yeah I didn't have a 1M one done on the same day. so used the two meter one. There is very little difference between the 1M and 2M readings if gated the same.
1st image is comparing the 1M and 2M measurements (taken on different days), and the second are both at 2M with the only difference being one is with the legs and the other without, which is why I reckon something is going on between the bottom of the speaker cab and the speaker stand
I zoomed a bit to make the differences more apparent.
edit: just saw your post, they are actually central, I didn't do any offseting, was something I considered but read some negative research duntech had done on this and decided not to. I will have to check the distances and report back, but right now I need to go to bed! Oh and you did read that right, those graphs ARE with only a 4.3uF cap on the tweeter and the MW144's running full range The slightly earlier graphs are with the notch filters which are quite a bit (but still not right) flatter.
Tony.
1st image is comparing the 1M and 2M measurements (taken on different days), and the second are both at 2M with the only difference being one is with the legs and the other without, which is why I reckon something is going on between the bottom of the speaker cab and the speaker stand
I zoomed a bit to make the differences more apparent. edit: just saw your post, they are actually central, I didn't do any offseting, was something I considered but read some negative research duntech had done on this and decided not to. I will have to check the distances and report back, but right now I need to go to bed!
Oh and you did read that right, those graphs ARE with only a 4.3uF cap on the tweeter and the MW144's running full range The slightly earlier graphs are with the notch filters which are quite a bit (but still not right) flatter. Tony.
Attachments
Last edited:
Well you seem to be on the right track for a very nice sounding system there Tony, Great to see it looking so promising. Keep us posted, (I'm sure you will).
My Three ways are coming along well and as a bonus (I think), by their very design, I can almost semi-hemishere the entire tweeter/mid section (massive chamfer), without compomising box strength for the woofers. Just ordered a cx 3400, should be here in the next few days, almost wetting myself with anticipation! Then I just need a third amp and I can get down to some serious buisness!
Mick.
My Three ways are coming along well and as a bonus (I think), by their very design, I can almost semi-hemishere the entire tweeter/mid section (massive chamfer), without compomising box strength for the woofers. Just ordered a cx 3400, should be here in the next few days, almost wetting myself with anticipation! Then I just need a third amp and I can get down to some serious buisness!
Mick.
Hi Tony
You are discovering what anyone measuring drivers needs to understand, measurements are very difficult! You must be able to remove all artifacts from the environment.
Never the less, your doing great. +/- 2.5dB is superb! It is what I would expect you to get. I suggest you celebrate. As you have discovered, your in room response will now swamp any improvements you achieve from here on in.
In fact, room treatments will help more than improving the speakers any further.
I strongly urge you to put the measuring gear away and use your ears for a while. I once devised a complex x-over for this combination. I got them down to +/- 1.5dB. The problem was it was the worst sounding speaker I had ever heard! That is why, with these drivers, I work on the Kiss principal, it is simply much better sounding.
Regards
Terry
You are discovering what anyone measuring drivers needs to understand, measurements are very difficult! You must be able to remove all artifacts from the environment.
Never the less, your doing great. +/- 2.5dB is superb! It is what I would expect you to get. I suggest you celebrate. As you have discovered, your in room response will now swamp any improvements you achieve from here on in.
In fact, room treatments will help more than improving the speakers any further.
I strongly urge you to put the measuring gear away and use your ears for a while. I once devised a complex x-over for this combination. I got them down to +/- 1.5dB. The problem was it was the worst sounding speaker I had ever heard! That is why, with these drivers, I work on the Kiss principal, it is simply much better sounding.
Regards
Terry
Thanks Terry!!! Both for the praise, and the reality check!! I tend to get a bit obsessive about things, and the current obsession has been trying to get as flat a response as possible (as well as trying to get a perfect finish on the baffles with some spray on polyurathane and lots of sanding) Last night I was not happy with the sound. The tweeter was too hot. apart from the 4.3uF on it, I had a 1 ohm in parrallel with a 3.3uf also in series, this clearly wasn't working. Tonight I upped the 1 ohm to a 2.2 ohm and stuck the 50uH coil in Parallel as well (basically a notch for the slightly higher levels between about 7 and 11K) This seems to have done the trick nicely and sounds MUCH better I think I could live with them now
For the first time since I have owned the CD, I could play Whitestripes Elephant at a normal listening level without wanting to dash for the volume control and turn it down! I think that is a good sign.
I'm listening to Ella and Louis at the moment and it sounds very nice
And all of this without a decent source! I'm just using the mythmusic player in my mythtv box which in turn is using the onboard sound card. Will have to get on with the B1 project so I can use my DV18 once more. I also watched "Video killed the radio star" on the ABC tonight and the dialog was so clear and the voices so natural I felt that this latest tweak was definitely right.
I think I can probably pretty safely now order some decent parts and replace the white coffins (which there are three of) and the 100uF electro in the 1K notch filter.
I might still make my own coils, but perhaps I will just buy some from speakerbug, as they have what appear to be some very good quality coils in the values I need for very reasonable prices.
I need another two 0.1mH coils to replicate what is currently playing on the one speaker which is Not being obsessively sanded and sprayed... so figure I might as well go the whole hog and get jantzen coils for the lot.
Will also need to order some more Axons, and some decent resistors. Nothing exotic or expensive but certainly a step above my "prototype" components The axon caps at parts connexion are an absolute bargain IMO!!! $16.95 for a good quality 100uF polyproylene cap I think is very hard to beat
Tony.
Last edited:
This is exactly why I do not like and use subjective listening tests for determining performance of a loudspeaker.
First, your ears are not calibrated nor do they function the same from moment to moment.
Second, the input source (music) is not calibrated, so how do you intend to reference what you hear?
You have to test and you have to have confidence in your test procedure and test results.
It is always wise to doubt your procedure and results, but use confirmation methods were possible to remove doubt.
Stick with measurements until you are satisfied that what you see for the results is real, then listen and enjoy your hard and good work.
Hi Loren, Whilst I like to take a scientific approach, in the end even if the measurements look right if it doesn't sound good to me I'll change it. I'm not an accuracy at all costs type
Having said that, the crossover that I was running with, was tossed together using non-optimal components to get an idea as to whether the sim would give me accurate results or not, something that it successfully demonstrated, at least that I could get meaningful if not dead accurate correlation. The coil I was planning on using on the tweeter was 80uH which I didn't have so for the measurement exercise I just decided to see how the 4K notch went, but tonight I decided to sim it with 50uH and it looked ok so decided to try it since I had the coil sitting around gathering dust. The results (though I haven't measured them) are a definite improvement, and I'm very certain that is not psycho acoustics at play!
In this case If I re-compare the actual measured result of the speaker with the crossover that was running last night (green) the simulation of that same crossover black, and the simulation of the crossover that I'm now running with tonight (blue) I think you will see why it doesn't sound so hot!! I believe if I do the measurements it won't be exactly the same but it will have cut the tweeter response in the region that was causing me uncomfort. I am particularly sensitive to hot tweeters and probably pad them down more than some would care to. But that is me, I'd rather have an "inaccurate" speaker that I can listen to happily, than one that I just want to turn off
The other issue is I DONT have a calibrated mic, so anything above probably 5 - 6K is a bit of a guess anyway, however looking at the published response curves of the DMS37 tells me that my mic is probably pretty flat past 5K (the WM60AY's generally are very flat).
Cheers,
Tony.
Having said that, the crossover that I was running with, was tossed together using non-optimal components to get an idea as to whether the sim would give me accurate results or not, something that it successfully demonstrated, at least that I could get meaningful if not dead accurate correlation. The coil I was planning on using on the tweeter was 80uH which I didn't have so for the measurement exercise I just decided to see how the 4K notch went, but tonight I decided to sim it with 50uH and it looked ok so decided to try it since I had the coil sitting around gathering dust. The results (though I haven't measured them) are a definite improvement, and I'm very certain that is not psycho acoustics at play!
In this case If I re-compare the actual measured result of the speaker with the crossover that was running last night (green) the simulation of that same crossover black, and the simulation of the crossover that I'm now running with tonight (blue) I think you will see why it doesn't sound so hot!! I believe if I do the measurements it won't be exactly the same but it will have cut the tweeter response in the region that was causing me uncomfort. I am particularly sensitive to hot tweeters and probably pad them down more than some would care to. But that is me, I'd rather have an "inaccurate" speaker that I can listen to happily, than one that I just want to turn off
The other issue is I DONT have a calibrated mic, so anything above probably 5 - 6K is a bit of a guess anyway, however looking at the published response curves of the DMS37 tells me that my mic is probably pretty flat past 5K (the WM60AY's generally are very flat).
Cheers,
Tony.
Attachments
I thought I'd post an update since the aesthetics are *almost* done. I had a mad rush to make some grills before my 2 1/2 year old got home and I managed to do it (albeit a bit roughly). I thought the idea was ok but unfortunately in reality it was a) hard to do (which I was expecting anyway, especially with my limited tools) and b) the fit was not great I ended up having to stack some of the 5mm rare earth magnets I used in order to get them to stay on! But in the end it's primary purpose (ie to keep curious fingers away from the drivers) has been a success.
The design idea was to make the grill frames out of some 16mm MDF but only have 7mm thickness above the drivers. in between the edge and the drivers I planned to use some 6.4mm wool felt. The way this was to be achieved was to route out the back of the frame with a 45 degree chamfer and some shelves to match the baffle. This sort of worked but armed with only a router, a jigsaw, a dremel and some 4X2 It was a challenge and the final fit wasn't great.
I've decided to scrap the whole 45 degree chamfer idea and instead just make a frame 7mm thick that goes all round. My father in law has offered to do it on the cnc machine at his work (the outside also has a roundover) so it should be a bit better than my (admittedly rushed) efforts.
Attached are some pictures as always. I'll take some decent ones once they are really finished
I still haven't made a final decision on the crossovers but I am considering at least buying the components I'd need to try out a 3rd order acoustic butterworth which seems to sim pretty well. Certainly for watching tv and movies I haven't been disapointed, and I was frankly surprised by the performance (in the bass) for movies without the 10" drivers helping.
I thought I had more pics of the making of the grill frames but I don't apparently. Probably was too disgusted with the result thought I might have just been a bit pressed for time.
1st pic is just showing the jigsawing (I can't cut straight to save my life!! Outside edges were trimmed using a router bit and the bit of 4X2 as a guide. The other pics show the finished box with grill and without. I ended up just using some polyurathane estapol spray for the front baffles. They aren't great but at least they are sealed. I did three coats on the first one (and it still needed more sanding and respraying) and two on the second because I ran out of time.
Tony.
The design idea was to make the grill frames out of some 16mm MDF but only have 7mm thickness above the drivers. in between the edge and the drivers I planned to use some 6.4mm wool felt. The way this was to be achieved was to route out the back of the frame with a 45 degree chamfer and some shelves to match the baffle. This sort of worked but armed with only a router, a jigsaw, a dremel and some 4X2 It was a challenge and the final fit wasn't great.
I've decided to scrap the whole 45 degree chamfer idea and instead just make a frame 7mm thick that goes all round. My father in law has offered to do it on the cnc machine at his work (the outside also has a roundover) so it should be a bit better than my (admittedly rushed) efforts.
Attached are some pictures as always. I'll take some decent ones once they are really finished
I still haven't made a final decision on the crossovers but I am considering at least buying the components I'd need to try out a 3rd order acoustic butterworth which seems to sim pretty well. Certainly for watching tv and movies I haven't been disapointed, and I was frankly surprised by the performance (in the bass) for movies without the 10" drivers helping.
I thought I had more pics of the making of the grill frames but I don't apparently. Probably was too disgusted with the result
thought I might have just been a bit pressed for time. 1st pic is just showing the jigsawing (I can't cut straight to save my life!! Outside edges were trimmed using a router bit and the bit of 4X2 as a guide. The other pics show the finished box with grill and without. I ended up just using some polyurathane estapol spray for the front baffles. They aren't great but at least they are sealed. I did three coats on the first one (and it still needed more sanding and respraying) and two on the second because I ran out of time.
Tony.
Attachments
Thanks Terry, it's been a long time coming... At least it didn't take 20 years though to get the cabinets finished like my three ways did I still find it hard to believe it was seven years ago I first emailed you regarding getting some morel drivers (it took me 12 months to actually put the order in). Hopefully I'll be able to say they are truely finished before the end of the year. I woke up this morning and was thinking about another project, not for me but my aunt, who is a real music lover. I fixed her amp a couple of weeks ago, and was not particularly impressed with the phillips speakers she had (sounded like the tweeters have blown). Was thinking about what I could do that wasn't too expensive and wouldn't take me for ever to do... seems the speaker building bug has bitten me
Tony.
I still find it hard to believe it was seven years ago I first emailed you regarding getting some morel drivers (it took me 12 months to actually put the order in). Hopefully I'll be able to say they are truely finished before the end of the year. I woke up this morning and was thinking about another project, not for me but my aunt, who is a real music lover. I fixed her amp a couple of weeks ago, and was not particularly impressed with the phillips speakers she had (sounded like the tweeters have blown). Was thinking about what I could do that wasn't too expensive and wouldn't take me for ever to do... seems the speaker building bug has bitten me Tony.
OK Well I did some more measurements on the weekend, The first since I had "tweaked" the crossover by ear. My first thought was I'd cut the highs too much (looking at the measurements) but then looking more carefully, I decided that apart from the hump between 1 and 2khz, overall the speakers were probably more balanced.
I also measured with and without the grills on, and whilst the grills do have an effect, it is not as big as I thought it might be.
I did some off axis measurements both horizontal, and vertical. Horizontal look good. Vertical look horrid. Probably due to the 2nd order bessel (accoustic) I currently have in place.
I decided to play around with the sim some more, and I'm leaning towards trying out a 4th order accoustic LR at 2Khz at the moment based on the sims. BUT I just noticed something which doesn't seem right.
According to the sim of the effect of the high pass filter, the freq is getting a boost between about 5k and 16kHz.... This doesn't seem possible to me.
Anyone care to comment whether this is in fact a possible outcome, or whether speaker workshop has lost the plot
graph attached, as well as the crossover circuit that is supposedly going to deliver this Bonus spl.
Black trace is the simmed response and the blue trace is the original measurement that the sim was run on...
I'm glad I spotted this because I was thinking I had a winner on my hands. pretty simple and the summed response looked good. Shame it is probably not real
Third plot is the combined response (of the sim) Was planning on implementing a little line level BSC in my active crossover, probably 2-3db.
Tony.
I also measured with and without the grills on, and whilst the grills do have an effect, it is not as big as I thought it might be.
I did some off axis measurements both horizontal, and vertical. Horizontal look good. Vertical look horrid. Probably due to the 2nd order bessel (accoustic) I currently have in place.
I decided to play around with the sim some more, and I'm leaning towards trying out a 4th order accoustic LR at 2Khz at the moment based on the sims. BUT I just noticed something which doesn't seem right.
According to the sim of the effect of the high pass filter, the freq is getting a boost between about 5k and 16kHz.... This doesn't seem possible to me.
Anyone care to comment whether this is in fact a possible outcome, or whether speaker workshop has lost the plot
graph attached, as well as the crossover circuit that is supposedly going to deliver this Bonus spl.
Black trace is the simmed response and the blue trace is the original measurement that the sim was run on...
I'm glad I spotted this because I was thinking I had a winner on my hands. pretty simple and the summed response looked good. Shame it is probably not real
Third plot is the combined response (of the sim) Was planning on implementing a little line level BSC in my active crossover, probably 2-3db.
Tony.
Attachments
Thanks Terry, just me being me!! Always trying to ring every last drop of performance out of something I put the frequency response and impedance data into jeff bagbies crossover simulation spreadsheet again and did a sanity check. The result in the sim matches the SW one, with the same components (there are no notch filters in this crossover so it does a good job), the difference is that the raw tweeter response is actually showing higher, speakerworkshop is doing something dumb when I overlay the original on the graph of the simulated response. So It wasn't magically bumping up the tweeter levels
I've attached the response measurement of the speakers as they are a the moment. I didn't like the sound with the notch filter that was reducing the peak between 1k and 2k. I think you will see it isn't as good as I had previously posted (objectively) but subjectively it sounds a lot better to me, the 1k notch was doing something nasty! I do think though it should sound better with that 1k-2k bump flattened out. Well I think it is worth having a shot at it anyway Yep 1.5db not worth worrying about, but I think (and could be wrong) an almost one octave wide 3db peak probably will be audible
The proposed crossover isn't too complex (I don't think). Time will tell I guess whether it is better or worse (or perhaps indifferent).
Note that the response graph is an attempt at getting vertical on and off axis measurements. black on axis, red 15 deg, blue 30 deg, lime 45 deg
The 45degree one is pretty much useless as the mic was only about 26cm off the ground so it was overwhelmed by floor bounce. I know that MTM's generally have a failry focussed vertical response, but I was a bit surprised that the 15 degrees off axis was so different. Measurements were taken at 1M so could be different further back.
Tony.
I put the frequency response and impedance data into jeff bagbies crossover simulation spreadsheet again and did a sanity check. The result in the sim matches the SW one, with the same components (there are no notch filters in this crossover so it does a good job), the difference is that the raw tweeter response is actually showing higher, speakerworkshop is doing something dumb when I overlay the original on the graph of the simulated response. So It wasn't magically bumping up the tweeter levels I've attached the response measurement of the speakers as they are a the moment. I didn't like the sound with the notch filter that was reducing the peak between 1k and 2k. I think you will see it isn't as good as I had previously posted (objectively) but subjectively it sounds a lot better to me, the 1k notch was doing something nasty! I do think though it should sound better with that 1k-2k bump flattened out. Well I think it is worth having a shot at it anyway
Yep 1.5db not worth worrying about, but I think (and could be wrong) an almost one octave wide 3db peak probably will be audible The proposed crossover isn't too complex (I don't think). Time will tell I guess whether it is better or worse (or perhaps indifferent).
Note that the response graph is an attempt at getting vertical on and off axis measurements. black on axis, red 15 deg, blue 30 deg, lime 45 deg
The 45degree one is pretty much useless as the mic was only about 26cm off the ground so it was overwhelmed by floor bounce. I know that MTM's generally have a failry focussed vertical response, but I was a bit surprised that the 15 degrees off axis was so different. Measurements were taken at 1M so could be different further back.
Tony.
Attachments
What I always do next is to set up for measurement and then take out each component one at a time and see what happens. If they don't make a difference leave them out. If they do put them back. Do the same for a listening session. It is amazing what you find out. Try it and see what you think.
Terry
Terry
Good Idea Terry! I did do some measurements varying caps and resistors but didn't try actually removing things!
I looked more carefully at the sim in Jeff's spreadsheet and it would appear that I'm chasing a pipe dream (ie +- 1.5db between 1k and 16k with a simple crossover). It gave the same results as SW but when I looked more closesly it is giving an artificial boost as well (around 2db)... the input was 93db and the output was 95db... somehow I don't think so
On thinking back the one area when I did my measurements that didn't match the sims well, was the higher frequencies which were down quite a bit. Now I know why!!
I think I will still have a go at the higher order network, but I will be a bit more cautious about the results of the sims
Cheers,
Tony.
edit: there is something sus about my off axis (horizontal) measurements too. fantastic to 4K then drop like a stone (on the 15 deg plot), off axis... perhaps it is time I invested in a protractor or read my d'appolito book again
thats on axis, 15 30 45 60 and 90 degrees off axis. can you tell I've been frequenting the measurements thread
I did do some measurements varying caps and resistors but didn't try actually removing things! I looked more carefully at the sim in Jeff's spreadsheet and it would appear that I'm chasing a pipe dream (ie +- 1.5db between 1k and 16k with a simple crossover). It gave the same results as SW but when I looked more closesly it is giving an artificial boost as well (around 2db)... the input was 93db and the output was 95db... somehow I don't think so
On thinking back the one area when I did my measurements that didn't match the sims well, was the higher frequencies which were down quite a bit. Now I know why!!
I think I will still have a go at the higher order network, but I will be a bit more cautious about the results of the sims
Cheers,
Tony.
edit: there is something sus about my off axis (horizontal) measurements too. fantastic to 4K then drop like a stone (on the 15 deg plot), off axis... perhaps it is time I invested in a protractor
or read my d'appolito book again thats on axis, 15 30 45 60 and 90 degrees off axis. can you tell I've been frequenting the measurements thread
Attachments
Last edited:
Good Idea Terry!
I looked more carefully at the sim in Jeff's spreadsheet and it would appear that I'm chasing a pipe dream (ie +- 1.5db between 1k and 16k with a simple crossover). It gave the same results as SW but when I looked more closesly it is giving an artificial boost as well (around 2db)... the input was 93db and the output was 95db... somehow I don't think so
On thinking back the one area when I did my measurements that didn't match the sims well, was the higher frequencies which were down quite a bit. Now I know why!!
I think I will still have a go at the higher order network, but I will be a bit more cautious about the results of the sims
Cheers,
Tony.
edit: there is something sus about my off axis (horizontal) measurements too. fantastic to 4K then drop like a stone (on the 15 deg plot), off axis... perhaps it is time I invested in a protractor
thats on axis, 15 30 45 60 and 90 degrees off axis. can you tell I've been frequenting the measurements thread
I am confused. Are these plots actual measurements or just simulations?
Hi Loren,
The two plots in post 92 are simulations. The blue line in the first one is the actual tweeter response, and the black is with the simulated crossover. The plot in post 94 is actual measured response off axis in the vertical plane. curves are smoothed 1/16th octave.
The plot in post 96 is also actual measured data, also smoothed 1/16th octave.
I think I've worked out where I went wrong with the measurements off axis (horizontal). I rotated the speaker on its centre axis, where I should have rotated it around an axis level with the front baffle (not the centre of the speaker). Because the tweeters are semi horn loaded the mic wouldn't have been "seeing" the actual dome and the effective angle would have been much greater. Doh.
Tony.
The two plots in post 92 are simulations. The blue line in the first one is the actual tweeter response, and the black is with the simulated crossover. The plot in post 94 is actual measured response off axis in the vertical plane. curves are smoothed 1/16th octave.
The plot in post 96 is also actual measured data, also smoothed 1/16th octave.
I think I've worked out where I went wrong with the measurements off axis (horizontal). I rotated the speaker on its centre axis, where I should have rotated it around an axis level with the front baffle (not the centre of the speaker). Because the tweeters are semi horn loaded the mic wouldn't have been "seeing" the actual dome and the effective angle would have been much greater. Doh.
Tony.
That I am Terry!! I think I've come a long way from the point where I was dreading trying to do the crossover! I should have posted a link to one of my blog entries the other day, it gives an insight into my problem with "going too far" The problem with wanting to get the best performance It's a bit long winded but hopefully either enlightening or entertaining
Cheers,
Tony.
I think I've come a long way from the point where I was dreading trying to do the crossover! I should have posted a link to one of my blog entries the other day, it gives an insight into my problem with "going too far" The problem with wanting to get the best performance It's a bit long winded but hopefully either enlightening or entertaining Cheers,
Tony.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.