AVR-Chat

For those interested in 'A' versions:

I have stuffed ~8000 ATMega16 onto boards, or at least they were supposed to be ATMega16. In reality, even though the package says 16, they are 16A (the die shrink).
Decapsulation confirms this, and Atmel confirms this. They were acquired through a reputable distributor, and the distributor is willing to help, so no blame there.

They are supposed to be functionally identical to non-A, only on the 16A I have, 95% of them can not run without capacitor on AREF. Atmel confirms that they should be able to do so.

This is what AREF looks like when switching from AVCC to internal 2.56V:

http://n1.taur.dk/avrfreaks/s16a9.png

neat, huh?

Is anybody else seeing this? These were dc 1036, on boards with awesomely clean power supplies and decoupling.

/Kasper Pedersen (KKP)

And to add a conspiracy theory, THIS is why avrfreaks is down ;-)

--- In AVR-Chat@yahoogroups.com, STEVEN HOLDER <s.holder123@...> wrote:

> Does not look good does it, i assume you do not get this with the non 'a' version ?

Yes, we have been using them since they were 'new'.

> Is the AREF pin left unconnected? i assume you are using the recommended filter circuit on the AVCC line and that there are no "switching" siganals that could couple into the circuit while the conversion is taking place ?

It is filtered, yes. So is VCC, and they are very clean, <500uV. While ADC needs to be enabled for the 2.56V reference to be output, it does not need to be converting.

> What is the clock frequency of the device ? and what is the converer prescaler ?

11.0592MHz. And /64 for ~176kHz. 

> The data sheet does by impication suggest a capacitor should be placed at the aref pin.

It says noise performance. And Atmel says it works without, just as it does in parts without an AREF pin (such as Tiny13).

And while I do have room for the capacitor, it means a delay every time I switch the reference. And since both the capacitance at which this part stops oscillating, and the sink current and thus settling time is undocumented, I will be relying on a double-undocumented fix. Just lovely.

/Kasper Pedersen

At 07:05 PM 18/07/2011, you wrote:
>95% of them can not run without capacitor on AREF. Atmel confirms 
>that they should be able to do so.

But you have made provisions for the cap on your PCB
and put it in as recommended right? :-)


Regards

John Samperi

********************************************************
Ampertronics Pty. Ltd.
11 Brokenwood Place Baulkham Hills, NSW 2153 AUSTRALIA
Tel. (02) 9674-6495       Fax (02) 9674-8745
Website  http://www.ampertronics.com.au
*Electronic Design * Custom Products * Contract Assembly
********************************************************

Hi,
 
Does not look good does it, i assume you do not get this with the non 'a' version ?
 
Is the AREF pin left unconnected ? i assume you are using the recommended filter circuit on the AVCC line and that there are no "switching" siganals that could couple into the circuit while the conversion is taking place ?
 
What is the clock frequency of the device ? and what is the converer prescaler ?
 
The data sheet does by impication suggest a capacitor should be placed at the aref pin.
 
Regards


--- On Mon, 18/7/11, kkpdk <kkpyahoogroups@kasperkp.dk> wrote:

From: kkpdk <kkpyahoogroups@kasperkp.dk>
Subject: [AVR-Chat] ATMega16A - AREF oscillating
To: AVR-Chat@yahoogroups.com
Date: Monday, 18 July, 2011, 10:05


  




For those interested in 'A' versions:

I have stuffed ~8000 ATMega16 onto boards, or at least they were supposed to be ATMega16. In reality, even though the package says 16, they are 16A (the die shrink).
Decapsulation confirms this, and Atmel confirms this. They were acquired through a reputable distributor, and the distributor is willing to help, so no blame there.

They are supposed to be functionally identical to non-A, only on the 16A I have, 95% of them can not run without capacitor on AREF. Atmel confirms that they should be able to do so.

This is what AREF looks like when switching from AVCC to internal 2.56V:

http://n1.taur.dk/avrfreaks/s16a9.png

neat, huh?

Is anybody else seeing this? These were dc 1036, on boards with awesomely clean power supplies and decoupling.

/Kasper Pedersen (KKP)

And to add a conspiracy theory, THIS is why avrfreaks is down ;-)








[Non-text portions of this message have been removed]

Kasper,
 
You are correct data sheet sheet does state that the cap at aref is for improved noise performance although in the register settings (ADMUX) it does say that when selecting the avcc or internal reference that a cap should be at aref 
(Internal 2.56V Voltage Reference with external capacitor at AREF pin).
 
Must admit i have always "assumed" that this cap should be there. You're right about the delay between switching references though.
 
The wave form at this pin is a little odd, it looks as though the waveform  is being integrated by something switching the reference, possibly through the on board reference select fet and some parasitic capcitance that is 'seen' at this pin, it seems to be very periodic (not noise) and consistant.
 
When using testboards i have sometimes left this cap off, and can't recall seeing something like this. Is the reference being switched on and off quickly, looks like around 2MHz ?
 
Regards
 


--- On Mon, 18/7/11, kkpdk <kkpyahoogroups@kasperkp.dk> wrote:

From: kkpdk <kkpyahoogroups@kasperkp.dk>
Subject: [AVR-Chat] Re: ATMega16A - AREF oscillating
To: AVR-Chat@yahoogroups.com
Date: Monday, 18 July, 2011, 11:55


  





--- In AVR-Chat@yahoogroups.com, STEVEN HOLDER <s.holder123@...> wrote:

> Does not look good does it, i assume you do not get this with the non 'a' version ?

Yes, we have been using them since they were 'new'.

> Is the AREF pin left unconnected? i assume you are using the recommended filter circuit on the AVCC line and that there are no "switching" siganals that could couple into the circuit while the conversion is taking place ?

It is filtered, yes. So is VCC, and they are very clean, <500uV. While ADC needs to be enabled for the 2.56V reference to be output, it does not need to be converting.

> What is the clock frequency of the device ? and what is the converer prescaler ?

11.0592MHz. And /64 for ~176kHz. 

> The data sheet does by impication suggest a capacitor should be placed at the aref pin.

It says noise performance. And Atmel says it works without, just as it does in parts without an AREF pin (such as Tiny13).

And while I do have room for the capacitor, it means a delay every time I switch the reference. And since both the capacitance at which this part stops oscillating, and the sink current and thus settling time is undocumented, I will be relying on a double-undocumented fix. Just lovely.

/Kasper Pedersen








[Non-text portions of this message have been removed]

--- In AVR-Chat@yahoogroups.com, STEVEN HOLDER <s.holder123@...> wrote:
> The wave form at this pin is a little odd, it looks as though the waveform  is being integrated by something switching the reference, possibly through the on board reference select fet and some parasitic capcitance that is 'seen' at this pin, it seems to be very periodic (not noise) and consistant.

The AREF waveform is the sourcing transistor turning on when AREF gets below 2.56V, sourcing lots of current into the pin capacitance, and overshooting by ~2V. The pin then discharges relatively slowly at around 140uA, until the cycle repeats. The 2MHz observed matches a 140uA discharge rate of 35pF, which seems like a reasonable ballpark figure for AREF internal capacitance.

This is the same waveform you get if you pick a voltage regulator that requires capacitive load on the output, and neglect to give it this load.

When I put capacitance on AREF, the slew rate of the charge/discharge changes (and thus the amplitude and frequency drops), and around 300pf-1nF, depending on the part, it becomes stable. 

/Kasper Pedersen

Kasper,
 
Looks like a capacitor on that pin is required then, no way round it.
 
Regards


--- On Mon, 18/7/11, kkpdk <kkpyahoogroups@kasperkp.dk> wrote:

From: kkpdk <kkpyahoogroups@kasperkp.dk>
Subject: [AVR-Chat] Re: ATMega16A - AREF oscillating
To: AVR-Chat@yahoogroups.com
Date: Monday, 18 July, 2011, 14:49


  





--- In AVR-Chat@yahoogroups.com, STEVEN HOLDER <s.holder123@...> wrote:
> The wave form at this pin is a little odd, it looks as though the waveform  is being integrated by something switching the reference, possibly through the on board reference select fet and some parasitic capcitance that is 'seen' at this pin, it seems to be very periodic (not noise) and consistant.

The AREF waveform is the sourcing transistor turning on when AREF gets below 2.56V, sourcing lots of current into the pin capacitance, and overshooting by ~2V. The pin then discharges relatively slowly at around 140uA, until the cycle repeats. The 2MHz observed matches a 140uA discharge rate of 35pF, which seems like a reasonable ballpark figure for AREF internal capacitance.

This is the same waveform you get if you pick a voltage regulator that requires capacitive load on the output, and neglect to give it this load.

When I put capacitance on AREF, the slew rate of the charge/discharge changes (and thus the amplitude and frequency drops), and around 300pf-1nF, depending on the part, it becomes stable. 

/Kasper Pedersen








[Non-text portions of this message have been removed]