Loading ...
Sorry, an error occurred while loading the content.

Re: [jallist] Re: getting full AD resolution between 2V and 8V

Expand Messages
  • Vasile Surducan
    ... Stef, a few month ago I have dissapointed Eur s... :) Now you are dissapointing me ... :) [joke] The full scale resolution ( represented in bits) of any AD
    Message 1 of 30 , Jul 1, 2003
    • 0 Attachment
      On Mon, 30 Jun 2003, Stef wrote:

      >
      >
      > a3veyk wrote:
      >
      > >How about increasing the resolution by doing this:
      > >
      > >First measure the signal with vref+=Vdd and vref-=Vss
      > >determine if the value is > Vcc/2
      > >if this is true then set vref-= Vdd/2 and Vref+=Vdd
      > >else set vref-=Vss and Vref+= Vdd/2
      > >measure again, resolution should be increased by 1 bit now
      > >
      > >Unfortunately you can not select Vref- and leave Vref+ connected
      > >to Vdd internally.
      > >
      > Brings me to a much simpler way:
      > take 2 AD values, add them, and you've gained 1 bit of resolution !!
      >
      Stef, a few month ago I have dissapointed Eur's... :)
      Now you are dissapointing me ... :) [joke]
      The full scale resolution ( represented in bits) of any AD convertor
      can't be changed. It's fixed by the producer. It can be changed only the
      analogic resolution for the full scale. (Vinput/digital_resolution)
      Splitting the full scale range in n subdomains, keeping one AD just to
      monitorize the subdomains and changing the gain of the input OA will
      allow to measure on other channel, with a constant and maximum resolution,
      any input voltage. Things which is not possible with one simple AD input.
      But the resolution of the AD is the same 10 bits ( talking about PICs).
      Multiplying, adding, dividing, averaging N AD
      measurements will never increase the PIC resolution !

      Vasile
    • Stef UCI
      ... Vasile, what about Sum(Signal+Noise) ? Ok signal must be constant, noise must have equal distribtion, and maybe some other constraints, but it really works
      Message 2 of 30 , Jul 1, 2003
      • 0 Attachment
        Vasile Surducan wrote:

        >
        >
        >On Mon, 30 Jun 2003, Stef wrote:
        >
        >
        >
        >>a3veyk wrote:
        >>
        >>
        >>
        >>>How about increasing the resolution by doing this:
        >>>
        >>>First measure the signal with vref+=Vdd and vref-=Vss
        >>>determine if the value is > Vcc/2
        >>>if this is true then set vref-= Vdd/2 and Vref+=Vdd
        >>>else set vref-=Vss and Vref+= Vdd/2
        >>>measure again, resolution should be increased by 1 bit now
        >>>
        >>>Unfortunately you can not select Vref- and leave Vref+ connected
        >>>to Vdd internally.
        >>>
        >>>
        >>>
        >>Brings me to a much simpler way:
        >> take 2 AD values, add them, and you've gained 1 bit of resolution !!
        >>
        >>
        >>
        > Stef, a few month ago I have dissapointed Eur's... :)
        > Now you are dissapointing me ... :) [joke]
        > The full scale resolution ( represented in bits) of any AD convertor
        >can't be changed. It's fixed by the producer. It can be changed only the
        >analogic resolution for the full scale. (Vinput/digital_resolution)
        >Splitting the full scale range in n subdomains, keeping one AD just to
        >monitorize the subdomains and changing the gain of the input OA will
        >allow to measure on other channel, with a constant and maximum resolution,
        >any input voltage. Things which is not possible with one simple AD input.
        >But the resolution of the AD is the same 10 bits ( talking about PICs).
        >Multiplying, adding, dividing, averaging N AD
        >measurements will never increase the PIC resolution !
        >
        Vasile, what about
        Sum(Signal+Noise) ?
        Ok signal must be constant, noise must have equal distribtion, and maybe
        some other constraints,
        but it really works !!
        Stef

        >
        >Vasile
        >
        >
        >
        >
        >
        >Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
        >
        >
        >
        >
        >
      • Stef UCI
        ... I did, but that was 20 years ago, fortunatly physics hasn t change that much ;-) ... Vasile, don t let some simple silicium spoil your summer holiday, the
        Message 3 of 30 , Jul 1, 2003
        • 0 Attachment
          >>>
          >>>
          >>Vasile, what about
          >> Sum(Signal+Noise) ?
          >>Ok signal must be constant, noise must have equal distribtion, and maybe
          >>some other constraints,
          >>but it really works !!
          >>
          >>
          >
          > Have you tested it?
          >
          I did, but that was 20 years ago, fortunatly physics hasn't change that
          much ;-)

          > Tell me man, let me go happy in my summer holiday!
          >
          Vasile, don't let some simple silicium spoil your summer holiday,
          the silicium on the beach can you give you much more pleasure !!
          Stef

          >
          >Vasile
          >
          >
          >
          >
          >
          >Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
          >
          >
          >
          >
          >


          [Non-text portions of this message have been removed]
        • Vasile Surducan
          ... Have you tested it? Tell me man, let me go happy in my summer holiday! Vasile
          Message 4 of 30 , Jul 1, 2003
          • 0 Attachment
            On Tue, 1 Jul 2003, Stef UCI wrote:

            >
            >
            > Vasile Surducan wrote:
            >
            > >
            > >
            > >On Mon, 30 Jun 2003, Stef wrote:
            > >
            > >
            > >
            > >>a3veyk wrote:
            > >>
            > >>
            > >>
            > >>>How about increasing the resolution by doing this:
            > >>>
            > >>>First measure the signal with vref+=Vdd and vref-=Vss
            > >>>determine if the value is > Vcc/2
            > >>>if this is true then set vref-= Vdd/2 and Vref+=Vdd
            > >>>else set vref-=Vss and Vref+= Vdd/2
            > >>>measure again, resolution should be increased by 1 bit now
            > >>>
            > >>>Unfortunately you can not select Vref- and leave Vref+ connected
            > >>>to Vdd internally.
            > >>>
            > >>>
            > >>>
            > >>Brings me to a much simpler way:
            > >> take 2 AD values, add them, and you've gained 1 bit of resolution !!
            > >>
            > >>
            > >>
            > > Stef, a few month ago I have dissapointed Eur's... :)
            > > Now you are dissapointing me ... :) [joke]
            > > The full scale resolution ( represented in bits) of any AD convertor
            > >can't be changed. It's fixed by the producer. It can be changed only the
            > >analogic resolution for the full scale. (Vinput/digital_resolution)
            > >Splitting the full scale range in n subdomains, keeping one AD just to
            > >monitorize the subdomains and changing the gain of the input OA will
            > >allow to measure on other channel, with a constant and maximum resolution,
            > >any input voltage. Things which is not possible with one simple AD input.
            > >But the resolution of the AD is the same 10 bits ( talking about PICs).
            > >Multiplying, adding, dividing, averaging N AD
            > >measurements will never increase the PIC resolution !
            > >
            > Vasile, what about
            > Sum(Signal+Noise) ?
            > Ok signal must be constant, noise must have equal distribtion, and maybe
            > some other constraints,
            > but it really works !!

            Have you tested it? Tell me man, let me go happy in my summer holiday!

            Vasile
          • Ian Forse
            Hi Vasile/Stef Take a look at application note AN018 at this site www.cygnal.com/support/applications.htm Cygnal produce 8051 based products. I have been
            Message 5 of 30 , Jul 1, 2003
            • 0 Attachment
              Hi Vasile/Stef

              Take a look at application note AN018 at this site
              www.cygnal.com/support/applications.htm

              Cygnal produce 8051 based products. I have been temporarily corrupted at
              work! 8051's are a lot different to PIC's.

              I ran some of there demo software on a 10bit ADC (200Ksps). 16000 readings
              allows 16bit resolution - RS232 O/P every second. For the demo the ADC is
              connected to an onboard temperature sensor which normally only has 1C
              resolution, with over sampling the resolution is 0.01C - It even detected
              someone blowing gently on it from about 30cm away. I did not believe it
              until I saw it.

              I am sure an F877 would produce similar results

              Regards

              Ian

              > -----Original Message-----
              > From: Vasile Surducan [mailto:vasile@...-cj.ro]
              > Sent: 01 July 2003 09:19
              > To: jallist@yahoogroups.com
              > Subject: Re: [jallist] Re: getting full AD resolution between 2V and 8V
              >
              >
              >
              >
              >
              > On Tue, 1 Jul 2003, Stef UCI wrote:
              >
              > >
              > >
              > > Vasile Surducan wrote:
              > >
              > > >
              > > >
              > > >On Mon, 30 Jun 2003, Stef wrote:
              > > >
              > > >
              > > >
              > > >>a3veyk wrote:
              > > >>
              > > >>
              > > >>
              > > >>>How about increasing the resolution by doing this:
              > > >>>
              > > >>>First measure the signal with vref+=Vdd and vref-=Vss
              > > >>>determine if the value is > Vcc/2
              > > >>>if this is true then set vref-= Vdd/2 and Vref+=Vdd
              > > >>>else set vref-=Vss and Vref+= Vdd/2
              > > >>>measure again, resolution should be increased by 1 bit now
              > > >>>
              > > >>>Unfortunately you can not select Vref- and leave Vref+ connected
              > > >>>to Vdd internally.
              > > >>>
              > > >>>
              > > >>>
              > > >>Brings me to a much simpler way:
              > > >> take 2 AD values, add them, and you've gained 1 bit of resolution !!
              > > >>
              > > >>
              > > >>
              > > > Stef, a few month ago I have dissapointed Eur's... :)
              > > > Now you are dissapointing me ... :) [joke]
              > > > The full scale resolution ( represented in bits) of any AD convertor
              > > >can't be changed. It's fixed by the producer. It can be
              > changed only the
              > > >analogic resolution for the full scale. (Vinput/digital_resolution)
              > > >Splitting the full scale range in n subdomains, keeping one AD just to
              > > >monitorize the subdomains and changing the gain of the input OA will
              > > >allow to measure on other channel, with a constant and maximum
              > resolution,
              > > >any input voltage. Things which is not possible with one
              > simple AD input.
              > > >But the resolution of the AD is the same 10 bits ( talking about PICs).
              > > >Multiplying, adding, dividing, averaging N AD
              > > >measurements will never increase the PIC resolution !
              > > >
              > > Vasile, what about
              > > Sum(Signal+Noise) ?
              > > Ok signal must be constant, noise must have equal distribtion, and maybe
              > > some other constraints,
              > > but it really works !!
              >
              > Have you tested it? Tell me man, let me go happy in my summer holiday!
              >
              > Vasile
              >
              >
              >
              >
              >
              > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
              >
              >
              > ---
              > Incoming mail is certified Virus Free.
              > Checked by AVG anti-virus system (http://www.grisoft.com).
              > Version: 6.0.491 / Virus Database: 290 - Release Date: 18/06/03
              >
              ---
              Outgoing mail is certified Virus Free.
              Checked by AVG anti-virus system (http://www.grisoft.com).
              Version: 6.0.491 / Virus Database: 290 - Release Date: 18/06/03
            • Stef UCI
              hi Ian, ... that s exactly what I mean, thanks (also for the bootloader link). Which Cygnal products do you use ? They have very nice controlers, even with USB
              Message 6 of 30 , Jul 2, 2003
              • 0 Attachment
                hi Ian,

                Ian Forse wrote:

                >Hi Vasile/Stef
                >
                >Take a look at application note AN018 at this site
                >www.cygnal.com/support/applications.htm
                >

                that's exactly what I mean, thanks (also for the bootloader link).

                Which Cygnal products do you use ?
                They have very nice controlers, even with USB (2.0).
                Do you use the USB ports of them ?
                And if so, how is the driver at the PC side for the USB ?

                cheers,
                Stef

                >
                >
              • bvwelch
                Greetings Vasile, I am just getting started with the 12F675 A/D. I have 1 channel working, but I haven t checked accuracy yet. You mentioned other chips 87x
                Message 7 of 30 , Jul 3, 2003
                • 0 Attachment
                  Greetings Vasile,

                  I am just getting started with the 12F675 A/D. I have 1 channel
                  working, but I haven't checked accuracy yet. You mentioned other
                  chips "87x ..." Would you care to list specifically those chips that
                  you have found working good/better than the 12F675? I have several
                  other parts, but the small size of the 12F675 appealed to me. But I
                  hope I am not disappointed.

                  The approach I took was to operate the CPU with the 4 MHZ internal
                  clock, but use the SLEEP mode and FRC while doing A/D operation.

                  I did notice the LSB jumping around, but I thought it was due to my
                  poor breadboard, and I don't have a VREF hooked up yet. I was just
                  working on the software at this point, later I plan to build a better
                  board for the "real" project.

                  Thank you in advance for any suggestions.

                  bvwelch

                  --- In jallist@yahoogroups.com, Vasile Surducan <vasile@s...> wrote:
                  >
                  > Nice thought. Will not work exactly as you said (Vref-= Vss and
                  > Vref+=Vdd/2 as far I remember does not work with 10bit
                  resolution)...
                  > But I'm using a PIC without vref-. :) like PIC12F675.
                  > BTW, I've discovered much more
                  > instability and offset in 675 AD than in 87x AD.
                  > Have anyone got the same conclusions ?
                  >
                  > top 10 wishes,
                  > Vasile
                  > http://surducan.netfirms.com
                  >
                  >
                  > On Sun, 29 Jun 2003, a3veyk wrote:
                  >
                  > > How about increasing the resolution by doing this:
                  > >
                  > > First measure the signal with vref+=Vdd and vref-=Vss
                  > > determine if the value is > Vcc/2
                  > > if this is true then set vref-= Vdd/2 and Vref+=Vdd
                  > > else set vref-=Vss and Vref+= Vdd/2
                  > > measure again, resolution should be increased by 1 bit now
                  > >
                  > > Unfortunately you can not select Vref- and leave Vref+ connected
                  > > to Vdd internally.
                  > >
                  > >
                  > > > And what about 14 bits from just one 10 bit AD-converter ??
                  > > > Or will it be too slow ;-)
                  > > >
                  > > > Stef
                  > > >
                  > > > >
                  > > > >
                  > >
                  > >
                  > >
                  > >
                  > >
                  > > Your use of Yahoo! Groups is subject to
                  http://docs.yahoo.com/info/terms/
                  > >
                  > >
                • Vasile Surducan
                  Maybe is just a wrong ideea, but 16F877, 16F876 seems to have (for me) a better AD converter (without external reference, just VCC and GND) than 12F675 which
                  Message 8 of 30 , Jul 15, 2003
                  • 0 Attachment
                    Maybe is just a wrong ideea, but 16F877, 16F876 seems to have (for me)
                    a better AD converter (without external reference, just VCC and GND) than
                    12F675 which have in average an offset of about 3-5% from the CS.
                    Tested on three PICs, all gift from Wouter. The fourth one, I've killed
                    with pleasure.
                    I have ordered here some more, to see if the problem is just me or
                    Microchip.
                    Probably just me. Probably not.

                    On Thu, 3 Jul 2003, bvwelch wrote:

                    > Greetings Vasile,
                    >
                    > I am just getting started with the 12F675 A/D. I have 1 channel
                    > working, but I haven't checked accuracy yet. You mentioned other
                    > chips "87x ..." Would you care to list specifically those chips that
                    > you have found working good/better than the 12F675? I have several
                    > other parts, but the small size of the 12F675 appealed to me. But I
                    > hope I am not disappointed.
                    >
                    > The approach I took was to operate the CPU with the 4 MHZ internal
                    > clock, but use the SLEEP mode and FRC while doing A/D operation.
                    >
                    > I did notice the LSB jumping around, but I thought it was due to my
                    > poor breadboard, and I don't have a VREF hooked up yet. I was just
                    > working on the software at this point, later I plan to build a better
                    > board for the "real" project.
                    >
                    > Thank you in advance for any suggestions.
                    >
                    > bvwelch
                    >
                    > --- In jallist@yahoogroups.com, Vasile Surducan <vasile@s...> wrote:
                    > >
                    > > Nice thought. Will not work exactly as you said (Vref-= Vss and
                    > > Vref+=Vdd/2 as far I remember does not work with 10bit
                    > resolution)...
                    > > But I'm using a PIC without vref-. :) like PIC12F675.
                    > > BTW, I've discovered much more
                    > > instability and offset in 675 AD than in 87x AD.
                    > > Have anyone got the same conclusions ?
                    > >
                    > > top 10 wishes,
                    > > Vasile
                    > > http://surducan.netfirms.com
                    > >
                    > >
                    > > On Sun, 29 Jun 2003, a3veyk wrote:
                    > >
                    > > > How about increasing the resolution by doing this:
                    > > >
                    > > > First measure the signal with vref+=Vdd and vref-=Vss
                    > > > determine if the value is > Vcc/2
                    > > > if this is true then set vref-= Vdd/2 and Vref+=Vdd
                    > > > else set vref-=Vss and Vref+= Vdd/2
                    > > > measure again, resolution should be increased by 1 bit now
                    > > >
                    > > > Unfortunately you can not select Vref- and leave Vref+ connected
                    > > > to Vdd internally.
                    > > >
                    > > >
                    > > > > And what about 14 bits from just one 10 bit AD-converter ??
                    > > > > Or will it be too slow ;-)
                    > > > >
                    > > > > Stef
                    > > > >
                    > > > > >
                    > > > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > >
                    > > > Your use of Yahoo! Groups is subject to
                    > http://docs.yahoo.com/info/terms/
                    > > >
                    > > >
                    >
                    >
                    >
                    >
                    >
                    > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
                    >
                    >
                  Your message has been successfully submitted and would be delivered to recipients shortly.