I have finally made some satisfactory acoustic measurements of my DIY speakers (fifteen years after finishing the boxes!). These are three-ways with Audax HM210Z0 and HM100Z0 bass and midrange, and ScanSpeak D2905-9300 tweeters. Crossovers are textbook fourth-order L-R, and I put some serious effort into linearising the driver impedances. After years of various tweaks and improvements, I am quite happy with the sound, which is quite natural and unfatiguing, but the bass (by comparison with my HD595 headphones) is inconsistent: it can come across as boomy with some material, but with other music it can sound slightly weak and lacking in impact.
For a while my only microphone was my trusty old analogue RS SPL Meter, but when I learned recently that this has serious discrepancies at mid-to-high frequencies it explained some oddities in the response I measured a year or so ago that were seriously at odds with my subjective impressions of my speakers. Anyway, I bought a Behringer ECM8000 microphone a week or so ago; it took me a few days to work out that it needed a phantom power supply and that my E-MU 0202 digital interface didn’t provide one, and once I had put together a suitable power supply I was ready to go.
I used Room Equalisation Wizard to produce the figure below, which shows the responses at one metre from the complete left-hand speaker (yellow trace) and from the individual drivers. I didn't really measure at 115dB, of course - I just didn't calibrate the SPLs! I did a quick measurement at 10cm on the tweeter axis to confirm that the high-frequency rolloff above 5kHz was a result of placing the microphone at listening height, horizontally in front of the midrange, which puts it significantly (perhaps 20 degrees) below the axis of the inclined tweeter. The D2905-9300 is actually very flat on axis.
The response above 200Hz looks pretty good to me – it is flat to within a decibel or two between 250Hz and 5kHz - and confirms that the decisions I made during the design, build and tweaking process were basically sound. The bass response, though, is awful, with peaks and suckouts of 5-10dB. The front baffle is about 50 cm from the wall behind it, which might explain the features between 100 and 220Hz, while the peak at 42Hz could be a room mode (the room is about 4.2m front to back).
Should I expect this uneven LF response from a simple sweep test like the one that REW provides? Given that the speakers are in our living room, and moving them around much is ruled out, is there anything I can do about the bass response?
Alex
For a while my only microphone was my trusty old analogue RS SPL Meter, but when I learned recently that this has serious discrepancies at mid-to-high frequencies it explained some oddities in the response I measured a year or so ago that were seriously at odds with my subjective impressions of my speakers. Anyway, I bought a Behringer ECM8000 microphone a week or so ago; it took me a few days to work out that it needed a phantom power supply and that my E-MU 0202 digital interface didn’t provide one, and once I had put together a suitable power supply I was ready to go.
I used Room Equalisation Wizard to produce the figure below, which shows the responses at one metre from the complete left-hand speaker (yellow trace) and from the individual drivers. I didn't really measure at 115dB, of course - I just didn't calibrate the SPLs! I did a quick measurement at 10cm on the tweeter axis to confirm that the high-frequency rolloff above 5kHz was a result of placing the microphone at listening height, horizontally in front of the midrange, which puts it significantly (perhaps 20 degrees) below the axis of the inclined tweeter. The D2905-9300 is actually very flat on axis.
An externally hosted image should be here but it was not working when we last tested it.
The response above 200Hz looks pretty good to me – it is flat to within a decibel or two between 250Hz and 5kHz - and confirms that the decisions I made during the design, build and tweaking process were basically sound. The bass response, though, is awful, with peaks and suckouts of 5-10dB. The front baffle is about 50 cm from the wall behind it, which might explain the features between 100 and 220Hz, while the peak at 42Hz could be a room mode (the room is about 4.2m front to back).
Should I expect this uneven LF response from a simple sweep test like the one that REW provides? Given that the speakers are in our living room, and moving them around much is ruled out, is there anything I can do about the bass response?
Alex
Hi Alex.
If you want to try to linearise bass response in your room there is more than one way to do it.
Multiple woofers is the most logical to me if you don't want to move them arround the living room.
Check this link: Earl Geddes on Mulitple Subwoofers in Small Rooms - YouTube
Cheers
If you want to try to linearise bass response in your room there is more than one way to do it.
Multiple woofers is the most logical to me if you don't want to move them arround the living room.
Check this link: Earl Geddes on Mulitple Subwoofers in Small Rooms - YouTube
Cheers
Yes, there is nothing wrong with the speaker, that's room modes! Very good summation at xo. Perhaps the bass level is low, too little baffle step compensation? Do you have a correction file for the microhone? To check the tweeter response, take measuremnts from other good speakers, nearfield at some 50mm (be careful with the volume knob!)
Last edited:
I think the good phase matching at the crossover points is mainly a result of my impedance correction of the drivers - this means that the LR4 crossover does something much closer to what it is supposed to (as I confirmed by measurement). I will check this by swapping the polarity of the tweeter and looking at the null, next time I get my microphone out.
Yes, I think the bass level is too low. When I chose the drivers for the project I made the mistake of thinking that, because the bass driver had a quoted sensitivity of 92 dB/1W/1m, it would match well with the 93 dB/1W/1m midrange. Unfortunately I hadn't noticed that on the datasheet the HM210Z0 has a steady drop off from 93 dB at 1kHz to 87 dB at 60Hz, and I only use it up as far as the crossover point at 420Hz anyway.
I was expecting to be able to add some baffle step compensation attenuating the midrange to the correct lower level, bearing in mind the +3B or so gain of the bandpass midrange crossover. In the end, though, I adjusted the mid to subjectively match the treble, and the bass is a little left behind.
Alex
Last edited:
I agree, and a while back I made some plans to build a pair, but unfortunately the construction of my terraced house is such that subwoofers would be deemed seriously antisocial (one neighbour has an HT system, so I understand this well).
Alex
Last edited:
To see what your woofer is really doing take a near field measurement of it. Place the mic very close to the dust cap without touching it, like 1/4" away.
Below is something that might help if acoustical measurement in small spaces is new to you:
The region where the acoustical performance of the space is mode dominated is bounded by the first mode of the space where f =.5C/RLD up to the highest where f = 3C / RSD. This is where RTA will lie to you, EQ can't help you, absorption is really your only weapon and your ears are your best tool.
Multiply the speed of sound by three, divide the answer by the rooms smallest dimension, eg 1130 X 3 =3390 / 8' =423.75 Hz.
Divide the speed of sound by two, divide the answer by the rooms largest dimension, eg 1130 / 2 = 565 / 20' = 28.25 Hz.
In the above example the region between 28 and 425 Hz is where modes dominate and cannot be controlled by EQ. Lowering certain bands of energy put into the room can only help by not aggravating the beast, but then you have limited the dynamics of the system.
RTA cannot give you good information in this region and EQ can't really help.
Hope this helps,
Barry.
Below is something that might help if acoustical measurement in small spaces is new to you:
The region where the acoustical performance of the space is mode dominated is bounded by the first mode of the space where f =.5C/RLD up to the highest where f = 3C / RSD. This is where RTA will lie to you, EQ can't help you, absorption is really your only weapon and your ears are your best tool.
Multiply the speed of sound by three, divide the answer by the rooms smallest dimension, eg 1130 X 3 =3390 / 8' =423.75 Hz.
Divide the speed of sound by two, divide the answer by the rooms largest dimension, eg 1130 / 2 = 565 / 20' = 28.25 Hz.
In the above example the region between 28 and 425 Hz is where modes dominate and cannot be controlled by EQ. Lowering certain bands of energy put into the room can only help by not aggravating the beast, but then you have limited the dynamics of the system.
RTA cannot give you good information in this region and EQ can't really help.
Hope this helps,
Barry.
That is very good blending between units, as confirmed by near 6dB gain at the crossover frequencies and no cancelation anywhere.
Note that that has very little to do with driver conjugating. Even with LR crossovers there is a bit of luck involved as the delay between drivers is somewhat arbitrary. LR is guaranteed to add to flat, only for electrical filters with no other delays.
"Better to be lucky than good."
David
Note that that has very little to do with driver conjugating. Even with LR crossovers there is a bit of luck involved as the delay between drivers is somewhat arbitrary. LR is guaranteed to add to flat, only for electrical filters with no other delays.
"Better to be lucky than good."
David
A quote from your web site:
"The bass enclosure needs to have plenty of internal volume while
being solid and well braced, with internal waves less of a problem."
Your internal bass enclosure height is roughly 70 cm which by calculation
for standing waves gives about 240 Hz. According to your FR plot, it seems
to me you have an internal standing wave still alive.
I did worry about internal resonances in the bass cabinet (as I mention in the page on building the passive crossover, linked to at the end of my main page), but I tackled these with extra shelves and some strategic damping. I managed to almost remove one close to this frequency, which I indeed matched with the main longitudinal dimension of the box.
Below are the impedance curves I measured for the woofer with and without impedance compensation. There is something at about 440Hz which is probably a standing wave between front and back, but very little sign of anything around 240Hz. My feeling is that the impedance plots suggest that I have more or less managed to deal with internal resonances.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Alex
Something that I forgot to mention earlier is that I realised a while back that there is a useful trick I can use to increase bass level: my amplifiers have doubled-up binding posts for bi-wiring, as well as a choice between 4 and 8 ohm output windings. If I connect the mid and treble to the 4 ohm output and the bass to the 8 ohm output, I get a handy 3dB boost in the latter. At present this simply adds to the issue of bass excess at certain frequencies, but if I manage to improve the room later on I could certainly use this as my default connection.
Alex
- Status
- Not open for further replies.