Searching \ for 'In System Calibration' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=system+calibration
Search entire site for: 'In System Calibration'.

Truncated match.
PICList Thread
'In System Calibration'
1999\05\04@100044 by Lawrence Lile

flavicon
face
I've been dreading a project that may require in-system calibration.  Here's
the scenario - I'm using some cheap, poor tolerance thermistors and
potentiometers in a controller.  I have a number of these built up, and
found their behavior changed quite a bit from unit to unit, ostensibly
because of tolerance variations.  (I'm not really sure)

I could tighten the tolerance by spending money, but the circuit must be
absolute minimum cost.  So I'm imagining a system that:

1.  Does a final check on resistances in the circuit  (and maybe a quality
control check too?)

2. Calculates two or three constants based ont he resistances in the circuit

3. Creates the proper code (would it have to COMPILE or just modify the
right bits in a hex file?)

4. Squirts the program into a PIC either in- circuit or just before
insertion

It sounds really complex and capitol-intensive.

Has anybody worked with these before?  Is there commercial equipment that
does this task automatically?

1999\05\04@110847 by Andy Kunz

flavicon
face
This sounds like you want a bed-of-nails to get the values to work from.
Expect to pay $1000 for the unit + engineering to design it.  Not cheap.

The alternative is to have the device talk to you and accept changes, which
it stores in an EEPROM.  This is the method I have used several times and
it is MUCH preferable.


>3. Creates the proper code (would it have to COMPILE or just modify the
>right bits in a hex file?)

Just modify the hex file.  You need to be able to read the correct line,
adjust the value, and rewrite it.  The reason for this is that the HEX file
has a checksum on each line.

>4. Squirts the program into a PIC either in- circuit or just before
>insertion

Either method works.  I do the in-circuit deal on my stuff.

>It sounds really complex and capitol-intensive.

It can be, but for really high volume it's cost-effective.  For small runs,
you are better to add a couple cents to the circuit so you have pads which
can be rx/tx data when in "not completely programmed" mode (ie, the EEPROM
is empty).

>Has anybody worked with these before?  Is there commercial equipment that
>does this task automatically?

Every one is unique.  That's part of the engineering.

Andy
==================================================================
  Montana Design Tech Support - http://www.montanadesign.com
==================================================================

1999\05\04@111915 by Michael Rigby-Jones

flavicon
face
We had to do something similar.  Fortunately we were using a Xicor quade2pot
for setting various parameters in the system and it happens to have a load
of spare registers that we use to store calibration info.

I guess you could pad the thermistors/pots to get a nominal value but I
don't suppose you want the hassle of SOT.

As long as you know the location of the calibration factors then you can
just modify the HEX file, but you will have to adjust the checksum at the
end of the line.

Good luck

Mike Rigby-Jones

> {Original Message removed}

1999\05\04@130439 by Lawrence Lile

flavicon
face
My thermistors cost US$0.15 and are (I think) 20% tol.   I need to get
quotes on 10% thermistors, that may be money well spent!

Supply voltage is pretty constant.

Thermistor is checked by using an AN512 style r-c time constant circuit in a
PIC.  It is only measured every ten second s or so to prevent self-heating.
The rest of the time there is zero (10e-8 amps?) current through it.

The tolerance on the capacitor in the RC , and power supply votlage s
compensated by measuring a standard carbon film resistor of known value,
then dividing by this value and multiplying by a constant.

The variations between PICS are probably a major contributor to tolerances,
especially variations in TMR0 trigger voltage.



{Original Message removed}

1999\05\04@135800 by Lawrence Lile

flavicon
face
-----Original Message-----
From: Andy Kunz <spam_OUTsupportTakeThisOuTspamMONTANADESIGN.COM>
To: .....PICLISTKILLspamspam@spam@MITVMA.MIT.EDU <PICLISTspamKILLspamMITVMA.MIT.EDU>
Date: Tuesday, May 04, 1999 10:09 AM
Subject: Re: In System Calibration


>This sounds like you want a bed-of-nails to get the values to work from.
>Expect to pay $1000 for the unit + engineering to design it.  Not cheap.



Nah.  Cheap.  I expect to pay $10,000 to $40,000 to tool this project
anyway.

1999\05\04@160830 by Craig Lee

flavicon
face
Another idea, Andy seeded here, is to program your standard hex file, but
save a place for storage later.  This way your hex file is never modified
and can remain read only in your document control tree.

Power up the circuit, do your measurements, power it down.

Now program the cal factors to the saved location.

Remember that 00s in an OTP can be written over.

Craig

{Quote hidden}

1999\05\04@171615 by Andy Kunz

flavicon
face
>Remember that 00s in an OTP can be written over.

No, FF's can be programmed over.

Andy


==================================================================
  Montana Design Tech Support - http://www.montanadesign.com
==================================================================

1999\05\04@181422 by Peter van Hoof

flavicon
face
-----Original Message-----
From: Lawrence Lile <.....lilelKILLspamspam.....TOASTMASTER.COM>
To: EraseMEPICLISTspam_OUTspamTakeThisOuTMITVMA.MIT.EDU <PICLISTspamspam_OUTMITVMA.MIT.EDU>
Date: Tuesday, May 04, 1999 11:05 AM
Subject: In System Calibration


>I've been dreading a project that may require in-system calibration.
Here's
{Quote hidden}

circuit
{Quote hidden}

dip the thermistor in icewater/boiling water and start autocalibration with
a dedicated input or unusual combination of already used input pins

store the calibration value in eeprom

Peter

1999\05\05@042505 by Benjamin Petersen

flavicon
face
> [@spam@PICLISTKILLspamspamMITVMA.MIT.EDU]On Behalf Of Craig Lee
> Remember that 00s in an OTP can be written over.
Isn't that 0xff or is that only for JW parts ?

>
> Craig
>


Regards
Benjamin Petersen

1999\05\05@053055 by mlsirton

flavicon
face
Hi,

On  4 May 99 at 11:59, Lawrence Lile wrote:
> My thermistors cost US$0.15 and are (I think) 20% tol.   I need to get
> quotes on 10% thermistors, that may be money well spent!

You could sort the thermistors before assembly.
If we assume uniform distribution of values you could pre-sort all
the thermistors to lets say 10 piles and then get 2% tolerance in
each of the piles.  Now each pile goes with its own software.
The question is if the temperature coefficient has good enough
tolerance.  You also have to pay attention to the ambient temperature
changes while sorting.

Hope this helps,
Guy - KILLspammlsirtonKILLspamspaminter.net.il

1999\05\05@094135 by paulb

flavicon
face
Lawrence Lile wrote:

> I've been dreading a project that may require in-system calibration.
...
{Quote hidden}

 Fascinating proposition.  As I see it, there are a few choices:

1} Measure the parameters externally and externally customise a program
to match.  Does not take into account PIC parameters, but can use cheap
OTP PICs.  May be performed by selecting parts before assembly, or in a
jig.

2} Include a test program as initial program, which performs a
calibrate function on the assembled unit, returning the data to the jig.
The test program is at the very end of memory.  The returned data is
used to adjust the main program which is then burned.  Can use OTP PICs
without EEPROM.

3} Variant; burn in the "full" program with a test mode which is then
used to make the calibration, the results are patched in.  Advantage:
test code is shared with normal code (temperature measuring
subroutines).

4} Use a PIC with (internal or external) EEPROM.  Test mode determines
calibration values and stores them in EEPROM, possibly using no external
interface.

5} Use a Flash reprogrammable PIC.  Program in the test code, make the
calibration measurements, customize the application program and burn
that in instead.  Advantage: Full utilization of program space.

 What other option have I forgotten?

 The Jig can, in all options but 1}, be either bed-of-nails or a ICP
header variant.

 Aside: Dataloggers with temperature and optional ADC channel must be a
popular PIC application.  One is in my surgery fridge at present, doing
a survey.  This has me thinking; *I*'d like to play with one.  It runs
from a lithium cell and is housed in - a film canister!  It uses SMD
components (of course) including Mchip PIC CPU and EEPROM.

 Anyone know of a source of these "raw" - as a kit preferably;
components and PCB but D-I-Y programming?  If not, might hassle Don with
a proposition, I'm sure everybody would love to play with one of these.

 Would they?
--
 Cheers,
       Paul B.

1999\05\05@095152 by Lawrence Lile

flavicon
face
oh HO!  Really?  I never thought of that!  I always assumed once an OTP was
written, the stone was chiseled forever!  HMMMM.  Perhaps I could get a
bunch of OTP's pre-programmed, with 00's in certain locations, stuff them,
solder them, power them up and do an initial simple functional test.

After passing the functional test (add 120V to the proper teminals and
listen for a beep signifying the chip has passed it's cleverly programmed
Power On Self Test)  have the chip do a measurement.  If I could read the
result of this measurement externally then I could adjust the caibration
values at that point.  Then write over the 00's....

Still seems pretty complex.


-----Original Message-----
From: Craig Lee <RemoveMEcleeTakeThisOuTspamATTCANADA.NET>
To: spamBeGonePICLISTspamBeGonespamMITVMA.MIT.EDU <TakeThisOuTPICLISTEraseMEspamspam_OUTMITVMA.MIT.EDU>
Date: Tuesday, May 04, 1999 3:09 PM
Subject: Re: In System Calibration





>Remember that 00s in an OTP can be written over.
>
>Craig
>

1999\05\05@095344 by Lawrence Lile

flavicon
face
Thanks for clearing that up, Andy!


-----Original Message-----
From: Andy Kunz <RemoveMEsupportspamTakeThisOuTMONTANADESIGN.COM>
To: PICLISTEraseMEspam.....MITVMA.MIT.EDU <EraseMEPICLISTspamMITVMA.MIT.EDU>
Date: Tuesday, May 04, 1999 4:17 PM
Subject: Re: In System Calibration


{Quote hidden}

1999\05\05@122753 by Craig Lee

flavicon
face
Correction:

Blank cells are actually 0xFF, not 0x00.  You might have missed
the correction.

To simplify:

I've never tried this, but you would set your cal block to 0xFFs,
and dump the program to OTP.  Then power it up, and if the cal block
is 0xFFs, your program would know it is in factory calibration
mode.  When your communication with the test jig is done, you can
fill the cal block with the cal factors with another programming cycle.

This way you get in-circuit measurements including the PIC variations.
You could also serialize it this way; placing a 'name' to your test data.

Craig

> {Original Message removed}

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...