Regulator spec sheets

[SeDeuced]

I've read many posts stating that adjustable 3-pin regs have lower noise than fixed value (78xx) regs. Looking at datasheets on modern parts, the difference seems to be small to non-existent for some manufacturers. Is the noise spec only considering a certain kind of noise?

For example, an LT LM317's RMS output noise spec is .001% of Vout while an ON Semi MC7818 is 10uV/Vout. At 18v, these are the same. Based on noise alone, it seems that there is no benefit of one over the other. Factor in ripple rejection and the story changes completely.

Compare those two to LT's LM3080 - the noise of the 3080 is much lower (40uV RMS to 100 kHz), but the ripple rejection is a mere 55 dB at 10 kHz (70dB for a LM317).

I'm replacing 78xx 3 pin regs in a guitar preamp that has audible white noise. The noise is coming from the power rails/regs. I'm thinking about building adjustable regs using 3080s, but the low ripple rejection concerns me, and the noise of the 317 is the same as the ON Semi's that I just put in it.

Any recommendations? There are .1uF and 22uF decoupling caps on the rails close to the preamps but not at the op amp Vcc pin. Would adding a .1 uF at each op amp Vcc pin help?

[SeDeuced]

Anyone?

[SY]

The noise in three pin adjustables can be greatly reduced by bypassing the set resistor and drawing sufficient current. See Errol Dietz's National Semiconductor note on 317 noise and impedance measurements- it was reprinted as an appendix to Bob Pease's "Troubleshooting" book. You can go somewhat quieter with discrete, but at a much higher cost; with some attention to design, you can definitely do better for noise with the adjustables.

[!]

I don't think your white noise has to do with the regulator chosen. How have you determined with certainty that it is the power supply subcircuit to blame?

How old is this equipment? What is the power source prior to the regulation IC?
Any chance you just have dried up electrolytic capacitors?

What does "There are .1uF and 22uF decoupling caps on the rails close to the preamps but not at the op amp Vcc pin" mean? Specifically "on the rails close to"? Yes, opamps perform best when there are ceramic or film decoupling caps within a few millimeters of their supply leads.

[SeDeuced]

Thanks for the replies.

The ckt. is from 1989. I changed all of the electrolytics first. Next, I upgraded the orig. 78xx regs. after reading about how noisy they could be. The ones I chose had much lower noise specs than the old ones. This made a noticeable difference in the hiss/white noise level.

At this point, I figured that I could possibly improve it more by getting a still lower noise reg.

The device is a crosspoint switcher that has 8 opamps on the input and 8 on the output. In both cases the op amps are lined up in a row and are fed by power buss' that run the length of the row. The ckt. layout has the gnd. buss running down the center of the row of 8 pin DIPs with +/-Vcc buss' running the down either side of the row of DIPs.

Note the positions of the two .1 uF ceramic decoupling caps in the photo. My thought was to add more .1 caps along that buss instead of just two for better decoupling. This would put a cap closer to each supply lead.

[Mooly]

I think your white noise is down to the actual audio and audio processing circuit design.

Any noise on the rails would have to be massive to find its way into opamp circuitry via the rails. Have you looked with a scope ?

Seeing a circuit might give some clues as to the gain the opamps are running and types and so on. Looking at the picture I'm guessing the large IC's are part of the audio processing too ?

[!]

Is that a row of mechanical controls in the picture? If so, I'd suspect them.

[SeDeuced]

The large ICs are the switching chips. The row of chips in the foreground are 5532 op amps for the input. Their gain is unity, as is that of the output op amps. The input Z of the input op amps is 1M, the input Z of the output op amps is 10K. The outputs are driving a 1K load resistor. All resistors are metal film.

The input op amps are running as inverting voltage followers at +/- 5 VDC. Their input comes stright from the input jack; their output goes through the switching chip to the output amps (identical configuration and gain) running at +/-18 VDC and on to the output jacks.

The strange power source config is because the switching chips in the middle of the ckt. are limited to operating on +/- 5 VDC.

I do have a scope, but it's old and not sensitive enough to show much. The supply rails have a small amount of ripple, and the trace itself appears slightly "unfocused" when compared to the normal trace. I take this to be the noise.

[Mooly]

Thanks for the diagrams...

You need some way to "prove" to yourself where the noise is coming from. I still find it hard to believe it's anything PSU related tbh.

A slight "blurring" of the scope trace may be a normal (if unwanted) result of grounds etc. If you short the probe tip and its ground lead the trace should be pin sharp. Now keeping the two shorted touch these to the ground on the working preamp. Does the trace thicken ?

I notice you have marked the diagram "replaced with 1 MΩ on the input opamps. That will degrade the noise figure considerably. The 20pf compensation would need looking at too but that's another issue...

[pierreohm]

I might be miss-reading:

-input gain of 10/39 where you replace with 1M resistor noise is increased as well as the noise gain
don't do it

-output gain is 39/10 you change it for 10/10 keeping 20pf constant, noise bandwidth is increased.
don't do it

-single ended design very sensitive to anything
live w/ it

-replace the power supply with batteries and listen/scope

-try to measure each outputs DC levels and compare differences

-change the op-amp to lm4562 ?

Good psrr is more important than low noise on the supply rail.

let us know