--- In lpc2000@lpc2..., "Stephen Pelc" <stephen@m...>
wrote: >
>
> How many would you like?
>
> We did consider Ethernet, but USB is much more convenient for
> site work. It may become an option ...
>
> Stephen
>
> --
> Stephen Pelc, stephen@m...
> MicroProcessor Engineering Ltd - More Real, Less Time
> 133 Hill Lane, Southampton SO15 5AF, England
> tel: +44 23 80 631441, fax: +44 23 80 339691
> web: http://www.mpeforth.com - free VFX Forth downloads
>
I get your point! :-)
Unfortunately I am no longer in a position to influence the purchase
of debug hardware, since I am recently retired. My former employer
bought large numbers of the Ethernet debuggers that I was refering to,
and we found the Ethernet connection to be very very handy. The
ability to debug from a distance was a highly used feature in that
work environment.
I don't think that it would all that difficult to enable remote
debugging using USB as long as a local workstation was available to
act as the local USB connection to the target?
From: "derbaier" <dershu@ders...> > I just wanted to note that you left out my
favorite as the most
> powerful of the JTAG to workstation connection methods.
>
> Ethernet!!!!
>
> The Lauterbach Trace32 Ethernet is the fastest and most powerful
> of all the JTAG setups that I have used, and extremely expensive.
> :-(
>
> One big advantage of Ethernet, is that the workstation and the
> debug target do not have to be in the same location. Of course,
> it is not as fast over the Internet as it is with a direct local
> connection. :-)
>
> Care to give them some competition in that arena? :-)
How many would you like?
We did consider Ethernet, but USB is much more convenient for
site work. It may become an option ...
Stephen
--
Stephen Pelc, stephen@step...
MicroProcessor Engineering Ltd - More Real, Less Time
133 Hill Lane, Southampton SO15 5AF, England
tel: +44 23 80 631441, fax: +44 23 80 339691
web: http://www.mpeforth.com - free VFX Forth downloads
Reply by Bertrik Sikken●January 3, 20062006-01-03
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
rtstofer wrote: >> We need two things. A reasonably priced USB
JTAG debugger supported
>> by a true "open-source" GDB/JTAG driver. Until then, the
debugging
>> situation will remain tenuous, at best (for those on limited
>> budgets).
>>
>> Cheers,
>> James P. Lynch
>>
>
> I posted a question re: using Insight under Linux on the Sparkfun
> Forum. I received this site as a reply: http://openocd.berlios.de/web/
>
> This may be a HUGE step in the right direction.
I played around with this in the past week and I think this is the
most promising open-source ARM7 JTAG project.
The source code looks well organised and extensible.
The latest version (download with subversion) is now able to run
under windows (with cygwin), it supports the ARM7TDMI-S micro and
works with various type of JTAG cables (parallel port and USB).
It successfully stepped through some simple programs in both
gdb and insight.
Kind regards,
Bertrik
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
>
>>Well then, if you want to do non-trivial debugging, you'll
certainly
>>hate Insight->OCD Remote->Wiggler clone. As James Lynch points out
>>in his excellent tutorial, there are days when it just won't work.
>>To be blunt, it sucks. It isn't worth the aggravation to get it
>>working; printf works better.
>>
>>Now, Tom Walsh is more than happy with his high dollar solution
>>(Abatron BDI2000) and if I were doing this for money, I would buy
>>the same interface he has. If I could find a distributor in the
>>US...
>>
>>And, if the US$ wasn't complete worthless when converted to British
>>Pounds, I might consider the Rowley CrossConnect package.
>>
>>
>
> > If I could find a distributor in the US...
>
>Try Ultimate Solutions ...
>
>http://www.ultsol.com
>
>I requested a quote for PowerPC 4xx and ARM development ...
>
>BDI2000+bdiGDB PPC4xx $2570
>bdiGDB for ARM7/9/9E $1000
> Interesting. I got the BDI2000+bdiGDB option for the ARM7/9/9E for the
same price. I suspect that the config files are all that is different
between the models. I'm still not sure what that "bdiGDB" is all
about...
>
> Well then, if you want to do non-trivial debugging, you'll certainly
> hate Insight->OCD Remote->Wiggler clone. As James Lynch points out
> in his excellent tutorial, there are days when it just won't work.
> To be blunt, it sucks. It isn't worth the aggravation to get it
> working; printf works better.
>
> Now, Tom Walsh is more than happy with his high dollar solution
> (Abatron BDI2000) and if I were doing this for money, I would buy
> the same interface he has. If I could find a distributor in the
> US...
>
> And, if the US$ wasn't complete worthless when converted to British
> Pounds, I might consider the Rowley CrossConnect package.
I requested a quote for PowerPC 4xx and ARM development ...
BDI2000+bdiGDB PPC4xx $2570
bdiGDB for ARM7/9/9E $1000
I just have to figure out how to get that passed the guys with
the money ...
Dave
Reply by rtstofer●January 3, 20062006-01-03
> We need two things. A reasonably priced USB JTAG debugger supported > by a true "open-source" GDB/JTAG driver.
Until then, the debugging
> situation will remain tenuous, at best (for those on limited
> budgets).
>
> Cheers,
> James P. Lynch
>
I posted a question re: using Insight under Linux on the Sparkfun
Forum. I received this site as a reply: http://openocd.berlios.de/web/
This may be a HUGE step in the right direction.
Richard
Reply by derbaier●January 3, 20062006-01-03
--- In lpc2000@lpc2..., "Stephen Pelc" <stephen@m...>
wrote: >
>
> Warning: Response from JTAG vendor: See
> http://www.mpeforth.com/jtagwidget.htm
> which is designed for hardwaare bring-up as well as software
> debugging.
>
> There are three aspects to consider:
> connection to PC host
> connection to ARM target
> software
>
> Host Connection
> ===============
> The host connection in rough order of performance and cost is
> PC parallel port
> USB
> dedicated PCI card
> Some parallel port interfaces are powered from the parallel
> port. Without great care in component selection and PC layout
> these can be unreliable. We have several, some work well, some
> don't.
>
> USB interfaces can draw power from the USB port. In general,
> these are the most convenient for site use. But the hardware is
> more expensive and software is required, probably both on the PC
> and on the JTAG adaptor. Software costs.
>
> PCI interfaces will give the highest performance, but are
> usually only found on dedicated testers.
>
> ARM target connection
> =====================
> As the JTAG speed increases, you need to be more careful with
> drive levels, cable length and cable impedances. We do not use a
> cheap HC244, we use a real level shifter on the JTAG side. It's
> rated to below 1.8 volts. Similarly, the lifetime and cost of a
> good quality 3M connector is much higher than that of a generic
> 20 pin connector.
>
> Software
> ========
> Once you move away from the parallel port connection, you'll
> most likely have to write software for the adaptor as well as
> for the PC.
>
> Conclusion
> ==========
> Better (performance, reliability ...) JTAG debuggers require
> more work, which has to be paid for somehow or other. Technical
> support costs time/money too. There's a lot more to ARM JTAG
> debuggers than at first meets the eye.
>
> Stephen
>
> --
> Stephen Pelc, stephen@m...
> MicroProcessor Engineering Ltd - More Real, Less Time
> 133 Hill Lane, Southampton SO15 5AF, England
> tel: +44 23 80 631441, fax: +44 23 80 339691
> web: http://www.mpeforth.com - free VFX Forth downloads
>
Nice analysis, Stephen.
I just wanted to note that you left out my favorite as the most
powerful of the JTAG to workstation connection methods.
Ethernet!!!!
The Lauterbach Trace32 Ethernet is the fastest and most powerful of
all the JTAG setups that I have used, and extremely expensive. :-(
One big advantage of Ethernet, is that the workstation and the debug
target do not have to be in the same location. Of course, it is not as
fast over the Internet as it is with a direct local connection. :-)
Care to give them some competition in that arena? :-)
-- Dave
Reply by Stephen Pelc●January 3, 20062006-01-03
> Date: Mon, 02 Jan 2006 21:01:04 -0500 > From: Sean <embeddedrelated@embe...>
> I have seen a lot of talk around about
Wiggler-based JTAG
> debugger boards, which can be built for < $1, and I see several
> commercial JTAG debugging pods that are > $1000. What is it that
> these things do that you can't do with a simple wiggler
> interface?
There are three aspects to consider:
connection to PC host
connection to ARM target
software
Host Connection
===============
The host connection in rough order of performance and cost is
PC parallel port
USB
dedicated PCI card
Some parallel port interfaces are powered from the parallel
port. Without great care in component selection and PC layout
these can be unreliable. We have several, some work well, some
don't.
USB interfaces can draw power from the USB port. In general,
these are the most convenient for site use. But the hardware is
more expensive and software is required, probably both on the PC
and on the JTAG adaptor. Software costs.
PCI interfaces will give the highest performance, but are
usually only found on dedicated testers.
ARM target connection
=====================
As the JTAG speed increases, you need to be more careful with
drive levels, cable length and cable impedances. We do not use a
cheap HC244, we use a real level shifter on the JTAG side. It's
rated to below 1.8 volts. Similarly, the lifetime and cost of a
good quality 3M connector is much higher than that of a generic
20 pin connector.
Software
========
Once you move away from the parallel port connection, you'll
most likely have to write software for the adaptor as well as
for the PC.
Conclusion
==========
Better (performance, reliability ...) JTAG debuggers require
more work, which has to be paid for somehow or other. Technical
support costs time/money too. There's a lot more to ARM JTAG
debuggers than at first meets the eye.
Stephen
--
Stephen Pelc, stephen@step...
MicroProcessor Engineering Ltd - More Real, Less Time
133 Hill Lane, Southampton SO15 5AF, England
tel: +44 23 80 631441, fax: +44 23 80 339691
web: http://www.mpeforth.com - free VFX Forth downloads
Reply by Anton Erasmus●January 3, 20062006-01-03
On 2 Jan 2006 at 23:25, Sean wrote:
>
> This is a very good point: why are they offering their software for
> free, knowing full well that other people are using it on
competitor's
> hardware?
The advertising value they get is immense. Anybody who has used a
wiggler or wiggler clone would at least give their more sophisticated
models a look if they are considering getting something a bit better
than a wiggler.
> Yes, baud rates are slower over parallel, although
you can get >
> 1mbit/sec on the parallel port itself, perhaps only 25KB/sec when
> using wiggler, but even there, even with a rather large code size of
> 512KB, this is only 20 seconds to download... For most debugging
> scenarios you won't need to download the full size anyway, and if
you
> split your program into modules you can get by with only debugging a
> module at a time, which means you only need to load a few dozen KB at
> a time?
The biggest issue with using the wiggler, is that it is temperamental. It
works
quite nicely when it works, but there are days when it just refuses to work -
normally
when you actually need it. Any JTAG interface where the JTAG signal timing is
independant of the PC hardware/software used are going to be more stable.
An open hardware / open source JTAG interface with USB / ethernet interface
would be great. Designing and building the hardware is not such a big
problem.
Writing the software is the biggest issue for such a project.
Regards
Anton Erasmus
> -- Sean
>
> At 10:21 PM 1/2/2006, you wrote:
> >Sean pointed out that the Macraigor Wiggler is $150 whilst the
> >Olimex clone is $19 and asks "what's the deal here?"
> >
> >Open up a wiggler and you'll see that it's a cheap
level-shifter chip
> >and a transistor built into a DB-25 shell. Admittedly, $19 seems
more
> >like a fair price, but Olimex didn't write the OCDRemote and
> >OCDCommander software and didn't have to ammortize any software
> >development (Macraigor does). Macraigor probably provides more
> >telephone support if the customer has trouble getting it to work.
> >
> >Even the fastest wiggler baud rate is slow in comparison to the
> >newer USB and Ethernet JTAG debuggers. This is not obvious in short
> >demo programs, but painful if you are downloading a very large
> >program.
> >
> >I worry about myself and others relying on the Macraigor OCDRemote
> >JTAG program that interfaces GDB to the JTAG pins. OCDRemote is NOT
> >open-source. Macraigor allows free downloads of the OCDRemote
> >executable but the source code has always been proprietary. They
> >could withdraw this "freebie" at any time.
> >
> >We need two things. A reasonably priced USB JTAG debugger supported
> >by a true "open-source" GDB/JTAG driver. Until then, the
debugging
> >situation will remain tenuous, at best (for those on limited
> >budgets).
> >
> >Cheers,
> >James P. Lynch
> >
> >
> >
> >
> >
> >
> >----------
> >>Yahoo! Terms of Service.
> >
> >
> >----------
>
>
> Yahoo! Groups Links
>
--
A J Erasmus
Reply by Leon Heller●January 3, 20062006-01-03
----- Original Message -----
From: "Sean" <embeddedrelated@embe...>
To: <lpc2000@lpc2...>
Sent: Tuesday, January 03, 2006 4:21 AM
Subject: Re: [lpc2000] JTAG debugging options
>
> Ok, the one I saw was
>
>
http://www.gwmicros.com/public_html/pages/products/iproducts/BDM_JTAG/Macraigor/BDM-JTAG_Macraigor_usbDemon.htm
>
> or
>
>
http://www.gwmicros.com/public_html/pages/products/iproducts/BDM_JTAG/Macraigor/BDM-JTAG_Macraigor_Raven.htm
>
> both 818EUR
>
> And their wiggler is 188 EUR. Yes, most others I see are in the $150 -
> $300 range.
>
> I'm just trying to understand here. I can see how originally there
were
> few (expensive) options, and there was a lot of design and support and
> software creation necessary, so yes, it was expensive. However from the
> point of view of anyone who is starting a design, be it a starving
student
> or a multinational corporation, what do the expensive emulators do that
I
> can't do with a $1 DIY one? Can I get by with a DIY one, or should
I
> actually fork out the cash for the name-brand ones? Let's assume that
I
> need to do non-trivial debugging here.
>
> I too am a fan of alternate methods for debugging, or creating modules
on
> the PC and then porting them, or loggers, or serial port dumps, however
> there are times when having an ICE makes it a lot easier to figure out.
A DIY one will be fine. I designed a PCB for one that may easily be made at
home with rudimentary facilities (artwork in jtag.zip in the Files area).
It'll probably take a couple of hours to make the PCB and get it
working.
I already had a genuine Macraigor Wiggler and mine works at higher speeds
than the real thing; my PCB layout is better. I've also got a Rowley
CrossConnect (99 GBP) with a USB connection to the PCB, and that works very
well. USB is more convenient than a parallel port connection, and faster.
Amontec sells their Chameleon interface, which is based on a reconfigurable
CPLD. It can emulate a Wiggler or Raven, as well as lots of other
interfaces, including JTAGs for Xilinx and Altera parts. I've got one of
those, as well. Worth getting if you want to play with FPGAs and other JTAG
MCUs.
