Just wondering if someone can clear up a quick question for me with regards to PSRR. I'm specifically thinking of chip amps in my case, but I assume it works the same way for any amplifier which is why I'm posting in the power supply sub-forum.
Suppose an amp has a PSRR of -65dB. I understand that this corresponds to an attenuation factor of 10^(-65/20) = 0.0006.
Does this mean that, say, 1 volt of ripple on the power rails will cause 0.0006 volts of ripple on the amplifier's output? Or does it mean that 0.0006 volts of ripple is effectively added to the amplifier's input, and is then subsequently amplified depending on the amplifier's gain?
Or is it something else entirely?
Suppose an amp has a PSRR of -65dB. I understand that this corresponds to an attenuation factor of 10^(-65/20) = 0.0006.
Does this mean that, say, 1 volt of ripple on the power rails will cause 0.0006 volts of ripple on the amplifier's output? Or does it mean that 0.0006 volts of ripple is effectively added to the amplifier's input, and is then subsequently amplified depending on the amplifier's gain?
Or is it something else entirely?
Keep it simple... the amout of PSU ripple rejection on the output waveform...expressed in dB... Meaning the amount of ripple that gets through the opamp onto the output, a higher PSSR means less ripple on output due to more rejection. This my brain's summary of it. Not trying to dive into the weeds on this. Others will gladly dive out into the weeds for sure. My assumption (key word assumption) is that class a is low, chip amps are higher, and class d is highest? Sounds good and in my head i think of PSSR being related to conduction. If a device is turned on constantly (class A) it would tend to be a window into the PSU. In switching applications (Class D) there is a smaller window into the PSU due to the switching. If you lived in antartica, could you reject the cold from outside better by leaving the door open 100% of the time? or opening and closing it 200,000 times per second?
I know my explanation is not techically correct, but this is how I understand things. I am sure others will correct me.
Quoted from here...Power supply rejection ratio - Wikipedia, the free encyclopedia
The PSRR is defined as the ratio of the change in supply voltage to the equivalent (differential) input voltage it produces in the op-amp, often expressed in decibels.[1][2][3] An ideal op-amp would have infinite PSRR. The output voltage will depend on the feedback circuit, as is the case of regular input offset voltages. But testing is not confined to DC (zero frequency); often an operational amplifier will also have its PSRR given at various frequencies (in which case the ratio is one of RMS amplitudes of sinewaves present at a power supply compared with the output, with gain taken into account).
I know my explanation is not techically correct, but this is how I understand things. I am sure others will correct me.
Quoted from here...Power supply rejection ratio - Wikipedia, the free encyclopedia
The PSRR is defined as the ratio of the change in supply voltage to the equivalent (differential) input voltage it produces in the op-amp, often expressed in decibels.[1][2][3] An ideal op-amp would have infinite PSRR. The output voltage will depend on the feedback circuit, as is the case of regular input offset voltages. But testing is not confined to DC (zero frequency); often an operational amplifier will also have its PSRR given at various frequencies (in which case the ratio is one of RMS amplitudes of sinewaves present at a power supply compared with the output, with gain taken into account).
Last edited:
I did see that Wikipedia article, which seems to suggest that the attenuated ripple appears on the input (and hence is amplified by whatever the amp's gain is), but it seems to talk exclusively about op-amps. I know that many chip amps are basically high-power op-amps, with inverting and non-inverting inputs, but plenty of chip amps just have one input (although I guess internally they are probably op-amp-like, just with hard-wired gain). I really don't know how similar various discrete or tube amps may be to op-amps. Thus I wasn't sure if the content of that article was generalisable to all amps.
I suppose it probably is, if only because a consistent definition is more useful. I guess the take home message then is that the importance of power supply cleanliness increases in proportion to gain, since the gain effectively offsets PSRR? If the PSRR and gain are equal (well, equal in magnitude, opposite in sign) then the output ripple and rail ripple should have the same amplitude.
I suppose it probably is, if only because a consistent definition is more useful. I guess the take home message then is that the importance of power supply cleanliness increases in proportion to gain, since the gain effectively offsets PSRR? If the PSRR and gain are equal (well, equal in magnitude, opposite in sign) then the output ripple and rail ripple should have the same amplitude.
Oh, actually, I didn't read your Wikipedia quote closely enough. Toward the end it says "often an operational amplifier will also have its PSRR given at various frequencies (in which case the ratio is one of RMS amplitudes of sinewaves present at a power supply compared with the output, with gain taken into account)." (my emphasis). So I guess it is based on output.
I'm curious now as to how gain is taken into account, since it can be set to whatever you like with an op-amp. I guess the publsihed PSRR specs must be for a particular gain setting only?
I'm curious now as to how gain is taken into account, since it can be set to whatever you like with an op-amp. I guess the publsihed PSRR specs must be for a particular gain setting only?
Really? So when an op-amp datasheet has a PSRR in it, how is that supposed to be interpreted, given that an op-amp can be used in countless different kinds of circuit?
It could mean one of two things:
1. the ratio of power supply rail ripple and output ripple, with the opamp used open loop - assuming that it is the output stage which dominates the PSRR (probably unlikely?).
2. the PSRR of a particular datasheet circuit - which may be given as a circuit diagram, or as a specified closed-loop gain.
I would guess option 2 is most likely.
1. the ratio of power supply rail ripple and output ripple, with the opamp used open loop - assuming that it is the output stage which dominates the PSRR (probably unlikely?).
2. the PSRR of a particular datasheet circuit - which may be given as a circuit diagram, or as a specified closed-loop gain.
I would guess option 2 is most likely.
Hmm, that's interesting. I suppose it must not vary too much across sensible circuits, otherwise if you build anything other than that reference circuit you'd have no idea what the value was.
It's surprising how much subtlety can be behind one datasheet parameter!
It's surprising how much subtlety can be behind one datasheet parameter!
In either case, it's usually input -referred, and falls with frequency.
This can really catch you out - see a nice headline figure then use said part at say 20-40dB gain, and at 10khz wonder why you've only got about 20dB of supply rejection? No, that's exactly what they said it would do.
Also why real attention to PSU rails may pay dividends...
This can really catch you out - see a nice headline figure then use said part at say 20-40dB gain, and at 10khz wonder why you've only got about 20dB of supply rejection? No, that's exactly what they said it would do.
Also why real attention to PSU rails may pay dividends...
It can be calculated for any given circuit very similar to the way input offset voltage and current can be accounted for.
op amp datasheet PSRR is a "input referred" value
to calculate the circuit's output PSRR you include a Vsource in series with one of the op amp inputs with amplitude = pwr supply ripple*PSRR - just like the Vnoise source
an interesting 5-pin op amp relation is that PSRR is usually the same as open loop gain for the rail that internal compensation C is referred to, the other rail PSRR may be much higher, less frequency dependent
in simple negative feedback op amp circuits you can just reduce the PSRR dB by the closed loop gain in dB to get the output PSRR
to calculate the circuit's output PSRR you include a Vsource in series with one of the op amp inputs with amplitude = pwr supply ripple*PSRR - just like the Vnoise source
an interesting 5-pin op amp relation is that PSRR is usually the same as open loop gain for the rail that internal compensation C is referred to, the other rail PSRR may be much higher, less frequency dependent
in simple negative feedback op amp circuits you can just reduce the PSRR dB by the closed loop gain in dB to get the output PSRR
Last edited:
- Status
- Not open for further replies.