Some updates from the last few weeks:
Our European distributor, Elektor, was contacted by German authorities who found some problems with the documentation being supplied with the Mooshimeter. Specifically, they said not enough safety information was distributed on paper with the device itself. We’re working to correct this and the next manufacturing run of Mooshimeters will include additional product safety information and regulatory compliance statements.
If you follow this blog, you will know the great pains we took going through radio and product safety testing in 2014, so we’re not worried that the product violates any standard. There are simply some requirements about the paper documentation that we must work harder to follow.
Apps and Firmware: Beta->Production
The beta apps and firmware will soon be released as production. For the last few weeks I’ve been hunting lingering bugs and have chased most of them down to the OS level – ie. many of them are not in Mooshimeter code, they are in iOS, Android, or TI’s BLE stack. I’ve submitted bug reports where relevant and tried to implement work arounds where possible.
Overall the stability of the beta is better than the production version, so I’ll be pushing it to production very soon. There are a flotilla of new features that I hope people will find useful, and the software is much better instrumented to provide bug data through Crashlytics, which has already proven its worth in beta testing.
The firmware presently in beta is being baked in to a new “golden image” which will be put in to Mooshimeters in the factory. One of the best tools I have for testing power consumption on the new firmware image is actually other Mooshimeters running the new firmware. Logging reliability is much improved, even at relatively high speeds. So for testing the power consumption of a meter in shipping mode, I just used another meter logging at high speed to an SD card.
Above is the current consumption of a meter in shipping mode. Every 10 seconds it wakes up to check if the C and Ω are shorted and the meter should wake up. If you look at the power consumption graph above, you’ll notice the meter is drawing significant current about 20% of the time – much too high! This is because of ADC settling time and a few optimizations I added to the firmware to improve measurement accuracy in the new firmware versions.
But we don’t need <1% accuracy for determining if the test leads are shorted. A few code revisions later, shipping mode power consumption looks like this:
Much better. Average current consumption is reduced by an order of magnitude. A Mooshimeter can sit on a shelf like this for years. And as the firmware image is nailed down, that leads me to:
New Contract Manufacturer and Preparation for another Production Run
Until now, we’ve been using Sencore Electronics for contract manufacturing, and I am extremely grateful to them. Their service has been absolutely phenomenal, the Mooshimeter wouldn’t have come as far as it has without them. But they are leaving the contract manufacturing business entirely to build their own products (they make video equipment). At their recommendation, future Mooshimeters will be manufactured at Electronic Systems Inc. (ESI). I’m very excited to get to know ESI, so far I’ve had only email and phone contact with them.
The next production run is scheduled to start in early August, so I will have more updates for you then from Sioux Falls, South Dakota (where both Sencore and ESI are located). That’s all for now, thanks for reading