Searching \ for 'high res adc or similar' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: techref.massmind.org/techref/io/atod.htm?key=adc
Search entire site for: 'high res adc or similar'.

Truncated match.
PICList Thread
'high res adc or similar'
1997\03\07@085822 by Peter Grey

picon face
I wish to use a pic to determine a signal which has a resolution of 10
microvolts in 200 millivolts. It also has a requirement for any component to
be in a very small package (ie surface mount and preferably in an 8-pin
SOIC). I can get a 16 bit ADC but the range is usually that of the supply
rail to ground. If I was to look at 10 microvolts in 5 volts then a 20 bit
ADC would be needed. As you can see I would need serial I/O for the digital
side. The ideal solution would be the LTC1298 but with a resolution of 20 bits.

Any ideas would be appreciated.



TIA,

Peter Grey
Neosystems
Australia

1997\03\07@092322 by Andy Kunz

flavicon
face
At 12:57 AM 3/8/97 +1100, you wrote:
>I wish to use a pic to determine a signal which has a resolution of 10
>microvolts in 200 millivolts. It also has a requirement for any component to
>be in a very small package (ie surface mount and preferably in an 8-pin
>SOIC). I can get a 16 bit ADC but the range is usually that of the supply
>rail to ground. If I was to look at 10 microvolts in 5 volts then a 20 bit
>ADC would be needed. As you can see I would need serial I/O for the digital
>side. The ideal solution would be the LTC1298 but with a resolution of 20
bits.

Peter,

I regularly get 12-16 bits of precision out of an 8-bit (PIC 71 or 73/74)
doing the following:

Take your samples are _very_ regular intervals, while applying a high then
low signal (at 50% duty cycle) such that the signal is pulled high or low
by 1-bit of resolution (1 bit of 8).

The values used for the resistor R1 and cap C1 are determined by the drive
of your analog signal, its level, and the amplitude of the square wave you
are providing.

(The little "o" is the junction of the components).

Wire this up and look at it with your scope (use a 1x probe).


               +----R1------< square wave source
               |
               |
               |
Analog In>----o------------> input to A/D converter
               |
               |
               |
               C1
               |
               |
               GND

Andy

==================================================================
Andy Kunz - Montana Design - 409 S 6th St - Phillipsburg, NJ 08865
         Hardware & Software for Industry & R/C Hobbies
       "Go fast, turn right, and keep the wet side down!"
==================================================================

1997\03\07@110027 by Wolfram Liebchen

flavicon
face
At 00:57 08.03.97 +1100, you wrote:
>I wish to use a pic to determine a signal which has a resolution of 10
>microvolts in 200 millivolts. It also has a requirement for any component to

Perhaps, it would be easier, to have corrections on the analog
side of the system.
Do the two signals (10 microV  vs.  200 mV) have different frequencies? Try
to filter them with an appropriate high-pass / low-pass including
amplification.
Then an 8-bit ADC may be sufficient.

regards,

Wolfram


+-----------------------------------------------------+
| Wolfram Liebchen                                    |
| Forschungsinstitut fŸr Optik, TŸbingen, Deutschland |
| spam_OUTliebchenTakeThisOuTspamffo.fgan.de                         |
+-----------------------------------------------------+

1997\03\07@112819 by Gerald Morrison

picon face
OK, I'm really intrigued but I admit to not understanding this.  I am
currently in a prototype stage using a PIC17C42 and an external ADC.  I
would really like to get more resolution.  Is there any chance of a more
detailed explanation?  What I don't understand is how the square wave
increases the resolution when the ADC (an 8-bit for example) can only
report 256 decision levels.  Thanks!


Regards,


Gerald Morrison
Product Development Engineer

Smart Technologies Inc.                Direct: 403-228-8557
#600, 1177 11th Avenue S.W.            Fax: 403-245-0366
Calgary, AB, Canada  T2R 1K9           email: .....geraldKILLspamspam@spam@cal.trlabs.ca
http://www.smarttech.com


> I regularly get 12-16 bits of precision out of an 8-bit (PIC 71 or 73/74)
> doing the following:
>
> Take your samples are _very_ regular intervals, while applying a high
then
> low signal (at 50% duty cycle) such that the signal is pulled high or low
> by 1-bit of resolution (1 bit of 8).
>
> The values used for the resistor R1 and cap C1 are determined by the
drive
> of your analog signal, its level, and the amplitude of the square wave
you
{Quote hidden}

1997\03\07@112830 by mbonner

flavicon
face
Peter Grey wrote:
>
> I wish to use a pic to determine a signal which has a resolution of 10
> microvolts in 200 millivolts. It also has a requirement for any component to
> be in a very small package (ie surface mount and preferably in an 8-pin
> SOIC). I can get a 16 bit ADC but the range is usually that of the supply
> rail to ground. If I was to look at 10 microvolts in 5 volts then a 20 bit
> ADC would be needed. As you can see I would need serial I/O for the digital
> side. The ideal solution would be the LTC1298 but with a resolution of 20
bits.
>
> Any ideas would be appreciated.
>
> TIA,
>
> Peter Grey
> Neosystems
> Australia

Try checking out Crystal Semiconductor's delta-sigma ADCs.
       http://www.crystal.com/
I have successfully interfaced them to PIC54s and 74s using modified I2C
routines.  They have 5V only devices with both 16 and 20 bit resolutions
and a variety of serial interfaces.  I'm having difficultly getting
delivery on military-grade parts but commercial/industrial parts should
not be a problem.

Matt

1997\03\07@154424 by Andy Kunz

flavicon
face
At 09:17 AM 3/7/97 -0700, you wrote:
>OK, I'm really intrigued but I admit to not understanding this.  I am
>currently in a prototype stage using a PIC17C42 and an external ADC.  I
>would really like to get more resolution.  Is there any chance of a more
>detailed explanation?  What I don't understand is how the square wave
>increases the resolution when the ADC (an 8-bit for example) can only
>report 256 decision levels.  Thanks!

Oh, yeah, sorry.

Take the sum of the samples.  For example, 16 samples will give you 4 extra
bits, 32 samples 5 bits, etc.

I usually have to throw away 2-3 bits, resulting in 4-8 times the number of
samples, but that's with a cheapo 8-bit A/D in these chips, with little if
any filtering of the analog line.  If you're using a better A/D and/or an
analog filter, you might to better.

Sorry for only giving you the hardware side.

Andy

==================================================================
Andy Kunz - Montana Design - 409 S 6th St - Phillipsburg, NJ 08865
         Hardware & Software for Industry & R/C Hobbies
       "Go fast, turn right, and keep the wet side down!"
==================================================================

1997\03\07@191339 by Peter Grey

picon face
At 09:21 AM 7/03/97 -0500, you wrote:
{Quote hidden}

Andy,

Thanks for this and I will certainly try this improved resolution. But with
the 16C71/74 I understand that the range input is that of the supply (5
volts) and so to get resolutions of 10 microvolts in 5 volts I need 20 bits.
If I could scale the range to read 0 to 200 millivolts then 16 bits is
enough. Alternatively I need 20 bits on an ADC.

This ADC/PIC/etc needs to be able to fit in a tube(with a pH probe,
conductivity probe and temperature sensor) which has external dimensions of
12mm diameter and 300mm long. This is why I require something small.

Tahnks again.


Peter
Neosystems

1997\03\07@191743 by Peter Grey

picon face
At 03:59 PM 7/03/97 +0100, you wrote:
{Quote hidden}

Wolfram,

Thanks for the suggestion. I am not sure what you mean. However, I would
like to have as few components as possible as I am trying to fit a pH probe,
conductivity probe and temperature sensor along with electronics, cable, etc
in a tube 12mm diameter and 300mm long.

The reason for the low resolution is the conductivity in a high pH environment.

Thanks again,

Peter
Neosystems

1997\03\07@222818 by Andy Kunz

flavicon
face
>Thanks for this and I will certainly try this improved resolution. But with
>the 16C71/74 I understand that the range input is that of the supply (5
>volts) and so to get resolutions of 10 microvolts in 5 volts I need 20 bits.
>If I could scale the range to read 0 to 200 millivolts then 16 bits is
>enough. Alternatively I need 20 bits on an ADC.

USE THE Vref PIN TO BRING YOUR REFERENCE DOWN!!!  You can go to 2.5V
(Vcc/2, actually).  I've found best results with Vref < 3.0V.

>This ADC/PIC/etc needs to be able to fit in a tube(with a pH probe,
>conductivity probe and temperature sensor) which has external dimensions of
>12mm diameter and 300mm long. This is why I require something small.

Cool.  My brother was working as a waste treatment plant chemist for a
while, and I did a model of their treatment system.  We had pH,
conductivity, temp, and all kinds of other sensors in the system, and a PC
to control it all.  He ended up with a neat gadget for measuring certain
pollutants in a handheld size, but never marketed it.

Andy

==================================================================
Andy Kunz - Montana Design - 409 S 6th St - Phillipsburg, NJ 08865
         Hardware & Software for Industry & R/C Hobbies
       "Go fast, turn right, and keep the wet side down!"
==================================================================

1997\03\08@012429 by John Payson

picon face
> OK, I'm really intrigued but I admit to not understanding this.  I am
> currently in a prototype stage using a PIC17C42 and an external ADC.  I
> would really like to get more resolution.  Is there any chance of a more
> detailed explanation?  What I don't understand is how the square wave
> increases the resolution when the ADC (an 8-bit for example) can only
> report 256 decision levels.  Thanks!

One approach I've used for 12 bit a/d conversion with about 3 counts' jitter
or 16 bit with about 16 counts' jitter is to use a comparator along with a
resistor and capacitor: take the output of a port pin, feed it through the
resistor to one leg of the cap.  Take the other leg of the cap and ground it.
Feed the first leg of the cap into one side of the comparator, and feed the
signal input into the other side.  The output of the comparator should feed
a second port pin on the PIC.  The resistor and cap should be chosen so that
the RC is fairly long (if you're using a 20KHz interrupt and you're trying
for 12 bit accuracy, an RC of a second or so is probably good).

Assuming the cap is on the inverting side of the comparator, your software
should every 50us or so (preferably faster, but the rate must be consistent)
copy the input from the comparator to the output pin.  In every 4096 inter-
rupts, count how many times you had to output a high and how many times you
output a low.  This count will be proportional to the input voltage on a
scale of 0 volts to Vdd.

While this A/D approach is neither as precise nor as fast as a true 12-bit
ADC, it is much cheaper and from experimentation it seems to work quite
well.  One warning, however: while the approach is less succeptible to stray
electrons or glitches than the integrator approach used in the MTA14000,
it will not work well with the built-in comparators on /Windowed/ 16C62x
parts at least not if the RC has a big R and small C.  Shining light on the
window causes the ADC reading to drift all over the place.  Since I've not
tried using OTP '620s for this (I just used an external comparator which saved
a port pin anyway) I don't know how well they'd work.

1997\03\08@044830 by Peter Grey

picon face
At 10:04 PM 7/03/97 -0500, you wrote:
>>Thanks for this and I will certainly try this improved resolution. But with
>>the 16C71/74 I understand that the range input is that of the supply (5
>>volts) and so to get resolutions of 10 microvolts in 5 volts I need 20 bits.
>>If I could scale the range to read 0 to 200 millivolts then 16 bits is
>>enough. Alternatively I need 20 bits on an ADC.
>
>USE THE Vref PIN TO BRING YOUR REFERENCE DOWN!!!  You can go to 2.5V
>(Vcc/2, actually).  I've found best results with Vref < 3.0V.
>
>>This ADC/PIC/etc needs to be able to fit in a tube(with a pH probe,
>>conductivity probe and temperature sensor) which has external dimensions of
>>12mm diameter and 300mm long. This is why I require something small.
>
>Cool.  My brother was working as a waste treatment plant chemist for a
>while, and I did a model of their treatment system.  We had pH,
>conductivity, temp, and all kinds of other sensors in the system, and a PC
>to control it all.  He ended up with a neat gadget for measuring certain
>pollutants in a handheld size, but never marketed it.

It may not be too late. I think it is a pity that these very useful devices
sit idle on the shelf and they are re-invented every so often. He may like
to take a stab at marketing it. What does it do?

Thanks for the VREF solution. I missed this a sI was not considering the
16c71/74 as they are too larger for my application. I would like one of the
new PIC12C5XX series with the ADC built in but understand they will not be
available until sometime in April/May. I still may be able to use these.

Thanks again,


Peter
Neosystems

More... (looser matching)
- Last day of these posts
- In 1997 , 1998 only
- Today
- New search...