Quietest voltage for testing

[Conrad Hoffman]

Forgetting long or even short term stability, what's the quietest voltage one could come up with for testing meter noise levels? Should be stable for a few minutes or so. I'm thinking a battery of some sort, but even those generate a certain amount of noise, depending on the type. It's got to be low impedance, lest one have Johnson noise. A pure reactance doesn't generate Johnson noise. A battery and some kind of LC filter? A charged super capacitor? I'd like something between 1 and 10 volts. Other ideas?

[CopperCone]

Battery noise is something I was always interested in.

Not much information about it. Some information relating to primary cells (i.e. cadmium mercury ones) does include noise discussion.,. but those are designed for voltage repeatability.

how about a peltier module with a negative delta (i.e. ice cube on one side, compressor on other side)... probobly high 1/f noise right?


Most things other then batteries make me thin of thermal phenomena.

Would a supercap be noisy since its granular?

a giant foil capacitor?
rectified RF? (again noisy because of semiconductor material.. unless some kind of a special diode?)

photovoltaics?

photoelectric current (i,e metal plate and UV light in a vacuum)

[Nusa]

This thread is basically a discussion of the same question.
https://www.eevblog.com/forum/projects/ultra-low-noise-voltage-reference/
Summary: Nicad battery chemistry had the lowest noise.

[alm]

If the voltage does not matter, then you can make a fairly low noise 0V reference . Granted, is not really suitable to quantify all aspects of stability like that of the reference and part of the ADC.

I would expect a good cap to be better than a battery, since discharging a cap is purely based on moving electrons around (very small quants and fairly homogenous thanks to the large numbers), while discharging a battery is based on a chemical reaction that may generate heat (affecting the reaction). I imagine for best performance you would need something very pure like a capacitor with air as dielectric, but that would be impractically large. Assuming a 10¹⁰ Ohm load, the capacitor would have to be in the order of 1 nF (and would not be particularly stable, but that was not the question).

So in addition to supercaps, I would also consider some high-quality film caps in whichever size is reasonably easy / affordable to acquire. Ceramic (except maybe NP0) is out due to microphonics. Sure, the voltage will drop exponentially during your measurements unless RC is many orders of magnitude above your measurement time, but this may be something you can correct in post-analysis if the meter represents a reasonably well-behaved load. It could be helpful to put a large-value resistor (~1 GOhm) in parallel to the meter input to make the input more resistive if the input impedance turns out to have a significant voltage coefficient.

[CopperCone]

This thread is basically a discussion of the same question.
https://www.eevblog.com/forum/projects/ultra-low-noise-voltage-reference/
Summary: Nicad battery chemistry had the lowest noise.

not really, this thread asks it in a different manner.

What if there is some crazy process that yeilds a low noise voltage for 2 seconds only?

YOu would not think about it if he asked about a voltage reference. This is like a antinoise reference. A quiet reference.

or perhaps a transient quiet phenomena

[CopperCone]

If the voltage does not matter, then you can make a fairly low noise 0V reference . Granted, is not really suitable to quantify all aspects of stability like that of the reference and part of the ADC.

I would expect a good cap to be better than a battery, since discharging a cap is purely based on moving electrons around (very small quants and fairly homogenous thanks to the large numbers), while discharging a battery is based on a chemical reaction that may generate heat (affecting the reaction). I imagine for best performance you would need something very pure like a capacitor with air as dielectric, but that would be impractically large. Assuming a 10¹⁰ Ohm load, the capacitor would have to be in the order of 1 nF (and would not be particularly stable, but that was not the question).

So in addition to supercaps, I would also consider some high-quality film caps in whichever size is reasonably easy / affordable to acquire. Ceramic (except maybe NP0) is out due to microphonics. Sure, the voltage will drop exponentially during your measurements unless RC is many orders of magnitude above your measurement time, but this may be something you can correct in post-analysis if the meter represents a reasonably well-behaved load. It could be helpful to put a large-value resistor (~1 GOhm) in parallel to the meter input to make the input more resistive if the input impedance turns out to have a significant voltage coefficient.

how about photo electric effect then? i.e. uv light on metal

[alm]

Funny that you pick the one experiment that was used to show the discrete nature of light. Conrad's next post : How do I make a very low noise voltage reference that can source enough current to power a UV light.

Not sure if moving the problem around is going to make it any easier to solve. Never mind the noise that will likely be in whatever light source you use.

Some other thoughts (which are probably not practical):
Johnson noise is proportional to output impedance. So if you increase the output voltage, you lower the relative Johnson noise (obviously will not work for series combinations like many battery cells). Of course now you need to figure out what to do with that high voltage. Resistive dividers are probably out .

What if you had a low(ish) noise voltage reference that is connected to a cap via a switch. Every say 10s, the switch would close for say 100 ms to charge the cap. This will likely cause a small step in voltage (due to voltage reference noise). The cap should be large enough to stay pretty much at the same voltage for that 10s. This can probably be a better quality cap than if it had to maintain the same voltage for minutes. If you sample at 100 NPLC, then you could get around 5 samples per step (obviously auto-zero disabled).

Repeat this a couple of hundred times, and calculate variance by averaging the variance calculated for each individual step over all steps. Because the expectation value of their means would be very close (on the level of the noise from the voltage reference, maybe 1 ppm?), you do not have to worry about standardizing the variances.This is probably one of those devil (charge injection has horns) is in the details problems, but it might work. Of course the relatively simple hardware and complicated analysis might not be everybody's cup of tea. And I am not sure if the short discharge time reduces of exacerbates the imperfections in the cap and the rest of the setup.

[zhtoor]

Hello,

how about a standard cell (weston) buffered by a low noise op-amp ?
or maybe just use a 2v lead-acid cell (high capacity).

regards.

[Alex Nikitin]

So if you increase the output voltage, you lower the relative Johnson noise (obviously will not work for series combinations like many battery cells).

Actually, if the batteries noise is uncorrelated (and that includes Johnson noise), connecting four batteries in series will half the relative noise level and so on. The temperature drift and other similar correlated influences won't improve though. 

Cheers

Alex

[EmmanuelFaure]

It tooks me 15 seconds with google :
https://www.mikrocontroller.net/attachment/100819/Measurement_of_Chemical_Battery_noise.pdf

[Vtile]

What about hall-effect created with strong magnetic field? I have no knowledge.. A big vacuum capacitor submerged to liquid nitrogen, hydrogen or dry ice submerged off the self solution. Edit. A lead shielded vacuum cap with a superconducting plates might be the ideal.

[Gyro]

It tooks me 15 seconds with google :
https://www.mikrocontroller.net/attachment/100819/Measurement_of_Chemical_Battery_noise.pdf

The paper seems to imply that the largest factor is Johnson noise related to internal resistance, NiCd being the lowest of the batteries tested. I notice that they didn't test any lead acid batteries.

Based on internal resistance I would have thought that a SLA, probably a cylindrical Cyclon cell (lower internal resistance than rectangular cells), might well be quieter still.

[Conrad Hoffman]

Hadn't seen that paper before- thanks! I'm thinking a stack of NiCd button cells (because I have a bunch) would get me up to a reasonable voltage. It sounds like additional filtering isn't really beneficial.

[Kleinstein]

With batteries, one has to be careful with mechanical movements. Mechanical force can change the internal setup by "damaging" surface films and just applying stress and this way change the voltage. Dave showed this with alkaline cells - it's quite strong there. I would expect the effect much smaller (as the voltage does not change much with discharging) with NiCd cells, but a similar effect could happen there too. Also some thermal insulation is likely a good idea, as the voltage changes with temperature.

Filtering is not helping with batteries.  For a small scale use, old Hg based cells might be better than NiCd, but they are not build anymore.

The old style weston cells are relatively high impedance and thus have some noise. There advantage is that the TC is rather low and the chemistry is reproducible even in 1000 years.

[MisterDiodes]

Conrad:  If you don't mind a very heavy boat anchor on the floor, an NiFe (Nickel Iron or Edison) battery  is a pretty quiet source.  They last forever, and have a fairly flat discharge curve compared to others.  Larger capacity  SLA batteries can work well also.

The hidden problem with using batteries as a measuring voltage excitation source - say for a radiometric resistance bridge - is the battery voltage is always a moving target.  At first you think that wouldn't matter..but...

On a battery-excited resistance bridge that means the excitation voltage is ever changing, and that means the power dissipation in the resistors is changing - without any guarantee that the resistances are changing exactly the same way.  Depending on what you're measuring this effect can cause a lot of problems.

For fairly short measurement periods a battery can certainly work - but it's no substitute for a real Vref. if you need accurate measures over time.

[Conrad Hoffman]

I've got some old Audel's books in my library of obsolete technology that describe the iron cells. Always wondered if anybody was using those anymore. Probably easy to construct.

What I want to do is look at short term noise (the auto-zero and dielectric absorption stuff in the SVT thread) on my HP3455A up near the top of the 1 or 10 volt range.  Current draw is near zero so a mechanically and thermally isolated battery should work OK.

[zhtoor]

I've got some old Audel's books in my library of obsolete technology that describe the iron cells. Always wondered if anybody was using those anymore. Probably easy to construct.

What I want to do is look at short term noise (the auto-zero and dielectric absorption stuff in the SVT thread) on my HP3455A up near the top of the 1 or 10 volt range.  Current draw is near zero so a mechanically and thermally isolated battery should work OK.

hello,

wet nickel iron batteries are still being manufactured, but the cells are *large*, you can get a 100AH 1.2V cell, the cost
is pretty steep also, ca USD 500+, electrolyte is a solution of Potassium Hydroxide (sometimes mixed with Lithium Hydroxide).

ps. nickel iron batteries were the *first* batteries to be made even before lead-acid batteries, they were called edison batteries,
they have an *extremely* long life.

regards.

[Nusa]

You can make your own canning-jar sized edison cells if you've got any kind of DIY ability. It's a classic science experiment.

[beanflying]

Bump.

Instead of starting a new thread on low noise power supplies this was about the best thread search turned up.

Running my current Voltage References on LiPo's but given they will stay in the shack 90%+ of there life it makes sense to look at a low noise supply for them but backed up by the battery on power failure or sending them on the road. I will be hooking up my newly acquired 34970's to look at some data over time including on charge/voltage float vs Mains.

With a couple of 399 ref's being made and waiting like the rest of the world for LTZ1000's most likely in temperature stabilised enclosures batteries will become fairly chunky at 5 or 6S 5000's.

I ran across this chip from TI and then the development board which I have on order from Mouser to try out rather than make a PCB with good thermals which has reasonable specs. (circa $20USD) http://www.ti.com/lit/ug/slvu741a/slvu741a.pdf

Anyone else care to chat about uber stable supplies

[beanflying]

Looks like an interesting read but to much maths for Midnight http://tf.boulder.nist.gov/general/pdf/1133.pdf