Klipsch R-41M Bookshelf Speaker Review

[QMuse]

Their conclusion "the tweeter's step has begun to decay..." is plain wrong, for the reasons mentioned. The assumption that the speaker's main lobe is tilted downward is most probaly invalid. One simply cannot infer such info from a step reponse, you either need to have vertical polars or individual mag/phase responses of the drivers.

The 600 has higher crossover frequency and, more importantly, a supposedly much better designed crossover with steeper slopes. Crossover parts are a big factor for BOM cost, that's why on the very cheap lines designers try to get away with the lowest part count and the cheapest stuff.

A quick sim with for the 600 (assumed flat to +-0.2dB) with 1.5kHz XO and 5th order (acoustically, as always) on the woofer (ported @50Hz) and 4th order on the tweeter (set back by 100us), both drivers in phase, gives the following step response which is not that far off.
In that sim I actually have some 60° phase offset at XO (not a good idea, but used for the sake of demonstration), the tweeter lagging, which makes the main lobe tilt upwards, not downwards, which illustrates the above point.
View attachment 50949
If you look hard, the initial rise of the woofer can still be identified

IMO response of both drivers looked something like this (midwoofer in red, tweeter in blue):

 

[KSTR]

It actually doesn't.

Main difference is in this case is that both drivers are coonected with positive acoustic polarity so once you invert R-41M step response they would become similar. Time difference between peaks of responses of tweeter and midwoofer is also similar with this graph, app 0.38 ms.

View attachment 50958

What you shouldn't compare is height and depth of those spikes as those are very much related to the # of cycles/octaves of frequency dependent windowing applied when generating graph like this. Same thing applies to comparison of any time-domain graphs (phase, GD, etc) - they all very much depend on applied FDW which makes comparison betweeen graphs very hard as their y-axis values are very much dependent on FDW.

This indeed a source of unknown variations but we can only hope the people doing the measurements apply only the necessary amount of gating/windowing/FDW to keep out annoying late (>50cm run legth or so) reflections.

 

[KSTR]

IMO response of both drivers looked something like this (midwoofer in red, tweeter in blue):

View attachment 50959

Yep.

 

[QMuse]

This indeed a source of unknown variations but we can only hope the people doing the measurements apply only the necessary amount of gating/windowing/FDW to keep out annoying late (>50cm run legth or so) reflections.

I'm affarid this is not the case as refelection-wise environments in which loudspeakers are measured vary greatly and there is no even unofficial "standard" in use with tie domain like when 1/6 or 1/12 smoothing is used with magnitude response. Quite often I have seen FDW of 2 or 3 cycles used with time domain measurements, more often 5-6 cycles are used, but some also use 12-15 cycles. In that context you cannot possibly speak of say GD of 20ms at 100Hz as you don't know which FDW was applied. Increase the FDW and that value can easilly become 80ms.

 

[KSTR]

Normally no windowing at all is (should be) used for measuring a bookshelf speaker on a stand because the first reflection should have about 1.5m delay (good for 4..5ms of reflection-free time) with no problem, at least for the school where I come from.
But you are correct, there doesn't seem to be any sort of standard.

 

[jaykay77]

My Peachtree DS4.5s would beg to differ! Lol. I don’t know if they qualify as “high fidelity” but they definitely sound better than those Klipsch 41Ms IMO.

It's difficult to sell subjectivism here lol...
But i'm with you I bet that no matter how those peachtrees measure, they probably sound much better than these klipsch to most people who are still capable of hearing.

 

[QMuse]

Normally no windowing at all is (should be) used for measuring a bookshelf speaker on a stand because the first reflection should have about 1.5m delay (good for 4..5ms of reflection-free time) with no problem, at least for the school where I come from.
But you are correct, there doesn't seem to be any sort of standard.

Sure, but here is what I actually meant..

This is IR of the measurement without much reflections:

And here are 2 step responses of that same measurement, green one is FDW of 5 cycles and blue one is without any FDW, both without any gating:

So, whenever I see a step response with smooth curve instead of a choppy one I know some FDW has been applied. What you don't know is how much as that is almost never stated on the graph, so any comparison with other graphs would be tricky.

 

[jaykay77]

It's worth it because it has good acoustic performance. The very reasonable price is a bonus.



That would be amazing if true - I'm sure everyone on here would love to see their measurements if you would be kind enough to send one of them to Amir for testing?

I think i'll do that.
Amir do you have a mailing address for sending? I'll send you an ADS speaker either L780/2 or a B7

 

[QMuse]

Btw, here are GD and phase responses of that same IR (green with FDW of 5 cycles and blue with FDW of 15 cycles, both with NO gating).

So, which one is "real"?

 

[mhardy6647]

Or an ASR essay contest: "Why I dislike these speakers more than any other speaker on the market!" The winner gets the pair, from here on out able to live in eternal sonic agony!

Hey, I'd enter for the Pioneers!

 

[KSTR]

And here are 2 step responses of that same measurement, green one is FDW of 5 cycles and blue one is without any FDW, both without any gating:

View attachment 50966

Well, your own measurements contradict your previous statement, don't they?
The relevant section with those"spikes" you mention is completely unaffected by the 5-cycle FDW (a rather strong windowing).

 

[gr-e]

Hi guys! Long time reader but I thought time has come to write my first post to try to clear the confusion about this.

Actually, there is nothing dramatic and strange happening with Klipsch's step response. In fact, apart from response being inverted, as both drivers are connected with negative acoustic polarity, this step response is actually very typical for a 2-way speaker.

Step response topic has been very well explained by Joe D'Appolito in this article so let's take a look at what he wrote and the graphs he presented.
This is how typical step response of a 2-way speaker looks:

View attachment 50915

Yep, that's the same one as the one measured with Klipsch, except this one has not been inverted. This graph actually consists of a sum of 2 step responses, tweeter and midwoofer. Luckilly, I can borrow that graph as well from D'Appolitos article:


View attachment 50916

And the mistery is solved!
Let me also do some very basic and ugly drawing just to point out midwoofer is legging behind tweeter app. 0.38 ms, which is also quite typical value for a 2 way systems. If the tweeter wouldn't be sitting in a deep waveguide the difference would be slightly larger but that small difference in timing would anyhow have no impact on perceived SQ. There is also nothing on this graph suggesting poor crossover integration - the "kink" in step response is coming from summation of the tweeter's part that is still oscillating with the smooth part of the midwoofer response.


View attachment 50922

That's exactly what I initially thought is happening, but it seems I was wrong

Their conclusion "the tweeter's step has begun to decay..." is plain wrong, for the reasons mentioned. The assumption that the speaker's main lobe is tilted downward is most probaly invalid. One simply cannot infer such info from a step reponse, you either need to have vertical polars or individual mag/phase responses of the drivers.

The 600 has higher crossover frequency and, more importantly, a supposedly much better designed crossover with steeper slopes. Crossover parts are a big factor for BOM cost, that's why on the very cheap lines designers try to get away with the lowest part count and the cheapest stuff.

A quick sim with for the 600 (assumed flat to +-0.2dB) with 1.5kHz XO and 5th order (acoustically, as always) on the woofer (ported @50Hz) and 4th order on the tweeter (set back by 100us), both drivers in phase, gives the following step response which is not that far off.
In that sim I actually have some 60° phase offset at XO (not a good idea, but used for the sake of demonstration), the tweeter lagging, which makes the main lobe tilt upwards, not downwards, which illustrates the above point.
View attachment 50949
If you look hard, the initial rise of the woofer can still be identified

So the tweeter arrives first because 600 has a lower XO frequency and because it uses a higher order filters?
Here's the board of 600m btw. 2nd order + a resistor for the tweeter. Sad to think that they used something even simpler on R-41m.

 

[QMuse]

Well, your own measurements contradict your previous statement, don't they?
The relevant section with those"spikes" you mention is completely unaffected by the 5-cycle FDW (a rather strong windowing).

Yes, FDW by itself doesn't affect the height of those initial spikes by much, but moving mic few cm left or right, up or down and especially closer to the speaker (this was taken from 120cm) will make much more change in captured step response.

 

[QMuse]

So the tweeter arrives first because 600 has a lower XO frequency and because it uses a higher order filters?

Tweeter always arrives first no matter where is the XO and of what order are the filters.

There are only 2 things you can do about that:

1.) push tweeter phisically backwards to delay it.

2.) use DSP to do the same

 

[gr-e]

Tweeter always arrives first no matter where is the XO and of what order are the filters.

There are only 2 things you can do about that:

1.) push tweeter phisically backwards to delay it.

2.) use DSP to do the same

@andreasmaaan and @KSTR say that it is pushed back enough by the waveguide, check their earlier posts.
Idk which is true at this point. I can't convince myself until I see measurements for each driver or the internal wiring.

 

[KSTR]

So the tweeter arrives first because 600 has a lower XO frequency and because it uses a higher order filters?

No, the woofer still physically starts first in absolute terms, though the relevant part of it (LF/MF) of course sees the group-delay of its lowpass function (with XO).

Here's the board of 600m btw. 2nd order + a resistor for the tweeter. Sad to think that they used something even simpler on R-41m.
View attachment 50982

We have no way of knowing what the acoustical crossover responses are without seperated measuements of woofer and tweeter paths. The electrical order of the filter circuit is irrelevant, other than that it always has gentler slopes unless you use two identical wide-band drivers ;-)

 

[QMuse]

@andreasmaaan and @KSTR say that it is pushed back enough by the waveguide, check their earlier posts.
Idk which is true at this point. I can't convince myself until I see measurements for each driver or the internal wiring.

From the step response it is obvious that tweeter is not pushed back enough to be time aligned with the midwoofer so it's response comes first.

 

[gr-e]

We have no way of knowing what the acoustical crossover responses are without seperated measuements of woofer and tweeter paths. The electrical order of the filter circuit is irrelevant, other than that it always has gentler slopes unless you use two identical wide-band drivers ;-)

Yeah I know, just decided to mention
Danny Richie made some measurements for RP-600M here

 

[QMuse]

Besides the physical mis-alignement the main cause of time mis-alignement is the passive XO.

For example, if you need to compensate for LR 24 XO at 2200Hz this is what you need to apply to the phase response:

 

[Belgarathian]

I feel like we've reached the point where we can do a roundup of the 10 worst reviews of the year.