Dan Miner wrote:
Yes, that was the basic idea behind what I wanted to do.
I don't even care if I can bypass a specific unit, I'd be happy
to program them all at once in a huge chain. Roughly: scan in
the first word for all devices in a huge string, program that
to flash, repeat with second word,... That way, each device
could have a unique program and/or EEPROM loaded but they could
all be programmed from only one JTAG connector.
The "simple" thing I need is the software. :-)
I just looked at the link Chris posted.
(see http://www-s.ti.com/sc/psheets/slaa149/slaa149.pdf)
I may be able to write my own software for this. Hmmm...
harry posted the link, i just added the we have implementation based on
that app note.
the PC based programming downloads a mini program to the target and
prigrams a chunk, wait until that program finishes, repeat until all is
programed. this meas that all the CPUs are not programmed at the same
speed and waiting until all are finished is required, i think that gets
very complicated (possible, but how much time/money do you want to spend
on developing this...)
if you have a dedicated hardware that generates the JTAG signals it
easier, then you can prgram flash directly. the application note from
above features such a device, and our Serial-JTAG box does that too.
you could also take the serial-jtag device and build a number of
hardware JTAG programers. they can flash as standalone device, over the
serial port or a combination of both. i'd connect each device and not
connect them in a chain (complexity...).
it's also possible to layout test points on the PCB and connect to those
with needles in a test bed. thats what we do in our production. that way
you dont need to have expensive connectors in the final product.
I was hoping I could get someone (Steve Underwood?) to answer
and say it is already supported with the current tools or that
it would be a trivial change and would be released soon.
i dont know. at least i'm not working on such a thing ;-)
Steve - could you please tell me if it is at least possible or
not with the public documents from above? (no NDA) If "N"
deviced are daisy chained together, can they be programmed in
one big chain? Does TDO always reflect a copy of TDI delayed
by X TCK cycles?
as far as i know, yes. its a JTAG interface and follows these rules. the
special part about it is that you can access the address and databus
inside the CPU, not only the pins as in a boundary scan (that part not
beeing public).
The gang programmer will be an acceptable alternative although
not nearly as nice for these 2 reasons:
1) All devices must be programmed with the same data
and? the MSPs have the so called information memory at 0x1000, that is
well suitable for calibration values and serial numbers. that way all
devices can have the same program and only differ in the information memory.
why dont you write the serial number later, e.g. in an end-test of the
product?
(jtag again or serial port e.g.)
programming different software on each device has other implications
too: you need to generate all these different programs, at least patch a
given one at the right places. dont forget that you have to write this
software too.
testing the system is also more complex, als well as error handling
(e.g. one device fails to program)
2) Only supports up to 8 devices at a time.
- Dan Miner
From: Harry Lemmens
Sent: Wednesday, January 28, 2004 7:58 PM
Thanks for that Chris.
The reason I asked is that (I think) by using the BDSL
"Bypass" instruction, a device effectively be frozen out of
a system during boundary scan. The data passes from the TDI
pin to the TDO pin, but does not access internal JTAG
"functions.
So by daisy chaining the TDO/TDI pins on multiple devices,
and paralleling TMS and TCK to all the devices, it would have
been possible to configure a string of MSP's (The last device
in the chain would be the first device requiring configuration.)
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Mspgcc-users mailing list
Mspgcc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mspgcc-users
