Class AB with op-amp input, audible noise - diyAudio
Go Back   Home > Forums > Amplifiers > Headphone Systems

Headphone Systems Everything to do with Headphones

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 9th June 2013, 11:43 AM   #1
diyAudio Member
 
Join Date: Oct 2012
Default Class AB with op-amp input, audible noise

Hi. This is the third time I design this particular amplifier, each time I try to improve it, but this last one have an awful hissing sound (like the clean static you get from your radio). I have a LM317 regulator, 470uF and 470nF input caps as well as a 470nF close to the op-amp. Here's how it looks like:

http://i42.tinypic.com/iodq4o.jpg
http://i42.tinypic.com/mmxzmf.jpg

And here is the schematic:

http://i43.tinypic.com/33wlau9.png

I did an attempt to make a start grounding with wires going to a common point on the other side of the board. The PSU is some 24VDC **** from Deal Extreme, it outputs a 20mV 2500hz sawtooth wave so I was forced to use a regulator. Adding the regulator fixed this, but I'm left with this constant hissing noise The hissing noise is also about 20-30mVpp, but given that I have a gain of -10V/V, I can't really scope it on the input (voltage is too low). I tried adding 470nF caps in parallel with R5 and R11 to buffer/stabilize my voltage divider, but oddly enough this introduced a bunch of other noises. With these caps, the 2500hz noise is present again, I get this metallic noise (same as when you touch the input of an amp) and on top of this, the whole amp started acting like a microphone :O I made this amp yesterday also, but that one didn't have a regulator (it was also without class AB biasing), I handled the 2500hz noise by adding caps in parallel with R5 and R11 and it worked quite well without adding any other issues. The only other difference between the two amplifiers is that I moved the common point for the grounds a little bit (to make it a little bit neater).

Could anyone help me out a little bit? I'm frustrated that my attempts to improve only makes it worse
  Reply With Quote
Old 9th June 2013, 12:08 PM   #2
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
it sounds like it could be oscillating, the AD822, whats the input/gate capacitance on the fet you are driving with it? the datasheet says it will drive capacitive loads up to 350pf, many mosfets can easily exceed this. not sure its got the bandwidth to cope with that switching noise, with only 1.9MHz unity gain bandwidth and not particularly great settling time.its PSRR isnt amazing, but not too bad.

thats all i've got without measurements mate

adding somewhat randomly chosen bypass caps isnt likely to be terribly effective and could set up a resonance with the package inductance, that could be what you are hearing when it changes too.
  Reply With Quote
Old 9th June 2013, 12:20 PM   #3
diyAudio Member
 
Join Date: Oct 2012
I was apparently a little sloppy with the schematic. The op-amp driving the class AB stages is a NE5532, the op-amp below (the one you are referring to) is a LM393 comparator that is in control of the output relays. This part of the circuit seems to be operating as it should though.
  Reply With Quote
Old 9th June 2013, 06:15 PM   #4
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Congratulations, you have found out that high-K dielectric ceramic caps are microphonic. Still, if adding a cap to ground (use a plain ol' 10-22 'lytic, low-ESR preferred) introduces power supply noise, your grounding scheme must be crap. (I haven't looked at it.) These caps are absolutely necessary for half-decent PSRR in this single-supply circuit. Otherwise you have half the voltage regulator noise on the noninverting input, and as shown the reg is not low-noise at all. Add a ~1-2.2 F cap across R19 to get noise down. Also add a few hundred F across the reg output.

From what you write, your PSU is a SMPS. These can be very noisy. You probably want a healthy common-mode choke at the input, a series choke would also be a good idea. You can probably afford some series R, too - which would also be easier on the reg that now has to drop about 10 V if I'm not mistaken.

I see the actual amplifier uses an inverting topology. I guess you were not aware of the noise implications. What sort of headphones are you testing with?

For the circuit as shown with no caps across R5/R11, I get a total input-referred noise density of 14.7 nV/sqrt(Hz) for resistors alone, with the volume pot turned all the way down. At the same time, noise gain (which is the same as signal gain in a noninverting configuration, i.e. inverting gain + 1) never drops below 6 anywhere. That's 12.5 Vrms of noise, absolute minimum. Still, this should not be audible on headphones below about 118 dB SPL / 1 V sensitivity. My bets would still be on voltage reg noise. Or maybe that plus PSU noise not filtered by the regulator.

Now you know why fancy headphone amps use split supplies (usually linear unless switching is absolutely necessary) and noninverting amplifiers.

Last edited by sgrossklass; 9th June 2013 at 06:34 PM.
  Reply With Quote
Old 9th June 2013, 06:58 PM   #5
diyAudio Member
 
Join Date: Oct 2012
Thanks for the reply I have to re-etch the board to add more electrolytic capacitors, but I still haven't bought my regulator. As you can see, there is a TO-92 regulator soldered to the board, I'm going to buy a SMD one so I could get a low noise type (do you recommend one?). If the source of the noise is probably from the regulator or PSU noise getting through, it's perhaps worth testing another regulator before re-etching the board?

I'm testing with both a AKG K-701 and cheap steelseries headset. The noise is way more audible in the 32Ohm steelseries headset than my AKG K-701.

The reason I went for a single supply solution is that I have bad experiences with virtual grounds, I always get the single supply designs to work It also gives me an excuse to use output relays I would love to make my own linear dual-sided supply, but that would call for a far bigger enclosure etc. I'm putting the amplifier inside this box: DIY Aluminum Project Box Enclosure Case Electronic NEW_ Small 110x70x24mm
  Reply With Quote
Old 9th June 2013, 07:13 PM   #6
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Had a look at the layout now.

I saw you have resistor values of 330 ohms and 5k6 on the reg. That works out to 22.5 V. Are you sure 1.5 V of dropout is sufficient? It seems to be the bare minimum for a LM317 at best.

Do I get this correctly that you have sort of a ground "bus" with a thin track where all these jumper wires originate? Now that would fit the definition of a crappy grounding scheme pretty well. Plenty of shared ground return there. You should redesign this with a proper star (/tree) ground, including star ground at the reg (read up on this, here's one thread). For now, reinforce that track with a piece of wire.

There's not too much to be said against an LM317 in a low-noise configuration with proper layout. IIRC you can get output noise down to ~10 V, which tends to be fine for any circuit that has half-decent PSRR (and opamp-based headphone amps normally do).

Last edited by sgrossklass; 9th June 2013 at 07:22 PM.
  Reply With Quote
Old 9th June 2013, 07:47 PM   #7
diyAudio Member
 
Join Date: Oct 2012
I reinforced the ground trace with a wire and I added a 470nF cap across R19. It's hard to remember how loud the hissing actually was before I did it, but I'm pretty sure it's better (still not acceptable though).

The PSU is giving out 24.5V, the voltage drop across the regulator is 2.2V which whould be within operating conditions

I was a little unsure about the "ground bus" that I made, I was thinking that anything would be better than my previous designs so I prioritized aesthetics. A proper star ground would be to get the common point as small as possible, no traces between them? And also try to get this ground point as close to the regulator as possible?
  Reply With Quote
Old 9th June 2013, 10:47 PM   #8
diyAudio Member
 
Join Date: Oct 2012
Also, about the choke you mentioned. What will a choke bring to the table that a capacitor can't? The frequencies in question aren't that high.
  Reply With Quote
Old 14th June 2013, 11:40 AM   #9
diyAudio Member
 
Join Date: Oct 2012
I came up with an idea, what if R21 and R23 are replaced with 12V zener diodes, won't that make the bias voltage for the op-amp completely stable?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Leaving class A how to make this amp leave class A to AB when volume is increased? rhythmsandy Solid State 2 15th June 2013 07:02 AM
Contest: Linear Power Amp in a mint tin (class Aa, class AB, or class B) danielwritesbac Chip Amps 206 19th January 2013 08:52 PM
Class AB input coupling cap question Bengali Solid State 13 27th December 2012 04:59 PM
Self-biasing class AB input stage Mr Evil Solid State 56 25th April 2009 07:53 PM


New To Site? Need Help?

All times are GMT. The time now is 01:40 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