I would be curious to hear the opinions of those out there using this family of parts, with the Metrowerks CodeWarior. Specifically, do you think it is easy to use, relatively free of bugs so that it is unlikely to cause signifcant time spent solving issues that shouldn't be there in the first place? I tried out the evaluation board for the 56803, with the 5.0.2 version of CodeWarrior (until my liscense ran out). I am new to C, but not high level languages or assembly, but I had lots of problems. I don't know if it was me, or the tools, but at this point I have pretty much given up and am now evaluating another chip (Zilog), which I am having no problems with. The last problem I had, that I could not find a fix for, was simply a long integer divide that did not work right. It seemed like it had to be something I was doing wrong, but how hard is it supposed to be to do a divide? My types were Ok, and so on, but it didn't work. Multiplys did work, which made me think it wasn't me, but who knows. Anyway, when I saw the list of fixes implemented in 5.0.3, I got a bit scared. I might have been misuderstanding the fixes, but the fixes seemed to be for very basic problems, and when I asked Metrowerks to claify the nature of these fixes, I didn't hear back from them. The chip seemed to function fine, with the exception that the a/d was in error by about 1.25% when a 2.5V signal was applied. (I measured the a/d reference when calculating my expected result.) The a/d's were noisy also, but I tend to believe Motorola when they told me that can be improved with board layout. So far they have no answer for the a/d inaccuracy, but they said they could look into it for me. Anyway, I wonder what the opinion out there is...is Codewarrior a bit immature for real product development at this time, or is does it work good enough? Thanks Greg Toman |
|
You opinion of Metrowerks and Motorola 56800
Hello, all I have similar problems with DSP56F807 ADC: measurements are noisy and inaccurate. I would like to use DSP56F827 in next project. Does anybody experience with ADC on 56F827? It is usable? Thanks, Richard Kis > -----Original Message----- > From: greg_toman <> [mailto:] > Sent: Monday, January 13, 2003 7:13 AM > To: > Subject: [motoroladsp] You opinion of Metrowerks and Motorola 56800 > I would be curious to hear the opinions of those out there using > this family of parts, with the Metrowerks CodeWarior. Specifically, > do you think it is easy to use, relatively free of bugs so that it > is unlikely to cause signifcant time spent solving issues that > shouldn't be there in the first place? > > I tried out the evaluation board for the 56803, with the 5.0.2 > version of CodeWarrior (until my liscense ran out). I am new to C, > but not high level languages or assembly, but I had lots of > problems. I don't know if it was me, or the tools, but at this point > I have pretty much given up and am now evaluating another chip > (Zilog), which I am having no problems with. > > The last problem I had, that I could not find a fix for, was simply > a long integer divide that did not work right. It seemed like it had > to be something I was doing wrong, but how hard is it supposed to be > to do a divide? My types were Ok, and so on, but it didn't work. > Multiplys did work, which made me think it wasn't me, but who knows. > > Anyway, when I saw the list of fixes implemented in 5.0.3, I got a > bit scared. I might have been misuderstanding the fixes, but the > fixes seemed to be for very basic problems, and when I asked > Metrowerks to claify the nature of these fixes, I didn't hear back > from them. > > The chip seemed to function fine, with the exception that the a/d > was in error by about 1.25% when a 2.5V signal was applied. (I > measured the a/d reference when calculating my expected result.) The > a/d's were noisy also, but I tend to believe Motorola when they told > me that can be improved with board layout. So far they have no > answer for the a/d inaccuracy, but they said they could look into it > for me. > > Anyway, I wonder what the opinion out there is...is Codewarrior a > bit immature for real product development at this time, or is does > it work good enough? > > Thanks > Greg Toman > > _____________________________________ > 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: > > To Post: > > To Leave: > > Archives: http://www.yahoogroups.com/group/motoroladsp > > More Groups: http://www.dsprelated.com/groups.php3 > ">http://docs.yahoo.com/info/terms/ > > |
RE: [motoroladsp] You opinion of Metrowerks and Motorola 56800
In my humble opinion, the hardware is really pretty good. The collection of peripherals, and their flexibility is very good. I am using the '807 chip, and am comfortable with the hardware. As I have said in the past, the SDK from Motorola is very good as a teaching tool, and as a framework to get a non-multitasking system up and running. Using a background loop with foreground interrupt routines is straight-forward with the SDK. Simple usage of the "C" language, or "Defensive Programming" as Art Johnson calls it seems well supported. I know there are issues with the choices made by Metrowerks in applying the "C" language to a Harvard architecture machine that "C" centric programmers are uncomfortable with. However, if you intend to push a workload onto the chip to fully utilize the available horsepower, you will probably be writing assembly language ISR's. To that end, the toolset provides adequate support for generating production quality code. I am pleased that simple ISR's can be written to execute in less than 0.6 micro-seconds in a real application (eg. SCI TX and RX ISR's) using the assembler. There have been long hours spent debugging seemingly good "C" and sometimes ".ASM" code, however, most of the time the errors are mine rather the toolset. So I am not one to complain a lot. The Optimizer settings seem frustrating to use, not obvious as to what to expect -- tight code, or fast execution. Project support seems good, there are a lot of "Soft" features for supporting multiple targets from the same code base. I give the chip and toolset a good to very good rating, regarding usability for generating production code. Jerry. -----Original Message-----
I would be curious to hear the opinions of those out there
using
.... Anyway, I wonder what the opinion out there is...is
Codewarrior a
Thanks
________________________________________________________________________ This email has been scanned for all viruses by the MessageLabs SkyScan service. For more information visit http://www.messagelabs.com ________________________________________________________________________ |
Greg, I replied to you privately via email and wanted to also post here on the newsgroup. As I said in the email, I would be happy to look into your issues/difficulties with our toolset and possibly try and help you out with your evaluation. Either respond to my email or re-post if you would like for me to do so. I will let others respond regarding Metrowerks and Motorola's opinion, as I'm sure my response would seem biased. I just want to make sure you have considered all options before abandoning your evaluation of both Motorola DSP568xx and CodeWarrior. Feel free to contact me or post and I will be glad to help you out. I will leave that at your discretion though........ Regards, John --- In , "greg_toman <grt757@a...>" <grt757@a...> wrote: > I would be curious to hear the opinions of those out there using > this family of parts, with the Metrowerks CodeWarior. Specifically, > do you think it is easy to use, relatively free of bugs so that it > is unlikely to cause signifcant time spent solving issues that > shouldn't be there in the first place? > > I tried out the evaluation board for the 56803, with the 5.0.2 > version of CodeWarrior (until my liscense ran out). I am new to C, > but not high level languages or assembly, but I had lots of > problems. I don't know if it was me, or the tools, but at this point > I have pretty much given up and am now evaluating another chip > (Zilog), which I am having no problems with. > > The last problem I had, that I could not find a fix for, was simply > a long integer divide that did not work right. It seemed like it had > to be something I was doing wrong, but how hard is it supposed to be > to do a divide? My types were Ok, and so on, but it didn't work. > Multiplys did work, which made me think it wasn't me, but who knows. > > Anyway, when I saw the list of fixes implemented in 5.0.3, I got a > bit scared. I might have been misuderstanding the fixes, but the > fixes seemed to be for very basic problems, and when I asked > Metrowerks to claify the nature of these fixes, I didn't hear back > from them. > > The chip seemed to function fine, with the exception that the a/d > was in error by about 1.25% when a 2.5V signal was applied. (I > measured the a/d reference when calculating my expected result.) The > a/d's were noisy also, but I tend to believe Motorola when they told > me that can be improved with board layout. So far they have no > answer for the a/d inaccuracy, but they said they could look into it > for me. > > Anyway, I wonder what the opinion out there is...is Codewarrior a > bit immature for real product development at this time, or is does > it work good enough? > > Thanks > Greg Toman |
Hi Greg, We use DSP56F807 with Codewarrior compiler. My opinion about the DSP56F807: a) The AD converter is extremely noisy (although we followed all the Motorola recommendations about board design) and alinear and has a lot of temperature drift (we use it from -10 to +70 degrees Celcius and have about 30 AD steps drift). To overcome these problems, we did the following: a1) Oversampling to eliminiate the noise (so take a lot of samples and calculate the average) In this way we also raised the resolution (not the accuracy). We take 512 AD samples, calculate the sum and divide the sum by 64. In this way we get 15 bit AD values. (So if you need 12 bit AD values, take 32 or 64 samples and calculate the average) a2) We do a 4 point calibration for each channel to overcome offsets and alinearity a3) We measure allways at a sepearate channel (with analog switches) a 250 mV and a 2850 mV reference value (also at calibration time). Temperature drift can be corrected by shifting and multiplying the scale dependant on these reference values b) As Art explained in some message, the startup behaviour of the crystal oscillator is not allways good at low temperatures. Workaround by external oscillator. My opinion about Codewarrior: I have never had a compiler with so much bugs as Codewarrior in my 17 years experience with embedded software. I lot of bugs have been solved in v5.0.3, but if Tasking had a compiler for DSP56F80x family, I would buy it immediately. My opinion is, that you can not perfectly repair buggy software. If you have spagetti software, the best you can do is to write at again, but then in a good way. And Metrowerks Codewarrior has all symptoms of spagetti software. Kind regards, Wim de Haan Exendis B.V. W.J. de Haan P.O.box 56, 6710 BB Ede Keesomstraat 4, 6716 AB Ede The Netherlands. Tel: +31-(0)318 - 676305 Fax: +31-(0)318 - 676319 mailto: URL: http://www.exendis.com <http://www.exendis.com/ -----Original Message----- From: greg_toman <> [mailto:] Sent: maandag 13 januari 2003 7:13 To: Subject: [motoroladsp] You opinion of Metrowerks and Motorola 56800 I would be curious to hear the opinions of those out there using this family of parts, with the Metrowerks CodeWarior. Specifically, do you think it is easy to use, relatively free of bugs so that it is unlikely to cause signifcant time spent solving issues that shouldn't be there in the first place? I tried out the evaluation board for the 56803, with the 5.0.2 version of CodeWarrior (until my liscense ran out). I am new to C, but not high level languages or assembly, but I had lots of problems. I don't know if it was me, or the tools, but at this point I have pretty much given up and am now evaluating another chip (Zilog), which I am having no problems with. The last problem I had, that I could not find a fix for, was simply a long integer divide that did not work right. It seemed like it had to be something I was doing wrong, but how hard is it supposed to be to do a divide? My types were Ok, and so on, but it didn't work. Multiplys did work, which made me think it wasn't me, but who knows. Anyway, when I saw the list of fixes implemented in 5.0.3, I got a bit scared. I might have been misuderstanding the fixes, but the fixes seemed to be for very basic problems, and when I asked Metrowerks to claify the nature of these fixes, I didn't hear back from them. The chip seemed to function fine, with the exception that the a/d was in error by about 1.25% when a 2.5V signal was applied. (I measured the a/d reference when calculating my expected result.) The a/d's were noisy also, but I tend to believe Motorola when they told me that can be improved with board layout. So far they have no answer for the a/d inaccuracy, but they said they could look into it for me. Anyway, I wonder what the opinion out there is...is Codewarrior a bit immature for real product development at this time, or is does it work good enough? Thanks Greg Toman |
Hi, Does anybody have an experience with DSP56F827 AD converter? It is the same quality like 56F807 ? Thanks, Richard > -----Original Message----- > From: Wim de Haan [mailto:] > Sent: Thursday, January 16, 2003 9:56 AM > To: 'greg_toman <>'; > Subject: RE: [motoroladsp] You opinion of Metrowerks and > Motorola 56800 > Hi Greg, > > We use DSP56F807 with Codewarrior compiler. > My opinion about the DSP56F807: > > a) The AD converter is extremely noisy (although we followed all the > Motorola recommendations about board design) and alinear and > has a lot of > temperature drift (we use it from -10 to +70 degrees Celcius > and have about > 30 AD steps drift). To overcome these problems, we did the following: > a1) Oversampling to eliminiate the noise (so take a lot of samples and > calculate the average) In this way we also raised the > resolution (not the > accuracy). We take 512 AD samples, calculate the sum and > divide the sum by > 64. In this way we get 15 bit AD values. (So if you need 12 > bit AD values, > take 32 or 64 samples and calculate the average) > a2) We do a 4 point calibration for each channel to overcome > offsets and > alinearity > a3) We measure allways at a sepearate channel (with analog > switches) a 250 > mV and a 2850 mV reference value (also at calibration time). > Temperature > drift can be corrected by shifting and multiplying the scale > dependant on > these reference values > > b) As Art explained in some message, the startup behaviour of > the crystal > oscillator is not allways good at low temperatures. > Workaround by external oscillator. > > My opinion about Codewarrior: > I have never had a compiler with so much bugs as Codewarrior > in my 17 years > experience with embedded software. > I lot of bugs have been solved in v5.0.3, but if Tasking had > a compiler for > DSP56F80x family, I would buy it immediately. > My opinion is, that you can not perfectly repair buggy > software. If you have > spagetti software, the best you can do is to write at again, > but then in a > good way. And Metrowerks Codewarrior has all symptoms of > spagetti software. > > Kind regards, > > Wim de Haan > Exendis B.V. > W.J. de Haan > P.O.box 56, 6710 BB Ede > Keesomstraat 4, 6716 AB Ede > The Netherlands. > Tel: +31-(0)318 - 676305 > Fax: +31-(0)318 - 676319 > mailto: > URL: http://www.exendis.com <http://www.exendis.com/ > > -----Original Message----- > From: greg_toman <> [mailto:] > Sent: maandag 13 januari 2003 7:13 > To: > Subject: [motoroladsp] You opinion of Metrowerks and Motorola 56800 > I would be curious to hear the opinions of those out there using > this family of parts, with the Metrowerks CodeWarior. Specifically, > do you think it is easy to use, relatively free of bugs so that it > is unlikely to cause signifcant time spent solving issues that > shouldn't be there in the first place? > > I tried out the evaluation board for the 56803, with the 5.0.2 > version of CodeWarrior (until my liscense ran out). I am new to C, > but not high level languages or assembly, but I had lots of > problems. I don't know if it was me, or the tools, but at this point > I have pretty much given up and am now evaluating another chip > (Zilog), which I am having no problems with. > > The last problem I had, that I could not find a fix for, was simply > a long integer divide that did not work right. It seemed like it had > to be something I was doing wrong, but how hard is it supposed to be > to do a divide? My types were Ok, and so on, but it didn't work. > Multiplys did work, which made me think it wasn't me, but who knows. > > Anyway, when I saw the list of fixes implemented in 5.0.3, I got a > bit scared. I might have been misuderstanding the fixes, but the > fixes seemed to be for very basic problems, and when I asked > Metrowerks to claify the nature of these fixes, I didn't hear back > from them. > > The chip seemed to function fine, with the exception that the a/d > was in error by about 1.25% when a 2.5V signal was applied. (I > measured the a/d reference when calculating my expected result.) The > a/d's were noisy also, but I tend to believe Motorola when they told > me that can be improved with board layout. So far they have no > answer for the a/d inaccuracy, but they said they could look into it > for me. > > Anyway, I wonder what the opinion out there is...is Codewarrior a > bit immature for real product development at this time, or is does > it work good enough? > > Thanks > Greg Toman > > > _____________________________________ > 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: > > To Post: > > To Leave: > > Archives: http://www.yahoogroups.com/group/motoroladsp > > More Groups: http://www.dsprelated.com/groups.php3 > ">http://docs.yahoo.com/info/terms/ > > |
Richard -
The DSP56F827 has the best ADC performance of the current DSP56F800 family. Its ADC has been modified slightly to achieve gains in linearity and noise performance. To achieve these results the maximum ADC internal clock frequency has been lowered from 5 Mhz to 2.5 Mhz. Thanks. - Bill
|