90S2343 startup/reset problem

Log in or register to post comments
Go To Last Post
12 posts / 0 new
Author
Message
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
#1
Posted by admin: Wed. Mar 7, 2001 - 09:31 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello,

I have a problem with resetting/starting a 90S2343 processor (exact printed on the chip: ATMEL 9819 AT90S2343 10PC).
When I power up my (simple) app, the processor doesn’t start (I use the internal clock). Puling down the reset line doesn’t help. A short short-circuit of the power lines does the trick, the processor starts. Also, if I disconnect the power supply en apply it again, it will start too. The power supply is a simple transformer, a bridge, capacitor and a 78L05. It is stable (no ripple). When disconnecting and connecting, the main capacitor isn’t fully discharged yet.
The same problem occurs with the AVR STK 200 evaluation board. To reset/start the processor, I also have to switch the power on, wait for a moment, switch off, wait 1 sec and switch on again. Now the processor starts.

I read the datasheet forwards and backwards three times, but couldn’t find an answer. Can someone tell me what’s wrong with a 78L05 power supply? Are there special reset considerations? Some other idea’s?

Regards Bert

admin's test signature
 

Tags:

AVR Microcontrollers, megaAVR and tinyAVR, AT90S2343
Top
fAVR
fAVR's picture
Level: New Member
Joined: Mon. Mar 5, 2001
Posts: 8 View posts
Posted by fAVR: Wed. Mar 7, 2001 - 10:02 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Bert!
Finally I fond someone able to understand my nightmare!
Yes, you described exactly what happend at 2343s running with internal osc. .
You have not red the ERRATA for 2343, get it at "Device" section in this site.
The only REAL trick seems to be... Change the device to Tiny22L.
At any rate I believe the real problem is not the device's misunderstanding about clock souce: even after statup (few power up and down needed) my devices hang, it seems 1-2ms after running, appling another startup it do not happens.
At first times I thougth it depends on my program, but at second startup (sometimes 3th, 4h) all goes right and, in my experience, code not running now, never will run. Then I saw at the programmer (nice, cheap, simple BA1F), nothing to do, then only trouble is in the device.
Unfortunatelly I need small devices and I'm not able to handle SOIC.
So I tried using an RC chain locked to RST, giving 1ms reset after power on, no good, I tried to power my devices by trasformers, Switchings, cells, sometimes even directly with 3 AA 1.5V batteries, same old story.
Some circuits for brown out detection, I never experimented, can they help??

At soon
Fabrizio

Top
morten
morten's picture
Level: Wannabe
Joined: Wed. Mar 14, 2001
Posts: 92 View posts
Posted by morten: Wed. Mar 7, 2001 - 10:04 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Bert,

There's a '2343 errata on this. I quote:

WRONG LATCHING OF RCEN FUSE:
-------------------------------------------
If V CC goes below GND and then up to the operating
voltage, the RCEN fuse can be read as unprogrammed
even if it is programmed. The result of this is that the
device starts looking for a clock signal on the external
clock input instead of from the internal RC oscillator,
making it look as if it “hangs”.

I'm sorry to have to say that there's no available workaround. The problem has been fixed in the replacement chip ATtiny22L, but this chip is not yet available.

If your application is too critical to allow having to switch the device off and on, an alternate option is to run the part on external clock.

Morten, AVR tech. support.

Top
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
Posted by admin: Wed. Mar 7, 2001 - 10:25 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Morten,

I looked for errata sheets, but didn’t saw them. Probably it was to late when I was searching….
When my app is powered off, there is no power is the app (no batteries), so all lines are 0V. Then when applying the power, there shout be a dip then before a rise….. seems strange. Would a shockley diode clamp help?
Unfortunately my memory scoop is broken and I cannot measure transients right now. My old regular scope won’t do the trick. This will be my last Philips scoop, for that matter.

Regards, Bert

admin's test signature
 

Top
morten
morten's picture
Level: Wannabe
Joined: Wed. Mar 14, 2001
Posts: 92 View posts
Posted by morten: Wed. Mar 7, 2001 - 10:39 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi again,

The errata sheets are available at this site! Just go to the "Device" section, find the device you intend, and all available documents will be listed. I use this site instead of the official one for info gathering :)

I haven't tried a diode clamp. Anyone?

Morten

Top
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
Posted by admin: Wed. Mar 7, 2001 - 10:40 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Fabrizio,

So I am not alone then…. I don’t now of I should be happy with that :o(
The strange thing that everyone who is using this processor, should be having the same startup problems. So, how are they solving the problem then?
It seems that a steep rise of the power supply voltage solves the problem, e.g. switching the +5V instead of the AC. I don’t like adding circuitry to do power switching; the idea of using this processor is building a cheap low component app. (I have to make > 5000 of them)
I think that the best approach would be finding a solution to lose my dependence of the internal RAM then. I could use the ATtine11/12 instead. I understand that I would not have startup problems.

Regards, Bert

admin's test signature
 

Top
danni
danni's picture
Level: Raving Lunatic
Joined: Wed. Sep 5, 2001
Posts: 2960 View posts
Posted by danni: Wed. Mar 7, 2001 - 11:08 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Bert,

this is exact the same problem, which I have seen on the first ATtiny22.
But in opposition to the errata sheet VCC was never reversed (never below 0V). It occur anytime if VCC was 0V some minutes.
The only solution I found, was to rise VCC very slow. With a 100R in series to VCC and 470µF parallel to the tiny22 it works, but then VCC not very stable, because some voltage drops on the 100R.

The new tiny22L with removed RCEN fuse is not avaiable now.

If you need a solution with SRAM instantly, please look at the Fairchild ACE1202. It works always on power on.
Some ACE program examples can you found on my web:

http://www.specs.de/~danni/ace/i...

Peter

Top
fAVR
fAVR's picture
Level: New Member
Joined: Mon. Mar 5, 2001
Posts: 8 View posts
Posted by fAVR: Wed. Mar 7, 2001 - 11:20 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey Bert,

this argument seems to be cool now.
My quantities are lower, but I used 2343 for my home automation (no other much mission critical to me!), I buit 70 units and I need probably 50 units yet (2.5x2.5cm each), they leave in each wall box, they control all ligths, themp and so on, BUT... If a power black out occurs, I need to open almost a half of wall boxes for manual restarting of devices!
So I'm using a car battery and a main line powering all devices.
Unfortunatelly I won't use EEPROMless devices, in the matter of fact I can replace single device's parameters remotely upon then network, and the program reached (not all finished yet!) over 950 words so, there is no substitute else than Tiny22.

Note:
2343 seems to have some internal capacitor, try this: when it doesn't start, get it out (phisically) from the circuit, manipulate a little (so discharging), and then put it back, and power up. Probably it will start.

If you will try the diode, let me know what the result.

Fabrizio

Top
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
Posted by admin: Wed. Mar 7, 2001 - 12:19 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,

It seems that the errata sheet problem is exactly what is going on (at least the symptoms).
I connected a function generator to pin 2. When switching on the app, nothing happens. Then when switching the generator on: the processor starts. After a power down and up, the processor runs on its internal oscillator (detectable by the speed change). Discharging everything to 0V, repeats the above procedure.
I connected a BAW62 diode (http://www.semiconductors.philip... - the fastest I have on hand right now, but the forward voltage drop is 0.7 V, which is maybe to high) directly over the power pins and over the reset pin. It doesn’t help.

Bert

admin's test signature
 

Top
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
Posted by admin: Mon. Mar 19, 2001 - 05:19 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello Morten,

Reading the signature bytes of my socalled ATtiny22L
shows that inside there is actually a 2343 ! (according
to the signature described in the datasheets).

So now I am stuck with a batch of tiny22 micros (production
week 9935) that are in fact 2343-s and won't start-up at
power-up !

Is this possible ???
Any comment ?

thanks,
Henk Boterenbrood, Amsterdam

admin's test signature
 

Top
newuser123
newuser123's picture
Level: New Member
Joined: Wed. May 28, 2003
Posts: 3 View posts
Posted by newuser123: Tue. Mar 20, 2001 - 07:58 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello Henk.
I heard some place that some of the first tiny22s had got the wrong signature byte programmed during testing. But it should not make any difference since both tiny22 and 2343 is programmed the same way (and have the same amount of memory)

John B

Top
Silver Markadmin
admin's picture
Level: Site Admin
Joined: Fri. Oct 18, 2013
Posts: 53184 View posts
Posted by admin: Tue. Mar 20, 2001 - 10:32 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi John,
of course I know the tiny22 and the 2343 are identical
but the point is that the 2343 is almost useless because of
this power-up problem it has (if it has been switched off
for some time and then switched on it is _guaranteed_ to
_not_ run, which is what I see in my system).
This problem was supposed to be fixed in the tiny22
(because there is no internal/external clock choice for this type), but as it turns out there is a 2343 in my tiny22 package!
So I have this start-up problem and I'm really annoyed by it,
because the micros are already installed on all my boards...
(about a 100)
How could ATMEL not have seen this problem during testing,
before they threw them on the market (as far as I'm concerned this bug is fatal for any product using this micro!).
It's a damn shame.

Henk

admin's test signature
 

Top