RevF Hardware Bringup Report
Here are some of the things that have happened in the last two weeks.
New Enclosure Prototypes
Custom connector mechanical samples
On Revision E, which was in our promo video, the connectors were hand-cut out of brass tubing and soldered directly to the circuit board. This is a point of concern for manufacturability and for assembly/disassembly, since those connectors had to slide all the way in to the housing and it’s easy to make a mistake and torque the board.
We found a competitively priced machine shop who can make us custom connectors that make the assembly and disassembly process much cleaner. The samples have arrived and make a world of difference.
SD card and Reset Button
As I covered in a previous blog post, the new enclosure moves the parting line between the two halves down to the PCB level to create a slot where the SD card and reset button can be accessed without disassembling the whole thing. The change works nicely and makes retrieving logs MUCH easier.
The new enclosure and mechanical features make the product easier to use. Nothing unexpected on this front!
I’ll give the bad news first: we’ll have to do one more run of prototypes before submitting for regulatory approval. There were two main issues:
- I made a a mistake in the new current source added at the $70k stretch goal. I confused the symbol for N type and P type FETs. It’s such a basic mistake… I feel dumb about it, but these things happen. I worked around it with some wire hacking that allows me to validate the rest of the design, but unfortunately we can’t submit wire-hacked boards for regulatory approval. Update: I didn’t confuse the symbols, Altium switched them on me! See bottom of post.
- The thermal offset and gain drifts on some of the new parts were higher than acceptable. The current sense resistor and fuse self-heating causes appreciable offset drift as the sense current gets in the 8-10A range. This was not a problem on Revision E because we did not use an extra current sense amplifier located close to the sense resistor – this was a Revision F add-on as part of the 70k stretch goal. We shouldn’t need to make any major topology change, just select a better current sense amplifier.
Besides those issues, everything else continued to work quite well. I took the time to characterize the non-linearity of the device and found it to be <0.1% across all ranges of voltage and current (until extended sense resistor heating becomes an issue around 8-10A). The DC gain accuracy before calibration was within 1% of my multimeter (which is only 1% accurate itself), so after cal I expect it will be even better.
Starting the API
We promised to release an API to allow you to interact with a Mooshimeter and integrate it in to your projects programatically. Our friends over at Technical Machine have a Revision E meter and have been doing some beta testing. They’ve also made the Tessel the first third party platform to successfully interact with a Mooshimeter! That’s very exciting for us… it’s one thing to make a device work, it’s another entirely to have other people start using it. Guiding them through the integration process taught us some valuable lessons about what the API documentation needs to say.
All change is risk. Our add-ons for the stretch goals were quite conservative, but they’re still change! Another minor revision should not push back our ship date though, the layout is already almost complete. More to come soon.
Shortly after publishing this, I remembered an error I encountered a few years ago with Altium switching parts in certain circumstances when annotating schematics. I went back through the revision history for RevF and found that, sure enough, I had placed the NMOS and PMOS transistors in the correct places.
When you tell Altium to annotate schematics (this means assigning every part a name), it switches the symbols in multi-symbol parts as it sees fit. Since we’re using dual N-P transistors, Altium decided to switch the N and P transistors. How frustrating! I should have caught it, but it makes me feel a little better.
Q5B is a PMOS.
Sure enough, it does. Now the circuit won’t work as designed.
For other Altium users, the way to get around this problem is to make sure this box is clicked on all your multi-symbol parts before annotating your schematics.
Great update, thanks for all the interesting detail. Quick question: is it possible to read the log files over bluetooth, without removing the uSD card?
Yes, though BLE is a relatively slow protocol. It will be possible, but it might take longer than you’d like.
Thanks for the update. Looks awesomely great!
No worries, we’re all humans and programs are all buggy.
Do the connectors still solder to the board?
No, they do not. This allows for much easier access.