Regulated PSU for chip amp, Which one is better sound quality that is produced, the regulator IC LM338 and LT1083 compared? the price is cheaper LM338 than LT1083, but from the datasheet value of PSRR (power supply ripple rejection) are both at 75 db it? and in real market I found the original LM338-made NS (shape metal can ) while the LT1083 printing rather vague? ..... presumably frinds who has experience can be shared?.....sorry my english is not good🙄
Im sure the sound quality would be pretty much the same to the ear ... much like an unregulated PSU for a Chip amp and a regulated one will sound pretty much the same .... Most chip amps have pretty good PSSR already so in most cases going regulated is a bit overkill in my opinion .....
Cheers
Cheers
In audio:
If you THINK it sounds better - it does...
This is important.
Market men are not able to sell anything if this was not "true".
If you THINK it sounds better - it does...
This is important.
Market men are not able to sell anything if this was not "true".
I could not agree more. A Solid State Amplifier that uses split rails does not need a regulated supply as it is a balanced amplifier. It just needs enough power!
When I wanted an LT1084, I found an LD1084 that was much cheaper but had very similar specs. Just FYI.
I remember reading that if you are not careful, you can over regulate a power amp supply, raising it's impedance to the point where It can badly effect sound quality.
I have noticed the effect a mains surge protector has on a power amp, it crushed dynamics and left it sounding dull even boring. This I am told is due to it increasing the mains supply impedance.
Anyway, why put in extra components if you don't need to?
I have noticed the effect a mains surge protector has on a power amp, it crushed dynamics and left it sounding dull even boring. This I am told is due to it increasing the mains supply impedance.
Anyway, why put in extra components if you don't need to?
I don't know whether the chipamp will benefit from a regulated supply. PSRR is decent. It will DEFINITELY benefit if you buffer the input and run inverted. Cordell describes in his book -- OPA2604 is the opamp of choice.
You can lower regulated supply impedance with a faster error amplifier, higher gain pass device -- but the pass device has have an SOA to match the current demands.
Hi everyone,
I'm currently using a regulated power supply with LM338T and I'm happy with the results, but I think that an unregulated power supply with large reservoir capacitors between 10mF-20mF per rail, is suitable for chip amps with higher PSRR.
A regulated power supply has less ripple and so less hum, but is quite inefficient 30-50 %
An unregulated power supply if not properly designed normally too long and thin ground paths and/or low value reservoir capacitors can have more hum, due to the ripple, but is much more efficient 80-90%.
I think that it depends mainly on the ripple voltage of your "raw" supply and on the PSRR of the audio chipamp.
PS: Always make wide and short ground paths and separate paths for power ground (return from speakers and sometimes power devices if single supply), the signal ground can be for the rest of the amp (CCSs and input reference), the center of the star ground must be a high value capacitor of at least 470-1000uF.
Best regards,
Daniel Almeida
I'm currently using a regulated power supply with LM338T and I'm happy with the results, but I think that an unregulated power supply with large reservoir capacitors between 10mF-20mF per rail, is suitable for chip amps with higher PSRR.
A regulated power supply has less ripple and so less hum, but is quite inefficient 30-50 %
An unregulated power supply if not properly designed normally too long and thin ground paths and/or low value reservoir capacitors can have more hum, due to the ripple, but is much more efficient 80-90%.
I think that it depends mainly on the ripple voltage of your "raw" supply and on the PSRR of the audio chipamp.
PS: Always make wide and short ground paths and separate paths for power ground (return from speakers and sometimes power devices if single supply), the signal ground can be for the rest of the amp (CCSs and input reference), the center of the star ground must be a high value capacitor of at least 470-1000uF.
Best regards,
Daniel Almeida
unregulated supply willbe sufficient for low volume playback. The reason is the ripply on main capacitor is funtion of average current in half cycle of supply voltage. When the amplifier in low volume, say 1Watt acctully quite loud on an average speaker, the current is only 0.5A, the ripple is only 0.5V on a 10000uF capacitor, so most people can get away with unregulated supply. If you play in 50W power, the supply ripple will be close to 4V, 8V for two channels. That is not so good. Especially if you consider cross modulation generated by signal and supply, which is beyond PSRR, the distortion will be proportional to ripple of the supply.
I saw some ill effects on some test data, but I have not seen sysmatic test data on cross modulation of supply ripple with large signal, even in simulation, most people just supply their amplifier with ideal voltage source, which means a perfect regulated supply, after the amplifier is built, I seldom saw any data though. There is however one IC amplifier with regulated supply on this forum with super good test results, I believe it was a LM3886 IC chip amplifier.
I saw some ill effects on some test data, but I have not seen sysmatic test data on cross modulation of supply ripple with large signal, even in simulation, most people just supply their amplifier with ideal voltage source, which means a perfect regulated supply, after the amplifier is built, I seldom saw any data though. There is however one IC amplifier with regulated supply on this forum with super good test results, I believe it was a LM3886 IC chip amplifier.
For LM3886, positive rail PSRR is excellent throughout the audio range but neg rail PSRR tracks CMRR and open loop gain (as seen with many chips, small signal opamps as well except of the newer breed).
Therefore, with a split supply, a regulator or a cap multiplier sure makes some sense on the neg rail because that is where most of the internals of the chip are referenced to during operation. Then again, you'd have to build a real excellent one which has lower and flatter broad-band impedance than a proper set of supply and bypass caps, most of which you still need with a regulator.
Regulated split supplies may have a small benefit of a "nicer" sounding clipping (clipped tops of the waveform not modulated with 100/120Hz supply ripple) at the cost of clipping earlier. To keep losses low, you'd need a fairly stiff mains transformer and slowly tracking regulator aka cap multiplier.
Overall, not so much net benefit to be had notably with fully regulated supplies, that's why we seldom find them in typical chipamps. Generous amounts of local supply and bypass capacitance and clever routing and placement is the better way, IMHO.
Therefore, with a split supply, a regulator or a cap multiplier sure makes some sense on the neg rail because that is where most of the internals of the chip are referenced to during operation. Then again, you'd have to build a real excellent one which has lower and flatter broad-band impedance than a proper set of supply and bypass caps, most of which you still need with a regulator.
Regulated split supplies may have a small benefit of a "nicer" sounding clipping (clipped tops of the waveform not modulated with 100/120Hz supply ripple) at the cost of clipping earlier. To keep losses low, you'd need a fairly stiff mains transformer and slowly tracking regulator aka cap multiplier.
Overall, not so much net benefit to be had notably with fully regulated supplies, that's why we seldom find them in typical chipamps. Generous amounts of local supply and bypass capacitance and clever routing and placement is the better way, IMHO.
Great points and analyses, everyone. Klaus, I know that you are a professional in these areas. I am not and I am still trying to understand some things about it.
What are the considerations for the extremely low audio frequencies, such as down at 20 or 30 Hertz? (And remember, this is diyaudio, with goals including "perfection at any cost". <grin>)
I remember calculating that in order to maintain a very low impedance, as seen by the chipamp's power/gnd pins, the necessary capacitance grew to be huge, as frequency got very low. It came out to be something between 5 and 10 Farads! (not to mention how to keep the inductance low-enough, in that case) At the time, someone suggested that was where they relied on regulation, instead of capacitance.
Is there even a case where the PSRR could require very low impedance at such low frequencies? I understand that impedance in this context is basically just the change in voltage (ripple) that would be caused by a change in the needed current. But I guess that I don't have a good feel for levels of the numbers that might be involved. (It's the same thing over and over, for me, in many different engineering contexts: There is a known effect or mechanism but is it significant or insignificant?)
What are the considerations for the extremely low audio frequencies, such as down at 20 or 30 Hertz? (And remember, this is diyaudio, with goals including "perfection at any cost". <grin>)
I remember calculating that in order to maintain a very low impedance, as seen by the chipamp's power/gnd pins, the necessary capacitance grew to be huge, as frequency got very low. It came out to be something between 5 and 10 Farads! (not to mention how to keep the inductance low-enough, in that case) At the time, someone suggested that was where they relied on regulation, instead of capacitance.
Is there even a case where the PSRR could require very low impedance at such low frequencies? I understand that impedance in this context is basically just the change in voltage (ripple) that would be caused by a change in the needed current. But I guess that I don't have a good feel for levels of the numbers that might be involved. (It's the same thing over and over, for me, in many different engineering contexts: There is a known effect or mechanism but is it significant or insignificant?)
Last edited:
- Status
- Not open for further replies.