About

Since the publication of my book, a number of people have requested me to create a place where we can carry on a discussion specifically related to the topics in that text and where they can directly address issues and questions to me. In addition to those specific topics in the book, my intention here is to set up a forum where people can talk about a range of issues generally related to design and optimization of sound reinforcement systems. I will have the ability to add features to this blog site, and so I am viewing this as the beginning of a sort of experiment. Naturally, the input of others in the field is welcome in regards to any of these topics. But, at least for now, we have set up this blog in a format in which I will write comments/articles, etc. from time to time and people can come in and directly comment on these or they can comment on any of the topics in  Sound Systems: Optimization & Design. For now, if you want to originate a topic, bring up a particular issue or some such thing, you can email me through the blog. I will try to respond on the blog, assuming it is general enough in subject matter as to be interesting to the professional audio public. If you want a question answered privately (i.e. you do not have a matter of general interest to discuss) you can contact me through my web site www.bobmccarthy.com, otherwise, we’ll keep it here.

16 Comments on “About”

  1. Kip Conner Says:

    Bob- love your book, thanks for taking the time to address many of the mythological creatures in the mysterious world of sound reinforcement. I like to consider myself a self taught student of the Amplitudic Arts, so when something comes up that I don’t understand I research in an unhealthy manner to find the answer. In this case I could not, so it would appear to be a lack on published knowledge.

    My question is this:

    What are group delays and specifically how do they pertain to the re-action of low frequency propagation. The thought returned to me after watching your great video on Time=Phase/Frequency.

    Are the phase wraps that occur over time the group delay that people speak?

    For examsple, if a -90º phase shift occurs from 50Hz to 150Hz thus resulting in 2.5ms of delay. Does this delay and of it’s subsequent wraps resulting in delay (until time equals a 0º phase shift) define our group delay?

    And/Or- (regardless of my misunderstanding on the topic)- Would you say that a system has a definable group delay. For instance, after doing a phase alignment at the crossover points of a 2 way speaker, thus producing more phase wraps via delay.

    thanks again for your contributions, it’s my hope to someday be in the financial position to attend one of your courses. I’ve been using the SIA product for years, but it seems the information is the same across the platforms.

    kip conner
    atlanta, ga

  2. 6o6 Says:

    Kip,
    Thanks for commenting. The term “group delay” is not one that I have ever used in print or in my seminars. My feeling about the term is that it’s definition seems too fluid for me to put to practical application. I have never been able to see something come on to my analyzer screen that says “I am group delay – distinct and separate from other forms of delay.” If I could ID it, I would search for solutions – since I can’t I don’t.
    There are, however two verifiable and distinct forms of signal transmission delay: frequency independent, and frequency dependent. These can both be termed “phase delay”

    FREQUENCY INDEPENDENT PHASE DELAY:
    The freq indepedent version is easy to characterize in time: X.X milliseconds, microseconds, nanoseconds, etc. The amount of time is fixed over frequency, and therefore the amount of phase shift is variable over frequency. e.g. 1.0ms of frequency independent delay creates 3600 degrees of at 10kHz, 360 @ 1kHz and 36 deg at 100 Hz. Constant time over freq creates a linearly based constant change of phase. As for the wraparaounds you speak of – the 10 kHz response would show 10 wraps, the 1 kHz would have 1 and the 100Hz 0.1 wraps. Wraparounds are simply the result of our display device’s limitations. The phase location (degrees) on the screen is like the minute hand of your watch, the wraparounds are like the hour hand. Taken together we can tell time.

    FREQUENCY DEPENDENT PHASE DELAY:
    The converse extreme is something that creates a constant amount of phase shift – such as 90 deg at all frequencies. This would be require a different amount of phase delay at every frequency. (2.5 ms at 100 Hz as cited in your example above) .25 ms at 1kHz etc. This is not something you will run into much in practical applcations. Common instances of freq dependent delay include transducer responses over frequency, filter functions, physical displacement between HF/LF drivers etc. In such cases the phase delay is defined by the number of wraps at the given freq range (as per your question above).

    Your next question: Does a system have a definable group delay? — I would answer that a system has definable phase delay, both types. A speaker I measure might have 23 ms of freq independent phase delay (the propagation time to the mic) and then we begin to parse out where the frequency dependent delays fall out. The 100 Hz response might be 5ms behind the 2kHz response (very typical) and the range below 100 will have more.
    As for phase aligning the 2-way XOVR – what we seek here is a matching of the phase delay of the two elements. If we can bring them to the same phase delay, they will give maximum addition. As a bonus, if we can mimimize the amount of phase delay over the whole spectrum, the more linear the system will be – and the better will be our transient response. Since lows (in transducers) tend to fall behind the highs, we will need freq depenedent delay in the highs to match the lows in time/

    I hope this helps and look forward to hearing more from you

    6o6

    • Kip Conner Says:

      Your last sentence-

      “Since lows (in transducers) tend to fall behind the highs, we will need freq dependent delay in the highs to match the lows in time”

      This is the area where I hear people discussing this idea as being a “group delay.” The idea that waveforms are longer and thus take longer to develop and that the size of the driver itself is completely inefficient as compared to a smaller driver producing a higher cycles. I’ve never read too much into this idea because all drivers are subject to transient movement in respect to their pass bands.

      However, I’ve always thought of the low frequency drivers as having “inherit delay” based on their response time (which would better be described as Phase Delay?), yet it doesn’t matter so much to me as long as I can align the phase at the crossover points within a sub-system and then time align to a source point on the whole system, thus reducing cancellations.

      I think that I wanted to place the phase delay in groups, just as you did to display consistent phase shifts (slopes) on the wrapped phase display.

      I’ve been tip-toeing around this stuff for years and the decrease in work opportunities has given me time to buckle down and put all the pieces of the puzzle together. I’m finding that I have been given bad information by my peers- not on purpose, of course. We think that we understand the principles and inadvertently pass along bad information. It can happen in any scholastic arena.

      I know how to utilize delay and the phase window in smaart to align the response between two drivers. It’s now time to get a better understanding of “the why” and less about “the how.”

      Thanks again for your kind offerings.

  3. 6o6 Says:

    Kip,
    Thanks again. A couple of clarifications:
    1) When I said “lows (in transducers)tend to fall behind…” I mean in contrast to lows in a wire (where highs and lows travel at ALMOST the same speed. Why the almost? Because the capacitance of the cable will filter the highest frequencies – the filter effect causes phase shift and where there is phase shift at some freqs there is freq dependent delay. On the other frequency extreme is cable inductance – which filters at the low end and again filtering causes phase shift etc…..
    2) When I refer to lows being slower in a tranducer I mean relative: e.g 800Hz is high to an 15 inch driver and low to a 4 inch compression driver. 800 Hz would be ahead of 100 Hz in the LF driver. 800 Hz would be behind 2kHz in the HF driver. The most important question is whether the phase delay is MATCHED at the point where the two drivers cross over. Before that – lets look at the nature of the delay……
    3) It is a common misunderstanding low frequencies “take longer to to develop” or that low big speakers (like an 18 in driver) cause extra phase shift because they are massive. Think of the following: if your job is to reproduce 50 Hz you need to complete it in 20ms. There is no bonus for completing the job ahead of schedule – if you did it would not be 50 Hz! So there is not an inherent delay in LF devices – but rather in the acoustic radiation- when the wavelength radiated is large compared to the transducer size, phase delay begins to accumulate. 100 Hz is huge compared to a 15 inch driver – so we get lots if delay. 800 Hz – is about the same size – so we get very little delay. 800 Hz is huge compared to the 4in. diaphragm – therefore it falls behind the 2khz response. Once the wavelengths are equal to, or smaller than the transduces size, the delay mostly levels of. therefore a normal tranducer phase delay response is flat over its upper range and then hits a knee around the transducer diameter and increases steadily below that.
    4) The concept that Lows take longer to develop (time or distance is widely misunderstood. Place your mic one inch from your subwoofer and you can disprove this. You will see plenty of 30 Hz – even though it is a 33ft (10m) wavelength. It is delayed for the reasons discussed above, but it does not need distance to develop. – Think about this – have any trouble hearing the subwoofer in your friends car with the super duper rapper stereo? No car has 30 ft of interior length but is can do 30 Hz.

    • Kip Conner Says:

      I’m only going to bother you with one more question and I’ll leave you alone for a few days.

      When we few the phase response of a 2 (or more) way system what the high frequency reverse mean? What are we looking at?

      We get to a point in the frequency response where Phase Delay=0 and then we see it “lead” (is that correct?) for the upper frequencies.

  4. 6o6 Says:

    Firstly, there is really no such thing as “phase delay = 0” in any transmission. This would require a transmission medium with infintie speed. When we measure the response of any audio transmission, there will always be some timing differential from input to output – although it could be insignificantly small.
    For our purposes we establish a “relative time zero” and base our conclusions on that. A typical full-range loudspeaker is spread some 10-20ms over its operating range. Where is time zero? It is where we make it. It we go from the first arrival (e.g. 8 kHz) then all other frequencies are behind. If we go by the middle then the highs are ahead and the lows behind etc. Think of a book written in 1930. World War I is in the past, WWII is in the future.

    Secondly, the phase delay can not be found be a single point. It is found by connecting a sequence of phase values over a frequency range. This creates a line, and the line has a slope. If the slope is horizontal, the phase delay is zero (relative). This is true regardless of whether the flat line is along the zero degrees point on the screen, the 180 deg or any other value. Flat phase line – 0 phase delay (relative).

    If the slope falls (left to right) with frequency there is delay (late). If it rises these frequencies are leading, relative to the flat sloped ranges.

    The trick of a crossoover is to get the slopes of the HF and LF components to lay over each other long enough to get through the shared freq range.

    Hope this helps

    6o6

  5. Kip Conner Says:

    What I was trying to describe is the area on the wrapped phase plot where our wraps to a reversal. I.E. the area where the phase has the “flat response” 4-6kHz.

    So, I was trying to figure out what we are actually seeing after we align the phase at all of the crossover points. What are we seeing that makes the big smiley face in the acoustic transmission?

    It sounds like what you’re saying is that it doesn’t matter that the high frequencies become leading on the phase response. I’m trying to understand what is exactly happening.

    Thanks for your patience with me on this subject, I realize that I’m really buggering the terminology!

    • 6o6 Says:

      A smiley face phase response for a speaker means the speaker’s time response is spread – this is normal, since the very low frequencies always have wavelenghths much larger than the transducers. The LF part of the smile is thus accounted for, but why the upturn on the HF side?
      The reason is that our delayfinder function – the mechanism that chooses how much internal delay we will put inside the analyzer to sync the measurement – has made a choice around the 4-8 kHz range (in the case you refer to) – The upturn in the phase tells us that 16 kHz etc. is slightly AHEAD of the 4-8 kHz response. If we modified the internal delay (reduced it slightly) the 16 kHz response would flatten and the rest of the response would tile downward.
      It is all about perspective. The analyzer can only sync to one timing – but the speaker is spread over time – we have to choose one place to sync to the speaker and therefore other frequency ranges will lead – or lag – from that point.

      No – I would NOT say that it doesn’t matter that the high freq are leading – it most certainly does. A speaker with flat amplitude AND flat phase will have the highest fideility to the input waveform – always our goal. However, for the purposes of aligning an XOVR, it is indeed the shared region – not the outer extremes that is the area of phase interaction interest.

  6. Matt Says:

    Bob,

    “A smiley face phase response for a speaker means the speaker’s time response is spread – this is normal, since the very low frequencies always have wavelengths much larger than the transducers.”

    I have always wondered why this is true. I asked Siegfried Linkwitz why a woofer caused phase shift and he said because it is an electro-acoustic high pass filter.

    Could you explain why the larger wavelengths relative to the transducer size create more phase shift.

    Thanks for your time.

    Matt

  7. Pepe Ferrer Says:

    Hi Bob,

    I would like to introduce a question on angular offset in an array of outfill.
    Your book is very well explained that knowing the rank of each array (PA & Outfill), we can calculate the necessary attenuation level and perform acoustic crossover adjustment.
    Up here all right, but I understand (perhaps I’m wrong) that the speakers although they may have different horizontal coverage angles, level of SPL should be similar.
    I think this because if Outfill speaker has a level below the main system, we no longer should have to change the level of outfill.
    And here begins my question: I use at major events a PA system 80 º horizontal coverage and a system of compatible Outfill in phase with the main system with a coverage of 100 degrees and lower.

    Unit Level 100 +80 / 2 = 90 °, this is my unit for the separation angle between arrays.

    But the question is: if I know my Otfill has a lower level, angular compensation should I?

    Can serve for the calculation of angular compensation, the differences in SPL of my prediccón software half of the distance of each arrangement?

    Thank yoy and regards

  8. Robert Says:

    Hi 606!

    I hope, my question is not too stupid … 😉

    I have a system with 2 tops flying over a proscenium. the lower hanging top is for the near seats, the higher for the back seats.

    So – now i am trying to tune them isolated. Manufacturers gave them a rise in the HF of 10 db. For “flat response” i have to remove this HF rise.

    If i tune both combined, i have to remove the LF build-up.

    This procedure works, but sounds not logical to me – especially the lowering of the HF in the first step.

    Regards!

    • 6o6 Says:

      Robert,
      It is not so easy to give a sure answer to this but I will do my best.
      When I evaluate a single speaker’s HF/LF level ratio I would focus 1st on the Xover area. I would look at this region from a close distance – 1-2m if I can and see about the relative level in the Xover region. It would not be so shocking if I need to turn down the HF 10 dB – especially if we have an HF horn and front-loaded LF. Once the Xover region is settled – in level and phase then we can move on to looking at the response out in the house. Then I would look at the upper solo (it goes 1st since it goes the long distance), then eq that one to flat – or to whatever level of tilt you like. The biggest part of the EQ will be LF cut since the room will favor tilting in favor of the lows. Now you look at the lower box solo and make it match – in level and eq. If it is a shorter throw – then it will be turned down. Also the shorter throw should mean less LF eq – so there will be less cut there. Now add them together – the horns stay separated – so the HF does not change much expect in the transition zone between upper and lower speakers. But the LF will be shared to both systems so expect a rise of several dB there. If the lower box is turned down 3 dB then yuou might see 4 dB of LF boost – which you can then EQ to bring it back to the single box response.
      Maybe it seems like a bit of push and pull but this is very normal (to me at least).
      I hope tho helps. Good luck

  9. Robert Says:

    Hi Bob. Thx for your answer. So you wrote “then eq that one to flat – or to whatever level of tilt you like” – so you suggest to “tone” the system and not to “tune”?

    • 6o6 Says:

      Tuning and toning are not exclusive. Tuning (optimization) is focused on uniformity of response over the space.How well that goal is achieved can be objectively assessed. Toning (the spectral shaping) is a subjective matter. You like apple, I like cherry. A well tuned system can have any tone – it just will have it for everybody. I system that is not well tuned will have different tones in different places, so defining its tone becomes less clear.
      A simple example: Cinema systems are tuned to a standard frequency response known as the “X curve” which rolls of the HF at 1 dB/octave above 1 or 2 kHz (can’t remember which at the moment and I am being lazy). It is NOT a flat response (tone) but it the goal is a uniform response over the room (tuning).

      Bottom line is this: I don’t argue with customers about tone. You want flat, OK you want tilt , OK – I just make sure EVERYBODY gets the same.

      I hope that clarifies

  10. Robert Says:

    What tilt, if you “tone” the system, do you prefer personally?

    • 6o6 Says:

      Rule#1 LF below 100 Hz should NEVER be below the midrange (Flat or boosted below 100 Hz)
      Rule#2 HF above 4 kHz should NEVER be above the midrange (Flat or rolled off above 4 kHz)
      Rule #3 Never forget rules #1 and #2

      The bend point at the LF is 100 Hz or below. The amount which I would allow the LF to tilt up scales with (a) venue size and (b) venue reverberation. If the room is small and dead it will be toned near to flat. If it is large and reverberant then +6 to 10 dB at the bottom is typical.
      On the HF the trend reverses – but it is really only about size – not reverb. Small room we let the HF run out flat. In a big room you can see a lot of HF loss due to the air absorption. You can only make so much of that back before the HF spills too much in the near field. in an arena +6 to 10 (50 Hz) and -6 to 10 (12 kHz) might be typical. At the HF a lot depends on how much loss you have and how much control and headroom you have to push the HF back there.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: