OK, i started this ARM venture with Atmel. got my feet wet and decided to move onto something with more gusto and better data sheets. so i moved onto NXP. i have several different flavors of them and all has been well until now. I recently bought another Olimex board, LPC-P2378. the board works but the main reason i bought it was to use the memory bus lines. i could not get the write line working to save my life, i then took a peek at the errata sheet. wouldn’t you know it Murphy strikes, there is a die problem with the version that is on the Olimex board in which the WR line does not work. the workaround is to toggle it manually they say.
SO, this is what i am pissed off about, how in the hell could NXP let something as obvious as a WR line not working get through QC. the QC department should be fired and replaced :x !!!
Moreover, i said no sweat and bought some LPC2378’s from digikey, different lot should be taken care of right? WRONG! the new chip i just replaced on my Olimex board suffers from the same problem. To top it off there is nothing i could find on NXP’s site that would let someone know what new offspring number has certain errata issues taken care of, so one does not keep buying LPC2378’s until the issue is resolved.
I know all companies have errata, but my new search is who has the least ? Who tests their products the most and lets the least slip through the cracks. what gets my goose, is your going to spend all that time and money to make a IC why in the hell wouldn’t you test the whole dam thing ?
There’s been some discussion of this sort of thing on the Yahoo LPC2000 forum - NXP FAEs lurk there and usually respond to such matters, but we have not yet had a reply.
IMO, yes. If you’re using their chips that have been released for a while, you’ll generally be fine. If you’re wanting to use their newer stuff, it’s a crapshoot. They’re just now releasing the new LPC2378 silicon that has the EMI fix. They’ve also finally acknowledged that their LPC23xx chips can’t reliably run full speed out flash, and can’t reliably use the MAM fully enabled. The LPC2468 still isn’t readily available, 8 months after it was slated to be released. The new chip with the integrated LCD controller (LPC2478?) looks really nice, but I have no confidence in it being released within the next 8 months, or even working when it does finally come out.
I’ll still use them for their simpler devices, like the LPC2148 and others, but with the newer chips, many of their supposed advantages over other manufacturers are riddled with errata. I’m using Atmel and ST for those applications instead. I’ll fully switch over once I feel annoyed enough by NXP. It probably won’t be long :lol:
well, i just got a response from their tek support on the issue.
i was given a pre-release user manual that has changed the bit settings for one of the registers i was using to configure the WR pin, i fixed that and now it works. I just wish they would have caught that before i wasted hours learning a new part and was convinced it was me that was the problem.
seulater:
OK, i started this ARM venture with Atmel. got my feet wet and decided to move onto something with more gusto and better data sheets.
Just out of curiosity, what is missing or wrong with the Atmel datasheets?
They are disjoint - you have to hunt around them to find all the info relevant to certain peripherals and sometimes the answer just isn’t there at all (having spent a lot of time reading the AT91SAM9261 data sheet of late).
seulater:
OK, i started this ARM venture with Atmel. got my feet wet and decided to move onto something with more gusto and better data sheets.
Just out of curiosity, what is missing or wrong with the Atmel datasheets?
They are disjoint - you have to hunt around them to find all the info relevant to certain peripherals and sometimes the answer just isn’t there at all (having spent a lot of time reading the AT91SAM9261 data sheet of late).
Robin
I've only used the AT91SAM7X256 - but the datasheet (even the preliminary one) was very complete and clear.
Not to mention most of the data sheets done even have an index. It may be a petty thing, but its very helpful when you need to find a particular thing.
seulater:
Not to mention most of the data sheets done even have an index. It may be a petty thing, but its very helpful when you need to find a particular thing.
Oh you print them? With a PDF no index hardly is a problem as a full text search is a lot easier … one window Acrobat second window an IDE/editor. Works really well for me and is a lot like most software documentation these days - one doesn’t get that printed any more
As for the hardware development, Atmel has a short guide with only the important information plus the reference/development kit documentation. That has always been sufficient for me.
no, not really. But many manufactures have the index in the pdf file on the left panel in acrobat, so when you want something it right there. I agree you could do a search, but my point is that Atmel needs to get with the program of making a respectable document.
no, not really. But many manufactures have the index in the pdf file on the left panel in acrobat, so when you want something it right there. I agree you could do a search, but my point is that Atmel needs to get with the program of making a respectable document.
I am looking at my copy of SAM7Sxxx.pdf and it has the index panel on the left hand side. Overall, I have been very happy with their documentation. I just downloaded the SAM9260 datasheet and it is indexed too although it was hidden by default.
In the quote I responded to, you stated they did not have an index in the pdf and you were ragging on Atmel for that. Look up 4 posts! Then you stated that an index was a new thing for Atmel - which isn’t true. I just looked in the back of the SAM9260 datasheet and guess what - it has a table of contents at the back that would print. Same as NXP does. And you know what else, the Mega128 datasheet from 2004 also has a table of contents at the back.
I like NXP too but you were asked why you don’t like Atmel. Thus far, you haven’t supplied much of a reason.
i don’t have a 2004 data sheet from them to verify, i will take your word for it. i have been an Atmel fan for several years, i have never liked their data sheets the way they explain registers. That my opinion.
This thread is not so much about atmel as it is about NXP. my reasons if you must know for leaving atmel were when i had a HUGE project complete all the boards were made and stuffed and in final test. i discovered a problem with the brown out detector not working properly on the MEGA8. At first i thought it was the board, i tried several all had the same problem. i then pulled the mega8 off and replaced it from another batch i got from another dist. with a different lot number. it then worked fine. i then pulled that new one off and replaced it from the older lot and it again failed the test. i once again replaced it with the newer lot number and it worked fine. i did this same test on 5 boards all had the same results. when i contacted atmel about it i was viewed like there must be something i am doing wrong. it was hard for me to barley get the time of day from them. they would not allow me to send them my boards with the different lot numbers IC’s to verify for themselves. meanwhile i got a HUGE order just sitting on the test line waiting…
i did much haggling with them to no avail. in the end they told me why don’t i just use the new lot. i had to throw out thousands of IC’s that day and Atmel would not even help out at all, not even take interest in it. For that i will NOT use them for nothing!
If a company is not there to support its customers they are useless in my eyes.
seulater:
i don’t have a 2004 data sheet from them to verify, i will take your word for it. i have been an Atmel fan for several years, i have never liked their data sheets the way they explain registers. That my opinion.
This thread is not so much about atmel as it is about NXP. my reasons if you must know for leaving atmel were when i had a HUGE project complete all the boards were made and stuffed and in final test. i discovered a problem with the brown out detector not working properly on the MEGA8. At first i thought it was the board, i tried several all had the same problem. i then pulled the mega8 off and replaced it from another batch i got from another dist. with a different lot number. it then worked fine. i then pulled that new one off and replaced it from the older lot and it again failed the test. i once again replaced it with the newer lot number and it worked fine. i did this same test on 5 boards all had the same results. when i contacted atmel about it i was viewed like there must be something i am doing wrong. it was hard for me to barley get the time of day from them. they would not allow me to send them my boards with the different lot numbers IC’s to verify for themselves. meanwhile i got a HUGE order just sitting on the test line waiting…
i did much haggling with them to no avail. in the end they told me why don’t i just use the new lot. i had to throw out thousands of IC’s that day and Atmel would not even help out at all, not even take interest in it. For that i will NOT use them for nothing!
If a company is not there to support its customers they are useless in my eyes.
Yes but they have no idea whether it’s your problem or theirs and I don’t know of any company that would spend any time finding out by debugging your board.
These Atmel parts are made by various manufacturing plants throughout the world, each of which has different processing limitations, so if you design is not very conservative, you will have problems like this.
Yes but they have no idea whether it’s your problem or theirs and I don’t know of any company that would spend any time finding out by debugging your board.
i don’t want them to debug the board for me. the simple fact that one lot works and another doesn’t clearly shows the something is wrong with one of the batches. having said that, you would think that would be proof enough for them to at least take an interest in the issue. at the very least they could have asked me to send them samples chips for the two different lots for them to verify. for them to not care just shows me the type of company they are.
P.S. i know a company should not look into every customer issue, however its clear to me that this is beyond some silly design issue on the customers end.
Fair enough reasons. Thanks for clarifying. I too had a problem with Atmel on the Mega128 that they would not admit to but was magically fixed by new lot number parts.
:roll: Specifically, the RAMPZ bit would randomly change with elevated temperatures (but within spec) causing all sorts of trouble.
I think all microcontrollers have issues of some sort that the manufacturer will never admit. But the problem was fixed and the project was successful. I still like Atmel.
I like NXP too. I would have used the LPC2148 on my current project but the peripherals line up for my application leaned towards the SAM7S256.
Personally, I love the fact that these two companies are in such close competition. I hope they keep it up.
