Be careful about the error detection being used and add your own if it's
doubtful. I was shocked to find that TI uses a single 16 bit CRC to
validate an entire image involved with one of its OTA schemes. There are
per-packet error checks and/or correction codes involved with most OTA
schemes. But I've seen too many cases where mother nature is clever and
the 1/2^n probabilities of uncaught errors is found to be inconveniently
far away from zero. So I'm very cautious about 16 bit CRCs.
I'm expecting to be using TI's 15.4 stack with sub-gig radios doing OTA
in coming months and may pass along some remarks. I've already settled
on a cryptographic hash to validate downloaded images in addition to the
off the shelf mechanism, but I'll be using a CRC to validate the
hashcode. After spending hours last night compensating for misbehavior
by what should be their bone-simple debug probe flashing tool I have to
say my expectations for TI are not very high (and this grieves me: I
have strong sentiments and experience with TI going back to the late
70s). It was very disappointing to get two new boards from Digikey at
10am and have one bricked a few hours later by a trivial command I've
used dozens of times to put a unique identifier string into the XDS110
debug interface. (Yes, I've tried the "ground pin 97" trick.) We shall see.
-Pete
On 2/26/19 2:49 PM, Alex Davis via TriEmbed wrote:
On Tue, Feb 26, 2019, at 2:40 PM, Jon Wolfe wrote:
Particle.io has a decent system for managing ota updates. It's not without
quirks, but it's got some nice bells and whistles. Of coarse then you are
committed to particle's platform, and all of it's quirks. I've only ever
used it with the (cellular) electron, I've not used it with the (wifi)
photon (still have the one I got at triembed a few months ago on my desk
waiting for my son and I to come up with a project for it.)
For my application, I don't need anything more beyond "Sonoff" functionality,
which has the added advantage of being an already-complete product and cheap.
I own a Particle Lepton. I have found it completely unusable in terms of
reliability. I've not tried the electron. For my intended application, WIFI is
adequate.
Alex
_______________________________________________
Triangle, NC Embedded Computing mailing list
To post message: [email protected]
List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
TriEmbed web site: http://TriEmbed.org
To unsubscribe, click link and send a blank message:
mailto:[email protected]?subject=unsubscribe
