EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: innkeeper on August 09, 2017, 07:28:39 pm
-
I'm in the middle of upgrading an old audio processor / mic preamp circa 1995. One of the issues i am running into is the overall noise level on the +/- 15v rails.
i'm seeing 40mV of hf noise on both rails.
Part of my upgrades to this will be replacing parts in the audio path that are known to add noise to the audio with updated quieter ones, but i cant imagine the hf noise on the power not adding noise along the way.
the regulators are lm78m15/lm79m05 and are bypassed currently at the output with .047uf cap and a 33uf cap.
generally these regulators should have an output noise in the 400uv range.
each audio processor chip and opamps have .047uf caps between the voltage rail and ground (there is a lot of chips in the audio path, approx 8 so noise effects are cumulative).
Still i am seeing 40mV of HF noise (not ripple) throughout the board. i can see variations in the noise as it is processing audio, but that might just be load related. The noise is there absent of any input, Even directly at the output of the regulators.
Any thoughts on what i might try or check to get that down??
i do plan on cleaning up the PS filter section using some low ESR caps, and higher values. but doubtful that is going to effect the HF noise.
-
I'd check for oscillation of the LM79m05 regulator. 79xx series regulators always took more careful decoupling for stability than the 78xx positive ones. Dig up the datasheet and check against recommended capacitor values. Oscillation of the -ve regulator would probably couple onto the +ve rail too. Check the recommended capacitor types too, the 79xx series don't like too low a capacitor esr.
The other thing that comes to mind is checking with the inputs shorted, not just disconnected in case it is RF pickup.
Just a couple of semi-random thoughts.
-
The other thing that comes to mind is checking with the inputs shorted, not just disconnected in case it is RF pickup.
Exactly, short the oscilloscope probe tip to the point where ground is attached.
What exactly are you using to measure the noise? The regulator specification is over a specific bandwidth.
You might need to invest in a differential probe to get reliable noise measurements.
A well designed audio circuit has considerable power supply rejection ratio so noise from the power supply is rejected.
-
You might need to invest in a differential probe to get reliable noise measurements
Now why didn't i think of that.....feeling a bit stupid....
So, i set up my scope using the 2 channels as a poor mans differential probe.
i read 1.7mV on one rail and 2mV on the other for the hf noise. So a lot lower.
this was measured on a 500uv/div scope.
Definitely operator error on those measurements I feel a lot better about those numbers over the 40mv i saw before.
The other thing that comes to mind is checking with the inputs shorted, not just disconnected in case it is RF pickup.
Though i did monitor the power with signals running though it, the posted numbers though were both with the input shorted so id not add to the noise noise :)
i didn't mention that on the original post.
Thanks for setting me straight.