Noise issue for headphone amplifier

[intuition]

Hi everyone, I'm a designer now on headphone amplifier. I've already made it a chip, but the psd noise level is as high as -100 dbv, there's my design spec:

dc gain : 130db
phase margin : 60 deg
GBW : 30MHz
structure : folded-cascode -- class-AB -- output buffer
stage gain : 70db -- 50db -- 10db
output load : 220uF or 16 ohm

And my issue is, the headphone amplifier noise level is usually required -120dbv or lower, but I just have -100dbv. The environment is clear: vdd, vcm, gnd, and signal noise level are all about -140dbv, it's obvious something wrong with op itself.

So here's some suspects I have, but I can't make sure what cause the op noise issue, could anyone give me some advice or reference about this topic?

1. Noise issue could be caused by the op non-stability, is that right?
2. My class-AB stage is operating at slightly cut-off for reducing power, could this cause op unstable and introduces noise ?
3. I've heard something like "the higher the gain, the worse the noise.", so the higher the dc gain, the more noise occurs on op, is that right?

appreciate for every reply, thanks! 8)

 

[scottieman]

Hi everyone, I'm a designer now on headphone amplifier. I've already made it a chip, but the psd noise level is as high as -100 dbv, there's my design spec:

dc gain : 130db
phase margin : 60 deg
GBW : 30MHz
structure : folded-cascode -- class-AB -- output buffer
stage gain : 70db -- 50db -- 10db
output load : 220uF or 16 ohm

And my issue is, the headphone amplifier noise level is usually required -120dbv or lower, but I just have -100dbv. The environment is clear: vdd, vcm, gnd, and signal noise level are all about -140dbv, it's obvious something wrong with op itself.

So here's some suspects I have, but I can't make sure what cause the op noise issue, could anyone give me some advice or reference about this topic?

1. Noise issue could be caused by the op non-stability, is that right?
2. My class-AB stage is operating at slightly cut-off for reducing power, could this cause op unstable and introduces noise ?
3. I've heard something like "the higher the gain, the worse the noise.", so the higher the dc gain, the more noise occurs on op, is that right?

appreciate for every reply, thanks! 8)

Would ike to ask questions first, before any suggestion:

1) how you use your designed amplifier to be a audio amp (e.g. in open-loop amp or in cloased loop with resistors and capacitors)

2) if your amplifier is used in opne-loop.. will your 30Mhz GBW a little bit too-large ... since audio mainly in the band up to 20 something kilo-Hz... the "too-much" bandwidth actually gives you more noise to the output

Scottie

 

[intuition]

Hi scottie

I'm using closed-loop with (negatively) resistive feedback to amplify input signal by R2/R1, and the noise level is also amplified with (1+R2/R1), since the noise been modeled as input-refered noise..

So if R2/R1=1, the noise level=-100dbv..it's too large for me. :evil:

 

[sunking]

what the value of the R1 and R2? it maybe too much!

 

[intuition]

I have 90k/90k for R2/R1.

 

[Borber]

How did you measured noise level? Was amplifier input terminated? What bandwidth you used for noise measurement?

 

[intuition]

Hi borber, I use fft plot machine to test the output of headphone amplifier. The noise level is plotted in power-spectrum-density, and because of audio application, the frequency interval is focused on 0~22kHz.

 

[Borber]

Have you tried different terminations on input of amplifier. For example: opened input, shorted input and 47kohm.

 

[intuition]

The closed input with 90k/90k feedback resistors has -100dbv noise level at output, and opened input has 6dbv improvement, say -106dbv noise level at output, it's obvious the noise is created by op itself. If I increase the closed-loop gain=8, say 160k/20k, I will also have (1+160k/20k)/2 times noise level.

I think to reduce resistance from 90k to 45k may reduce noise a little bit, but the resistors have already inside the chip.

So my point is, the noise seems to be created by op, and what cause the noise in op? thermal noise? Generally the thermal noise is about the level of -150dbv.. so I guess the stability of op may be the main issue for noise.

So..has anybody has experience between op topology and noise level?.. or what should I consider if I want to design an op with -130dbv~-140dbv noise level ??