IMPLEMENTED SO FAR

- Support for 4x20 LCD Display and large number display
- Brightness and contrast adjustment with remote
- (OPUS/Wolfson WM8741) DAC volume control: remote and rotary encoder
- (OPUS/Wolfson WM8741) DAC random filter selection 1 to 5 with remote
- (OPUS/Wolfson WM8741) DAC upsampling selection (L, M, H -this is the OSR setting)
- I2C level shifting (5V to 3.3V)
- Optimized power-up sequence

Monday, March 15, 2010

Comparing Noise Figures in Linear Regulators

Just like phase noise in clocks, it is difficult to compare noise values among linear regulators because there is no common ground in specifying noise figures. Some companies report noise density, others RMS V noise, and yet others % of Vout. The frequency range for the reported noise figures also varies from company to company.

The LT1763 familiy is a favorite for audio projects because it has low noise figures. Among linear regulators it is probably universally preferred by audio diy aficionados.

In order to compare other regulators to this benchmark device, I decided to calculate the Vrms noise in each of the fequency ranges provided by the chart. The Vrms noise is basically the product of the noise density times the frequency delta.

The results is shown in the graph below. Total Vrms = 23.2 uVrms for the frequency range 10Hz to 100KHz. This approximation is very close to the specified value of 20 uVrms.

I then looked as the specified noise of several common regulators and matched the values to the corresponding value of the LT1763 device.

A couple of observations: The LM340 is actually very good, in fact better than the beloved LM317 according to spec. The LM723 seems of lower noise than the LT1763, at least in the reported frequency range.





We can approximate the total noise Vrms for the 10Hz-100KHz interval by noticing that each frequency range contributes a percentage of the total noise. In the case of the LT1763, we notice that the 10Hz-10KHz range contributes about half of the total 10Hz-100KHz noise. The table below compares all the regulators in the 10Hz-100KHz range.

Here is a very good paper from TI explaining noise in linear regulators: [link]
According to the paper,
"The dominant source of noise in an LDO is usually the
bandgap. In most cases this is solved by adding a large
low-pass filter (LPF) to the bandgap output so that none
of the noise makes it into the gain stage."
Unfortunately, it is not easy to access the bandgap output line in most integrated regulators...

5 comments:

Russ said...

While low noise is certainly desirable, the point of diminishing returns is hit pretty quickly in digital circuits once you get into the 10s of uV your in great shape for most of today's digital ICs.

I only use the LT1763 for digital supply pins of DACs and the Crystek clock. Some would argue you need the lowest possible noise for clocks. And this is generally true. The reason the LT1763 is a good fit for the clock we use is because that clock in essence has a form of local regulation and decoupling internally. It is in fact a small circuit board. They need it that way because of the way they generate their output signal at low phase noise. :) I won't say much more than that. But suffice it to say it is a very well designed clock. if I were using a lesser clock I might choose a lower noise supply, but would anyone notice? Who knows.

It is key to look for line an load regulation (output impedance) when evaluating voltage regulators for digital use. Also, proper bypassing in digital circuits cannot be ignored, and it is far more critical than the choice between any two good voltage regulators.

As you can see no small amount of thought goes into choosing these things. And the measurements and listening tests bear this out.

Cheers!
Russ

The Lazy Engineer said...

Russ, appreciate the comments. Just today I was looking at this report: http://pstca.com/spice/ripple/ripple.htm and made me think how important it is simulation and test in designing a good regulator, and that low noise is not the only criteria.

Thomas & Betts said...

get good results from these products.

Pawel said...

Hello I think youre noise calculations aren't right you can't just add the values of integral noise just as you did. The proper formula is Vnoise=sqrt(Vn1^2+Vn2^2)

Best Regards
Pawel

Kuii said...

What about Micrel LDOs?