I tried to short C19 pins. It gave very small reduction of amplitude of an oscillations. Slightly less oscillations if to short pin3 of LME to ground. And no oscillations with volume full on.
interesting, may be then it caused by stray capacitance of volume regulator that can depend on its position.. datasheet states that lme49990 can drive up to 100pF that is rather big number for such a small pot.. is oscillation stops just on the full-on or at some regulator position, not exactly at the end?
anyway, checking/fixing this means cutting PCB's tracks thats not good..
anyway, checking/fixing this means cutting PCB's tracks thats not good..
Its starts / stops just before volume maximum. I tried to desolder output pins 1 and 7 from the adapter. In such case, with open NFBL, there is no oscillations. But opamp still quite hot - 62C* @ room temp 27C*.
Now from datasheet, the Iq = 9-10mA (quite lot to my eyes). The voltage supply is+-11V, then:
T=Tamb+Tresistance*P
27C+145C/W*(11*0.01)=42.95C
But there is 2 chips on the small adapter, so we can to multiply P dissipation by 2:
27C+145C/W*(11*0.01)*2=58.9C*
So looks like temperature rice is predictable due to hi current consumption of that opamp.
But the plot of output voltage still looks poor.
Now from datasheet, the Iq = 9-10mA (quite lot to my eyes). The voltage supply is+-11V, then:
T=Tamb+Tresistance*P
27C+145C/W*(11*0.01)=42.95C
But there is 2 chips on the small adapter, so we can to multiply P dissipation by 2:
27C+145C/W*(11*0.01)*2=58.9C*
So looks like temperature rice is predictable due to hi current consumption of that opamp.
But the plot of output voltage still looks poor.
Yes, but it operates at 62-64C* with all assembled. Looks like it works in class-A
Last edited:
to clarify you can desolder output pins AND connect them (with thin wires) to neg in-s.
BTW if you will do that and if opamps will be stable - then you may do some more microsurgery and put a ~600R SMD resistors between opamp's out-s and corresponding adapter's pad.
BTW if you will do that and if opamps will be stable - then you may do some more microsurgery and put a ~600R SMD resistors between opamp's out-s and corresponding adapter's pad.
no caps, just build an unity gain buffers directly on adapter and connect it with o2's circuit via 600R resistors - to isolate stray capacitance of main PCB from opamp's output.
Here is something interesting.
I created the unity gain buffer, and noise still be here, even with output disconnected from PCB.... But ! Noise gone when I removed both 4556. Looks like those are root of an evil
I created the unity gain buffer, and noise still be here, even with output disconnected from PCB.... But ! Noise gone when I removed both 4556. Looks like those are root of an evil
There is no 3.3u caps in o2 schematics as I can see. There is only 2.2u, assuming thats you're telling about. Putting 200pF to ground after VR will change a bit frequency response at high frequencies and mostly with middle (by resistance) VR position thats not fair. And still its not clear why oscillation happens.
Let summarize
no oscillation: njm2068 + njm4556, nothing + njm4556, lme49990 + nothing, lme49990 + njm4556 with RC on input
oscillation: lme49990 + njm4556, lme49990 as unity gain just on adapter effectively not connected to PCB + njm4556.
Did you check what happens on power rails?
Let summarize
no oscillation: njm2068 + njm4556, nothing + njm4556, lme49990 + nothing, lme49990 + njm4556 with RC on input
oscillation: lme49990 + njm4556, lme49990 as unity gain just on adapter effectively not connected to PCB + njm4556.
Did you check what happens on power rails?
no oscillation: njm2068 + njm4556, nothing + njm4556, lme49990 as unity buffer , not connected + nothing.
oscillation: lme49990 + njm4556, lme49990 as unity gain + njm4556 ,lme49990 as unity gain + nothing, lme49990+ nothing.
Same 100MHz sinusoida on the power rails.
oscillation: lme49990 + njm4556, lme49990 as unity gain + njm4556 ,lme49990 as unity gain + nothing, lme49990+ nothing.
Same 100MHz sinusoida on the power rails.
Last edited:
What about amplitude of signal on power rails? Isn't it larger that on opamp outputs? I afraid that maybe 'power managment' unit is taking part in all that mess.
Can you add 10+uF of ceramic/low-inductive film (Epcos are good) in parallel to C10/C17/C18? Or better - same fat 10uF on each of opamps?
Can you add 10+uF of ceramic/low-inductive film (Epcos are good) in parallel to C10/C17/C18? Or better - same fat 10uF on each of opamps?
is oscillation with same or higher amplitude observed on voltage regulators (U5/U6) outs? if yes, and if you have battery-able modification - can you check with running on only batteries?
With batteries - oscillations at same frequency, but slightly less amplitude - about 34 milli volt.
About frequency - its interesting.
Do following 'test' - connect DSO to output, power on and while watching oscillation - put finger randomly on opamps pins 🙂, watching at same time at oscillation - if it changes much or even disappear. Considering your finger as RC - this can give some hint where oscillation root located.
PS Finger must be dry🙂
Do following 'test' - connect DSO to output, power on and while watching oscillation - put finger randomly on opamps pins 🙂, watching at same time at oscillation - if it changes much or even disappear. Considering your finger as RC - this can give some hint where oscillation root located.
PS Finger must be dry🙂
- Home
- Amplifiers
- Headphone Systems
- The Objective2 (O2) Headphone Amp DIY Project