Hello ezkcdude,
Across theses tests, you can see there no simple solution who sounds better than others on all parameters. Even small component as a 10nF capacitor changes sound. Adjusting tonal balance is not easy. You have to do your own choice!
Many articles on LM317 speaks about enhancing noise rejection, but none of them speak about results on sounding. (It is why I do theses tries).
I'm always spectical on using chemical capacitor. You certainly agree with me when I say that using chemical capacitor in signal path is not a good solution. How this bad component should be good when it uses in the power supply?
Chemical capacitor is a necessary harm, because we could not use only polypropylene capacitors everywhere!
Eric
If you look at the graphics on my article's page you see that increasing Cout capacitor can be good for bass level and worse for other parameters. Using small Cout increase definition, details and subtleties, but reduce bass level !ezkcdude said:
Across theses tests, you can see there no simple solution who sounds better than others on all parameters. Even small component as a 10nF capacitor changes sound. Adjusting tonal balance is not easy. You have to do your own choice!
It certainly be a good solution to reduce LM317's noise and enhance rejection, but it have other (negative or positive) impact on sound.ezkcdude said:
Many articles on LM317 speaks about enhancing noise rejection, but none of them speak about results on sounding. (It is why I do theses tries).
I'm always spectical on using chemical capacitor. You certainly agree with me when I say that using chemical capacitor in signal path is not a good solution. How this bad component should be good when it uses in the power supply?
Chemical capacitor is a necessary harm, because we could not use only polypropylene capacitors everywhere!
Eric
Eric, high Q caps like polyprop or film can cause ringing which sounds really bad.
So I prefer stinking normal electrolytics around LM317/337.
No Oscon either!
So I prefer stinking normal electrolytics around LM317/337.
No Oscon either!
Hello QSerraTico_Tico,
Using electrolytic capacitors is certainly the easy way to save troubles. You can obtain better sound from LM317 using other configuration.
Using incorrect capacitor's values can cause ringing. The LM317 rings very well! Rings depend on schematic diagram, PCB…
On my test configuration, I give a steady sample.
Rings risk decrease when heating time increase. In my tries I could test special configuration without rings. With LM317 at ambient temperature, special configuration immediately rings at startup.
Actualy, I use Cadj=10µF MKP and only Cout=4.7nF MKP. When Cout is solder near LM317, it rings. When Cout is solder far from LM317 (10cm) it is steady!
Regards
Eric
QSerraTico_Tico said:
Using electrolytic capacitors is certainly the easy way to save troubles. You can obtain better sound from LM317 using other configuration.
Using incorrect capacitor's values can cause ringing. The LM317 rings very well! Rings depend on schematic diagram, PCB…
On my test configuration, I give a steady sample.
Rings risk decrease when heating time increase. In my tries I could test special configuration without rings. With LM317 at ambient temperature, special configuration immediately rings at startup.
Actualy, I use Cadj=10µF MKP and only Cout=4.7nF MKP. When Cout is solder near LM317, it rings. When Cout is solder far from LM317 (10cm) it is steady!
Regards
Eric
Eric, is the ringing audible or something that you can measure? I've never heard ringing on my setup. Can you give an estimate of the amplitude/frequency of ringing? What would you see on an analyzer?
Winfield Hill has noted that a LM317 reg has an effective output inductance of about 5uH with the 'Adjust' pin bypassed as the datasheet recommends (10uF). So for a given cap on the output - say 47uF:
Impedance at resonance =SQRT(5uH/47uF) = ~0.3R.
So we se that a good-quality electrolytic cap of 0.3ohm ESR will be critically damped; and cheapo electrolytic, Xc=1ohm, will be more-than critically damped (Q<0.7, a Good Thing) but a 47uF Oscon, maybe 0.015ohm ESR, represents a high-Q circuit (Q ~20) at f = 1/ (2*pi*SQRT(L*C)), about 10Khz, or a tendency to ring at that frequency. Hard /forward /sibilant treble anyone?
So we can see Eric's 4n7 styrene cap will ring at 100Khz, but the Q is likely to be low - since the 317 has little gain that high, and losses in the PCB traces probably become significant. So
Now try the numbers for the 1uF cap shown in the original National Semiconductor LM117 datasheet and you'll find X=2.7ohm - which is exactly the resistor value they show in series with the 'ideal' cap, with no explanation why...
Impedance at resonance =SQRT(5uH/47uF) = ~0.3R.
So we se that a good-quality electrolytic cap of 0.3ohm ESR will be critically damped; and cheapo electrolytic, Xc=1ohm, will be more-than critically damped (Q<0.7, a Good Thing) but a 47uF Oscon, maybe 0.015ohm ESR, represents a high-Q circuit (Q ~20) at f = 1/ (2*pi*SQRT(L*C)), about 10Khz, or a tendency to ring at that frequency. Hard /forward /sibilant treble anyone?
So we can see Eric's 4n7 styrene cap will ring at 100Khz, but the Q is likely to be low - since the 317 has little gain that high, and losses in the PCB traces probably become significant. So
- looks about right to me.Eric said:
Now try the numbers for the 1uF cap shown in the original National Semiconductor LM117 datasheet and you'll find X=2.7ohm - which is exactly the resistor value they show in series with the 'ideal' cap, with no explanation why...
Hello Martin,
Thanks for all theses details.
To answer to ezkcdude, the LM317 rings at few megahertz and can have very high amplitude (more than 10Vpp). Generally, you can't miss it. But in particulars conditions, amplitude could be very small and not visible on the scope.
Eric
Thanks for all theses details.
To answer to ezkcdude, the LM317 rings at few megahertz and can have very high amplitude (more than 10Vpp). Generally, you can't miss it. But in particulars conditions, amplitude could be very small and not visible on the scope.
Eric
- Status
- Not open for further replies.