That was a lot of work. Every single instruction was wrong. The 355 was a giant disaster in 2019, when it came out. All the engineers wanted to use it for general machine monitoring. That would be when you mount off the machines and look at the general vibration. I had proposed many such things for the old company, and always rejected.
Anyway, the chip was as flaky as trumpypants. They complained to the manufacturer, and got nothing. In the US you have to do that, or get sued to death. There have been several new versions, and I got two cards of newer ones. I am waiting for the latest one that promises machine monitoring. When the chip works, it is 24 bit, which means that +-2g is divided up into 16 million pieces.
The chip has a lot of internal noise, so I am hooking up two of them, and averaging the results. That helps smooth things out. The original chip could be knocked out by normal handling.
The card still needs a weird pull of the power lead to reset it, over and above powering off the rp5. Hopefully, the new chip irons all that out.
ps. new board just came in! Up against the old boards. These things are soon going to be too tiny to see.
ps on the bench I'll have two AI driven micro robot arms, with an ultra-thin laser attachment thingie. And they say I'll be out of a job - ha!
ps the new card works perfectly at $40 cdn. I want the price to go lower.
No comments:
Post a Comment