Maintaining Steady Voltage

I have an electronics project in which I have to maintain as stable a DC voltage as possible. The voltage has to be kept constant, with little to no ripple, at 9 volts. What is the best way of achieving this? How would using a 9 volt battery compare with using a DC power supply in how stable the voltage would be. Which would give better results?

It depends on whether you need a highly regulated voltage (exactly 9.000 volts) or low noise (9.0 volts with 0.00001 volt of noise).

Low resistance batteries (like NiMH or LiPoly) produce very low noise outputs if you don't need exactly 9 volts but you need low noise. If you need an exact output, you can use a low dropout regulator on one of these batteries to get a good compromise of both.

Typically, the voltage output will vary slightly by temperature and load current. Output noise from a low ESR battery is pretty low, but once you add a regulator you will get a more constant output voltage (across different loads and battery discharge states) but the LDO will also add noise.

A DC power supply can be designed to effectively eliminate ripple, but designing for very low output noise (

It would help if you specified more precisely what you want, how constant, what sort of stability, how much noise and ripple, over what sort of environmental conditions, what load current variation. Some idea of the application would help, since it could well define more precisely some of the values you need, and how much tolerance that there may be on then (maybe more than you believe), otherwise people like Dean cannot give you a better answer,
cheers,
Richard

"It depends on whether you need a highly regulated voltage (exactly 9.000 volts) or low noise (9.0 volts with 0.00001 volt of noise)."

The answer is that it doesn't have to be highly regulated, but it does have to be low noise. So for example, if I were to set the voltage to 9 V, but it actually was 8.932 V that would be fine, but I would want it to remain steady at 8.932 V despite varying the resistance of the load by a few kOhms or the temperature of the environment by ~10 degrees C. I would want the noise to be as small as possible, and most important of all, I wouldn't want the voltage to drift over time.

With regards to the load, its total resistance wouldn't always be the same, but would be somewhere around 100-400 kOhms and may increase by about 10 kOhms every day or so ( I would want the voltage across this load to remain the same after this resistance increase takes place). With regards to much of the other questions, I don't yet know the answer to them. All I know is that I will need the voltage to be very constant, but I don't yet know to what tolerance I'll need. Its good to here that batteries work for this since this is a cheap option that I can just go an try out.

Hi there,

Then what you need is an LDO. [Low-dropout linear regulator].

-Very low noise, can be as low as 60 uVrms.
-Regulates well too.

-the drawback: low efficiency

I'm not sure of charge pumps and buck converters because they're noisier.

Regards,
Justin

From what you say, it does have to be highly regulated, but the absolute value is not as important. From your figures, you imply a stability of better than 1 part in 10,000, over 10 degrees C. Do you have a means of actually measuring this, what is the resolution, temperature coefficient and drift on the meter you have. Some battery types have significant temperature coefficients, so unless you can hold the temperature constant(or you can compensate for it), they will not meet your needs. You need to look up the curves of voltage versus temperature for the various battery types. Load is high, so self heating is not an issue. I think an Alkaline cell is around 1mV/4 degrees C, so you would get away with it if the temperature doesn't shift more than that temperature range, you need to check that out though.
There are regulators and references with noise down at the uV level, as Dean said, good references may have temp coefficients

Hi there,

I was just wondering, are there references that regulate well
under a significant load?
If Mr. Don were to use a reference then it may not regulate well when
he loads the 100-400 kohms or so. Or maybe I'm not informed of
any well-regulating references (sorry, I'm still quite inexperienced).
I once tested the load regulation of a bandgap circuit and it didn't provide
a good curve (maybe because of the current limitation of the test pin/port).

It is really nice to learning a lot of things here in the forums, specially those
lessons that can only be learned through experience.

Regards,
Justin

a Load of 100K to 400K was specified. The figure for deltaR was around 10K/day increasing.
A poor reference will have perhaps 50ppm/mA variation with load, a good quality one