Click to view our Accessibility Statement or contact us with accessibility-related questions
Showing 1 of 1309 conversations about:
Parak
89
Input Club
Jul 21, 2015
bookmark_border
Hey folks,
So on the soldering side, this may have been mentioned before, but we are switching PCB fabs due to the inconsistency of the previous one. We are QAing them carefully and made sure that they are IPC certified for assembly and soldering quality, as well as packaging. That should take care of issues that came up with some boards on the previous (second) drop, such as diodes missing from the board.
The USB connector, a Molex 67503-1020, is a fairly widely used part on many a controller and other keyboards (for example Pokers used it or similar). While the Massdrop support only had a case or two of it falling off (at least partially due to the aforementioned fab issue), I felt that making it a bit stronger wouldn't hurt, so for the current drop (3) that ended the connector will be reinforced via through hole soldering of the shield tabs. I would still suggest that folks reflow each of the 4 shield leads of the USB connector if they can, but they also shouldn't worry about it too much.
On the matter of the issue that @chilofthehorn was having, it's definitely a strange one that we've been so far unable to replicate, and as per Massdrop support nobody else has reported. We do spec proper nongeneric components (usually picked out at Mouser and Digikey), but I'll also be taking greater care to avoid even the possibility of generics for future assembly runs. Unfortunately, I do not know for certain if any generics made their way into the previous runs, but based on the numbers so far I don't believe many did if at all. I'd prefer to avoid usage of extra components because of how unreproducible it so far is, although we're definitely open to the suggestions listed and may implement TVS diode protection later on especially if we ever implement USB host or hub.
0402 parts are used primarily because it's still easier to route with them used as opposed to larger components, and also because we can keep portability of the BOM from board to board using the same component sizes, regardless of how much room we have. There's absolutely no reliability difference between large or small parts that are of the same spec, nor should any just randomly fail like a catalytic cap (which are not used) and require replacement. They certainly can still be hand soldered, though right equipment is needed like hot tweezers or hot air, but again we don't expect anyone to do that and they should contact support if they suspect a failed component.
Of course, we are keeping all these things and more in mind for the Infinity Ergodox :)
Jul 21, 2015
WHPGH
32
Jul 22, 2015
bookmark_border
Parakwill drop 3 participants have the new PCB fabs? or are they stuck with the same issues previously mentioned
Jul 22, 2015
Kaan
297
Jul 22, 2015
bookmark_border
ParakThanks for the analysis and partial transparency, I understand it's more of an issue with part sourcing and manufacturing
I think the most honourable course of action would be to address all of these and send new PCB's to everyone that requests one, from MassDrop's side of view - or at least report which runs had generic C4/C6's that are susceptible to failure
Even if we see past the soldering issues (I look at the microcontroller, and as far as my eyes can see, none of the microcontrollers seems to be held by solder), the minor plate imperfections, which are superficial considering the component issues, if these mixed generic components fail at one point, they will brick the keyboard and ruin all the efforts that went to it - as I understand, even a small surge could be enough to brick an infinity PCB, especially during programming, with multiple point of failures, the weakest link being the unverified components (not even including the lack of protective circuitry of the keyboard)
I spend at least 10 hours building a single keyboard, likely spend up to 400$'s each to find a set of keycaps that I like, and now I gather that my PCB's are the weak links, it's not easy to handle, after spending months waiting, researching, importing infinity's (each pcb and case has a unique plate, so I just can't desolder and move things to a new pcb either) - There are also very unique creations like Victor_L's - who waited and worked much longer than us on his keyboard(s)
I have 6 PCB's, and after this point I don't think I have enough energy to look for and import new PCB's, so I will probably move on with my builds and hope no electrical anomalies hit my keyboards or the keyboards that I built and gifted, but I have to say I'm very disappointed
I really wish you designed these PCB's to last as long as an IBM M/F does, but if the situation is as dire as portraited, we would be lucky if the keyboard works at all
Jul 22, 2015
Parak
89
Input Club
Jul 22, 2015
bookmark_border
KaanI'm honestly not sure what aspect of that was partial transparency, as I've tried to explain the issues as clearly as possible. Massdrop support does take care of people's issues as they may come up, but as I mentioned, having obtained these numbers from them directly, issues like the one that @chilofthehorn experienced is currently at a grand total of 1, out of several hundred shipped boards. Since the failure is both hard to replicate and non-pervasive, it's also hard for us to point fingers at any particular root cause.
From what I know so far, and I mentioned this already, any generic components is pure conjecture at this point. The PCB fab was given a proper BOM, with said C4 and C6 actually being Murata japanese caps. Anything else past that, given the amount of potential component failures (1), is very difficult to substantiate. I do have some ideas on how to thoroughly disprove it, but it will take a bit of time.
Furthermore, I should point out that all of the enthusiast produced keyboards, including well known projects like the GH60, Gon's keyboards, and so on, use similar schematics as far as things like ESD protection go. By that I mean essentially not having it. This is actually an interesting subject for me, so if you can come up with a list of keyboards, either community, Korean custom, or commercial ones that include something like TVS diodes, I'd be quite interested in that list. Keep in mind that most of the projects in the first two categories use an Atmel Atmega32u4 (or similar), which is not even ESD rated, where as the ARM chip that we use on the Infinity is properly rated and clamped on USB and analog pins for a few thousand volts.
The top thing we know currently to be a significant issue is the soldering, which we are working hard to fix for this third drop. As I've mentioned, people whose infinities are otherwise working normally can reinforce the USB connector with a reflow of the shield tabs, which will take care of the most stressed part as far as soldering goes. We're also looking into the stabilizer issues, which a number of people mentioned.
I know you're invested in the project, which I really appreciate, but we (Input Club and Massdrop) are here in case of any issues and will try to fix them as best as possible to everyone's satisfaction. However, making hypothesis and inflated prognosis about things that have yet to be properly verified or have an incidence of >1 really does nobody any favors and is borderline FUD.
Jul 22, 2015
Kaan
297
Jul 22, 2015
bookmark_border
ParakHi Parak
I said partially transparent as I was under the impression that a subset of the PCB's had low-quality components, and it's unclear whether that's really the case and to what extent
I really really appreciate that you take the time and effort to investigate and reply to these concerns
I agree with you in terms of comparisons, I also didn't notice any amateur manufacturer do anything out of the ordinary for extra protection, but I similarly didn't hear or read of any PCB failures from keyboards that use the regular Atmega32u4 - that kind of creates a, maybe false, sense of confidence in them, while our ARM infinities feel like a bit experimental - from that aspect, 1 in 100's is worse than none in thousands - as far as I perceive
>> So, when you inspect a random Rev 1.2 infinity, are all the components within the intended specs of your design?
I dismissed my personal concerns, you showing this kind of confidence in the PCB's is enough for me, it would also be great if you could share the website of "Input Club" - so we could ask for advice, let's say in 10 years, if any issues occur, or just to keep up with new PCB's/keyboards and stuff
Edit: If you decide to pursue surge-proof/lifetime-guaranteed pcb-designs, I think there is a market for them, there are a lot of people who pay $500's for gon keyboards, I'm sure a decent pcb could fetch even $200 easily with proper design and marketing
Jul 22, 2015
Parak
89
Input Club
Jul 23, 2015
bookmark_border
WHPGHYep, batch three will be in the new fab.
Jul 23, 2015
Parak
89
Input Club
Jul 23, 2015
bookmark_border
KaanI would certainly hope for Atmel boards to not have a fantastic failure rate just because they're not as well protected as an ARM K20 chip. There's unfortunately not enough information about all the different customs out there that use said chips and their failures, merely the ESD differences that are telling via the datasheets. There's a large amount of designs out there, with all sorts of component types, so I'd be surprised if there wasn't ever even a single one that has died under mysterious circumstances. On the other end of the scale, reports of various commercial keyboards dying are all over various forums, but those are made in even higher quantities of course.
Website is coming soon. We hope to make it a more compact place for all this information, as well. On the second 'market' question, there may or may not be something in that respect down the road ;)
Jul 23, 2015
WHPGH
32
Jul 23, 2015
bookmark_border
ParakGlad to hear! I'm now looking forward to my infinity arriving, instead of being anxious
Jul 23, 2015
View Full Discussion
Related Products