diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Power Supplies (http://www.diyaudio.com/forums/power-supplies/)
-   -   PS Noise - Solutions and Suggestions (http://www.diyaudio.com/forums/power-supplies/227318-ps-noise-solutions-suggestions.html)

Hengy 8th January 2013 04:02 PM

PS Noise - Solutions and Suggestions
 
Hi everyone,

I am in the process of building a np100v12 headphone amp Link. Right now, most of it is on a breadboard - and works - except that there is a tiny bit of hf noise from the power supply.

I have included many diagrams and screen shots from my scope, in hopes that it aids in finding a solution!

I am using an older laptop smps power supply - 18.5V, 3A. This is, right now, followed by 3500uF of low-ESR (Nichicon HE) E-caps, and a 0.1uF ceramic. Next are two voltage regulators: a 7812 for the heater, and a LM317 for the rest of the circuit. Both of these are followed by 1uF low-ESR ceramics. Also, the LM317 has a 10uF E-cap on the adj pin.

This is the schematic of the filtering and regulation after the input from the smps:
http://matthengeveld.com/np100v12/PS_Schematic.png

This is the raw input and from the smps (top, yellow, 50mV/div), and the output from the LM317 (bottom, blue, 5mv/div). Only C5. Circuit is loaded with 150mA - about 1/4 of what the final circuit will use.
http://matthengeveld.com/np100v12/NoInCaps.bmp
Very noisy input, and still lots of HF noise on the output.

And this is the input and output with 3.5mF input caps and 0.1uF output caps. Both are now 5mV/div. Circuit is loaded with 150mA:
http://matthengeveld.com/np100v12/2mFCap.bmp
2mV/div
http://matthengeveld.com/np100v12/1uFOutCap.bmp
Much cleaner, but upon close up, still much too noisy for audio. It is the best I could get. Adding more caps, both on the input and output didn't help :(

Finally, after adding adding the heater on the 7812 output:
http://matthengeveld.com/np100v12/Heater_SMPSGround.bmp
Doesn't really add noise, but increases the frequency a little (?).

The HF noise doesn't surprise me, especially after reading this article.

At this point, I figure I need a LC filter after the smps. Trouble is, I know very little about inductors. For some reason, it was never even mentioned in my university courses :confused:

From what I've learned on the internet, I propose putting around 10-20uH in series, after the smps, and before the input caps. Would something like this inductor work? According to an online calculator, this gives a cutoff frequency of 220Hz, and about -20dB @1KHz.

Any suggestions would be much appreciated, especially on the design of the LC filter!

Thanks,
Hengy

knutn 8th January 2013 05:25 PM

If you read the data sheet for the 317 regulator, it says an RMS noise from the output typically 0.003% of the output voltage. So it does not help very much to filter the input. The answer is: use another regulator, or filter the output voltage from the 317 regulator.

Hengy 8th January 2013 05:47 PM

Thanks for your reply knutn,

Not sure if you took a look at this link: Using 3-pin regulators off-piste: part 2 It shows that smps switching noise can pass through a lm317 regulator! I am fairly new to p2p on perfboard, so the less components the better.

I'm pretty set on building a LC filter, and that's what I need help on. The long post was meant to show what I had already tried, and that I'm not just asking for help without trying anything myself. ;)

The above link suggests an inductor preceding the lm317, but the only detail it gives is that it should be "10s of uH". I think I need a little guidance before going into that unknown! (unless I can just randomly start plunking in inductors in series after the smps without ill effects?)

Thanks,
Hengy

CharlieLaub 8th January 2013 05:51 PM

You might try using an RC or CRC filter before the linear regulators. The spiky noise from the SMPS is probably too much for the reg to filter on its own. Caps alone are not useful for noise filtering - you need a low-pass filter, RC or LC, but with an LC filter you can inadvertently form high frequency resonances so try RC or CRC configuration first.

Hengy 8th January 2013 05:55 PM

Knew I shouldn't have done a long post! :mad: I know I need an LC filter, I just need help/suggestions on building one!

Hengy

tauro0221 8th January 2013 06:01 PM

Hi,
What about increasing the C6 to about 50uf or 100uf . Also remember that the PC power supplies are noisy due to they are switching power supplies. I would not use it for an amplifier. Why not use a transformer with a bridge?

Hengy 8th January 2013 08:01 PM

tauro0221,
I did try combinations of different sized ceramic and Ecaps, all low ESR types.

Here is 100uF and 470uF Ecaps, as well as an additional 10uF ceramic cap on the output, respectively:
http://matthengeveld.com/np100v12/1&100uF_PMInCap.bmp http://matthengeveld.com/np100v12/1&...owESRInCap.bmp http://matthengeveld.com/np100v12/1&...wESROutCap.bmp

No additional capacitance makes a difference.

I have never built a xformer-based power supply, and I don't think I want to start on this project.

Thanks,
Hengy

knutn 8th January 2013 09:04 PM

1 Attachment(s)
Quote:

Originally Posted by CharlieLaub (Post 3316652)
You might try using an RC or CRC filter before the linear regulators.

The size of the resistor R depends on how much current you are going to draw. The higher the value you can choose, the better the filtering.

tauro0221 8th January 2013 09:24 PM

Hi,
I missed the scope settings. Those are high frequency pulses and very low voltages. So you may need to design a filter to get rid off the pulses. Maybe somebody with expertise in simulation can come up with a filter base on the frequency pulses.

rikkitikkitavi 8th January 2013 10:36 PM

The regulators does nothing to remove this high frequency ripple, it is beyound their bandwidth by far.

The noise has mainly two different sources:
the ripple up and down in a triangular shape is due to the current flow through the SMPS inductors, this ripple current multplies with ESR in the output cap etc to give this ripple voltage. Thus we can learn that the SMPS operates at about 50 kHz (it is late and my eyes are tired...)

Second is spikes, that is due to leakage/stray inductance (nH is enough) in the high current switching path. There is probably some ringing in the tens of MHz due to parasitic capacitance/inductance here and there in it to.

First is best removed by an extra LC filter, say 10uH and 100uF.

Second is harder, because any stray capacitance couples this . Best is probably a very low selfinductance cap (SMD ceramic) and then adding ferrite beads to both the positive and ground lead of the output, causing a higher impendance at high frequency. Beads, beads, beads...


All times are GMT. The time now is 07:10 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2