I've made a headphone amplifier that give off a low frequency thump sound when powered on. The sound is such that I can almost feel the headphone membranes moving heavily, it's not like a fine pitched pop. It makes no sound when powered off.
The amplifier is powered from a 30VA +-15V dual sided transformer with a snubber circuit on the primary side, there are no output capacitors. It has a NE5532 acting as a input stage with MJE3055 and MJE2955 bjts as a class AB output stage (biased well into class A). The offset voltage is like 5mV so I doubt that this is enough to cause this kind of sound when powering up. I'm wondering what would cause this thump sound
The amplifier is powered from a 30VA +-15V dual sided transformer with a snubber circuit on the primary side, there are no output capacitors. It has a NE5532 acting as a input stage with MJE3055 and MJE2955 bjts as a class AB output stage (biased well into class A). The offset voltage is like 5mV so I doubt that this is enough to cause this kind of sound when powering up. I'm wondering what would cause this thump sound
All circuits need a fraction of a second to stabilise. Rails rise, input and feedback caps need to allow DC conditions to settle. Its all normal but a good design should incorporate some form of muting on the output to prevent the user ever experiencing the effect. Same at switch off too. It should be silent.
Have to see the circuit to comment.
Best method is to use symmetrical circuits throughout and D.C. coupling wherever possible.
Thumps are nearly always due to capacitors charging in the circuit during switch-on. The best "common mode" circuits for avoiding this are those with e.g. differential input-differential VAS and current mirror in the VAS. The worst culprits tend to be single sided VAS/current source where the current source drags the output low until the VAS turns on which may be delayed by an RC network in the feedback loop.
JOhn
Best method is to use symmetrical circuits throughout and D.C. coupling wherever possible.
Thumps are nearly always due to capacitors charging in the circuit during switch-on. The best "common mode" circuits for avoiding this are those with e.g. differential input-differential VAS and current mirror in the VAS. The worst culprits tend to be single sided VAS/current source where the current source drags the output low until the VAS turns on which may be delayed by an RC network in the feedback loop.
JOhn
Schematic:
R22, R23, R26, R27 - 0 Ohm jumpers
R1, R2, R8, R9 - 6900Ohm
R17, R18, R19, R20 - 300Ohm
R15, R16 - 9100Ohm
R12, R21 - 10kOhm
R3, R4, R7, R10 - 1Ohm
R5, R13, R24, R25 - 120kOhm
R6, R14 - 10kOhm
C1, C4 - 3.3uF ceramic SMD
C6, C3 - 470nF ceramic SMD
C7, C8 - 2.7nF ceramic SMD
C2, C9 - 4700uF alluminum electrolytic
T1, T5 - MJE3055
T2, T4 - MJE2955
IC1A - NE5532
Alright, so I have a little trouble understanding what is being said here. I don't know what a VAS is, but are you saying that another type of op-amp would be better?
There are three capacitors being charged at the moment the amplifier is turned on, it's the input DC blocking capacitor, the 4700uF current reservoir capacitors, and the small cap in the feedback loop (for the bass-boost functionality). I have a switch for the bass-boost function. While the thump sound is louder if the switch for the bass-boost is open, the thump is still there, so it doesn't fully explain it. I did change some resistor values to reduce the offset voltage from about 20mV to 5mV, but that didn't change the thump sound in any remarkable way, so I'm pretty sure that a offset this small isn't enough to cause this sound. This leaves me with the charging of the 4700uF caps, but I don't understand how changes in the supply voltage of the op-amp is present at the output to such a degree.
I know that adding reed relays to the output of the amp would fix this problem, but I'm in no way interested in doing something like that.
About DC blocking and offset voltage. I don't believe that a constant DC wattage of 1uW into a pair of 32Ohm headset is any issue what so ever so I don't understand why I would need DC blocking on the output.
An externally hosted image should be here but it was not working when we last tested it.
R22, R23, R26, R27 - 0 Ohm jumpers
R1, R2, R8, R9 - 6900Ohm
R17, R18, R19, R20 - 300Ohm
R15, R16 - 9100Ohm
R12, R21 - 10kOhm
R3, R4, R7, R10 - 1Ohm
R5, R13, R24, R25 - 120kOhm
R6, R14 - 10kOhm
C1, C4 - 3.3uF ceramic SMD
C6, C3 - 470nF ceramic SMD
C7, C8 - 2.7nF ceramic SMD
C2, C9 - 4700uF alluminum electrolytic
T1, T5 - MJE3055
T2, T4 - MJE2955
IC1A - NE5532
Alright, so I have a little trouble understanding what is being said here. I don't know what a VAS is, but are you saying that another type of op-amp would be better?
There are three capacitors being charged at the moment the amplifier is turned on, it's the input DC blocking capacitor, the 4700uF current reservoir capacitors, and the small cap in the feedback loop (for the bass-boost functionality). I have a switch for the bass-boost function. While the thump sound is louder if the switch for the bass-boost is open, the thump is still there, so it doesn't fully explain it. I did change some resistor values to reduce the offset voltage from about 20mV to 5mV, but that didn't change the thump sound in any remarkable way, so I'm pretty sure that a offset this small isn't enough to cause this sound. This leaves me with the charging of the 4700uF caps, but I don't understand how changes in the supply voltage of the op-amp is present at the output to such a degree.
I know that adding reed relays to the output of the amp would fix this problem, but I'm in no way interested in doing something like that.
About DC blocking and offset voltage. I don't believe that a constant DC wattage of 1uW into a pair of 32Ohm headset is any issue what so ever so I don't understand why I would need DC blocking on the output.
Non linearity due to DC offset moving a driver cone is something I once wondered over,
http://www.diyaudio.com/forums/solid-state/116731-can-d-c-offset-cause-non-linear-distortion.html
Headphones are so sensitive, such that I don't think you will ever get a noise free switch on from an amplifier as you have shown.
Your series output resistor (that for some reason is only shown on one channel) needs a relay to be "normally shorting" on its output so that the output is shunted to ground. The relay would open after a couple of seconds. You also need to then trim the offset of the amp to a true 0.000 volts DC which isn't straightforward with 5532's. You have a huge imbalance in the input bias currents that doesn't help either and with no way around that as you are switching the feedback networks.
The only way to correct the DC offset is to have a preset to inject a current into the non inverting opamp input. That's doable. NE5534's or most single opamps have an offset null facility that could be used.
If you ditch the 5532 for a FET opamp all the bias current issues disappear.
VAS would normally mean the "voltage amplifier stage" which is just one part of the internal circuitry of the 5532, typically just one transistor.
http://www.diyaudio.com/forums/solid-state/116731-can-d-c-offset-cause-non-linear-distortion.html
Headphones are so sensitive, such that I don't think you will ever get a noise free switch on from an amplifier as you have shown.
Your series output resistor (that for some reason is only shown on one channel) needs a relay to be "normally shorting" on its output so that the output is shunted to ground. The relay would open after a couple of seconds. You also need to then trim the offset of the amp to a true 0.000 volts DC which isn't straightforward with 5532's. You have a huge imbalance in the input bias currents that doesn't help either and with no way around that as you are switching the feedback networks.
The only way to correct the DC offset is to have a preset to inject a current into the non inverting opamp input. That's doable. NE5534's or most single opamps have an offset null facility that could be used.
If you ditch the 5532 for a FET opamp all the bias current issues disappear.
VAS would normally mean the "voltage amplifier stage" which is just one part of the internal circuitry of the 5532, typically just one transistor.
Yes, but as I said, this isn't caused by the offset voltage as increasing or decreasing the offset makes no difference to the loudness of the thump sound. About the output resistor, look at the parts list, it's just a 0Ohm jumper.
I don't really need a noise free amplifier, I just want to reduce it a little bit (enough to not think that it's harmfull). It's hard to say if the thump sound is harmful as it is now, but it isn't pleasant.
Looking at mouser I couldn't find any fet input op-amps that can support a supply voltage of more than +-20V. Perhaps I'm not looking closely enough?
I don't really need a noise free amplifier, I just want to reduce it a little bit (enough to not think that it's harmfull). It's hard to say if the thump sound is harmful as it is now, but it isn't pleasant.
Looking at mouser I couldn't find any fet input op-amps that can support a supply voltage of more than +-20V. Perhaps I'm not looking closely enough?
Offset in itself is inaudible under static conditions. My suggestion of muting would still need the offset to zero though because the step change from 0.00 to 0.005v would appear as a definite click. If you are not using a series output resistor then muting by shorting to ground isn't suitable. It would have to be series and in the output line.
Headphones, and particularly running direct of an amplifier output are just to sensitive. I doubt you would get a noise free switch on tbh no matter what you tried.
OPA2604, -/+24 volts and a superb sounding device.
The bottom line is that a DC coupled headphone amp requires DC precision and stability. This is why people are recommending fet input op amps; they will increase DC precision by a magnitude or two.
The fact that your design has full DC gain exacerbates this issue. Reducing DC gain to 1 will reduce the thump by several decibels. And if you add in an offset adjustment circuit (Walt Jung outlines several in his "IC Op Amp Cookbook"), your circuit may provide satisfactory performance.
I prototyped several headphone amp circuits (still looking for the holy grail) and was able to obtain exemplary performance with a circuit very similar to yours, with just DC gain = 1 and offset adjustment. I'm still trying to come up with the "ultimate" practical circuit.
The fact that your design has full DC gain exacerbates this issue. Reducing DC gain to 1 will reduce the thump by several decibels. And if you add in an offset adjustment circuit (Walt Jung outlines several in his "IC Op Amp Cookbook"), your circuit may provide satisfactory performance.
I prototyped several headphone amp circuits (still looking for the holy grail) and was able to obtain exemplary performance with a circuit very similar to yours, with just DC gain = 1 and offset adjustment. I'm still trying to come up with the "ultimate" practical circuit.
DC servo?
I would do 2 things:
- Add DC servo integrators using the second NE5532 or some even more low-offset chip;
- Use class A output with current source - heat dissipation requirement is much lighter than for a big power amp, so from my point of view it makes sense.
Capacitors serially with R6 and R14 also minimize DC offset, but I don't like this option because of compromising the quality...
I would do 2 things:
- Add DC servo integrators using the second NE5532 or some even more low-offset chip;
- Use class A output with current source - heat dissipation requirement is much lighter than for a big power amp, so from my point of view it makes sense.
Capacitors serially with R6 and R14 also minimize DC offset, but I don't like this option because of compromising the quality...
Read what Sajti says, his suggestion is correct. 22uF would be sufficient for -3dB = 10 Hz.
Edit: Also Fast Eddie and vzaichenko
Last edited:
If you "hear" a capacitor in that application, you're using the wrong capacitor. Capacitors introduce distortion when they are a significant impedance in the circuit. An electrolytic in a tone control (bass control) is a terrible choice; but you will see it in a lot of consumer grade equipment.
Cyril Bateman did exhaustive research on this topic. You should check it out.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.