Since the LM4562 has been mentioned, what might be a more suitable decoupling capacitor type in the common B configuration - X7R ceramic or C0G?
*Examples in this case might be:
-Vishay K104K15X7RF5UL2 100nF X7R ceramic.
-Wima MKP2D031001F00JSSD 100nF polypropylene.
*Examples in this case might be:
-Vishay K104K15X7RF5UL2 100nF X7R ceramic.
-Wima MKP2D031001F00JSSD 100nF polypropylene.
Last edited:
Yes. It appears to be a data entry error on a sale site, and I inadvertently copy-pasted it. Fixed.
Some say X7R is good enough, but use the much better NPO/COG type. All are types of ceramic.
https://www.electronics-notes.com/a.../ceramic-dielectric-types-c0g-x7r-z5u-y5v.php
There are plenty examples of cheap op amp circuits that use minimal cacacitor decoupling. They usually have one thing in common - poor audio performance.
Many low ESR electrolytics do not work well with an adjacent small bi-pass capacitor. If small caps are local to the amplifier better to have a bigger physical distance to the bigger decoupling caps.
The OP amp may be requied to actually drive the following circuit wih a little current. if that is the case larger decoupling caps actually complete the current path.
Many low ESR electrolytics do not work well with an adjacent small bi-pass capacitor. If small caps are local to the amplifier better to have a bigger physical distance to the bigger decoupling caps.
The OP amp may be requied to actually drive the following circuit wih a little current. if that is the case larger decoupling caps actually complete the current path.
Ok. Maybe I've got this wrong. If option B is used the bypass capacitor should be connected to the same ground plane as the signal at the PCB, thus making the trace length shorter?
I thought that these decoupling capacitor should be connected to a separate ground at the pcb, and then to the central ground star or similar. This to avoid noise to be injected to the signal.
Wellif that's the case, option B does not complicates the pcb layout very much more than option A, if this is how it's suppose to be done. I've actually build a few circuits like this before and it worked ok. But it was quite small circuits using one or to opamps, so I thought it was just luck then...
The local decoupling capacitor's function is to provide a low inductance path from the op amp supply pin to ground.
Parasitics like trace inductance are included in this role. A ground plane will simplify the layout, not complicate it.
https://www.engineering.com/story/decoupling-capacitors-are-crucial-in-pcb-design
Parasitics like trace inductance are included in this role. A ground plane will simplify the layout, not complicate it.
https://www.engineering.com/story/decoupling-capacitors-are-crucial-in-pcb-design
Last edited:
Well that pretty much answers my question. I was thinking of designing a two layer pcb, so one groundplane. No need for separate grounds either, since both the signal and decoupling capacitors should be connected to same ground plane.
However I think option A will be enough anyway. But interesting to know for future design.
However I think option A will be enough anyway. But interesting to know for future design.
Self is wrong. The correct way to decouple OPAs is the use B with cheapo Aluminium ELECTROLYTIC caps; AS CLOSE TO THE OPAs as possible. And yes. You will need separate CLEAN & DIRTY earths and work out where to join them.
This is of little importance if you have only one 5532 but if you have 2 or more, you will better THD bla bla if you do this.
This is of little importance if you have only one 5532 but if you have 2 or more, you will better THD bla bla if you do this.
I prefer to trust the ones that actually make the op amp. They say 0.1uf low-esr ceramic bypass capacitors as close as possible to the power and ground pins (in layout B).
Not electrolytic.
Not electrolytic.
If you can measure very low levels of HF THD with reactive loads, why don't you check out the difference between
- just 100n ceramics
- 100n ceramics and cheapo electrolytics
- just cheapo electrolytics
There is option D: that is A with half of B. So one 100 nF cap from + to - and only one 100 nF cap from + to GND. Of course all very close to the opamp.
Please think about the benefits 🙂
Please think about the benefits 🙂
I'm very tempted o do actually, when I get the time. At least between option A and B would be very interesting, also trying different op-amps and see if there's any difference.
Mean while I was reading some in R. Slone's book "The Audiophile's project sourcebook". He uses a different decoupling of opamp, with 4,7 uF tantalum in parallel with 0,1uF ceramic between each supply pin and ground. Then 0,1 uF ceramic in parallel with a 0,1uF mylar (= polyester?) between each supply and ground per board.
It probably worked for him, but I guess it's overkill for the NE5532 at least. He uses TL074/TL084 in some design I've seen. I also studied the PCB layout he uses, and I can conclude that he does not uses a separate ground for the decoupling capacitors, as I first intended to do.
Finally I got the right answer I was looking, regarding how to ground the the decoupling capacitors, after reading this thread;
https://www.diyaudio.com/community/threads/op-amp-decoupling-best-practice.274567/page-2
The ground path or the decoupling capacitor should not be to long from the load that the op amp is driving. Meaning there should not be a separate ground for the signal and decoupling capacitors at the board. There might be I risk for noise being injected into the the ground of course, but for now I will use this design then I'm designing my next PCB, let's see how it works.
Even though I'm quite convinced that option A should work just fine for the NE5532 (also according to Elliot and Self). I think I'll use option B, just to have the possibility to change the op amps to others in the future. Nothing really says (including the datasheet) that option B shouldn't be used for NE5532....
That is a good plan. "A" was due to a quirk of the NE5532. "B" is better in all other cases.bengtssk said:
Ed
This is not the semi mfr recommended method, nor is it standard practice.
Two ceramic caps, one from VCC to GND, one from VEE to GND with their ground terminals ideally placed right next to each other and connected to a solid ground plane. Works well on the bottom layer right below the OpAmp with SMD capacitors but can also be implemented on the top layer. Move both caps to the same side of the IC and then make the connection to the GND plane together and have an imaginary "keep out area" around that connection point (don't use it as a reference). At the low currents in line level OpAmp circuits and the low impedance a solid ground plane provides that shouldn't be a problem in my limited experience. As soon as you are starting to drive heavier loads and higher currents are involved, start thinking more about what should go where to keep the high currents localised. PCB design is like real estate: Location, location, location.