Jump to content

Double Stack ESS AMT-1 with Wings--Possible Kit for Heritage


Chris A

Recommended Posts

Just a comment on this subject:  the objective of the DSP crossover configurations that I use is to flatten the phase across the entire audible band.  If I can do that with first order crossover filters and/or shelf filters, then I do it.  So there isn't a right/wrong approach to doing it--it's the results in terms of flattened phase response that count.

 

One other thing I should mention: most crossover filter designs for loudspeakers don't pay attention to phase response--especially the "cookbook" methods found in all books on the subject of DIY loudspeakers.  This subject--flattening the phase response of loudspeakers--is what I believe what makes loudspeaker designs like the Danley Synergy series, Dunlavy Audio Labs/Duntech loudspeakers, the much earlier B&O "phase link" designs, and even full-range drivers sound so realistic in my experience, as well as most studio monitors that are sold today for higher-end studios that use active studio monitors employing DSP crossovers with FIR filtering.

 

I couldn't believe what I was hearing when I first applied the "zero phase" crossover filters to the Jubilees using PEQs only.  I was stunned, and the experience ranks as probably the most revealing discovery for me about loudspeakers (and not in any way to downplay the full-range directivity and low modulation distortion of horn-loaded loudspeakers like Roy and PWK designed). 

 

It turns out that there are very good reasons why flat phase or even linear phase loudspeakers sound so lifelike and realistic (i.e., subconscious sound quality), and the reason is related to how humans hear the harmonics of musical instruments and voices, and even spoken word--even more strongly.  I recommend reading Dave Griesinger's presentation on clarity (look closely at slides 12-19, and particularly slides 17-18). This simple explanation is missed by just about everyone in the audio engineering world, it seems. 

 

So for the purpose of this thread and others where I talk about flattening phase--it's the flat phase response that's important, and not so much how that is achieved.  All I did was apply what other smart guys have been saying for some time to DSP crossover configurations.  The results are, in my experience, startling.

 

Chris

Link to comment
Share on other sites

@Chris A, thank you for the continuing explanations.  In your other thread on the subject, you gave us this method of setting the PEQs.

 

"So all you have to do is put the HF and LF drivers together without phase shifts:

  1. Don't use the "crossover filters" that come with DSP crossovers--clear any crossover filters if they're set.
  2. Set the HF or LF channel delay to get perfect impulse response in the time domain--as seen in the spectrogram view.
  3. Flatten each driver's SPL response within their pass bands.
  4. Match the channel gains between flattened phase drivers.
  5. Use output channel PEQs to trim off response on each end of the bass and high frequency drivers until you've got overall flat SPL across the crossover interference band and smooth handover of SPL vs. frequency.  The drivers themselves will tell you where that transition/crossover should occur.  [If you're using MEHs, you'll have to use multiple PEQs to attenuate the bass bin peaks in response above the first notch frequency.]
  6. Use the input channel PEQs to further flatten the overall response within the interference band to correct any dips or peaks in response within that band.

Voila!  Flat phase.  It's really that easy."

 

It seems simple enough for the smart guys.  In my case, my question concerns how many PEQ filters to use and what attenuation is appropriate.  Looking at the settings on @Thaddeus Smith post, the attenuation seems rather modest for the Heils. -6dB @ 640hz.  My question is, what does the Heil diaphragm see when it gets say a 50hz signal?  I guess that's the part I'm having difficulty understanding.  I always understood that part of the function of a xover was to protect those drivers that could not handle certain frequency ranges....such as a tweeter getting an LF signal. 

 

Sorry to be so dense guys.

 

 

Link to comment
Share on other sites

18 minutes ago, Rudy81 said:

My question is, what does the Heil diaphragm see when it gets say a 50hz signal?  I guess that's the part I'm having difficulty understanding.

The AMT-1 is a bandpass device--it doesn't respond to 50 Hz, nor does it respond to 40 kHz (if your system could do that).  That's what Tom Danley was saying about the drivers in his Synergy designs--just notch out the near bandpass response that's not needed.  That's also been my experience. 

 

My one failure of an AMT-1 diaphragm was due to manufacturing quality control problems, and I wasn't the only one that this happened to, apparently.  It failed at less than 2W input, and it's rated for 40W continuous, 160W peak.  That's 1/20th the rated power due to quality control issues that existed late last year and earlier this year (that they hopefully have gotten under control since then), not the basic design of the driver which is rated for much more power input without any high pass filtering required (and none is specified in their documentation).  All the experience with older versions of the AMT drivers say that they are quite rugged and last indefinitely--even drivers from the early 1970s (almost 50 years ago). 

 

Now if you're trying to drive your AMT-1s using 100 or 1000 watts/channel to drive them, then you've created a problem.  In that case, I'd simply turn on the limiting feature on the DSP crossover--because the DSP crossover cannot only catch all the peaks above rated power of the AMTs, it also catches short term total power surges over time (like infrasonic issues that you're worried about).  This is the way to protect your drivers--not by using shelving filters or blocking capacitors. 

 

Chris

Link to comment
Share on other sites

@Chris A, thank you.  Starting to make sense now.  It just takes time to understand new information that seems to contradict everything you thought you understood to be correct.  I will play around with the settings this afternoon.

Link to comment
Share on other sites

44 minutes ago, Rudy81 said:

It just takes time to understand new information that seems to contradict everything you thought you understood to be correct.

Exactly.  I haven't lost any drivers in operation, only that one AMT-1 diaphragm that had issues earlier this year.  I think audio enthusiasts confuse home hi-fi operation with PA operation, and proceed to destroy their drivers (as well as the dramatically lower reliability of tubes--that tend to fail with DC on the output--which really do need blocking capacitors to guard against that event.  I've never experienced DC output failure mode on a SS amplifier--or any SS amplifier failure, for that matter.  It's always the drying out electrolytic capacitors that become issues with SS amplifiers after decades of aging.)

 

I think that the issue with too much amplifier power is actually something that can easily be accommodated using the DSP crossover.  It isn't really available when using passive crossover filters, except perhaps zener diodes which were a staple in PWK's crossover designs during the time when "rock-roll" enthusiasts first discovered Klipsch loudspeakers and started to blow K-77 tweeters using amplifiers like Phase Linear, etc. from the late 1960s-early 1970s (i.e., Bob Carver's ultra-high-output SS amplifiers).  The K-77 tweeter (nee EV T-35s), can handle only 4W input power (music power).  If instead those same "rock-roll" enthusiasts were using DSP crossovers then, it's merely a matter of a few minutes to calculate the settings for the DSP crossover channels to protect the tweeters, etc...and then seconds to dial it into the crossover settings.

 

11 minutes ago, Thaddeus Smith said:

Thanks for indulging us Chris. It's good to have these mini brainstorm/learning sessions in the midst of broader discussions.

These are just as important as the discussions on the wing assembly dimensions and manufacturing, IMO.  Understanding the DSP settings and the way that they're set up ultimately determines the sound quality of the resulting loudspeakers, in my experience.

 

Chris

Link to comment
Share on other sites

Once you get to ~100w/channel, the benefits of higher power amplifiers basically disappear for high efficiency loudspeakers: https://community.klipsch.com/index.php?/topic/170074-crown-xli1500-a-modern-day-staple/page/5/&tab=comments#comment-2501529

 

image.png

 

I can easily understand higher power for subwoofer designs (which tend to be much lower efficiency due to the frequencies involved and the owner desiring a smaller subwoofer package for smaller listening rooms).  But for the vast majority of Klipsch designs it's basically a double-edged sword: you've got to protect the drivers in the loudspeakers at some level of amplifier output power from the (idiot) party guest that turns the volume control all the way up while the owner is not in the room--something that's not necessary to protect against with amplifiers having perhaps 40w/channel output capability or less. 

 

Chris

Link to comment
Share on other sites

1 hour ago, Rudy81 said:

"So all you have to do is put the HF and LF drivers together without phase shifts:

  1. Don't use the "crossover filters" that come with DSP crossovers--clear any crossover filters if they're set.
  2. Set the HF or LF channel delay to get perfect impulse response in the time domain--as seen in the spectrogram view.
  3. Flatten each driver's SPL response within their pass bands.
  4. Match the channel gains between flattened phase drivers.
  5. Use output channel PEQs to trim off response on each end of the bass and high frequency drivers until you've got overall flat SPL across the crossover interference band and smooth handover of SPL vs. frequency.  The drivers themselves will tell you where that transition/crossover should occur.  [If you're using MEHs, you'll have to use multiple PEQs to attenuate the bass bin peaks in response above the first notch frequency.]
  6. Use the input channel PEQs to further flatten the overall response within the interference band to correct any dips or peaks in response within that band.

Voila!  Flat phase.  It's really that easy."

 

 

@Chris A I've been doing #2 after #5. How would you set delay correctly before implementing the no name crossover in #5? I must be missing part of the concept.....not the first time. I thought it was basically

  • Flatten Highs
  • Flatten Lows
  • Gain Match
  • Cross where the drivers tell you to be closely matched in phase and use PEQs to do that high/low crossing
  • Use group delay and spectrogram to set the delay (adjust polarity as necessary). This is still the most difficult for me. I spend many hours messing with spectrogram. Any additional pointers on how to attack that in a more systematic way would be greatly apprciated
  • Flatten more with input PEQs
  • Listen and wonder why you didn't make the jump to Active DSP sooner
Link to comment
Share on other sites

Using a spectrogram view, you can see if the driver/horn is leading or lagging the other crossing driver/horn.  But point taken. 

 

I usually do an initial delay setting before trimming the frequency response, and I also flatten the response of individual drivers/horns before anything else (because it saves a lot of time).  After getting the combined output flattened, I go back and fine tune the delay settings, then fine tune the PEQs for flatter SPL response (I try to achieve ±1.5 dB flatness of the loudspeaker with the measurements taken at 1 m in front on-axis. 

 

I think the list of actions is notional, i.e., use whatever method works best for you.  If you're trying to dial in the loudspeakers via receiving emailed REW measurements and emailing back DSP crossover presets files (like I find myself doing often), the exact sequence of events is probably different than if you're doing it in your own listening room with everything hooked up to your PC/laptop, DSP crossover, and calibrated microphone.  In the second case, I'd probably do things in a different order and to a different set of tolerances in each step in order to blow through perhaps more steps in the round-robin series of events, but resulting in fewer PEQs used and flatter SPL response, and better time alignments, etc.

 

Chris

Link to comment
Share on other sites

18 minutes ago, Chris A said:

Using a spectrogram view, you can see if the driver/horn is leading or lagging the other crossing driver/horn. 

 

By the dashed line jumping left or right abruptly, not having a smooth curve to the right and down, correct?

 

 

19 minutes ago, Chris A said:

if you're doing it in your own listening room with everything hooked up to your PC/laptop, DSP crossover, and calibrated microphone.  In the second case, I'd probably do things in a different order and to a different set of tolerances in each step in order to blow through perhaps more steps in the round-robin series of events, but resulting in fewer PEQs used and flatter SPL response, and better time alignments, etc.

 

As I am hardly good enough at this to help myself let alone anyone else, I find myself doing just that....round robin iterations. PEQs, XOs, Delay, repeat, repeat, repeat to arrive at the flattest phase and least amount of PEQs. Lately I've had a hard time matching up what the spectrogram and Group Delay are telling me. So if I end up with GDs below 2ms I just let that be and use the Spectrogram exclusively to set the delay. Is that a reasonable approach?

Link to comment
Share on other sites

8 minutes ago, rplace said:

By the dashed line jumping left or right abruptly, not having a smooth curve to the right and down, correct?

Yes.

 

10 minutes ago, rplace said:

Lately I've had a hard time matching up what the spectrogram and Group Delay are telling me.

Me, too.  I usually choose the excess group delay curve to correct to and not the spectrogram's peak energy time curve.  I find the excess group delay curve tells me more of what I need to know from a delay setting standpoint. 

 

The other issues with the peak energy time curve not being smooth and vertical are usually associated with the drivers/horns themselves.  If I had FIR filtering capability implemented, I would try to correct those.  I need to acquire a more powerful PC that can run JRiver upstream of the preamp to hear what can be done to the AMT-1s at their lower frequency end--to straighten out the peak energy time curve a lot more (i.e., the phase growth of the AMT-1s at their lower frequency end).  So many things to do, so little time...and I don't have an 8-to-5 job.  ;)

 

Chris

Link to comment
Share on other sites

50 minutes ago, Chris A said:

I find the excess group delay curve tells me more of what I need to know from a delay setting standpoint. 

 

Remind me again how to do that, please? I seem to remember when I was using an ordered/named crossover you looked at that value say 400Hz for example, and placed your cursor on one of the curves to read the amount of milliseconds it was off. Not having a High Pass and Low Pass at an exact value I don't recall how to do it.

 

Thanks!

Link to comment
Share on other sites

Here are the REW results for my current setup.  I have a 1ms delay on the AMT stack. Some EQ on the open baffle H frames.  I am ignoring the HF rolloff above 10kHz since that is due to mic placement.  I can get relatively flat response in the HF, but the mic then is set very high off the floor relative to the bass OB. 

 

The important thing to note relative to the current discussion is the 'crossover' region.  I created that region using a combination low shelf and PEQ for the HF and a high shelf and PEQ for the LF.  Overall, the results sound excellent, but I want to get closer to @Chris A and his technique of avoiding the use of the shelf and go with just PEQ.

 

 

Original SPL.jpg

Original GD.jpg

Original Spectrogram.jpg

Link to comment
Share on other sites

Group delay plot.jpg

 

The items in yellow are the ones that I added to your chart in the "Using REW to Determine Time Delays Between Drivers" thread: https://community.klipsch.com/index.php?/topic/182892-using-rew-to-determine-time-delays-between-drivers/page/3/&tab=comments#comment-2418399

 

Basically, you're trying to move the excess group delay values below the crossover frequency back to zero ms (vertical scale) by delaying the higher frequency channel.  Most excess group delay curves don't look like the one in your figure, i.e., they are flatter and don't increase in excess group delay so quickly below the center crossover frequency.

 

Chris

  • Like 1
Link to comment
Share on other sites

@Chris A, so far, all my testing shows virtually no difference in GD, spectrogram and SPL between using two PEQ points and using a shelf and one PEQ.  I expected to see some big change in the phase profile, but not so.  The phase continues to look as it does in the plots I posted above.  Seems to me that using the shelf has little or no effect on the overall speaker response. 

 

Timing differences, on the other hand, make a HUGE difference in the GD if you go too far away from ideal. 

Link to comment
Share on other sites

Look at the phase plots closely, looking for relative phase growth of different DSP filter configurations.  Phase growth is just as bad as GD growth.

Link to comment
Share on other sites

2 hours ago, Chris A said:

Look at the phase plots closely, looking for relative phase growth of different DSP filter configurations.  Phase growth is just as bad as GD growth.

 

Unless I am doing something incorrectly, I don't see much in the way of phase changes.  This plot is the original setup with a shelf and  single EQ, vs. various timing changes....all with the two PEQ method and NO shelf. I don't see much in the way of phase change.

 

 

Phase changes.jpg

Link to comment
Share on other sites

Very interesting discussion. 

 

Sorry to beat a dead horse - but I want to go back to @Rudy81 question about what does the AMT-1 see at say 50hz? 

 

Some of my questions may seem elementary because I haven't even done much with active systems / PEQ. 

 

My understanding PEQ was that one has the specify the target frequency, gain/trim adjustment, and Q. 

 

If one was to "crossover" the AMT using solely PEQ  say at 600hz, - would one not need multiple PEQs?- say at 600hz, 500hz, 400hz, 300hz, 200hz, - all the way down? One can broaden the impact of the PEQ by lowering the Q, but then you run to risk of impacting frequency about the target crossover region. 

 

What am I missing? 

Link to comment
Share on other sites

14 hours ago, Rudy81 said:

 

Unless I am doing something incorrectly, I don't see much in the way of phase changes.  This plot is the original setup with a shelf and  single EQ, vs. various timing changes....all with the two PEQ method and NO shelf. I don't see much in the way of phase change.

 

 

Phase changes.jpg

 

 

I'm also curious about the same question @Rudy81 you pose above. My understanding was that phase was dependent on the frequency response. Whether you use PEQ or crossover to achieve a target curve  - your phase response would be identical.    <---- This is what I always assumed. 

Link to comment
Share on other sites

I have now spent the better part of a day working strictly on optimizing the crossover region without using 'named' xovers, while monitoring effects on phase and GD.  So far, I cannot detect much (if any) of a difference in using just PEQ vs. using a single PEQ and a shelf.  In my trials I find that using the shelf and a single PEQ makes setting the xover region much easier.  For the time being, I will leave it as is.  I could not better the spectrogram or reduce phase growth using just the PEQ technique. I did manage to optimize the overall EQ settings to get the entire SPL within +/- 2 db, which in a room is pretty good IMHO. 

 

Hopefully I am not missing some major point that would vastly improve performance. 

 

I will note that when working with my system, an AMT stack and dual 18" woofers, mic placement takes a lot of trial and error.  A couple of inches movement up or down at 3 feet can cause all sorts of anomalies in the GD plot.  I am guessing there is some comb type filtering going on near the stack, causing the plot anomalies.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...