! I'll just trust that it's a joke, no worries! It's a trap for non-native speakers, beauty!

German_EE let's meet on our way there 

[thm_w]

From the recent Great Scott video I managed to locate this adapter: https://www.ebay.com/itm/Super-Mini-ZYPDS-DC-mini-20V-trigger-support-millet-65W-PD-power-supply-/132714794782
Which is intriguing as its using a single TSSOP part (no PD chip, they have other devices with the PD ICs too). I don't see any markings, it could be a micro or a standalone PD IC.
Also it is only capable of 15 or 20V, not sure if that was intentionally chosen or a limitation of the IC.

I'm very curious as to why this mega-library is so big. Anyone got a link to it? Found very little on google.

Edit: Found the part of the video mentioning closed source and undocumented registers, wtf? Why even have a (partial) register listing if the only realistic way of getting it working is a bloated binary blob?

The STUSB1602 library fits in about 45kB or so, you can find it here if you want to look: https://github.com/st-one/X-CUBE-USB-PD
A lot of it is the USB comms itself, the PD stuff doesn't look too big.

But anyway, as mentioned above, PD covers a LOT and implementing all of these options in a library takes space, even if you only use one of the features. Sometimes the compiler can strip out the unneeded stuff and save space, but it depends how its been written.

I had a similar problem and came to a similar solution.  I also performed a few experiments to try out different techniques.

You can see my results here:
http://www.brysonics.com/we-dont-need-no-stinking-bodges/

TQFP one is awesome.

For a single line of pads, small LCDs seem to have standardized on a pad with a via or two that worked well when I tried it: https://imgaz.staticbg.com/images/oaupload/banggood/images/C1/CA/42bf26a4-a71a-46d1-a09e-f3abf4ed7ceb.JPG

[Xenoamor]

Definitely hyped for more uSupply videos!

[Nominal Animal]

if you use the 'Polly put the kettle on' phrase to describe a piece of plastic once more I will jump up and down on your head

While you're there, would you mind telling Stefan Gotteswinter to pronounce surface as "sur-fis" (/'sɜːfɪs/ or /'sɝːfəs/), not "sur-face" (/'sɝːfeɪs/)?

Always reminds me of a person I knew who pronounced bicycles as "bi-cycles" (/'bɪsaɪkəl/), not /'baɪsɪkl̩/.