On Thu, 16 Aug 2012 22:19:49 -0400, Roberto Waltman <usenet@rwaltman.com>
wrote:

>rickman wrote:
>>I understand that perfectly.  I don't know why they are doing a board=20
>>spin, but they don't want to touch any code they don't have to.  Using =
a=20
>>different MCU chip can wreak havoc on code if it turns out to have=20
>>unsuspected hardware dependencies.
>
>Precisely. The new and old boards share 80% of the peripherals, and
>that means a lot of the code is already written, tested and known to
>be reliable, if we stay with the same CPU.

Well alrighty then.  Family compatible could be potentially acceptable
then.  Depends a lot on just which peripherals are onboard the MCU.

?-)

rickman wrote:
>I understand that perfectly.  I don't know why they are doing a board 
>spin, but they don't want to touch any code they don't have to.  Using a 
>different MCU chip can wreak havoc on code if it turns out to have 
>unsuspected hardware dependencies.

Precisely. The new and old boards share 80% of the peripherals, and
that means a lot of the code is already written, tested and known to
be reliable, if we stay with the same CPU.
--
Roberto Waltman

[ Please reply to the group,
  return address is invalid ]

On 8/16/2012 3:09 AM, josephkk wrote:
> On Tue, 14 Aug 2012 18:38:13 -0400, Roberto Waltman<usenet@rwaltman.com>
> wrote:
>
>> Uwe Bonnes wrote:
>>> How about using a uC with built-in unique ID? E.g. STM32F?
>>
>> Valid for a new design. This is a respin of an existing product, and
>> the CPU (untouchable) does not have an ID.
>
> Wait a minute, they are doing a respin and the old uC is untouchable???
> Hand them a flashlight and a crowbar.  They are in dire need.  Even in
> aerospace and medical any respin is effectively a new design.  New pass on
> ALL qualifications.
>
> ?-)

I understand that perfectly.  I don't know why they are doing a board 
spin, but they don't want to touch any code they don't have to.  Using a 
different MCU chip can wreak havoc on code if it turns out to have 
unsuspected hardware dependencies.

"There's many a slip, twixt cup and lip."

Rick

On Tue, 14 Aug 2012 18:38:13 -0400, Roberto Waltman <usenet@rwaltman.com>
wrote:

>Uwe Bonnes wrote:
>>How about using a uC with built-in unique ID? E.g. STM32F?
>
>Valid for a new design. This is a respin of an existing product, and
>the CPU (untouchable) does not have an ID.

Wait a minute, they are doing a respin and the old uC is untouchable???
Hand them a flashlight and a crowbar.  They are in dire need.  Even in
aerospace and medical any respin is effectively a new design.  New pass =
on
ALL qualifications.

?-)

Uwe Bonnes wrote:
>How about using a uC with built-in unique ID? E.g. STM32F?

Valid for a new design. This is a respin of an existing product, and
the CPU (untouchable) does not have an ID.
--
Roberto Waltman

[ Please reply to the group,
  return address is invalid ]

In comp.arch.embedded Roberto Waltman <usenet@rwaltman.com> wrote:

> For a project I am working on, I would like to give boards fresh from
> manufacturing a distinct "identity", before they are assigned a serial
> number, or have a MAC address or IP address programmed, etc.

> This could be provided by some devices, such as Maxim's DS2411
> "Silicon serial number" ( with a "Unique, Factory-Lasered and Tested
> 64-Bit Registration Number" ) or DS18B20 temperature sensor, ( "has a
> Unique 64-Bit Serial Code Stored in an On-Board ROM" )

> Looking for the least expensive chip with such an ID, with an I2C
> (preferred), SPI or 1-wire interface. Don't care what other
> functionality that chip may have, I just want the unique ID.
> A device that report its own serial number would be OK.

> Any recommendations?

How about using a uC with built-in unique ID? E.g. STM32F?

Bye
-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

In article <TsOdnZ0rq-7G-LjNnZ2dnUVZ_rSdnZ2d@earthlink.com>, jchisolm6
@earthlink.net says...
> On Fri, 10 Aug 2012 13:47:30 -0400, Roberto Waltman wrote:
> 
> > For a project I am working on, I would like to give boards fresh from
> > manufacturing a distinct "identity", before they are assigned a serial
> > number, or have a MAC address or IP address programmed, etc.
> > 
> > This could be provided by some devices, such as Maxim's DS2411 "Silicon
> > serial number" ( with a "Unique, Factory-Lasered and Tested 64-Bit
> > Registration Number" ) or DS18B20 temperature sensor, ( "has a Unique
> > 64-Bit Serial Code Stored in an On-Board ROM" )
> > 
> > Looking for the least expensive chip with such an ID, with an I2C
> > (preferred), SPI or 1-wire interface. Don't care what other
> > functionality that chip may have, I just want the unique ID. A device
> > that report its own serial number would be OK.
> > 
> > Any recommendations?
> > 
> > Thanks,
> 
> I use the Microchip 25AA02E48 in several products.  It's SPI but the
> 24AA025E48 is I2C.  They will run you about $0.23 in a SOT-23
> package.  Get them from Mouser,Digikey,Avnet or direct from Microchip.
> Good news is you also get some EEPROM along with the MAC address.
> 
> 
I second the recommendation, I use both flavors of these guys a regular 
basis.  As easy to use as any serial EEPROM and cheap.

Leo Havm�ller wrote:
>Does the board have a flash? Then maybe you already have a 64-bit unique ID 
>available.

Thanks,  I am aware of those and no, the only flash is the CPU's
internal memory.
--
Roberto Waltman

[ Please reply to the group,
  return address is invalid ]

> For a project I am working on, I would like to give boards fresh from
> manufacturing a distinct "identity", before they are assigned a serial
> number, or have a MAC address or IP address programmed, etc.

Does the board have a flash? Then maybe you already have a 64-bit unique ID 
available.

Leo Havm�ller. 

Nico Coesel wrote:
>Nowadays a lot of MCUs come with a unique serial number.

I know of a few, such as NXP's LPC1311.  But the processor in this
project does not have this feature.  (Can not change that)
--
Roberto Waltman

[ Please reply to the group,
  return address is invalid ]