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.
PCBway Flexible PCB quote: bit.ly/2Xmphrs
Forum: www.eevblog.co...
#FlexPCB #USBPD #uSupply
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
AliExpress Affiliate: s.click.aliexpr...
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.co...
T-Shirts: teespring.com/s...

[@lty1993]

Why not do a rigid flex PCB, which rigid part only over the original pad. Then there is initial height for the flex part, and probably no bending required for flex over resisters.

[@EEVblog]

Because there is no physical or electrical need for that, it would just be increased cost and lead time.

[@aramgaribyan6248]

Can put FR-4 stiffeners instead of making it rigid.

[@EEVblog]

Maybe if it was a production solution you would consider that

[@3lilmonkeys289]

I had to design a flex pcb for a diy project I was doing and rigid flex designs are HILARIOUSLY overpriced. If you absolutely needed it it would be cheaper, much cheaper, to design a normal pcb and a flex pcb separately and join them together yourself.

[@btouw8558]

There seem a lot of space on the board. Why dont redesign the pcb so it fits the other micro. It comes over like a hitech bodge job to mee. I dont understand why you are taking this approache

[@MasterAlexei1]

17:16 Dave, you looking in wrong datasheet for USB. The USB stuff are in the STM32G0x1 and your Datasheet corresponds to STM32G0x0. Try to look another datasheet.

[@kagu]

Found mistake in the video - little disappointed (but not so much - everyone make mistakes). Found polite remark about this mistake - I'm so proud of this community.

[@valamilamasag]

uSupply : a project which is in the works for more than 7+ years

[@EEVblog]

Closer to 9 years. I can remember showing Leo Simpson a prototype when he visited Altium. I left Altium in March 2011.

[@ryebis]

@@EEVblog Still using the LT3080 ? I was going to spend some time over the Christmas break building one based on Rev-C but with LM317 and ICL7660 to provide a negative bias to get the output voltage to zero, but I might wait for you to finish it :)

[@Audio_Simon]

That's okay I also have a project going for 10 years. Something else always takes my time.

[@Hasitier]

EEVblog I buy/back a uSupply as soon as it’s available. I like the design with the keypad input. Reminds me of my old lab power supply from Voltcraft which also has a keypad to dial voltage and current in.

[@godfreypoon5148]

That's why it's called "you supply""!

[@ExarchNZ]

Facepalm Dave! You did not read the that ST Micro table properly at all! You are looking at the datasheet for the STM32G0x0 line!!! The G0x1 line has the PD PHY!! If you read the table properly, none of the MCUs the G0x0 line does not have a PD PHY at all! G071 is about 50c more in volume. Compared to the $1.2 increase to the G4 series. I will concede that it is misleading that the G0x1 row shows USB PD phy, but not all have the PHY (only the QFP and larger pin count models do). But doing a Ctrl F on the datasheet for the wrong micro is a facepalm moment!!

[@AlexB-yx5no]

Am I the only one shouting at the screen at 44:29? The option says "Min hole size/ Pad size(diameter) > 0.15/0.35mm". To me, that means min hole size is >0.15mm and min pad size is >0.35mm, so Dave's design should be alright in terms of min hole size as his hole is 0.3mm, no pun intended!

[@AllElectronicsGr]

I stoped work with STM when I called them for support on a sample board and no one knew the responses until some one said I would need to attach the board to an oscilloscope and debug the problem of that crap board and crap libraries. Today for me it is a pleasure to work with Atmel ARMs. Beautiful documentation and great support.

[@sugarbooty]

Nothing but good experiences with microchip for me, too

[@emmettturner9452]

Keep in mind: a flex PCB mod/adapter board will even let you fold it back over into the footprint where it is connected. No reason to hunt for space all around that. ;) I often solder very small flex PCBs to a dense QFP, sometimes several a day: UltraHDMI mod for Nintendo 64. ;) The castellated edges of the flex mesh perfectly with one side of a QFP160 and it works incredibly well.

[@sugarbooty]

Thank you for sharing some wisdom with me!

[@ivanlevitskiy8058]

Hey Dave! Can agree with you that the st naming scheme is really messy but their table clearly shows that the STM32g0x0 doesn't have usb pd while the stm32g0x1 has it. Look at the stm32g071 datasheet. It's in there.

[@jovangrbic97]

and that just shows you how one wrong glance costs a project hundreds of $ more than it should have!

[@pr0engineer873]

Really love these type of videos; not only to get more insight into how the uSupply is evolving, but also to get the good real-world experience and tips when it comes to design, layout. Great stuff, Dave!

[@MrJef06]

About the undocumented registers and data sheets under NDA: seriously I don't understand why manufacturers do that. Are they trying to protect some kind of implementation secrets? I can't believe there can be anything to protect here -- anybody can define such a programming interface. They're just making it harder for people to adopt their solution. Oh perhaps they want to charge more to release the information or sell libraries?

[@sarowie]

Well: Fully documenting it would show to the world how painful it is in realty is. That might lead you to choice a manufacturer where setting up USB power delivery is shown in just a five minute youtube video... Offcourse when getting the part you have to deal with the NDA bullshit and undocumented crap, but at that point you are already locked in, so success for them and suck ass for the engineer.

[@squelchedotter]

What I especially don't understand is why designers accept parts that are under NDA. Like, that's an immediate "nope, won't use" for me.

[@zvpunry1971]

WeNeedMoreFarads: It is often not the choice of the engineer but the choice of someone else.

[@d3stinYwOw]

They do NDAs not only to hide painful stuff, also to scare smaller players in electronics field.

[@Necrocidal]

They might also hide some details so it's harder to infer the implementation, and thus harder for patent-trolls to attack them. There was patent troll (in Aus IIRC) who had a patent on portable devices with multiple power rails from a single battery source. I shit you not.

[@LydellAaron]

I want to take a moment to thank you for your contribution to society. Your content and consistency, curiosity and exploration capture the essence of a technical generation. Your blog series is so information packed and deserves to go in the (video) Library of Congress. Thank you from the bottom of my heart.

[@hempbear]

I liked your uSupply idea when it had a knob and own batteries.

[@TheDefpom]

Fold the flex over and secure the new package directly on top of where the original was supposed to sit, so it’s not on top of other parts.

[@martinda7446]

You know those 'bodge' wires Dave complains about.....This is a very sophisticated bodge wire...

[@felipemakara5067]

By the length of the video and the number of boards, delivery time and cost, I would argue that super-glue-ing the chip on the top of the U2 and hand-solder each pin is a viable option. (Done myself with simpler circuits, and seen done with a Ethernet PHY that was placed with the wrong footprint)

[@nicholasuloth6530]

yes handsoldering wires is quicker for a couple prototypes

[@ketandesai5326]

I appreciate the videos in Altium. Hope to keep seeing more design videos in the future!

[@testgearaddicts4607]

just imagine the Dave from 10 years ago, reviewing this board 3 years from now in a real teardown video. "there's some interesting engineering here for sure" "I wonder why they just didn't do it right from the start" "whoever designed this should be...." I miss that Dave.

[@RC-1290]

Normally when I watch EEVblog videos, I think "Ahhhh, something I find interesting, but know nothing about." Then suddenly it's about binary size. My pet peeve. Let me guess, it depends on 15 massive libraries?

[@jendib]

Crapy software depending on hundreds or even thousands of libraries are really the plague of modern software development. I look at you Electron...

[@SLLabsKamilion]

Would you believe me if I told you they forgot to turn THUMB emission on? (It's a joke, because all of the Cortex M series can only execute THUMB instructions, not the full fat ARM A-series instructions...) And no, since this is embedded MCU dev there's not as much library dependency hell, especially not when you're targeting 16-bit binaries to 32KB of flash. Might be able to get away with that on AVR8, but ARM's code density isn't nearly as high.

[@0xbenedikt]

@@jendib Yep, Electron is an absolute failure

[@Necrocidal]

How long until a vendor-supplied "embedded" software package contains a zipped node_modules directory.....

[@AndrewSkow1]

Thought I was going to see some curved traces!

[@EEVblog]

I'm lazy

[@StreuB1]

@@EEVblog The reality is, the reality. When you need it to just work, no reason to gild the lily.

[@lcdconsultant5252]

EEVblog is admitting your lazy hard? Being lazy is easy.

[@juanmanuelgranda17]

Dave, you said that in flex PCBs you would't do regular 45° angles in the traces, but then you routed them like that...

[@TomStorey96]

It's a prototype bodge fix. And you certainly wouldn't want angled traces if the flex was expected to move often. In this case it's highly unlikely to move very much once glued and soldered down.

[@lcdconsultant5252]

Lazy or “just prototype” is no excuse for 2nd rate designs. Extra 2 minutes to clean it up. Especially when you present it to the public. This is how China does design “good enough”.

[@tntodorov]

You could also find a surface-mount connector that fits the SO20 package footprint and have the matching connector put on the flex PCB. Something similar to Amphenol 20021221-00020T4LF or Hirose FX4A1-20S-1.27SV(71)... Molex and Semtec also have a lot of these high speed connector types. Some TFT displays - especially if they are MIPI-DSI - use those on flex PCBs.

[@spagamoto]

That binary blob will be reverse engineered so fast if that PD chip gets used in an eevblog thing, thanks to the forums. I'm hoping the manufacturer's hush hush strategy backfires a bit.

[@Audio_Simon]

OMG thank God you pulled back that anulus ring diameter! I can relax now. =)

[@StreuB1]

I'm sorry....at 13:40 when Dave said "So, we're gonna layout blah blah blah" I totally got excited. I am a mechanical designer and always felt I missed the bus with electronics and electrical and always wanted to learn about it. So, sitting shotgun watching this about to go down is hella exciting for me lol. Edit: Ok, just finished. That was badass. You could do more of things like this and it would be A OK with me.

[@TomStorey96]

He has done other videos where he lays out boards. He did one for his Nixie tube subscriber count display, and some time ago he did one for a power supply.

[@skyfly200]

Thanks for this! I would love a full flex PCB tutorial!

[@tomboxyz5564]

I've got an old Casio scientific calculator (25-30yo, my dad got it as a gift) and the whole circuit is on a flex PCB, membrane buttons, the main IC, screen is soldered directly to it, and there's a small hole to fit 2 THT diodes (leads are straight, just cut to lenght) for the solar panel, it's absolutely beautiful and insanely light, using it without the protective case feels very weird, since the back behind the buttons is recessed, so the flex rests on the enclosure

[@tookitogo]

TomboXYZ He’s shown teardowns of those on a few episodes.

[@viperidaenz1]

Dave, put your i2c pullup resistors on the flex. Bending problem solved.

[@SwitchAndLever]

You mention using rounded traces on flex boards, is there a reason why you didn't use that on this flex but instead used the 45 degree angled traces?

[@EEVblog]

Lazy Dave.

[@SwitchAndLever]

@@EEVblog haha! Gotcha! A "do as I say don't do as I do" kind of situation huh? ;)

[@MrJef06]

32KB? You should be able to fly to the Moon with that much memory! Seriously, is USB PD so complicated it can't be done in a few KB? Starting from scratch rather than trying to trim a bloated library? (I must admit I know absolutely nothing about the protocol though...) Edit: 18:55 apparently it is :-/

[@EEVblog]

Yep, no reason it needs to be that big I'm sure. Probably uses a ton of pre-written libraries and non-optimised stuff.

[@nicolaedumitrache9246]

Probably is written in C++ with templates...

[@dimtass]

There are many ways to minimize the size by just using compiler and linker flags. But yeah in general when I comes down to firmware, it gets more and more difficult to find optimised code and best practices. The reasons for that is another topic, though.

[@andymouse]

More about u-supply please!

[@jort93z]

If you can shrink a 1mb library down to 32k, your library was poorly made in the first place.

[@jojoposter]

They hated him 'cause he told them the truth

[@GGigabiteM]

As computers got faster and had more memory available, there was less incentive for programmers to write efficient and compact code. That incentive basically vanished when huge "do everything" libraries like .NET and Java came along where programmers could just use easy functions instead of writing their own, so you ended up with programs that are small, but get massively bloated from the gigabytes of dependencies that are included. If you want a small and efficient program/firmware/whatever, you'll need to find a programmer in the 40+ age group who still remembers the days of having to make every last byte count. A modern programmer is going to have a tough time weening themselves off of Microsoft's tit to be able to program on a microcontroller with just a few dozen kilobytes of memory.

[@666Tomato666]

@@GGigabiteM I mean, sure, when I'm doing Java, Python, or C++ with destination of server or desktop, I don't care if the lib uses 32k or 1M. But for embedded use? That's incompetence.

[@GGigabiteM]

@@666Tomato666 Not caring at all about executable size on any platform is incompetence, it's one reason why operating systems have become bloated nightmares in the past 13 or so years.

[@666Tomato666]

@@GGigabiteM I didn't say "don't care at all" what I meant is that normal, unoptimised code (but not enormous) will likely be a drop in a bucket compared to sizes of data processed. So usually it's better use of the time to make it faster than to make it smaller

[@Osmanity]

cant wait to see the result

[@Petertronic]

I can't wait to see more about this product!

[@Karshilistics]

Weird flex, but okay.

[@666Tomato666]

Son of a bitch. He stole my line.

[@doningram3978]

Dave, BTDT with legacy package replacement on industrial systems. Use 0.8 FR4 with 2 risers mounted perpendicular to the surface, no castellation required but ensure copper extends to the edge, no flex required... Use vias both ends near the pads to bring the traces up both sides of the risers. Mount the CPU board on top off in line from the pad rows, toward the top of the image, and extend a tongue to couple to the risers on their inside edges. All in a low cost quick turn FR4 process from any fab in the PRD. Glue some braces to support the assy.

[@enercotec350]

Dave, you will have problems with the annular ring for the 0.35 mm hole and the pads with the 0.5 mm pitch, I have worked with PCBWAY by long time pushing their limits in rigid boards and their standard annular ring for rigid is 0.15 mm, and I think you do not meet that...

[@z_deniz]

I think you can use qfn or bga packet to fit without over flow the old footprint. Especialy qfn packets has lower price tag in general.

[@Rong2k]

Have you noticed how specifying 5 and 10 units have the same cost? In my experiments, you can order up to 220 of 22x12 mm PCBs for the same price as 5 would cost. Might be a calculator error, though.

[@sarowie]

no, not really. When they make a whole panel just for you, that it is that way. I was able to order 1000 small Flex pcbs for 1000 Euros (which was the full initial series of the device), as the prototype price was also 1000 Euro, but just 100 boards, guess what I took. Note that even gave us the benefit of having the prototype board in series production quality in series production processes.

[@lcdconsultant5252]

Usual the cost is just a shipping cost difference.

[@TheVirindi]

"Libraries from the manufacturer are huge" and use up all your space? Gee, I wonder why that would be. On a completely unrelated note, I bet they just happen to offer a uC with more flash for a bit more money.

[@ThePlacehole]

They must have done something wrong. I'm a moron and even I've got the stm's usb stuff working on a similar chip. There was a lot of cussing, but it wasn't impossible. I'd be weary of using their prototyping libs in a real product anyway, that stuff is vile...

[@urdnal]

Just a couple of months ago, messing around in Kicad (mostly when I should have been doing my actual, unrelated day job) I designed a TSOP 44 to SOP 40 adapter for the Neo Geo MV-1C BIOS. Flex PCB, holes in the pads and everything. Surprised and glad to see my ideas are sound. Maybe I will have them made after all...

[@TheEmbeddedHobbyist]

you have to remember that if the flex is made out of Kapton you need to bake the PCB before use to allow any absorbed water to be removed before soldering. we used multi-layer flex pcb's to provide screened twisted pair's to replace wire. solid screens except where they bent, at this point it was xhatched. could reduce 20 twisted pair cableform down to one PCB from the connecter at the front to the motherboard at the rear.

[@povicollege3055]

Interesting video Dave. Thanks again! Please keep up the great work!!

[@brylozketrzyn]

We had to replace WLAN module in small batch of much more expensive boards (just for evaluation purposes). I've ordered rigid PCBs based on 0.3mm substrate and made adapter hidden under tested module completely. This way I created a breakout board for existing module, which was indeed a breakout board for LGA module on top of it; the LGA module was indeed a breakout board for final IC (how many breakouts do you need?) In your case I'd probably go for some WLCSP package and make small PCB for it.

[@Adi-lv9bo]

I'm really looking forward to see result pcb and how is assembled. You have done a nice and interesting work!

[@SkylerOlsen]

Enjoyed it Dave. Would love to hear more about the USB-C Power Delivery. Currently designing a product using PD with the FUSB307B chip and it's been a learning experience to say the least.

[@MaX271]

are you not worried about poor power supply decoupling with those thin traces? you couldn't get one of the small QFN/chip-scale BGA to fit inside the SSOP footprint instead of the TQFP?

[@remotepeak]

On flex board, next to power pin, should be placed decoupling cap. Mandatory. Capacitor on rigit board is too far away.

[@fabianfeilcke7220]

I used ST micros for a lot of projects in the past but lately they are pumping out new Controllers without properly documenting them. Back in 2011 we tried to implement a Flexray-CAN FD Adapter using a SPC56XL70-Series controller. In the end we gave up on this as the datasheet was not complete and several registers for Flexray were completely unexplained. Other Register descriptions were plain wrong. The correct datasheet came out in 2013! As for your flash problem. I usually do the software first on the biggest available Controller. Once thje software is final i reduce the flash size for production as far as possible.

[@etome8]

Let David2 rant on ST, at least on the second channel. I'll watch till the end.

[@_ATHONOR]

Note: You can model bend lines in AD19! Use the example one to try it out on

[@Nebulorum]

Like the practical example. Wondering if it wouldn’t be easier to have a raiser SOT20 to clear all components, like they use to do with DIP packages.

[@breakalegfpv9532]

I watched the whole thing and I still don't get it. I am going to watch it again though because I feel smarter when I do, thanks.

[@emmettturner9452]

You must have early access considering this hour long video is an hour old and so is your post.

[@breakalegfpv9532]

@@emmettturner9452 this posted 10 hours ago

[@emmettturner9452]

@@breakalegfpv9532 Not for me. Even now it has only gone up to 2 hours though it's gotta be pushing 3. I refreshed. Got the notification pushed to my phone around then too.

[@breakalegfpv9532]

@@emmettturner9452 that's HI TEC 4 YA ;)

[@darkwinter6028]

“You? Tryin’ to flex on me? Don’t be silly...” 🤪

[@jorditribo94]

Nowadaws you don't need solder mask expansion. In fact some manufacturers remove it from the gerbers.

[@michelfeinstein]

AFAIK crosshatching is used just to increase the flex PCB flexibility, as a solid slab of copper won't flex as much, so crosshatch it if you think you need the flexibility.

[@kstxevolution9642]

oh god i hate the letters thing on ST Micros. I had an stm32f103c8t6 and it took me MONTHS to read that it was c8t6 on the silicon and not cBt6. couldnt tell the bloody things apart

[@Arek_R.]

I hope this time I won't miss the kickstarter campaign, last time I wanted 121GW, sold out before I was able to get it...

[@Alexandermichel1994]

Thanks for the Video! Very interesting stuff. Will there be some Videos coming about the rest of the uSupply design process and choices? Would be interesting to see more of the development. -Alex

[@gotj]

Move the decoupling capacitors next to the chip!!

[@AnotherBrokenToaster]

Did you consider ordering from OSHPark, they charge 10$ per sq inch for flex prototypes.

[@Roobotics]

For figuring out complex routing around components, just copy all the parts in the afflicted area to a new design, and create mech layer keep-out boxes around them, delete parts, make flex PCB outlines. Keep outlines as absolute measurements so you can avoid 'eyeballing it' when the design inevitably doesn't fit the first try through.

[@vincei4252]

Putting out incomplete data sheets for parts that you can actually buy from a vendor is a kind of scummy practice. I guess only people with preproduction samples get the actual datasheets under an NDA? wtf.

[@RoterFruchtZwerg]

I have a Fingerprint Reader module with red and green LEDs. There is no docs how to control these LEDs and the red one is never lit. I asked the vendor and he said he signed an NDA with an other customer and cant tell me how to use the red led 🤦🏻‍♂️

[@redtails]

honestly why go for flex? small normal rigid pcb with some pins sticking out. the pins provide clearance to the pads and you can solder them to the board

[@EEVblog]

Height is an issue. Also, it's a nicer solution, and excuse for a flex video.

[@moertelruehrer]

I think the USB PD implementation would also be a very nice video!

[@protonjinx]

Why not use "rigid-flex", the rigid part only covering the main board footprint, and then flex out the new chip? reduce stress on the flex/soldering and maybe simplify the interface between main board and adapter board?

[@TheHuesSciTech]

Because that would be a completely, utterly unnecessary overcomplication?

[@protonjinx]

@@TheHuesSciTech well, if you want to go down that path... changing chips n making patch-pcb's just because you cant fit code in 32k or less sounds like youre doing something wrong. its a powersupply, how complicated do you need it to be, really?

[@EEVblog]

Because it's not for production, it's just a quick hack for handful of prototypes.

[@Micetticat]

there might be space to make two unplated small holes in correspondence of the capacitor and the ICs that are on the way. That way the flex PCB would lie flat even with the capacitor mounted.

[@dman8734]

I had a hunt, and the bit you missed in your video was the G0x1 series has USB-PD, while the G0x0 series (that you showed) does not. However, none of the range seems to include USB FS/HS so if you want to have data, you'd need a different chip! The thing the suppliers don't want to do is provide support - and as you go up the value chain it gets worse and worse. You need to convince them to let you in which relies on demonstrating capability. Terrible datasheets is a good way to make sure people can't use your part. ST, on the other hand, was probably releasing a bit earlier than they should have. There's a lot of moving parts in releasing a chip and clearly someone was later than they were expecting. In my experience, ST's documentation is so good I haven't needed to contact them.

[@pavlitt355]

Great video, Dave. Thanks for covering flex with a real life use. As things get smaller, the more we have to design things just for design development purposes! P.S. Hole size was fine wasn't it? Manufacturer's capability was 0.15mm min hole with 0.35mm min land I think.

[@vandalest]

Add that bypass cap on the FPC

[@a1nelson]

Hmm. I must be one of those screaming viewers. The 0.3mm vias are probably fine, but I would not feel totally confident that this flow-through plan would result in quality joints. I’m not seeing why one wouldn’t simply add ~3 vias (potentially smaller for lower impedance to the traces) to provide sufficient heat flow to the lower layer, with a plan to simply perform the actual flex-to-baseboard assembly with solder paste. Seems like the “flex” is still rigid enough to lay flat on the pads. A little bit of fixturing/helping hands or whatever will be required in either case. Maybe another commenter can clear up any misconceptions. Thanks for the quality vid.

[@EEVblog]

Two vs three holes, meh, hardly any difference.

[@DFlame69]

Add holes for plastic clips so you can deal with vibration stress.

[@vincei4252]

New to Altium using 19.1.8 . Struggling thru the online tutorial where after creating the components and generating a PCB file the component outlines show but the air wires didn't. Spent several hours trying to find out why and shutdown the PC in frustration several days ago. Saw this video with you mentioning v19 , reopened Altium, and lo and behold the air wires show. Go figure.

[@envisionelectronics]

Vince I Altium professional here: the missing air wires bug has bitten me on a production run. It’s truly unusable at times.

[@yngndrw.]

You could route holes out around U2, U21, C88, R60, R62, etc so that the flex PCB will sit flush with the board and place the new microcontroller between U2 and U11, then you don't have to remove the capacitor or worry about the bend radius. You could also route small slots over the U7 solder mask for easier alignment.

[@yodjisuzuki9033]

it's just freestyle by DJ

[@EEVblog]

Sure, several different ways to do this like I said.

[@greglaletin]

40:50 light bulb moment, I was cringing on you finishing the traces before picking that up.

[@robk5969]

17:06 no usb controller, thats the STM32G070CB, which fits in the "G0x0" (x is wildcard) row of that features table, the top row, doesnt have usb power delivery the STM32G071CB (g0x1, the row that has usbpd ticked) does have usb pd, on page 33 www.st.com/resource/en/datasheet/stm32g071cb.pdf

[@michelfeinstein]

I am quite sure you don't need to round the outer edges of a flex PCB, they won't be subject to any mechanical stress, just the inner edges will I believe. (Dave shows this later on the video, although in the beginning he shows something different).

[@pavloskokozidis8233]

Can you please make a video for ESD PCB design of a device or a tutorial ? Realy love to see it in combination with all those !

[@Daniel-ib5bx]

Interesting schematic layout. The no grid is making me a bit anxious haha

[@electrolance7505]

why there's no bypass cap for the uC on the flex ?

[@marten.wikman]

Some people just want to watch the power rails collapse.

[@tomgeorge3726]

Hi Dave, great video. [soapbox] As a service and repair engineer I keep seeing schematics of the type you have there. They are great for PCB design, but bloody hopeless mate when it comes to using them for hardware trouble shooting. I call that type of schematic a "search-a-word" schematic. For example, you have a connection cc1; 1). Where does it go? 2). How many components does it go to? I have to play "search-a-word" over the ENTIRE schematic to make sure I have ALL the cc1 connections. Data and Address busses are not a problem due to the size of their presence in a schematic. I don't want to have to have "Paul Daniels" Apple schematic software or the expensive CAD it was made with to do the job. We still need a CONNECTED type schematic, OR a way of stating where cc1 goes on the schematic (not a table/legend at the side of the schematic). [/soapbox]

[@Blue7Alien]

On the STM32G0 topic, I use it in my LiPow USB PD Lipo Battery Charger. I don't need normal USB functionality, only PD, so it worked out. I had initially designed it with the STM32F0 + external USB PD controller like you did on the usupply. But after attending an ST info session last fall, I switched to the G0. It turns out that it was not very simply to get the PD middleware up and running for just receiving the power profiles and setting one. Their "middleware" for the PD on the G0 was specifically written for their G0 demo board. Not intended to be portable... I did get it up and running with a few hacks. You can see the code in the repo. Also, it can be dual role, not sink only like you stated in the video. github.com/AlexKlimaj/LiPow-Firmware

[@sanityd1]

Cool video Dave, I never would have thought of the flex pcb solution. When I first read a ST uC datasheet I thought 1000 pages was a bit steep, a month later I was like where are the other 5000 pages. Sadly I get the impression most silicon vendors are the same, fringe Bluetooth uCs are just absurd, you need to read every datasheet the manu has written in in the last 5 years just to understand the basics of one product. I looked at the ST G series chips after you first mentioned them, went to the datasheet, found nothing for USB PD. Most of my USB power needs are low, unlike you who is actually trying to use the features of the USB PD spec. The USB spec and all the subspecs are a mess, I blame them along with silicon vendors.

[@MichalCanecky]

Don't you need a decoupling cap on the flex board for the micro?

[@jonathanwilhelm8891]

Can I give 2 big thumbs up. This plays right into my business card idea

[@MSP_TechLab]

Why wouldn't you search for some SMD connector? Probably you can find one which almost fits pattern of old stm (or perhaps connector on another adapter board). In this case you'll solve problem with soldering of flat flex.

[@Topolis]

Hi Dave, love your PCB design videos. How about using a thicker FR4 with castellated pads for U2 chip and cutout slots to clear discrete components and putting your new chip on this daughter board somewhere closer to USB port? You might even add extra castellated pad for more stability and solder it to pad next to your switch, sorry can't quite make out the silk screen part number. I'm only 14 min in so this might have come up later.

[@GlennHamblin]

I'm glad you changed the board around and re-routed it. It was pretty ugly on the first go. 😁

[@tamyboy1]

Good video

[@dGroupcom]

great example of when design goes sideways. Actual life is not perfect. great job.

[@Mtaalas]

Why in the heck was that library so large? :D You gotta make a video or write a blog about it... i mean would be a wonderful example to firmware developers that "hey... don't make libraries unnecessarily large and bloaty and this is how you make them smaller!" :D

[@MetalheadAndNerd]

Probably the library contained workarounds for non-standard Behavior of lots and lots of USB-Hostcontrollers and probably support for several proprietary protocols which device manufacturers use to negotiate the power delivery.

[@Mtaalas]

@@MetalheadAndNerd In that case they could have at least make the damn library modular :D Though maybe they were thinking that "if someone puts a usb-c onto something, they have gigabytes of memory anyway!" :D

[@MetalheadAndNerd]

@@Mtaalas Dave never mentioned noteworthy programming skills. Maybe his discussion with the manufacturer was him being told which includes he should leave out?

[@andymouse]

@@Mtaalas modular...that makes to much bloody sense.

[@Mtaalas]

@@andymouse Oh...right, i'm really sorry for suggesting anything so sensible... ;D

[@vmiguel1988]

Jesus! Dave break that corners!

[@ANTALIFE]

Have seen people use rigid PCB for mods on larger boards (just like yous wrong footprint), but have never seen a flex PCB being used. Am super interested to see how it goes Dave

[@OmarMekkawy]

Hi Dave, I saw before a FR4 PCB which a flex connected to the top of it ( all in one piece ! ) without even soldering them together. This might help you for making the mod. You could find it in the moving head of the DVD/CD Writer.

[@AMalas]

That is a rigid flex, as discussed in the video, they are quite expensive!

[@SLLabsKamilion]

Yo, the Value line doesn't have PD or USB like you said at 16:40. You need the Access line. STM32G071 not STM32G070.

[@Graham_Wideman]

21:48 "caps and LEDs". Pretty sure those are resistors in series with the LEDs :-)

[@gusbert]

Questions I always ask: 1. Is this micro available with bigger memory options in the same package? Software always expand to fill the available space. 2. Whenever I see a line of vias or pads, I ask "Can I route a track between them?" 3. Where the f**k is pin 1 on a connector (in this case the pads for the PCB IC) especially when it is rotated and/or flipped.

[@JGunlimited]

Follow up? Did the flex PCBs arrive from the manufacturer? What do they look like?

[@lennynnnnnn]

Hi Dave, Can you talk about the choice for locating banana jacks on bottom, and USB on top? Why not the other way around, all connectors on top or all connectors on the left? Just curious as to the usability decisions you guys made and how other trade-offs such as board layout impacted this.

[@EEVblog]

Power supplies are usually located *above* the product you are using/testing/designing, so you want the output connectors on the bottom obviously.

[@sarowie]

I would have "anchored" the pads in the flex with traces into the flex material "just to be safe".

[@jasonhowe1697]

i would think going the route of press in chips you be a benefit unless you have a height limitation requirement for the board you are using