Forums

DSP hang up with EMC disturbance

Started by mvigano September 11, 2003
We use 56F807 with internal watchdog (COP) enabled but we experienced
a stop of the DSP when we apply EMC disturbance like fast transient
or
elettrostatic
discharge.
Our application is an high reliability one so we cannot simply reset
it
with an external watchdog.
The problem seems to came from the JTAG port and specially
from /TRST signal
(we have connected it in the same way used on the evaluation board).
The only solution we found is to connect togheter /RESET and /TRST
and keep
very short
the distance (no more than one inch) to the reset generator.
In this arrangement we cannot use the emulator with the production
board.

Somebody has found the same sensivity on /TRST signal?

Marco Vigano'

Microelettrica Scientifica S.p.A.
via Alberelle 56
20089 Rozzano (MI)
Italy

Tel. :+39-02-575731
Fax :+39-02-57510940
mailto:



The /trst line resets the internal JTAG registers. If you do not provide a active /trst the JTAG registers can end up in a random state possibly with a value that puts the processor in a debug state. The /trst needs to be driven active at power up. You can also tie it permantly active in a system that you are not using with the debugger. You can use a jumper or switch to allow debugger and non debugger operation similar to how the EVM operates.

[Hutchings William-p23437]  -----Original Message-----
From: mvigano [mailto:m...@yahoo.com]
Sent: Thursday, September 11, 2003 9:50 AM
To: m...@yahoogroups.com
Subject: [motoroladsp] DSP hang up with EMC disturbance

We use 56F807 with internal watchdog (COP) enabled but we experienced
a stop of the DSP when we apply EMC disturbance like fast transient
or
elettrostatic
discharge.
Our application is an high reliability one so we cannot simply reset
it
with an external watchdog.
The problem seems to came from the JTAG port and specially
from /TRST signal
(we have connected it in the same way used on the evaluation board).
The only solution we found is to connect togheter /RESET and /TRST
and keep
very short
the distance (no more than one inch) to the reset generator.
In this arrangement we cannot use the emulator with the production
board.

Somebody has found the same sensivity on /TRST signal?

Marco Vigano'

Microelettrica Scientifica S.p.A.
via Alberelle 56
20089 Rozzano (MI)
Italy

Tel.  :+39-02-575731
Fax  :+39-02-57510940
mailto:m...@microelettrica.com


_____________________________________
Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer.  You need to do a "reply all" if you want your answer to be distributed to the entire group.

_____________________________________
About this discussion group:

To Join:  m...@yahoogroups.com

To Post:  m...@yahoogroups.com

To Leave: m...@yahoogroups.com

Archives: http://www.yahoogroups.com/group/motoroladsp

More Groups: http://www.dsprelated.com/groups.php3


">Yahoo! Terms of Service.


YES, I have been having problems related to EMC with this processor!
I am developing a switching power supply which switches at 50 kHz
(harmonics out to several MHz), and handles several tens of amps, and
up to a few kV. I am using the 56F807 for all feedback loop control
(DSP implimentation of filters), ADC's, PWM generation, and user
control. The processor and associated signal conditioning electronics
are galvanically isolated from the high power circuitry.

Having the processor "hiccup" is *NOT* good in this application. I am
in the process of debugging these issues. At this time, I can say
that it appears that some of these problems *ARE* related to the JTAG
circuitry - but I'm not sure if that's everything. I will post an
update within a few days once I do some further troubleshooting.

Regards - Bill

--- In , "mvigano" <mvigano@y...> wrote:
> We use 56F807 with internal watchdog (COP) enabled but we experienced
> a stop of the DSP when we apply EMC disturbance like fast transient
> or
> elettrostatic
> discharge.
> Our application is an high reliability one so we cannot simply reset
> it
> with an external watchdog.
> The problem seems to came from the JTAG port and specially
> from /TRST signal
> (we have connected it in the same way used on the evaluation board).
> The only solution we found is to connect togheter /RESET and /TRST
> and keep
> very short
> the distance (no more than one inch) to the reset generator.
> In this arrangement we cannot use the emulator with the production
> board.
>
> Somebody has found the same sensivity on /TRST signal?
>
> Marco Vigano'
>
> Microelettrica Scientifica S.p.A.
> via Alberelle 56
> 20089 Rozzano (MI)
> Italy
>
> Tel. :+39-02-575731
> Fax :+39-02-57510940
> mailto:mvigano@m...




How is your JTAG lines configured  in your target system. The /trst should be driven low to hold the JTAG port in reset when you are not using this port. If the /trst is not connected in your target system and is left to float this will cause intermittant problems.
-----Original Message-----
From: Bill Schintler [mailto:w...@mchsi.com]
Sent: Thursday, September 11, 2003 9:16 PM
To: m...@yahoogroups.com
Subject: [motoroladsp] Re: DSP hang up with EMC disturbance

YES, I have been having problems related to EMC with this processor!
I am developing a switching power supply which switches at 50 kHz
(harmonics out to several MHz), and handles several tens of amps, and
up to a few kV.  I am using the 56F807 for all feedback loop control
(DSP implimentation of filters), ADC's, PWM generation, and user
control.  The processor and associated signal conditioning electronics
are galvanically isolated from the high power circuitry.

Having the processor "hiccup" is *NOT* good in this application.  I am
in the process of debugging these issues.  At this time, I can say
that it appears that some of these problems *ARE* related to the JTAG
circuitry - but I'm not sure if that's everything.  I will post an
update within a few days once I do some further troubleshooting.

Regards - Bill

--- In m...@yahoogroups.com, "mvigano" <mvigano@y...> wrote:
> We use 56F807 with internal watchdog (COP) enabled but we experienced
> a stop of the DSP when we apply EMC disturbance like fast transient
> or
> elettrostatic
> discharge.
> Our application is an high reliability one so we cannot simply reset
> it
> with an external watchdog.
> The problem seems to came from the JTAG port and specially
> from /TRST signal
> (we have connected it in the same way used on the evaluation board).
> The only solution we found is to connect togheter /RESET and /TRST
> and keep
> very short
> the distance (no more than one inch) to the reset generator.
> In this arrangement we cannot use the emulator with the production
> board.
>
> Somebody has found the same sensivity on /TRST signal?
>
> Marco Vigano'
>
> Microelettrica Scientifica S.p.A.
> via Alberelle 56
> 20089 Rozzano (MI)
> Italy
>
> Tel.  :+39-02-575731
> Fax  :+39-02-57510940
> mailto:mvigano@m...



_____________________________________
Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer.  You need to do a "reply all" if you want your answer to be distributed to the entire group.

_____________________________________
About this discussion group:

To Join:  m...@yahoogroups.com

To Post:  m...@yahoogroups.com

To Leave: m...@yahoogroups.com

Archives: http://www.yahoogroups.com/group/motoroladsp

More Groups: http://www.dsprelated.com/groups.php3


">Yahoo! Terms of Service.


Hi,
I'm working on a similar hardware design. It has 4 current loops
that operate at up to 30 amps off of a 300 volt DC bus.

The inteface to the JTAG is exactly like that on the EVM board.

I had problems with the DSP hanging up due to glitches on the JTAG
port.

I found the solution to be connecting a 10k ohm pull down on the
PORT_RESET (this is on pin 2 on the DB25 connector)The resistor is on
the PCB near the 74LCX244 that receives the signals from the parallel
port.

I also found I had to add a good solid braided ground strap from the
housing of the 25 pin connector to the ground plane of the PCB. (the
DB25 uses a small molex connector with a short adapter cable to
connect to the board so the shield was not connected on the PCB end)

Kirk