Exactly, and this is the big clue. Also you say no boost on the compression driver. But it is bi amped, not passive. And there are two compression drivers. So to me there is a boost on the high frequency compression driver compared to the mid frequency compression driver.
Normally one would have quite a boost to keep the upper limits (20K) in line with the rest of the HF, 5K. The fact that 20K is higher in level than the 3k to 6k octave, and the fact that 7K is higher than 5K by about 10db says that he has the relative gain of the HF spectrum to mids. If it were a crossover point problem it would be much more narrow band, as Art hedged in his first response.
To my way of looking at it, if the highs were 10 db lower, then it would be fairly flat from 3K to 10K, and start rolling of there. And need the CD boost to bring 20K back up. To have a high point in the 2K area (broad band) that needs taming is quite in line with my experience of this driver.
You are all seeing a dip in the 3-6k range, I am seeing a boost in the 1.5-3K range that needs to come down, and a relative gain in the highs to the mids.
The BMS 4590 is a dual concentric compression driver. It has a mid and high diaphragm. 3.5" VC for the mids, and 1.75" VC for the highs. 500hz to 6,3 on the mids, and 6,3 on up on the highs.
Hi Guys,
Last night I had some technnical issues with the MiniDSP and the woofer amp. I changed the computer GUI from Mac to Windows and it wasn't working well, so I need a few days make that work again and improve it.
I need to buy a longer microphone cable, to have a propper measurement setup to do this well, so I can repeat measurements faster.
I will be back as soon as I have a better setup, better measurements and more info...
Hi Kindhorman, The driver has two voice coils too. It's justdrivers in one, with a common 2in output:
http://images.thomann.de/pics/prod/155588_datasheet.pdf
Best!
Last night I had some technnical issues with the MiniDSP and the woofer amp. I changed the computer GUI from Mac to Windows and it wasn't working well, so I need a few days make that work again and improve it.
I need to buy a longer microphone cable, to have a propper measurement setup to do this well, so I can repeat measurements faster.
I will be back as soon as I have a better setup, better measurements and more info...
Hi Kindhorman, The driver has two voice coils too. It's justdrivers in one, with a common 2in output:
http://images.thomann.de/pics/prod/155588_datasheet.pdf
Best!
Last edited:
![a>](https://www.diyaudio.com/forums/<a href="<a href=http://www.flickr.com/photos/7934164@N02/9016550717/" target=_blank>http://www.flickr.com/photos/7934164@N02/9016550717/"</a> title="P1020184 by Jackan!, on Flickr"><img src="<a href=http://farm8.staticflickr.com/7339/9016550717_2a8be53121_c.jpg" target=_blank>http://farm8.staticflickr.com/7339/9016550717_2a8be53121_c.jpg"</a> width="800" height="800" alt="P1020184"></a>)
The sound from the mid diaphragm goes off the back side of the diaphragm and goes through slots that are behind it. The sound goes through tubes, and into the center, where you can see the tubes terminate as fins where they meet the pole piece. In the photo that is more straight on, you can see the ring in the very back behind the fins for the mid range tubes. That is the exit for the HF, the sound from this diaphragm merges with the sound from the MF diaphragm near the pole piece where the slots meet the pole piece. The HF diaphragm is set back slightly, so that the HF and MF are time aligned.
Attachments
OK guys. I improve the measuring setup. This allows me to measure and listen in a more agile way so I plan to go on with more measurements.
1) First of all, I discovered why I had a big dip in the 5khz-7khz zone. It was because the HF driver has less sensitivity than the the MF driver. The first time I try to setup somthing, I just increased HF driver volume, HF sounded very clear, but this produced the big dip.
2) Blue is the sum of the 3 drivers with equal volume level (woofer is an exception because it uses a 10W amplifier).
3) Green is the same response with an increase of 15db in the volume (i did this just to confirm what was going on in my original measurement).
4) Please be aware that I am already filtering bands (arbitrary) for each driver based on BMS specificacion of his crossover ( LowPass 12db/oct & HighPass 18db/oct). I guess I should relax those even more to have a better idea of how they behave in the crossover region, but I am a bit shy about stressing drivers and broking something.
1) First of all, I discovered why I had a big dip in the 5khz-7khz zone. It was because the HF driver has less sensitivity than the the MF driver. The first time I try to setup somthing, I just increased HF driver volume, HF sounded very clear, but this produced the big dip.
An externally hosted image should be here but it was not working when we last tested it.
2) Blue is the sum of the 3 drivers with equal volume level (woofer is an exception because it uses a 10W amplifier).
3) Green is the same response with an increase of 15db in the volume (i did this just to confirm what was going on in my original measurement).
4) Please be aware that I am already filtering bands (arbitrary) for each driver based on BMS specificacion of his crossover ( LowPass 12db/oct & HighPass 18db/oct). I guess I should relax those even more to have a better idea of how they behave in the crossover region, but I am a bit shy about stressing drivers and broking something.
Now, this his the full response (blue, including woofer) versus the responses of the Mid & High drivers:
2) also, why at higher frecuency spl decreases. Is it something about biradial horns?
3) Again: Please be aware that I am already filtering bands (arbitrary) for each driver based on BMS specificacion of his crossover ( LowPass 12db/oct & HighPass 18db/oct). I guess I should relax those even more to have a better idea of how they behave in the crossover region, but I am a bit shy about stressing drivers and broking something.
p.s. I am measuring at arround 4 meters away from the horn (kind of in axis).
p.p.s. Now I am going to check what EQ methods are available in the Minidsp to see if one is suitable for this...
An externally hosted image should be here but it was not working when we last tested it.
1) So the question is now, what would be a better and more natural way to correct this.
2) also, why at higher frecuency spl decreases. Is it something about biradial horns?
3) Again: Please be aware that I am already filtering bands (arbitrary) for each driver based on BMS specificacion of his crossover ( LowPass 12db/oct & HighPass 18db/oct). I guess I should relax those even more to have a better idea of how they behave in the crossover region, but I am a bit shy about stressing drivers and broking something.
p.s. I am measuring at arround 4 meters away from the horn (kind of in axis).
p.p.s. Now I am going to check what EQ methods are available in the Minidsp to see if one is suitable for this...
Last edited:
Something like this would work in my very humble opinion.... advice is welcomed 
I will test it tonight.
I will test it tonight.
An externally hosted image should be here but it was not working when we last tested it.
Charly,
This is much better, but there is still work to be done.
Note the word "gating" in your first graph on the frequency axis. That is the frequency below which everything is garbage. Only the graph to the right of the marker is realistic.
You can play with this gate in the impulse window. As you slide the slider to the right after the first impulse, more and more reflections come in. You can see the reflections as spikes in your impulse response.
In Holm, for the particular measurement, there is a button called "Options". Inside, you can choose how to view the measurement. You can choose frequency domain smoothing. Input 3.0 as your smoothing. Now this view is valid for the entire frequency range, i.e., 20 Hz to 20 kHz. You can see where your bass level is compared to the mid. But remember, you have now asked Holm to take the impulse response for all time and smooth the response in frequency domain. So, it will include all the reflections. So, you will likely see some peaks and dips, but some of them will go away because of the 1/3rd octave smoothing. You can lessen the smoothing by inputting 6.0, 9.0, 24.0, 48.0.
It looks like your phase is reversed. Try reversing the leads of the supertweeter on the BMS coax, i.e., connect your amp positive to tweeter negative or vice versa. Then measure again and see if the dip in the HF has gone away.
A word about MiniDSP. I use it too, and it is wonderful. Superb. You don't want to boost anything in MiniDSP. it is a poor preamp, and it will run out of dynamic range very fast. A better thing to do is to cut. So, for the above post, you can say filter type is 'low shelf', gain -16. It should produce the same filter, but now, you are cutting the lows instead of boosting the highs.
This is much better, but there is still work to be done.
Note the word "gating" in your first graph on the frequency axis. That is the frequency below which everything is garbage. Only the graph to the right of the marker is realistic.
You can play with this gate in the impulse window. As you slide the slider to the right after the first impulse, more and more reflections come in. You can see the reflections as spikes in your impulse response.
In Holm, for the particular measurement, there is a button called "Options". Inside, you can choose how to view the measurement. You can choose frequency domain smoothing. Input 3.0 as your smoothing. Now this view is valid for the entire frequency range, i.e., 20 Hz to 20 kHz. You can see where your bass level is compared to the mid. But remember, you have now asked Holm to take the impulse response for all time and smooth the response in frequency domain. So, it will include all the reflections. So, you will likely see some peaks and dips, but some of them will go away because of the 1/3rd octave smoothing. You can lessen the smoothing by inputting 6.0, 9.0, 24.0, 48.0.
It looks like your phase is reversed. Try reversing the leads of the supertweeter on the BMS coax, i.e., connect your amp positive to tweeter negative or vice versa. Then measure again and see if the dip in the HF has gone away.
A word about MiniDSP. I use it too, and it is wonderful. Superb. You don't want to boost anything in MiniDSP. it is a poor preamp, and it will run out of dynamic range very fast. A better thing to do is to cut. So, for the above post, you can say filter type is 'low shelf', gain -16. It should produce the same filter, but now, you are cutting the lows instead of boosting the highs.
Last edited:
2) Yes, horns that are not exponential will experience this. (Exponential horns will have a flatter response, but only on axis. Tom Danley explains this very well, you can find it almost everywhere he writes about horn response.)
3) Try measuring the midrange only, with no crossover on the high end. (High frequencies will not stress the mid diaphragm.)
4) Measure one horn only, about 1M from the horn.
PS, I do not see it in the graph right now, but if you are doing 12db/octave one way, and 18 db/octave the other way, you are probably out of polarity as RA7 says. and you should try it both ways.
Last edited:
Thanks.
Based in advice I equalized the 3-10khz area reducing the signal.
Green is original, Red was with equalization, and green was with equalization but also inverting tweeter polarity.
Here is the detail of the response with the equalization and tweeter inversion. Signal was smothed by 1/3oct.
and here I attenuated Mid and High at the level of the woofer.
I look at it and I think it could be done better, I am thinking about giving it another try, but I am going to listen to it first....
Based in advice I equalized the 3-10khz area reducing the signal.
Green is original, Red was with equalization, and green was with equalization but also inverting tweeter polarity.
An externally hosted image should be here but it was not working when we last tested it.
Here is the detail of the response with the equalization and tweeter inversion. Signal was smothed by 1/3oct.
An externally hosted image should be here but it was not working when we last tested it.
and here I attenuated Mid and High at the level of the woofer.
I look at it and I think it could be done better, I am thinking about giving it another try, but I am going to listen to it first....
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Thanks Ra7,
To be honest I am not very happy, It sounds more correct, more like a wall with all frecuencies there, but not sure, something missing, like it sucked life? sorry, not sure about how to describe it in technical terms since I am not sure what is going on. Could be too many things, so the only option is keep experimenting.
1) I am presumming that maybe I just like highs higher, so I can fell all the micro detail and bronzes, even if it's not flat...
2) other posibility is that I am equalizing too aggresively, since I am applying tons of dB EQ here and there. So I want to review all I have done and see If I can apply less EQ. I am thinking that maybe, move from a 10W amp to a 100W amp for the woofer may put the woofer beside the horn, so there will be less necesity to attenuate the horn. Also want to do as much attenuation as possible in analog.... I felt like, maybe, dynamics suffered a bit... but I am only GUESSING at this point. more work to do...
3) Pending is the work with the woofer, and it's crossover point is now at 500hz. This horn/driver is suppoused to go far lower so I expect to be measuring again and doing some comparisons.
4) I think I need to think and listen a lot about what is going on here...
To be honest I am not very happy, It sounds more correct, more like a wall with all frecuencies there, but not sure, something missing, like it sucked life? sorry, not sure about how to describe it in technical terms since I am not sure what is going on. Could be too many things, so the only option is keep experimenting.
1) I am presumming that maybe I just like highs higher, so I can fell all the micro detail and bronzes, even if it's not flat...
2) other posibility is that I am equalizing too aggresively, since I am applying tons of dB EQ here and there. So I want to review all I have done and see If I can apply less EQ. I am thinking that maybe, move from a 10W amp to a 100W amp for the woofer may put the woofer beside the horn, so there will be less necesity to attenuate the horn. Also want to do as much attenuation as possible in analog.... I felt like, maybe, dynamics suffered a bit... but I am only GUESSING at this point. more work to do...
3) Pending is the work with the woofer, and it's crossover point is now at 500hz. This horn/driver is suppoused to go far lower so I expect to be measuring again and doing some comparisons.
4) I think I need to think and listen a lot about what is going on here...
Yes... your thinking about it in the right way. Listen. Adjust. Measure. Listen.
Life sucked out likely means fundamentals are missing. This is the 200 to 400 Hz range. If it is lower than the rest of the spectrum, it just sounds bad. You should be as flat as possible from about 200 Hz, all the way up to 5000 Hz. Of course, what you measure is not always going to tell you how it sounds. For example, the floor bounce cancellation may be some part of the 400 Hz dip. You bring it up too much and it sounds congested. Getting the 200 Hz to 500 Hz range right in relation to bass and highs is very very important for music.
Don't worry about how low the compression driver 'should' go. I'd say that the woofer will outperform the CD below 500 Hz. Just my thought.
Look at this post:
Altec VOTT Reborn!!
The "after" is the balance you should be shooting for in the listening position. But it's very hard to get a clean measurement below 500 Hz. What I do is take about 10 measurements (left 8", center, right 8", 25" high, 36" high, 45" high) around the listening spot, then take a power average. You can do that in ARTA. This will average out the floor bounce and room response in the region below 500 Hz. You can then compare this with the target.
But before getting the overall balance right, focus on the crossover from the mid to low.
Regarding dynamics, if you are cutting the lows in EQ, and using only 10W, it will not have good dynamics.
Life sucked out likely means fundamentals are missing. This is the 200 to 400 Hz range. If it is lower than the rest of the spectrum, it just sounds bad. You should be as flat as possible from about 200 Hz, all the way up to 5000 Hz. Of course, what you measure is not always going to tell you how it sounds. For example, the floor bounce cancellation may be some part of the 400 Hz dip. You bring it up too much and it sounds congested. Getting the 200 Hz to 500 Hz range right in relation to bass and highs is very very important for music.
Don't worry about how low the compression driver 'should' go. I'd say that the woofer will outperform the CD below 500 Hz. Just my thought.
Look at this post:
Altec VOTT Reborn!!
The "after" is the balance you should be shooting for in the listening position. But it's very hard to get a clean measurement below 500 Hz. What I do is take about 10 measurements (left 8", center, right 8", 25" high, 36" high, 45" high) around the listening spot, then take a power average. You can do that in ARTA. This will average out the floor bounce and room response in the region below 500 Hz. You can then compare this with the target.
But before getting the overall balance right, focus on the crossover from the mid to low.
Regarding dynamics, if you are cutting the lows in EQ, and using only 10W, it will not have good dynamics.
Carlos,
One thing to consider in your gain structure is the basic nature of DSP, if you reduce level too much, there is less "information" to retrieve, resulting in less detail, more "grainy" sound.
Ideally, it is best to run the DSP as close to clipping on peaks as possible, and reduce level after, preferably at the amplifier's input gain stage.
As well as sounding better, this will also improve the signal to noise ratio.
Also, since you "acclimatized" your hearing to listening with the mid/highs sounding twice as loud (+10 dB) as the lows, it will take a similar interval of time before a flat response again sounds "normal" to your hearing.
That said, 10 watts per side in the bass department limits your LF peaks to under 110 dB clean at the listening position, while the mid/highs are capable of 120+ dB, adding 10 dB headroom (going to 100 watt amps) would be a good idea if you like dynamic realism and low distortion.
Art
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.