Thank´s for your reply!
Is there any risk (for worser sound) if combining 0,1uF or 100nF/47uF for each voltage pin to ground with 0,1uF between + and - (pin 4 and 8)? Can they harm each other if using both princeples at the same time?
Should I remove either of them?
mixing capacitors with little resistance (damping) between them risks ringing or worse on the supply lines.
I never put any decoupling capacitors in low noise audio circuit cards. It generates noise on the local ground reference, and reduces PSU noise rejection.
Stéphane
Stéphane
OpAmps need a stable, low noise low impedance supply as seen from their power pins. Strict local decoupling is not mandatory to get there if you want to avoid that for any reason.
With 4 dual opamps arranged like the spokes of a wheel and shunt regulators placed at the hub you can power eight "audio grade" (moderate GBW and SR) opamp channels sans local decouplers quite well with a good compressed layout (4-layer preferred). With both sides of PCB stuffed with components this doubles to 16 channels.
With 4 dual opamps arranged like the spokes of a wheel and shunt regulators placed at the hub you can power eight "audio grade" (moderate GBW and SR) opamp channels sans local decouplers quite well with a good compressed layout (4-layer preferred). With both sides of PCB stuffed with components this doubles to 16 channels.
That's a good way to get oscillation and noise coupling. Good layout can keep the ground local to the opamp, not disturbing the rest of the circuit.
Do you mean the phono stages and microphone amplifiers?
Depends on how solid your local ground reference is, the way your feedback is referenced, etc. If everything is done right, it reduces the rails noise, rather than generates noise on the local ground reference.
There are many ways to skin a cat, as they say.
Extending the scheme I mentioned above the number of channels could be doubled again to 32 total using 4ch opamps and still have no need for local decoupling other than intrinsic by the supply layers. Would need beefy and well behaved supply regulators at the hub, and a tight layout (SMT only). Never tried this but it looks perfectly feasible. I'd still place rail-to-rail decouplers at each package if board space allows...
Extending the scheme I mentioned above the number of channels could be doubled again to 32 total using 4ch opamps and still have no need for local decoupling other than intrinsic by the supply layers. Would need beefy and well behaved supply regulators at the hub, and a tight layout (SMT only). Never tried this but it looks perfectly feasible. I'd still place rail-to-rail decouplers at each package if board space allows...
Last edited:
Below is a Picture of one of the two channels from the schematic of the HD970. Just replace the OP275GS and OPA2134UA with THS4032 and adding a 0,1uF bipolar cap for each between pin 4 and 8 (+ and -).
An externally hosted image should be here but it was not working when we last tested it.
It does´nt seems the link for the picture is working, but here it its:
https://1drv.ms/i/s!Ap_55U2rYJrvqhoqhCS2bUPTnWNE
Very unusual to specify a bi-polar capacitor for MF decoupling duty.
I wonder if the bipolar capacitor has a higher esr. That may give sufficient damping to control supply rail ringing.
Did they adopt any HF decoupling for the supply rails?
Last edited:
You will find the link in my latest post from today.
And maybe we are missunderstanding each other about what kind of cap. I´m using - here is a picture of a similar mod I´ve done:
https://1drv.ms/i/s!AoSW4mjyvjGAgyZNuyRuZdLikhfV
Well, anyway, an OPA1642 has better datasheet specs than an OPA2134---about 4 db less noise, 4 db less distortion, and ~ 1 db more headroom. Both are FETs. Has anyone done any direct listening comparison between these two?
The OPA164x does look like the "altogether better" equivalent replacement for the OPAx134 series. Pretty much the only place I could see the latter is when you need a DIP8 package. If you need the small extra DC precision, you're not looking at the right parts.
Has anyone had any thoughts on the LME49990. I've tried this for an MC (RIAA) pre-amp and got very good results.
regards
Graham
regards
Graham
- Status
- Not open for further replies.