[Norm Dresner]

I need a very low-noise reference voltage, something around .5v or so for
input to a very high input impedance (FET) comparator. While checking and
changing the AA batteries in a phone today I found that 2 of the three were
discharged to around .8v -.9v Surely this is about as low-noise as it gets
and a simple few-megohm pot could be adjusted to provide the required
voltage if the battery voltage remained stable.

How constant would the output voltage of a partially discharged Alkaline
battery remain with a, say, microampere load over a few months?

Thanks in advance to all who answer for even considering this crackpot idea
worthy of comment.

Norm

 

[Fred]

I need a very low-noise reference voltage, something around .5v or so for
input to a very high input impedance (FET) comparator. While checking and
changing the AA batteries in a phone today I found that 2 of the three were
discharged to around .8v -.9v Surely this is about as low-noise as it gets
and a simple few-megohm pot could be adjusted to provide the required
voltage if the battery voltage remained stable.

How constant would the output voltage of a partially discharged Alkaline
battery remain with a, say, microampere load over a few months?

Thanks in advance to all who answer for even considering this crackpot idea
worthy of comment.

Norm

Won't a few megohms resistor produce quite a lot of noise?

What sort of bandwidth are talking of here?

 

[Reg Edwards]

Not very good performance versus time with 1 micro-amp. Not much better
with zero current. Changes occur for no apparent reason at all but due to
uncertain internal chemical action.

There's considerable variation between cells of different types, even
between cells of the same type of different ages.

Try discharging the cell with 1 milliamp for a few hours, crudely equivalent
to 1 micro-amp for a few thousand hours.

But such things as variation with temperature will have a FAR greater
short-term effect.

After a short somewhat heavier load the voltage never recovers to its
previous value.

An ordinary zener at low current, shunted with a capacitor, will be far
better.

Before zeners I think there were mercury-based cells used off-load for cheap
long-term voltage stability.

Then there were expensive Standard Weston Cells which were often housed in
temperature controlled ovens and lasted for many years.

 

[Spehro Pefhany]

Then there were expensive Standard Weston Cells which were often housed in
temperature controlled ovens and lasted for many years.

They can last decades if treated with a bit of care. I had one in a
"portable potentiometer" for calibration of thermocouple instruments.
You only drew current from it (and then only uA) when zeroing the
mechanical galvo.

But they are hermetically sealed in glass...

Best regards,
Spehro Pefhany

 

[John Larkin]

They can last decades if treated with a bit of care. I had one in a
"portable potentiometer" for calibration of thermocouple instruments.
You only drew current from it (and then only uA) when zeroing the
mechanical galvo.

But they are hermetically sealed in glass...

Best regards,
Spehro Pefhany

The Weston cell TCs were actually pretty bad. Lots of instruments used
to use mercury cells as a kinda-stable reference; at least the hum was
very low. A handheld Fluke is now a lot better than a tabletop full of
Wheatstone bridges and volt boxes and and galvos once were.

The Tektronix 7S14 dual-channel 1 GHz sampling plugin used an
avalanche transistor to generate the sampling pulse (no SRD!) and four
mercury batteries to back-bias the two sampling bridges. They tended
to die in about 10 years, and are the very devil to try to replace.

John

 

[Norm Dresner]

Won't a few megohms resistor produce quite a lot of noise?

What sort of bandwidth are talking of here?

Other answers have pretty much convinced me of the futility of the
concept, but I was assuming that a capacitor would be used to filter out at
least the higher frequency shot noise.
Norm

 

[CFoley1064]

Subject: I should probably be fired for even asking this question BUT ...

From: "Norm Dresner" [email protected]
Date: 6/17/2004 11:07 AM Central Standard Time
Message-id: <[email protected]>

I need a very low-noise reference voltage, something around .5v or so for
input to a very high input impedance (FET) comparator. While checking and
changing the AA batteries in a phone today I found that 2 of the three were
discharged to around .8v -.9v Surely this is about as low-noise as it gets
and a simple few-megohm pot could be adjusted to provide the required
voltage if the battery voltage remained stable.

How constant would the output voltage of a partially discharged Alkaline
battery remain with a, say, microampere load over a few months?

Thanks in advance to all who answer for even considering this crackpot idea
worthy of comment.

Norm

Hi, Norm. Discharged batteries get flaky -- they're not reliable. Apart from
noise issues, their output voltage seems to suffer from a step-type function
when in deep discharge -- it goes down pretty quickly, then remains stable for
a while, sometimes but not always a long while. I think a lot of that has to
do with the chemistry being affected by temp changes.

If anyone else hasn't suggested it, you might want to try the easy thing and
use two fresh AA batteries with an LM385BXZ-1.2 and a 110K series resistor.
You'll be burning 15 uA for the shunt reference to go along with your 1 meg pot
and small cap. That shouldn't affect battery life significantly, and you won't
have to worry about noise introduced by the voltage reference. You'll have
temperature stability and noise figures much better than your pot.

Good luck
Chris

 

[Norm Dresner]

Hi, Norm. Discharged batteries get flaky -- they're not reliable. Apart from
noise issues, their output voltage seems to suffer from a step-type function
when in deep discharge -- it goes down pretty quickly, then remains stable for
a while, sometimes but not always a long while. I think a lot of that has to
do with the chemistry being affected by temp changes.

If anyone else hasn't suggested it, you might want to try the easy thing and
use two fresh AA batteries with an LM385BXZ-1.2 and a 110K series resistor.
You'll be burning 15 uA for the shunt reference to go along with your 1 meg pot
and small cap. That shouldn't affect battery life significantly, and you won't
have to worry about noise introduced by the voltage reference. You'll have
temperature stability and noise figures much better than your pot.

Yeah, that's a good scheme too. I guess I was just being a little too
"cheap" and trying to use what other people would term garbage to save a few
bucks. Thanks for the suggestion.
Norm

 

[Tim Shoppa]

I need a very low-noise reference voltage, something around .5v or so for
input to a very high input impedance (FET) comparator. While checking and
changing the AA batteries in a phone today I found that 2 of the three were
discharged to around .8v -.9v Surely this is about as low-noise as it gets
and a simple few-megohm pot could be adjusted to provide the required
voltage if the battery voltage remained stable.

I fail to see how this is low noise (with a few megohm pot in there,
you've now got a lot of resistor noise, even if the battery were somehow
perfect!)

Instability is a kind of noise, I suspect that maybe you mean that
instead of the traditional thermal noise.

How constant would the output voltage of a partially discharged Alkaline
battery remain with a, say, microampere load over a few months?

1 Microamp * 24 hr/day * 100 days = 2.4 mA*hr. That's small
compared to the rating of a fresh AA battery (more than 1000mAH). The
sweet spot would depend on cell chemistry, and you probably don't want a
completely fresh battery or a nearly empty battery - something in between
would be best, where the discharge curve is nice and flat.

I suspect that temperature variations will be the limiting factor... cell
voltage dependence on temperature depends on chemistry and a bazillion other
things, but generally several hundred to several thousand ppm/deg C unless
you have a mercury cell (and even these aren't good compared to a cheap
regulator IC).

But why not just get one of those little adjustable micropower shunt
regulators, and a couple of high-precision fixed resistors? Has to be
cheaper than a pot.

Tim.

 

[Norm Dresner]

I fail to see how this is low noise (with a few megohm pot in there,
you've now got a lot of resistor noise, even if the battery were somehow
perfect!)

Instability is a kind of noise, I suspect that maybe you mean that
instead of the traditional thermal noise.


1 Microamp * 24 hr/day * 100 days = 2.4 mA*hr. That's small
compared to the rating of a fresh AA battery (more than 1000mAH). The
sweet spot would depend on cell chemistry, and you probably don't want a
completely fresh battery or a nearly empty battery - something in between
would be best, where the discharge curve is nice and flat.

I suspect that temperature variations will be the limiting factor... cell
voltage dependence on temperature depends on chemistry and a bazillion other
things, but generally several hundred to several thousand ppm/deg C unless
you have a mercury cell (and even these aren't good compared to a cheap
regulator IC).

But why not just get one of those little adjustable micropower shunt
regulators, and a couple of high-precision fixed resistors? Has to be
cheaper than a pot.

Cheaper than a pot and a thrown-away-battery??? I doubt it.

But I've already been convinced that it was a hairbraned idea based on
the instability.

Norm

 

[Tim Shoppa]

Cheaper than a pot and a thrown-away-battery??? I doubt it.

You can get a pot and a battery and a battery holder for 30 cents? That's
how much a shunt regulator and two resistors cost. Never mind the additional
assembly cost for wiring up the battery holder to the PCB, the screw to hold
the battery holder down, the cost of aligning the pot, the cost when the
pot gets bumped and shifts value by 5% and the unit gets returned, etc.

Tim.