Voltage reference for A/D conversion

[Glenn Pure]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

I was thinking of using the 5 volt supply for the PIC (from a 7805)
but suspect it will drift over time (eg due to temperature variation)
and load. Does anyone have advice on this?

Ideally I want to measure to a resolution of 10 millivolts with
reasonable accuracy.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

 

[Peter Howard]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

LM336-2.5v precision voltage reference diode?
The NatSemi data sheet downloadable here
http://www.national.com/pf/LM/LM336-2.5.html#Application Notes
has a suggested design combining LM336 with a LM308 op-amp to create a 5v
reference. I'm not as smart as some here on PICs or electronics in general
so you'll have to decide if it will be accurate enough for you. However, I
have used LM336 (and Nationals design ideas) with success as voltage
reference for an automotive oil temp thermometer.

PH

 

[Franc Zabkar]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

Try the MAX6350 (5.0V) or MAX6325 (2.5V):
http://pdfserv.maxim-ic.com/en/ds/MAX6325-MAX6350.pdf

"The MAX6325/MAX6341/MAX6350 are low-noise, precision
voltage references with extremely low, 0.5ppm/°C
typical temperature coefficients and excellent, ±0.02%
initial accuracy. These devices feature buried-zener
technology for lowest noise performance. Load-regulation
specifications are guaranteed for source and sink
currents up to 15mA. Excellent line and load regulation
and low output impedance at high frequencies make
them ideal for high-resolution data-conversion systems
up to 16 bits."


- Franc Zabkar

 

[Adrian Jansen]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

I was thinking of using the 5 volt supply for the PIC (from a 7805)
but suspect it will drift over time (eg due to temperature variation)
and load. Does anyone have advice on this?

Ideally I want to measure to a resolution of 10 millivolts with
reasonable accuracy.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

We use Natsemi LP2951 for both supply and A/D ref. Very stable, low
dropout regulators. Excellent results.


--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.

 

[swanny]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

I was thinking of using the 5 volt supply for the PIC (from a 7805)
but suspect it will drift over time (eg due to temperature variation)
and load. Does anyone have advice on this?

Ideally I want to measure to a resolution of 10 millivolts with
reasonable accuracy.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

You don't mention your input range.
Do you need 5V input range or will 2.5V do.

You also don't mention the A/D resolution. Is it 8, 10 or 12 bits?

What kind of accuracy are you after, 2%, 1% 0.1% ?
Are you after absolute accuracy or relative accuracy of your measurement?

cheers,
Geoff

 

[David L. Jones]

I'm working out the best solution for a voltage reference to be used
with an A/D converter in a PIC microcontroller.

I was thinking of using the 5 volt supply for the PIC (from a 7805)
but suspect it will drift over time (eg due to temperature variation)
and load. Does anyone have advice on this?

Ideally I want to measure to a resolution of 10 millivolts with
reasonable accuracy.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

A REF02 5V reference works as a 0.2% precision regulator (8V to 40V
input) and can source around 10-20mA.
Available from several manufacturers and Farnell have it.
A great solution for PICs and ADCs which are referenced from the power
rails.
Not the cheapest solution though.

Regards
Dave

 

[David L. Jones]

We use Natsemi LP2951 for both supply and A/D ref. Very stable, low
dropout regulators. Excellent results.

They are only stable with a minimum value low ESR output cap,
otherwise they are completely unstable beasts and guaranteed to
oscillate.
Absolute accuracy ain't that great either, they are an LDO, not a
precision reference.

Regards
Dave

 

[Rob]

LM336-2.5v precision voltage reference diode?
The NatSemi data sheet downloadable here
http://www.national.com/pf/LM/LM336-2.5.html#Application Notes
has a suggested design combining LM336 with a LM308 op-amp to create a 5v
reference. I'm not as smart as some here on PICs or electronics in general
so you'll have to decide if it will be accurate enough for you. However, I
have used LM336 (and Nationals design ideas) with success as voltage
reference for an automotive oil temp thermometer.

PH

Sounds good PH, the LM336-2.5 with an op amp voltage follower for a 2.5V
ref, works well.
rob

 

[Glenn Pure]

Thanks all for advice.

I'm sure I have more than enough now to get a good solution.

The PIC has a 10 bit ADC. I'm measuring voltage across an LM335
temperature sensor so voltage range I will be measuring is about 2.5 V
to 3.3 V.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

 

[David L. Jones]

Thanks all for advice.

I'm sure I have more than enough now to get a good solution.

The PIC has a 10 bit ADC. I'm measuring voltage across an LM335
temperature sensor so voltage range I will be measuring is about 2.5 V
to 3.3 V.

Cheers
Glenn
Glenn Pure
Canberra, Australia
Web page: http://www.evans-pure.net

In that case a 0.2% 5V reference will give you a +/-1 degC absolute
error just due to the reference, so you'd be looking at a 0.2%
reference at a minimum I would suspect.
The LM335 is about +/-1 degC again, at best, calibrated. Uncalibrated,
it's a fair bit worse than that. You can do a lot better than the
LM335.
What accuracy are you after?

Dave

 

[swanny]

In that case a 0.2% 5V reference will give you a +/-1 degC absolute
error just due to the reference, so you'd be looking at a 0.2%
reference at a minimum I would suspect.
The LM335 is about +/-1 degC again, at best, calibrated. Uncalibrated,
it's a fair bit worse than that. You can do a lot better than the
LM335.
What accuracy are you after?

Dave

And since you are only interested in a tiny range (0.8V) from the output
of the temperature sensor, it would also pay to scale it up and level
shift it to, say, 0-4V (ie x5).

As Dave says, the untrimmed accuracy of teh LM335 and LM336 are pretty
rough. It will depend on your application as to what is best suited.

Geoff