Hi All - here’s an update on our analysis results from PCB assembly testing -
We’ve studied the occasional intermittency we saw on some TREG units, and worked to determine if it could be screened out through certification testing, or if it required preemptive replacement of the flex PCB’s.
We reviewed sample units from different batches of thousands of TextBlades. Through a battery of different testing techniques, we were able to identify some but not all of the intermittency cases.
By its nature, intermittent and gradual-onset phenomena is tricky to positively identify in each case where it may be latent. We were able to narrow it down, but not conclusively rule it out through the certification tests.
Because an intermittent connection effect can alter the analog data stream being measured from your fingertips, if it happens, it can materially affect input. When it’s sporadic, it’s also tougher to know whether an error was you or the machine, especially when it’s a fleeting effect inside the hardware.
Given TextBlade’s new technology that can interpret human intent at very high speed, it’s especially important to eliminate any noise source. The integrity of the signal stream is fundamental to what TextBlade can do.
Reviewing these considerations collectively, we’ve decided that the most rational course is to preemptively replace the flex PCB’s. The installation handling stress that had caused this issue could have occurred with almost any of the units we already built, so the conservative path is to replace them on all the units in our volume production inventory.
We’re also deploying new fixtures on our lines that protect the flex from exposure to any handling stress. This effectively provides guard rails in addition to the new training of the operators.
We’re proceeding now with component orders for a large volume of new PCB’s and core silicon to cover replacements in our entire inventory of KeyBlade assemblies. (Right and left Keyblade PCB’s are mirror images of each other).
The SpaceBlade has a different mechanical structure, so it does not present any similar stress conditions during assembly. SpaceBlade inventory is testing good, so replacement of those PCB’s is not needed.
Within September the focus will be on getting these new KeyBlade PCB components fabbed, assembled and tested.
After completion of the new flex boards, we’ll start installation into our KeyBlade inventory in batches, and do end-to-end checks to confirm that the new PCB’s and assembly procedures show no signs of intermittency. We’ll use both our automated thumpers with logging tests, and also perform destructive tests on some samples from each batch to verify good stress limits.
When the tests confirm that we’ve eliminated the stress intermittency, we expect to start releasing from those batches to customers within Q4. We’ll do another update with more calendar info as we get more data on the finished goods output from the process.
For the technically inclined, you might find these recent articles of interest -
They describe a similar issue with PCB stress that is now affecting Apple’s most successful product, the iPhone 6. Hundreds of millions of iPhone 6 pcb’s are out in the field right now. Apple will figure this out and they’ll make it right, but it is a fresh reminder that even the biggest folks also have to deal with the same stuff.
Apple definitely didn’t skimp, and put an army of great talent on testing their flagship product. Even so, it was possible for an issue like this to present, with a vexing, delayed-action effect.
Especially with all-new tech, it pays off very well to be conservative, and perhaps a little paranoid about testing it. That’s what we’re doing here, to nip whatever we find in the bud, in-house, so it’s fully settled ahead of deliveries. That’s the smartest path for both manufacturer and customers, and the best way to deliver a great new experience. Many thanks to our TREG users for helping us find this.