Headphone amplifier. Can't find the source of a weak audible humming

[Plecto]

I've made a few of these class A headphone amplifiers, but I got a weak humming in my latest design that I can't quite figure out the source of. I have a star grounding and a torodial transformer. The humming is constant, it does not depend upon the setting of the volume pot and it doesn't even depend upon gain! I have a button that switched between a gain of 5 and 15 (if I remember correctly), but pressing this button doesn't change the volume of the humming If this was a grounding issue, the humming should have changed with gain so I'm thinking that the noise is coupled directly to the output or something, but I'm not even sure if that's possible Here's the schematic and board layout:

I've highlighted the ground scheme. Mind that the component values in the schematic are NOT correct. I will upload a new schematic with the correct values if it becomes relevant. R22, R23 and R27 are 0Ohm jumper resistors. The op-amp is a TLE2142.

 

[barry]

The fact that the gain makes no difference would imply that the hum (I assume this is line frequency) is getting in somewhere AFTER the gain stage. Does the hum go away if you short the input? You might try decoupling the power supplies right at the output stage (it doesn't look like you've done this). Also, your schematic is a little difficult to understand. Some connections are shown with dots, yet others, which I ASSUME are connected (like R14 and pin 6 of IC1B) show no dot.

 

[FANT]

Because it does not depend on settings, for me the problem is concentrate on the + terminal of the IC'S.
Try to shield the components related on the + terminal and filter all the supplies that are connected on that terminal.

Consider that any noise on the pin is multiplied by the gain.

Mandi

 

[kam1787]

otherwise it works properly?
hum in both channels?

 

[Plecto]

The fact that the gain makes no difference would imply that the hum (I assume this is line frequency) is getting in somewhere AFTER the gain stage. Does the hum go away if you short the input? You might try decoupling the power supplies right at the output stage (it doesn't look like you've done this). Also, your schematic is a little difficult to understand. Some connections are shown with dots, yet others, which I ASSUME are connected (like R14 and pin 6 of IC1B) show no dot.

What do you mean by decoupling the supply after the output stage? I agree that the hum is getting in after the gain stage, but I'm having a hard time understanding how The input pot effectively shorts the input to ground, this has no effect on the hum what so ever.

otherwise it works properly?
hum in both channels?

The hum is to weak for me to actually measure it, but to me it seems equal in both channels. Other than this, the amp works like a charm.

 

[barry]

By decoupling I mean add capacitors from the power pins to ground--at the collectors of your output transistors.

 

[Plecto]

I could try that, but I want to understand what difference that could make. First off, how is the humming coupled to the output? Are we talking air-coupling because of short trace spacings? How will adding some capacitors make any difference? The ripple on the supply will be more or less unchanged by adding small ceramic caps.

 

[barry]

I could try that, but I want to understand what difference that could make. First off, how is the humming coupled to the output? Are we talking air-coupling because of short trace spacings? How will adding some capacitors make any difference? The ripple on the supply will be more or less unchanged by adding small ceramic caps.

So, add big aluminum caps.

 

[betwixt]

Can you clarify something please: are you using a center tapped transformer? It makes a big difference to the way ground currents will flow.
Personally, I would flood fill the grounds on the PCB or even make it double-sided with a ground plane.

Brian.

 

[Plecto]

So, add big aluminum caps.

I already have big aluminum caps. I know that reducing the ripple further by increasing the capacitance of my rectifying caps will decrease the hum, but it won't point to where the ripple gets into my headphones.

Can you clarify something please: are you using a center tapped transformer? It makes a big difference to the way ground currents will flow.
Personally, I would flood fill the grounds on the PCB or even make it double-sided with a ground plane.

I'm using a +-15V 30VA center tapped torodial transformer, yeah. My experience, ground planes has often lead to more humming. Depending on the pcb layout, the feedback return ground has joined the rectifying caps ground and continued down to the pot ground. Looking at this specific pcb, a ground plane might not cause further humming, but I don't see what difference it would make. Using a star ground with separate traces for each return ground will ensure that the supply ripple doesn't enter the amplifier, a ground plane can't always ensure this. Double sided boards are more costly and that would require me to drill a bunch of holes and make vias.

 

[FvM]

I'm using a +-15V 30VA center tapped torodial transformer, yeah.

Seriously speaking, I won't have guessed that from the schematic.

Can you fill a few more voids?
- What's the actual DC supply voltage (said "+/- 5V"), what's the peak-peak AC ripple?
- Value of power supply filter capacitors
- Output transistor bias current
- Impedance and sensitivity of the headphone that shows the hum

- - - Updated - - -

P.S.: Are you sure that the hum is not caused by capacitive or magnetical coupling to the circuit, in other words it's not affected by external shielding or moving the transformer?

 

[Plecto]

Can you fill a few more voids?
- What's the actual DC supply voltage (said "+/- 5V"), what's the peak-peak AC ripple?
- Value of power supply filter capacitors
- Output transistor bias current
- Impedance and sensitivity of the headphone that shows the hum

- - - Updated - - -

P.S.: Are you sure that the hum is not caused by capacitive or magnetical coupling to the circuit, in other words it's not affected by external shielding or moving the transformer?

DC supply is 43V. Ripple is 900mV
Value of the supply capacitors are 2x2200uF + 2x470nF ceramics.
Output transistor bias current is exactly 70mA for both channels (after it has heated to a stable temperature)
I'm testing with two different headsets. One has an impedance of 62Ohm, sensitivity 105dB/V. The other one is a cheap steelseries headset, but it doesn't say what model it is. It's clearly more sensitive as the hum is more audible.

Moving the transformer around makes no difference.

 

[Audioguru]

Your schematic does not show a center-tapped transformer and does not show the center-tap connected to ground. Is the center-tap of the transfortmer connected to the ground of the circuit?

What is the value of the big power supply filter capacitors? I use 1000uF/25V supply bypass capacitors on 9V battery powered circuits and the capacitors are fairly small.

The output transistors are powered and biased directly from the supplies that have some hum on them. The negative feedback through the opamps tries to reduce the hum but not perfectly. That is why most headphones amplifiers use voltage regulators or batteries.

 

[FvM]

I assume that the power supply rejection ratio and amplifier loop gain at 100 Hz isn't sufficient to completely suppress the 900 mV supply ripple. I've never seen a quality headphone amplifier without voltage regulators or transistor hum filter stages.

 

[Plecto]

Your schematic does not show a center-tapped transformer and does not show the center-tap connected to ground. Is the center-tap of the transfortmer connected to the ground of the circuit?

What is the value of the big power supply filter capacitors? I use 1000uF/25V supply bypass capacitors on 9V battery powered circuits and the capacitors are fairly small.

The output transistors are powered and biased directly from the supplies that have some hum on them. The negative feedback through the opamps tries to reduce the hum but not perfectly. That is why most headphones amplifiers use voltage regulators or batteries.

The center tap is the circuits signal ground, yeah. The value of the caps are 2200uF. The PSSR of the TLE2142 is 106dB TYP, the graphs in the datasheet shows about the same value at 100hz. 0.9V/(10^(100/20))=9uV. Am I calculating this correctly? I'm having a hard time believing that 9uV of 100hz humming can be audible

 

[FvM]

Yes, the auditory threshold at 100 Hz can be expected at some 100 µV with the said headphone sensitivity. But the hum is possibly not 100 Hz, the dominant component may be at higher 50 Hz harmonics. I'm not sure if you can clearly distinguish the frequency range, a skilled sound engineer of course would. If you have an audio oscillator, you can try to determine roughly an equivalent hum level and frequency at the headphone input.

There are also other possible circuit problems like missing bypass capacitors at the OP. It might be that the output stage is oscillating at ultrasound frequencies, resulting in a respective amplification of supply voltage ripple.

I believe that the problem cause can be localized by well considered tests, but hardly from a distance.

 

[Plecto]

I tried a tone generator to see if I could find the dominant frequency (whatever sounded the most like the humming from the amp). To me it seems like 50hz is the most dominant, 100hz sounds higher pitched than what I'm hearing.

Scoping the output of the amp shows no oscillations, I've never had that issue with this design.

I've made other amps like these in the past with less humming. The previous one I made is quite similar, the difference being a slightly different PCB layout, a standard iron core transformer instead of a torodial, and the use of a NE5532 instead of a TLE2142. This amp has hissing noise and humming when turning the volume up, but the source of this is clearly from the transformer as an unshielded input wire goes directly past the iron core transformer. The changes the PCB is that the gain setting has been moved to the front of the pcb and can be changed by a push-button instead of a DIP switch and the bass-boost functionality can be adjusted with a pot instead of a push-button. Everything that has to do with the rectifying caps, op-amp and output stage hasn't been changed at all and that includes every trace. I can't see how any of these changes could introduce hum.

 

[betwixt]

Time for some experments. The PCB connections are not labeled but I guess the two highlighted pads in the top left are where the transformer center tap connects to. The headphone ground is connected between there and the two big capacitors so there is some voltage drop directly in the headphone ground connection. Can you try removing the ground connections from those pads and wire them directly to the capacitor pads to see if that helps.

Brian.

 

[kam1787]

did you use shielded wire for the cable to push-button switch and pot? is the shielding grounded at one end?

 

[schmitt trigger]

One of those experiments would be, to power the circuit from batteries...don't laugh, many years ago I had a similar problem, and it turned out that the problem was being induced (radiated?) by electrical wiring in my work room...
The way to prove it was to feed the circuit from a pair of 12 volt batteries. The hum remained, and not until I completely turned off the mains breakers, did the problem disappear.