Forums

If the TEST pin is left floating, the JTAG pins revert to normal port pins.
This may or may not have influence on the JTAG performance (leakage currents, impedances etc.).
So my answer is an authoritative "maybe" :)

Thank you Jens-Michael for your quick answer!

Anyway I don't believe that no one ever tried to do something like that :) Just wanted to know if it works or not to avoid wasting time building a pcb.

I think it is useless test it with some PDIP package MSP430s on a breadboard since I don't know if the 5xxx or 6xxx families will behave the same way.

 

Hints ?

 

Thanks,

Mário

The typical application for multiple MSPs attached to one programmer is production programming wiht a GANG programmer. But here, usually each attached processor is on its own PCB and therefore separately connected.

On the experimenters board with a 4x and 2x processor, there are two independent JTAG headers, not just a jumper. So it seems that indeed nobody has tried that yet.
Earn a badge as multi-processor-JTAG pioneer. :)

Hello!

For future reference this kind JTAG sharing works!

You just have to place a jumper on each TEST pin to choose what uC you want to flash. I've used two F5436A and one F5638 and it worked with no problem. I've placed  also a jumper on RST pin to be able to reset only one uC at a time (usefull for debug).

;-)

Hello,

Did someone of you try to do that JTAG sharing with uCs with dedicated JTAG pins (no TEST pin)?

In this document:
http://www.ti.com/lit/ug/slau320d/slau320d.pdf

Page 22, section 1.3.1.2 "Fuse Check and Reset of the JTAG State Machine (TAP Controller)" it says:

 “Each MSP430 family device includes a physical fuse used to permanently disable memory access via
JTAG communication. When this fuse is programmed (or blown), access to memory via JTAG is
permanently disabled and cannot be restored. When initializing JTAG access after power up, a fuse check
must be done before JTAG access is granted. Toggling of the TMS signal twice performs the check.
While the fuse is tested, a current of up to 2 mA flows into the TDI input (or into the TEST pin on devices
without dedicated JTAG pins). To enable settling of the current, the low phase of the two TMS pulses
should last a minimum of 5 μs.”

 

So it means that on the devices that have dedicated JTAG pins (no TEST pin), the  TDI line is acting more or less like TEST line on devices with shared JTAG pins?
If we disconnect TDI line from MCU, then the fuse checking will fail and the access to flash memory won’t be granted – JTAG for this specific MCU will be turned off.

But what about the other lines, for instance TDO. If I disconnect TDI and fuse checking will fail for one of the uC, will there be  some feedback signal on TDO from MCU after fuse checking failure? Then it could override the TDO line between programmer and uC that I want to have acces to.

I will appreciate any helpful tips.

 

In general, this scheme is not going to work. If one of the target devices uses dedicated JTAG pins, it will hog those pins and the others cannot use them. For target devices that multiplex JTAG functions with other I/O functions on the same pins, not connecting TEST pin will make those pin do the “other I/O functions” which more likely than not will interfere. For target devices with real fusible JTAG fuse, I do not even want to try “blow the fuse”.

Thank you for the answer.

So, there is no way to share the JTAG in some simple and easy way (by switching just 1 line instead of 4)?

How about putting one of the MCUs into the reset state while the other one is being programmed? Will it work?

Mariusz Wysoczanski

So, there is no way to share the JTAG in some simple and easy way (by switching just 1 line instead of 4)?

Not with switching. All JTAG interfaces support the Pass-through, even if the fuse has been blown.

So in theory, you can daisy-chain all JTAG peers. The software, however, needs to precede the JTAG commands by the appropriate number of pass-through commands then. And this is not supported by MSP430LIB and the FETs.

Mariusz Wysoczanski

How about putting one of the MCUs into the reset state while the other one is being programmed? Will it work?

I don't know what state the dedicated JTAG pins have when the processor is in reset state. I don't remember reading about it in any datasheet.