Post by thread:[PIC]: PIC14000 ADC Problems

I posted on PIC14000 ADC problem sometime back. At that time I had problem of getting ADC to work. That was solved. At that time few had specifically told me not to use the part. But I am stuck with it. Due to reasons beyond my control I have to use it. Now the problem is fluctuations in ADC count. At the time I was using general purpose board and had only 6 count variation. Now after making PCB I get 40 count variation. I tried everything hardware wise I could think of. But no success. And I have connected 1.5Volt battery as input to make sure of pure DC input, but still more than 40 count variations occur. I tried averaging 10 readings but no success. To me it seems, this is 50Hz pickup problem. In other ADCs normally by selecting ADC clock in multiple of 50Hz, this problem is avoided. That solution is not available here. External 16bit ADC is also not a solution, as this project is very price sensitive. Is it at all possible to get steady count from PIC14000? If yes, how? Vyas -- http://www.piclist.com hint: To leave the PICList spam_OUTpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu

Hi I designed with PIC14000 four years ago. The A/D is crap. Really forget 16 bit resolution, i guess 14 is tops. Now i don´t remember the details, but it sits with very stable supply, good caps on inputs and reference of course. It is also worth experimenting with integrating capacitor and AD clock speed. Remember to use a very good capacitor for integrating (i.e polycarbonate, polypropylene) IIRC i used 22nF Naturally, place capacitors close to the PIC - also the integrating cap, so no noise is picked up. And think about layout caused interfernece of any nearby power. Also try put all the rest of the PIC to sleep, wake up on ADC done. (minimizes internal noise much) If you have some of the every first produced like i had, the A/D clock selector don´t work (a never documented bug, byt fixed early) I still had to use heavy filter, "home made" idea, that stores the values into an array, while discarding the most positive and negative values, then calculates the mean value of the middle ones. The idea is that all noise peak induced values are trashed, then averaging the good samples. (i had switching cirquitry nearby) Search for MOMA filter in PIC archives november 1998. Regards /Morgan Hej Vyas. Tack för ditt meddelande 11:44 2002-10-03 enligt nedan: {Quote hidden}>I posted on PIC14000 ADC problem sometime back. At that time I had problem >of getting ADC to work. That was solved. At that time few had specifically >told me not to use the part. But I am stuck with it. Due to reasons beyond >my control I have to use it. Now the problem is fluctuations in ADC count. >At the time I was using general purpose board and had only 6 count >variation. > >Now after making PCB I get 40 count variation. I tried everything hardware >wise I could think of. But no success. And I have connected 1.5Volt battery >as input to make sure of pure DC input, but still more than 40 count >variations occur. I tried averaging 10 readings but no success. > >To me it seems, this is 50Hz pickup problem. In other ADCs normally by >selecting ADC clock in multiple of 50Hz, this problem is avoided. That >solution is not available here. > >External 16bit ADC is also not a solution, as this project is very price >sensitive. > >Is it at all possible to get steady count from PIC14000? If yes, how? > >Vyas > >-- >http://www.piclist.com hint: To leave the PICList >.....piclist-unsubscribe-requestKILLspam@spam@mitvma.mit.edu -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

Hi Morgan, Sorry for delayed response. I am using 0.1uF polypropylene cap, soldered right on the pins. Clock is internal 4Mhz. I tried using Sleep mode, but it not does not wake up. Somehow the capture interrupt is not working. I had this problem from beginning. So I am using simple wait till !OVFIF. So sleep is out. You have mentioned about AD Clock speed selection, I think you mean AD current selection, because I can't find anything in data about selecting clock speed. The AD cap selection and AD current selection go hand in hand. I will have go with MOMA filter. Whole purpose of using PIC14000 was for 16 bit adc, if I can't get 16 bit resolution, I will have to trash the project. Thanks. Vyas >Hi >I designed with PIC14000 four years ago. >The A/D is crap. Really forget 16 bit resolution, i guess 14 is tops. >Now i don4t remember the details, but it sits with very stable supply, good caps on inputs and reference of course. >It is also worth experimenting with integrating capacitor and AD clock speed. >Remember to use a very good capacitor for integrating (i.e polycarbonate, polypropylene) > >IIRC i used 22nF > >Naturally, place capacitors close to the PIC - also the integrating cap, so no noise is picked up. And think about >layout caused interfernece of any nearby power. > >Also try put all the rest of the PIC to sleep, wake up on ADC done. (minimizes internal noise much) > >If you have some of the every first produced like i had, the A/D clock selector don4t work (a never documented bug, >byt fixed early) > >I still had to use heavy filter, "home made" idea, that stores the values into an array, while discarding the most >positive and negative values, then calculates the mean value of the middle ones. The idea is that all noise peak >induced values are trashed, then averaging the good samples. (i had switching cirquitry nearby) Search for MOMA >filter in PIC archives november 1998. > >Regards >/Morgan -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

LSB of a 16 bit value is 1/2^16, or 1/65536, or aprox 0.0000153, or 0.00153 %, or 15 ppm. You'll probably need *very* advanced analog curcuit to reach that kind of resolution !! Note that a typical Polypropylen cap (from Farnell) have a temp coeff of 200 ppm/deg C. You can get a Polystyrene cap with down to +- 60 ppm/deg C. Still much higher then the LSB of a 16 bit value. Even 10 bits of resolution gives a LSB that is aprox 0.098 % of full scale. Even that is hard to get in a repeateble way. May I ask, *what* needs a 16 bit A/D value ? Jan-Erik Svderholm. >Whole purpose of using PIC14000 was for 16 bit adc, if I can't get 16 bit >resolution, I will have to trash the project. > >Thanks. >Vyas Jan-Erik Svderholm S:t Anna Data tel : +46 121 42161 mob : +46 70 5241690 -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email .....listservKILLspam.....mitvma.mit.edu with SET PICList DIGEST in the body

At 07:07 PM 10/7/02 +0200, you wrote: >May I ask, *what* needs a 16 bit A/D value ? Weighing systems, for one, especially when dollars (euros, dinars whatever) are involved. Also if you need wide dynamic range (eg. hi-fi audio). The PIC14000 was a first cut attempt, if you want high resolution, use a delta-sigma converter with a microcontroller or use one of those other micros that has an on board converter (two types, both of them 8051 core). You can get a 24 bit converter. Microchip will eventually put these converters on-chip, but maybe not for a year or two, in the meantime, if you want to use a PIC use an external. Microchip sells them, also LTC and others. Best regards, Spehro Pefhany --"it's the network..." "The Journey is the reward" EraseMEspeffspam_OUTTakeThisOuTinterlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservspam_OUTmitvma.mit.edu with SET PICList DIGEST in the body

Hej Vyas. Tack för ditt meddelande 15:37 2002-10-07 enligt nedan: >Hi Morgan, > >Sorry for delayed response. >I am using 0.1uF polypropylene cap, soldered right on the pins. Seem OK for 4MHz. >Clock is internal 4Mhz. Stable freq is mandatory for slope converters, thus Xtal is better! >I tried using Sleep mode, but it not does not wake up. Somehow the capture >interrupt is not working. I had this problem from beginning. So I am using >simple wait till !OVFIF. So sleep is out. That should be good enough. I dont remember how i did. >You have mentioned about AD Clock speed selection, I think you mean AD >current selection, because I can't find anything in data about selecting >clock speed. Ah, yes you are right. You can only select internal or external osc. (That switch was locked in one position in early silicon, BTW) I got better results with external Xtal osc. I used max possible 20Mhz for max sample rate, then filtered to increase precision... or more correctly reduce noise.. i guess precision is not state of the art... > The AD cap selection and AD current selection go hand in hand. >I will have go with MOMA filter. > >Whole purpose of using PIC14000 was for 16 bit adc, if I can't get 16 bit >resolution, I will have to trash the project. Unless you really want your nose bleeding, connect every PIC14000 you have to mains voltage directly. Then pop some for me too, hehe. My nose still bleeds... The only solution is a nice delta sigma, like LTC2400. 24 bit, or faster with less bits. Even does 50/60Hz rejection at appropriate clock speed. Beware of early silicon there too, there was a bug in readout in one of it´s modes, a particular timing, and of course I used that combination... Also avoid MAX110/111 and some other MAX (internal test lockup bugs). Whats up with AD converter quality anyway, every A/D i have used have shot me in my back!! AAARGH! (For bug details search PIC archive on LTC2400 and MAX110 i posted a couple years ago) /Morgan -- http://www.piclist.com hint: To leave the PICList @spam@piclist-unsubscribe-requestKILLspammitvma.mit.edu

>Even 10 bits of resolution gives a LSB that is aprox 0.098 % of >full scale. Even that is hard to get in a repeateble way. > >May I ask, *what* needs a 16 bit A/D value ? > >Jan-Erik Svderholm. Thanks for your input. But if it was so difficult to get repeatable result in 10bit ADC, why would there be 24 bit ADC? made? I have used 4andhalf digit ADCs (7135) and it has no problems of repeatability. 7109 which is a 12 bit ADC also has no problems. > >May I ask, *what* needs a 16 bit A/D value ? > > Weighing systems, for one, especially when dollars (euros, dinars whatever) > are involved. Also if you need wide dynamic range (eg. hi-fi audio). > > The PIC14000 was a first cut attempt, if you want high resolution, use > a delta-sigma converter with a microcontroller or use one of those other > micros that has an on board converter (two types, both of them 8051 core). > You can get a 24 bit converter. >Spehro Pefhany --"it's the network..." "The Journey is the reward" May I ask which one? I have used 8051 core micros, but haven't come across one with 24 bit adc. (Though I must confess, I haven't searched for it). I was told to use PIC14000, so I didnot look beyond it. Regards, Vyas -- http://www.piclist.com hint: To leave the PICList KILLspampiclist-unsubscribe-requestKILLspammitvma.mit.edu

Vyas, >May I ask which one? I have used 8051 core micros, but haven't come across >one with 24 bit adc. >(Though I must confess, I haven't searched for it). I was told to use >PIC14000, so I didnot look beyond it. Check into either TI, MAXIM or Analog Devices. The Analog devices parts are called "MicroConverter" and I would recommend them. We've had pretty good luck with them as well as the support from Analog Devices was great! Best of luck! Steve Steven Kosmerchock Radio Frequency Systems Phoenix, Arizona USA (WORK) http://www.rfsworld.com http://www.geocities.com/researchtriangle/lab/6584 "Great spirits have always encountered violent oppposition from mediocre minds."--A.Einstein -- http://www.piclist.com hint: To leave the PICList spamBeGonepiclist-unsubscribe-requestspamBeGonemitvma.mit.edu

Are the 50 cycle mains radiation causing noise on your board? When the power goes off through out the entire city, does the count become steady? How about a zero crossing detector for the radiated 50 cycles? Some over amplified comparator or transistor hooked to a input pin? Then start conversion after zero crossing so that the noise signal is same each time? It seems like 40 counts may be a very small amount of noise, could some component be light sensitive and picking up the strobing effect from florescent lighting? Bill ----- Original Message ----- From: "Vyas" <TakeThisOuTvyasEraseMEspam_OUTSJVYAS.COM> To: <RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU> Sent: Thursday, October 03, 2002 2:44 AM Subject: [PIC]: PIC14000 ADC Problems snip..... > Now after making PCB I get 40 count variation. I tried everything hardware -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads