I'm working on designing an op-amp based headphone amp, and I'm currently working on building the power management section of the circuit. I'm using a +/- 12V switching split power supply, which is then fed into +/- LDOs - LT3080 and LT3015. I can see on my oscilloscope that both the pre- and post- regulated power signals have high frequency ripple, about 75mV peak-to-peak @ 100Mhz. This makes sense since the LDOs have very poor ripple rejection at such a high frequency.
My question is, is this signal worth filtering? In my mind, such a high frequency signal is way out of the audible range and thus shouldn't affect the output of the op-amps in any audible way. However, I admit that I'm a bit of a novice when it comes to the power supply side of things, so I could be overlooking something obvious. Thanks in advance!
My question is, is this signal worth filtering? In my mind, such a high frequency signal is way out of the audible range and thus shouldn't affect the output of the op-amps in any audible way. However, I admit that I'm a bit of a novice when it comes to the power supply side of things, so I could be overlooking something obvious. Thanks in advance!
Last edited:
You want to do a graceful dance.
While somebody runs a jack-hammer under the stage.
Even though the hammer-rate is faster than a dance, it can upset the dance and reduce grace.
While somebody runs a jack-hammer under the stage.
Even though the hammer-rate is faster than a dance, it can upset the dance and reduce grace.
Thank you for the analogy. Out of curiosity, I implemented a simple RC low-pass with a very low cutoff (10Hz) to try to filter the noise. To my surprise, the 100 MHz ripple I was measuring remained unchanged. I'm starting to become skeptical that the ripple is from supply, especially since reading that most switching supplies have switching frequencies in the KHz to ~5 Mhz range. Unfortunately, the spec sheet of the one I'm using doesn't include the frequency. Perhaps the ripple I'm seeing is from the scope itself or some other source.
Thanks for the help! I'll keep investigating.
Thanks for the help! I'll keep investigating.
You want to do a graceful dance.
While switching noise has you in its trance
Unexpected conduction on PN conjunction
Can cause quite an unexpected eruction.
With apologies to Dr. Seuss who has been dis-platformed.
Thank you for the analogy. Out of curiosity, I implemented a simple RC low-pass with a very low cutoff (10Hz) to try to filter the noise. To my surprise, the 100 MHz ripple I was measuring remained unchanged. I'm starting to become skeptical that the ripple is from supply, especially since reading that most switching supplies have switching frequencies in the KHz to ~5 Mhz range. Unfortunately, the spec sheet of the one I'm using doesn't include the frequency. Perhaps the ripple I'm seeing is from the scope itself or some other source.
Thanks for the help! I'll keep investigating.
Capacitors behave differently at different frequencies.
Have a look at a higher order LPF or an active LPF. You could even put an LC in - two inductors and three capacitors as a higher order butterworth LPF and it would remove the noise.
I dont know - one guy many years ago told me of an experiment where he deliberately injected signal into the +/- rails of an op-amp - and found the amp soundly rejected everything but the most ridiculous levels, such as any normal power supply would never present. This was back when people at DEC were discussing the sonic benefits of shunt regulators over series - probably the mid 80s...
Nothings to say you cant go beyond reason and make the effort to filter 75mV of 100Mhz noise, regarding its possible influence on the passage of kHz signals... Keeping circuits tidy given what you've observed is always best practice in DIY - when cost adders are "who cares" compared to a commercial implementation.
Nothings to say you cant go beyond reason and make the effort to filter 75mV of 100Mhz noise, regarding its possible influence on the passage of kHz signals... Keeping circuits tidy given what you've observed is always best practice in DIY - when cost adders are "who cares" compared to a commercial implementation.
...RC low-pass with a very low cutoff (10Hz) to try to filter the noise. To my surprise, the 100 MHz ripple I was measuring remained unchanged.....
Real filters do not slope-off to infinity. 10Hz suggests a monster capacitor. It may go below tenths-Ohm at 10kHz and rise back up in MHz just because the two legs are so far apart.
However I agree that 100MHz is probably not a "power" frequency but stray reception. It is one reason I prefer 0.45MHz 'scopes to 15MHz 'scopes.
Are there any FM broadcast transmitters nearby? Do you still see the ripple when the supply is off?
Could it be common mode noise?
This is definitely possible, especially if it were caused by EMI. See below.
Are there any FM broadcast transmitters nearby? Do you still see the ripple when the supply is off?
I do still see the ripple even when the supply is completely disconnected, though the peak-to-peak voltage is lower. And well, I just searched on google maps for radio stations. Turns out there is an FM broadcast station literally 500 feet (150m) away from my apartment.

Last edited:
& how are you measuring: at 100MHz - exactly how are probing with a 'scope matters , deeply. What are 'seeing' on an o'scope at 100Mhz might just be down to the length of the probe 0v lead; useful results in that sort of bandwidth are way out of accurate range of a cheapie 'probe' with the typical 0v croc clip on a short loop - that loop-area has long become a problem..!
My go-to reference, my first & always recommendation as similar to other becasue I have learned so much from it, is the utterly-wonderful Linear Tech An-47 written by the late Jim Williams. This is a definitive compendium not only on implementing good analogue design; but also the appendices detail how to know you are measuring the right thing.
Here is a link, with informed appreciation:
Reading Jim Williams: Best App Notes
My go-to reference, my first & always recommendation as similar to other becasue I have learned so much from it, is the utterly-wonderful Linear Tech An-47 written by the late Jim Williams. This is a definitive compendium not only on implementing good analogue design; but also the appendices detail how to know you are measuring the right thing.
Here is a link, with informed appreciation:
Reading Jim Williams: Best App Notes
Last edited:
I would highly suggest Jim Williams as well, AN101F covers switching noise in linear post regulators. and shows that ferrite beads are highly effective
& how are you measuring: at 100MHz - exactly how are probing with a 'scope matters , deeply. What are 'seeing' on an o'scope at 100Mhz might just be down to the length of the probe 0v lead; useful results in that sort of bandwidth are way out of accurate range of a cheapie 'probe' with the typical 0v croc clip on a short loop - that loop-area has long become a problem..!
This. Forget about the damn probe lead, it's evil. Some oscilloscope manufacturers have Youtube videos on proper probing in addition to written guides which are available from many sources. Dave Jones from the EEVblog also made many videos on the topic. I'll attach a few examples.
EEVblog #594 - How To Measure Power Supply Ripple & Noise
EEVblog #1266 - PSU Probing Screw Up!
EEVblog #1367 - 5 Types of Oscilloscope Passive Probes COMPARED
- Home
- Amplifiers
- Power Supplies
- Is High Frequency Switching Noise Worth Filtering?