advertisement


Speaker/Room Measurement Witchcraftery

Setting aside the discussion on envelope distortion for a moment, I would welcome comments on the Impulse response below:

Impulse-Responce.jpg


Please could the experts tell me how this looks for time coherence and where the various drivers contribute.

To be honest it looks very odd and I'm not sure I have seen anything like it.

There is the sharp tweeter peak, but the other drivers clearly aren't in line with it. But I'm not clear on the why the lower levels then increasing.

Are those drivers levels reduced or off ;) ?

TBH I don't know, but it sure not a response you want
 
Even if you don't have the kit to measure at the moment swap the phase of the sub and see if the sound is more full. If as you say the sub and mid bass are out of phase it would certainly explain the 45 Hz dip
 
Well I think the point is more that, whether you believe it or not, better measurements do lead to better sound.

No, it's pure speculation. You have never heard my system and you are trusting my novice measurements, which were made with no attempt to make them look nice. It's easy to make them look nicer, but I chose not to.

There is nothing wrong with your measurements, you did a good job. They do exactly what I would expect for a room with those dimensions and I have measured and listened to many similar rooms, including my own.

OK, problem with that definition is that Im afraid its your speakers that have the poor... correction, no its not poor per-se, its actually very typical..... time alignment issues, something which the Kii doesnt. So you need to look for a different explanation

Your time alignment - sharp tweeter impulse followed b y wider mid/bass, followed by wider sub. So a low frequency fundamental having harmonics that play through the mid will be about a 1.5 mS early and through the tweeter 3 ms early.

imp2_zpsqdnphgb0.png


.


Setting aside the discussion on envelope distortion for a moment, I would welcome comments on the Impulse response below:

Impulse-Responce.jpg


Please could the experts tell me how this looks for time coherence and where the various drivers contribute.

Well I think the two impulse responses look quite similar.
The first one is ZRB + CAOW1
The second one is LS50 (open port, used as a 2-way).

JA's Impulse response measurment for the LS50 looks totally different:

1212KEF50fig8-2.jpg


So I think we have to question some of these assumptions:
- My measurements were competently made
- Your interpretation is correct
- REW is doing the maths right (most likely)
- We know how to drive REW properly
etc.

On the topic of phase/polarity. All drive units except the tweeters were checked with an AA cell, besides which it's easy to hear if the ZRBs have incorrect polarity compared to the mains.
 
Perhaps get in touch with the manufacturer and ask exactly what you have been sold?
The LS50’S tweeter is connected with negative polarity that Is why the initial peak is downwards then the woofer with positive.


Keith
 
Well I think the two impulse responses look quite similar.
The first one is ZRB + CAOW1
The second one is LS50 (open port, used as a 2-way).

JA's Impulse response measurment for the LS50 looks totally different:

1212KEF50fig8-2.jpg


So I think we have to question some of these assumptions:
- My measurements were competently made
- Your interpretation is correct
- REW is doing the maths right (most likely)
- We know how to drive REW properly
etc.

On the topic of phase/polarity. All drive units except the tweeters were checked with an AA cell, besides which it's easy to hear if the ZRBs have incorrect polarity compared to the mains.

JAs isn't totally different, its just inverted compared to yours. Yours starts to make a lot more sense if you turn it upside down. The ZRB impulse is different showing a lower frequency impulse delayed from the tweeter.
I thought you said the mid was wired out of phase? Did you mean with respect to the tweeter and not the bass?

Can you post some pictures of the ZRB and the caow1 as I have no idea what they are.
 
S-Man, the files you sent me are labelled LS50 ZRB and CAOW ZRB. I didnt look at the LS50 ones previously. I just have and they seem too similar to the CAOW to be a different speaker. Measurement of the same speaker in a slightly different position yes. Did you get them mixed up? Do your CAOW also have the tweeter connected in reverse polarity because the impulse responses are also too similar?

FR_zpstt0jns73.png
 
So if we invert your impulse for your LS50 +ZRB measurement it looks very similar to JAs measurement apart from the additional delayed impulse which my best guess is the ZRB

IMP_zpsnnyeqrhz.png


1212KEF50fig8-2.jpg


A few things to consider and check here, as I am making the assumption JAs measurement is correct (I had no idea the LS50 tweeter was in reverse polarity).

  • Check if your measurement microphone/amp has correct polarity.
  • Check to see if your hifi amp has absolute polarity (some dont regardless of speaker terminal markings)
  • Speaker wiring polarity.
 
Last edited:
I've never really looked at the impulse response before. Here's the step response of my Tannoy Monitor Gold 12 taken with the mic on-axis with the HF horn:
32203674897_aa12cc1282_o.jpg


And here's the step responses of my Celestion Ditton 66, taken on-axis with the tweeter and then on-axis with the mid unit:
33270286668_b583b01107_o.jpg

32203674977_0519a09d34_o.jpg


I was expecting the Dittons to measure poorly due to the physical distance between the drive units and using an ABR to augment the bass response.

As a newbie to impulse response I'd be grateful if someone could interpret the above graphs and speculate/advise as to which peaks correspond to which drive units?

All that I'm able to deduce is that being on-axis with the MF driver gives a slightly better impulse response than being on-axis with the HF driver (frequency response measurements and subjective listening tests support this, - the presentation sounds smoother and more integrated when the MF drivers are elevated to ear height).

PS - My BK XXLS400 subwoofers are not included in any of the above measurements.
 
Last edited:
Here are the impulse responses of each individual drive unit from the Celestion Ditton 66 BEFORE they were connected to the crossover and installed in the cabinet. The measurements were taken with each driver connected directly to my amp, laying face up on the carpet (pointing at the ceiling), with the mic suspended 1 metre above them. Can we infer anything further from these measurements, or are they totally meaningless because the drivers aren't connected to a crossover?

47148649181_2763ecb9b4_o.jpg

46425215894_279b1f7e4e_o.jpg

47148649301_115d2b1c37_o.jpg
 
Check if your measurement microphone/amp has correct polarity.
  • Check to see if your hifi amp has absolute polarity (some dont regardless of speaker terminal markings)
  • Speaker wiring polarity.

Unfortunately I cannot check the measurement setup. I cannot remember exactly how I did the measurements, other than I used a B&K microphone with its associated PS/preamp, a Focusrite 2i2 and a laptop.

All my power amps are DIY. Only one has an inverting gain stage and I conneced the output leads swapped over inside the amp case to correct that.
Speaker wiring polarity is correct - checked with AA cell and the usual colour coded wires.

Just in case I made a mistake earlier:
Both CAOW1 and LS50 have the mid/bass unit wired in phase and the tweeter with inverted polarity.
When using the (stereo) ZRBs, the XO from bass to mid is a quasi-LR4 so the drive units need to be connected in the same polarity. Which they definitely are and were when the measurements were taken. Both had the ports blocked when used with the ZRB.

I would expect the CAOW1+ZRB and the LS50+ZRB measurements to be very similar, they are wired the same and have a similar sized bass/mid in the same position! The room dominates the measurements. The significant difference is the extra ~3 dB of SPL around 2KHz of the LS50s, which is responsible for the "live" but slightly tiresome quality of the LS50s and the fact that they are now with a new owner.

There is absolutely no reason why the ZRBs should have a few mS of delay versus the main speakers. The "3rd peak" you are attributing ot the ZRB is much more likely to be the reflection off the front wall - as I have already explained. The hard evidence for this is that the 3rd peak you attribute to the ZRB is still apparent with the LS50 run full range (no ZRB)!!
 
Well you have an inversion in your system somewhere.

Just curious, were the drivers designed for closed box use? They have to have the correct TS parameters to do so. What drivers are in the CAOW?

Different drive units in different boxes with different crossovers will not be that similar. The room will have significant effect below schroeder but the mid is too similar.

The 3rd peak could be the wall. Whats the distance? Crossovers and different acoustic centres will move the timing, so its actually unlikely that they will all match up without active intervention.

At least its making more sense now that we have the correct polarity on the plots. :)
 
What drivers are in the CAOW?

Different drive units in different boxes with different crossovers will not be that similar. The room will have significant effect below schroeder but the mid is too similar.

All the info is in the thread.
Why shouldn't they be that similar, they are both very well designed speakers using similar parts? Although IMO, 2 or 3 dB is quite dissimilar.
It's well known that room effects will mask individual loudspeaker measurements in this type of plot. I am beginning to suspect your level of expertise is no better than mine, and quite possibly worse. The difference is that I am prepared to question my knowledge.

The 3rd peak could be the wall. Whats the distance?

Once again, read the thread!
 
All the info is in the thread.
Why shouldn't they be that similar, they are both very well designed speakers using similar parts? Although IMO, 2 or 3 dB is quite dissimilar.
It's well known that room effects will mask individual loudspeaker measurements in this type of plot. I am beginning to suspect your level of expertise is no better than mine, and quite possibly worse.

The difference is that I am prepared to question my knowledge.



Once again, read the thread!
Oh dear it really wouldn't be too difficult for you to just say so. A few words, no too difficult eh?

Sorry but different speakers even in room do not measure that similar. 200 to 1500Hz they are practically identical. Let's not forget that it's your inverted measurement that caused the confusion in the first instance, so sorry if I now question the rest of the data. Have I not acknowledged that the second hump could be wall reflection? Would that not constitute questioning my previous conclusions drawn from the erroneous data?

Well I perhaps should leave you to it then.



Edit: I have just skimmed the thread again and I can't see where you talk about what drivers are in the caow.

So what drivers are they so we can look at their raw response?

Can you post a photo?
 
Last edited:
So I found where you talked about the wall reflection

I reckon the 3.3mS hump is the bass reflecting off the front wall. The drivers are 60cm from the wall , so the path difference is 120cm, which is 3.5S (at 343m/S).

OK, however this is your COAW + ZRB impulse and the hump is at 1.9ms. Which doesnt tie up with what you say above.

imp_zpsac6n4ng3.png
 
Hmm, when I said Oh Dear I was accused of “So less of "oh dear" patronising nonsense please.”

I think that interpreting one’s own system’s room plots can be tricky but trying to interpret someone else’s is fraught with problems which is why I would have to be very confident of someone’s knowledge, expertise and experience before posting my plots. Not to say that some of our posters don’t qualify but it is difficult to tell sometimes. As for me, I have many years of experience with tricky ;) speakers in a limited number of rooms. As for knowledge and expertise it would need someone with more knowledge than me to judge! :)

With our own systems, if we see something that looks a bit odd we can change one thing at a time, remeasure and compare plots and then listen carefully to see if the changes are audible to us. Another problem is that the plots give the conclusion, so to speak, but don’t necessarily indicate the cause, particularly if more than one thing in the room is causing a problem unless of course you have a totally symmetrical room with no furniture and consistent walls, floors and ceilings. With any sort of fault finding multiple causes can make analysis challenging and frustrating. Inevitably, so many manuals illustrate how to diagnose a single fault at a time rather than combinations.
 
Last edited:
Yes, that's why it's important.
Yes and interpreting the graph is not dependant on knowing the room.The impulse is indicating the driver outputs plotted against time. That said the measurements need to be correctly taken in the first place and the context needs to be clear for them to be useful.
Ive been here before and concluded phase and/or time alignment were off but was informed it sounds better this way (or words to that effect).
 
Last edited:
Hmm, when I said Oh Dear I was accused of “So less of "oh dear" patronising nonsense please.”

I think that interpreting one’s own system’s room plots can be tricky but trying to interpret someone else’s is fraught with problems which is why I would have to be very confident of someone’s knowledge, expertise and experience before posting my plots. Not to say that some of our posters don’t qualify but it is difficult to tell sometimes. As for me, I have many years of experience with tricky ;) speakers in a limited number of rooms. As for knowledge and expertise it would need someone with more knowledge than me to judge! :)

With our own systems, if we see something that looks a bit odd we can change one thing at a time, remeasure and compare plots and then listen carefully to see if the changes are audible to us. Another problem is that the plots give the conclusion, so to speak, but don’t necessarily indicate the cause, particularly if more than one thing in the room is causing a problem unless of course you have a totally symmetrical room with no furniture and consistent walls, floors and ceilings. With any sort of fault finding multiple causes can make analysis challenging and frustrating. Inevitably, so many manuals illustrate how to diagnose a single fault at a time rather than combinations.

I think my reaction was reasonable considering the unhelpful reaction and comments. If you dont want input from a forum then dont start a thread on your measurements.

Yes it is challenging and as we have seen above it is easy to be mislead. We were initially unknowingly looking at an inverted impulse response which led us up the garden path. The plots do tell us whats going on but yes you do need the whole range of additional information available to you plus a process of diagnostic elimination. Something I obviously cant do from here.

With above I would take measurements without the sub and move the speakers to see if that second hump moves with them.
 


advertisement


Back
Top