Need recommendation for Op-Amp with better PSRR

Hi Anthony,

Because power supply rejection ratio rolls-off as frequency increases, the specification given in the data tables of a datasheet is really a "best case" scenario. To really narrow down a suitable opamp, it would be helpful if we knew what the frequency content of the noise in your power supply is, as well as what level this noise needs to be reduced to. For example, if this noise is coming from a switching power supply, what is the frequency that this supply is running at and what is the magnitude of the output ripple.

It is also possible that for very high frequency noise, no opamp will have adequate PSRR to reduce the noise to acceptable levels, in which case you would need to add some additional regulation to your power supply rails.

Hello, thanx for the response.

The noise is from an variable frequency oscillator circuit on the same board, frequency range is variable from 1 hz to 10khz. The noise on the supply lines is from this oscillator circuit. The rest of the circuits on the board are fine. The supply lines are bypassed at each op-amp on the board. The supply lines are regulated well. I believe the problem is actually because of the circuit now that I think about it... I am using the LM358 to buffer a +10V Zener reference. The LM358 is running from +/-12V supply, so the +10V reference is probably saturating the LM358. If I set the zener ref to 8V, I no longer have the power supply rejection problems, the buffered 8V ref is clean enough. I suppose I might benefit from a Rail to Rail Op-Amp?

T

Have you thought about using a Linear Regulator to filter the noise coming from your power supply?

Most operational amplifiers spec PSRR at DC.

PSRR over frequency, in most cases, is only simulated or omitted because most of the time opamps can't effectively filter noise coming from the power line. What many signal chain customers had used in this cases is a high PSRR/ low Noise, high voltage LDO like TPS7A4901DGN (positive voltage) and TPS7A3001DGN (negative voltage).

Well, to start one possible culprit for this noise is the zener itself. Although we think of them as nearly ideal voltage sources (zero DC resistance) they do have a resistance associated with them, typically called the zener resistance. This resistance can be surprisingly high for higher voltage zeners and is also dependent upon the bias current of the zener. For instance, assuming the zener is being reverse biased by a 1k resistor to the positive supply, and has a zener resistance of 10 ohms, then the noise from the power supply would be reduced by 10/(1k+10) = .0099 or -40dB as opposed to the >60dB reduction in noise from the opamps power supplies. For lower bias resistor values this will be worse. Two possible solutions are to place a capacitor in parallel with your zener and, even better, bias your zener with a current source.

I would disagree that PSRR over frequency is omitted from datasheets because opamps can't filter noise on their power lines. Most new opamps feature this curve in the datasheets as it is considered a standard datasheet characterization, however this hasn't always been true and older parts may neglect the data.

Yeah, I've considered using a Linear regulator, but since I need +/- Ref voltage at only about 5mA, I figured I'd try using an op-amp and zener, that way I get both ref. voltages, buffered at a cost that is less then half that of just the TPS7A4901DGN (I'd need the TPS7A3001DGN). Not to mention I already have the LM358 in the BOM, so it is fewer parts to source.

Thanx for the tips though, I will keep those TPS7A3001DGN parts in mind for future designs.

T

John, the Zener Ref. circuit looking into the non-inverting pin of the op-amp buffer is OK. The Supply noise is sufficiently filtered out the RC circuit formed by the current limiting resistor feeding the zener and a 1uF and .1uF cap in parallel with the zener.

Check the output voltage swing from the datasheet, LM358 cannot swing close enough to the positive rail to 

buffer +10V from a +12V power supply. The datasheet states that you can get +26V (to 2kOhm load) from +30V supply, so

which about matches getting 8V from 12V supply.

Use a rail-to-rail opamp or an amplifier that can go closer to the positive rail.

Anthony,

Before you try a different opamp, you may want to consider using a lower voltage zener diode and applying gain using the opamp. I believe what is happening is that the high input voltage (in relation to the positive supply) is not allowing sufficient voltage drop across the internal current sources of the opamp to allow them to properly reject power supply noise. Using a lower voltage zener (e.g. 5.6V) and configuring the opamp for a gain of 1.79 will still give a 10V output, but may allow the opamp to provide better PSRR and CMRR.

Definitely something to try before switching to a more expensive opamp!