ATMEGA163 DON'T EVEN THINK ABOUT IT! WARNING!

Go To Last Post
10 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This e-mail I just sent to ATMEL is self explanatory.
Mr Jacob L. Lassen

Further to my previous e-mail

I have spent a lot of time on this, to say I am rather upset would be an understatement.

ATMEGA163L power save mode:

Using a 3.6864 crystal, 3.6 v lithium battery 2,400 mAh (hey the smaller ones go flat too quickly!!)
No external components connected at all only the two crystals 3.6864mhz and 32khz 12pf(what an application hey!)
ports set as input, pull up enabled.

Yes the chip is properly decoupled etc, I am 55 years old, I was there when the 8080 arrived!

500ua is the best I can achieve.

Using 100k pull ups on all ports same as before no external connections on ports except crystals.

500ua is the best I achieve.

The PART I have has the following markings (I have two parts, cost me an arm and a leg!)

ATMEGA163L
4PI
0037

underneath it has;
a small circle with 67 in it
also; laser etched the following markings
OG4698-19906F
1KOREA-H OG0035

Now as if that is not sufficient to shatter my confidence in ATMEL parts forever, there's more.

Using an STK500

Using the ADC measuring one channel at a time.

I wired porta to a couple of switches using the 2 cable connector, from say PA0-PA1 to SW0-SW1

most channels when I press the switch and connect 150ohms to gnd, read 13mv to 18mv

HOWEVER CHANNEL 7 reads 0 mv (it is the only one that is correct!)

SO I measured the actual pin and sure enough, that is exactly what they are most read 13 to 18mv and channel 7 reads 0 mv

Wow this is certainly not nice now is it?

When I observe it with an oscilloscope, all pins are high, except channel 7 it is low.

When I press reset and hold it, all ports are low, data sheet says they are tri-stated and float!!

Both chips that I have do exactly the same thing, so at least it isn't random thank god!

I have spent $1500, wasted nearly 2 weeks and do not have any confidence in this chip, none, sit, schil, nada, Zemi, UNDERSTAND!!!

I read the posting on AVR Forum, (after I started all this wish I had read it before!!!) warning of the pitfalls in using AVR from someone calling himself AVR GUY, I have to totally agree with him, this stuff is badly if at all documented, and it is preliminary till when? when ATMEL can actually make one or till like "Intel" the data sheet is change to match reality.

The panic and scarcity of parts will work for a while, and then no one will want your non functional parts!!

Unfortunately in the real world I can't simply mark my product PRELIMINARY, SORRY, WORKAROUND AVAILABLE ON REQUEST!!

An honest detailed (more than one paragraph saying OH YEAH, sorry, it is preliminary etc) would be the least I would expect.

I spelled checked this using Microsoft Outlook and it sugested changing ATMEL to ATHEL, sounds appropriate don't you think?

Sort of kind regards (how is that new job application going?).
Joseph Abram

admin's test signature
 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I hope the Mega103 design I am supposed to deliver end Mar'01 does not misbehave.
Only section not used is the SPI - all the others must work.
Any detail I find/work out I'll forward.

Bob Rust

admin's test signature
 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My understanding is that the mega163 is not released yet. How and when did you get your samples ?

I have used the Mega103 in production for more than 16 months without having any problems. In the early days when mega103 was at preliminary stage I had some issues, but as soon as it became available in the commersial market it was 100% predictable. I have used about 40K units total in two products.

I assume the same thing is true with the mega163 as well. It seams like an extreamly powerful part !

I have seniors in my group that have done the decoupling failure several times so I suppose Mr. Lassen have reasons for his suggestions...I am very happy with the support I get from Atmel FAE's and hope they are active on this site.

Ian Fischer

admin's test signature
 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I obtained my samples (I did pay for them) together with an STK500 and ICE200 from the local distributor in Sydney Australia.

He did say that if I wanted chips in quantity there was a 16 week lead time.

I also purchase CodeVision AVR from Pavel Haiduc.

Now CodeVision AVR is something I DO RECOMMEND, the guy is a genius extraordinaire!!

I have solved the problem of making this chip work at low power 7ua, details published at http://www.ja.olm.net

Joseph Abram

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I left off the last part of the URL

The details of atmega163l, working schematics and all are at

http://www.ja.olm.net/rtc

Joseph Abram

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think it's inappropriate to chuck about electric specifications before the part is actually released. Compare it to someone bashing a software product that is in Beta.

Dave

admin's test signature
 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok yes, but!

No one told me it wasn't released.

But surely that doesn't prevent ATMEL from telling me (precisely, exactly using a simple schematic, ala TEST CIRCUIT) how to obatin the same low power figures they got, after all does all the stuff have to be handed down from father to son after a secret handshake or what?

In any event the 8535 is the same, yet I don't see any definitive app note on how to use the rtc, NO NOT THE SOFTWARE, but the hardware, with all the complications of having some parts powered up while others are powered down while you're trying to make it all run of the smell of an oily rag.

Anyway I have solved it. But there is more to the subject than meets the eye initially, so it would be nice to see a real HARDWARE app note from ATMEL.
Joseph Abram

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi all!

The ATmega163(L) has been released, and datasheets are available at www.atmel.com (as indeed on this very site).

Morten, Atmel FAE

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am surprised!!!!

ATmega163 have been available in samples since early 2000 (I dont know the exact date)
the Datasheet was released from NDA in end of October 2000, also first ATmega163 did
arrive on public sales (on off qty) on 20 October 2000, they have been available before
that of course, but in non production quantities.

I think the first 'real production run' was actually in end of December 2000, but you can not
say that the part is not released.

Sure the current part has 'silicon bugs in it' ASFAIK all the ATmega163's produced so far
(I havent checked the latest silicon runs)

The bugs know are

Problem with onchip charge pump shut off timing: if you try to verify a code page
after writing it then you wil get up to 10% programming failure rate.

Solution: program all array (no verify) then verify all array.
Program Block, Verify Block does not work! (up to 10% failure)

It was possible to erase the signature (maybe this is fixed)

I also have been able to make FULL CHIP ERASE whilst in internal Boot Rom, by any
specification this can not happen. I have seen it happen. The second time it occoured
I called an eye whitness to look from side and verify what I was doing, and yes the
total self erasure was confirmed.

My friend has had also several WEIRD things happening with ATmega163 but he
has just 'used other ways' to get around so I am not sure if the thigns reported
are real bugs or not, we didnt have time to go trouble hunting.

As of the above bugs, some of them have been reported to Atmel by several
Parties and ASFAIK Atmel has not admitted them possible. They kept telling me
that what I see is impossible. And then stopped to respond at all.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

On your web site you say that disabling AVCC helps. This is not a recommended way of doing it, the data sheet specifies maximum 0.3 volts difference between VCC and AVCC. A larger voltage difference may be harmful.

I think the reason for your high power consumption might be that you did not disable the ADC before entering the power down/power save mode. It is stated in the data sheet that you must do this.

Hope this helps. Sorry about your lack of confidence...

Arne

admin's test signature