Careful on the input

May 8, 2021
 by Paul McGowan

One of the ways we designers make good sounding amplifiers is to lightly limit the input frequency while at the same time extending its high-frequency response.

That's something that might seem counterintuitive but it works.

For example, at the input of a power amplifier, I like to form a light low pass filter of around 30kHz but within the amplifier's circuitry, extend its bandwidth to as high as is practical—hopefully somewhere close to 100kHz.

This combination of limiting what the amp has to deal with while making sure what does come in is easily handled makes for a wonderfully open and easy presentation of music.

I like to think of it as a car with more power than it needs, and then a light foot on the accelerator pedal.

Easy in so the amp never breaks a sweat.

Subscribe to Paul's Posts

21 comments on “Careful on the input”

  1. Careful on the input Eugene 😉

    Since 'reply notifications' are not currently functioning on this site,
    I figured that this would be a good way (the only way) to give you
    some feedback about your 11:11pm - "Let's recap" reply a couple
    of days ago ("Sacred cows" May 6, 2021)

    Well thought out, well pointed out & thanks for posting!

    1. I see the Pink Floyd influence continues to reach out into unexpected corners. Sometimes I only have to see the word ‘careful’ and it’s done.

  2. I’m certainly not going to question or criticize an amplifiers design considerations. Though I guess as a consumer I will sit in judgment when choosing one.

    I kind of understand the thought process when it comes to limited bandwidth in and then high bandwidth the rest of the way thru the amps stages. I would ask, why not a 40KHz filter? (Just because I like multiples of 2)

    As far as breaking a sweat, to me that implies driving a load with no added distortion and 100% faithfulness to the input signal. So depending on how tough my load is, determines what amp designs I sit in judgement of.

  3. It’s always interesting that also in amp design and probably everywhere good sound is ensured by tricks and knowhow that sometimes contradicts a simple logic puristic approach (in=out etc.), even if here we talk of above 30kHz.

    It’s also interesting to see, that, as appropriate, we partly tend to argument for the theoretical/puristic and partly for the “as long as it sounds better/more real, let’s do what helps, even if not puristic and we don’t exactly know why it does” approach. As long as the whole chain from media production to playback is not perfect, I think the latter approach is more constructive.

  4. Someone (jimk?) pointed out the other day that all analogies can be torn apart so perhaps I’m being unfair but anyway. I get the bit about the car having more power than it needs but a light foot on the accelerator pedal? Surely you get the most dynamic performance when you boot it.

  5. Interesting, as that's the precise opposite approach to that Spectral Audio takes, which is why you HAVE to use their cables on input and output.

    For example, their DMA-300 amplifier is speced at:

    ±0.1 dB, DC-150 KHz
    ±1 dB, DC-1 MHz
    ±3 dB, DC-1.8 MHz

    1. Due to the high-speed megahertz bandwidth capability of a Spectral amplifier the network cable acts as a low pass filter minimizing extreme rf from entering the amplifier which could potentially cause oscillation and take out an amplifier module.

      This is one application where Cardas Cables may not be the best interface choice.

  6. I will never argue with an engineer who knows miles more than I do about designing a great sounding amplifier. I will leave that argument between engineers and my ears.

  7. I find this shocking to say the least. We know that leading edges of music signals such as drum strikes, cymbal strikes, finger snaps have frequency content that goes out to 96 KHz or higher. Limiting what goes into a power amp to 30 KHz is going to make all those leading edges less sharp. For someone who expounds the virtues of DSD this seems like a very contradictory position.

  8. As I am not an audio engineer, perhaps you can further enlighten me on what a light low pass filter of around 30kHz actually does.

    1. Input signals over 30kHz to 40kHz, some might take it as high as 100kHz, eliminates ultrasonics that can upset an amplifier's delicate balance and cause audible harm.

      Not everyone agrees with me and I don't always follow this rule either - because I do this on an amplifier design by ear. Some rolloffs I have used in the past were much higher—100kHz or so, while other circuits benefited from lower ultrasonic protection. One thing of note is the slope of the filter. It really needs to be a single slope -6dB/Octave so that wherever you start you don't add any phase distortion.

      While we cannot hear anything over 20kHz (and most of us cannot hear even that) we are extremely sensitive to phase distortion. So this is why it is critical to make sure any filtering happens gently and very much out of range of any audible phase shift.

  9. The reason for using high bandwidth is to minimize distortion. Reproducing high frequencies is only a side effect.

    Carefully selected, great sounding low-pass filters are an incredible tool in today's RFI wilderness.

  10. I too am not understanding the post. Not being an electrical engineer, I'm probably way off base when I guess that the low pass filter gets rid of a lot of ultra-high frequency noise, sorta like what a tube with its limited high frequency output does. When in the amplification circuit you then bump some of the under 30kHz up to around 100kHz, it is artificial, not high frequencies that were in the original input signal. The ears and brain sense those artificial high frequencies, for whatever reason, as they interact with the lower audible frequencies and interpret the composite sound as having air and space, or "openness." Does this frequency boost follow a harmonic structure? Otherwise, wouldn't it just be reintroduction of noise?

  11. Perhaps the key term here is "light". Finesse, moderation if you will, can often accomplish things in a more effective, even graceful, manner than a heavy handed. brute force approach. Sometimes brute force actually is necessary, but then don't be surprised when collateral damage also happens.

  12. Great engineers allow an an amplifier to wear swimming trunks without ever having to jump in the water. 🙂

    Also Gentlemen and possibly ladies. Am I inept at using the new PS audio website where by I cannot find the “subscribe to updates in comments and notifications “ for these threads? Are people just leaving the browser window open to check if anyone had commented or replied to your thoughts?

    I kind of miss the old website for that functionality.

      1. Oh okay. Great to know, Paul. It is a nice feature and I think allows for more creative banter in the threads. It will also enable me to bug you more. I kid of course. 🙂
        Happy birthday today. 😉

  13. I wonder how much better the sound would be if the the input was not limited. Would it not be better to come up with an input that does not need limiting? With the claims made by manufacturers it would not be asking for too much. Regards.

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram