All,
I have created a 3 way ½ active speaker system, as shown in the picture below. The results are quite impressive in most respects, ie dynamic range and accuracy, but the system does not give the presence of the mini monitors alone. Soundstage in particular, is massivley degraded. The active X-over is ESP 24dB/oct LR at 300Hz between the bass and mid. I thought driver output phase could be a problem, but moving the units relative to each other does not seem the have an appreciable effect. Could this just be the added complexity of going from 2 to 3 way? Any ideas?
Regards
Mark
I have created a 3 way ½ active speaker system, as shown in the picture below. The results are quite impressive in most respects, ie dynamic range and accuracy, but the system does not give the presence of the mini monitors alone. Soundstage in particular, is massivley degraded. The active X-over is ESP 24dB/oct LR at 300Hz between the bass and mid. I thought driver output phase could be a problem, but moving the units relative to each other does not seem the have an appreciable effect. Could this just be the added complexity of going from 2 to 3 way? Any ideas?
Regards
Mark
Attachments
Yes, that's a part of the added complexity. With a three way system you are facing about ten times as much problems compared to a decent two way system.
You are using some off the shelve 24dB/oct crossovers. This is a electrical 24dB/oct, not acoustical. You have to include the drivers responce in the calculation of the filters. The midrange in your small enclosure probably has a roll off at 80 to 100 Hz with 12 dB/oct. To get a acousticaly 24dB/oct at 300Hz you only might need a 12 or 18 dB/oct electrical crossover, maybe with some correction at the low end. The same counts for the woofer. This one has probably a 3 to 6dB/oct high frequency rise that needs to be compensated in order to get a acoustically 24dB/oct at 300 Hz or else you will get only a crossover of somewhere between 18 and 20 dB/oct with all the phase differences.
Moving the top speaker to the back or front does not solve these problems.
For example. If your satelite has an acoustically roll off at 80Hz with 12 dB/oct, adding a second order butterworth filter at 80 Hz will give an acoustical fourth order Linkwitz filter at 80 Hz.
You are using some off the shelve 24dB/oct crossovers. This is a electrical 24dB/oct, not acoustical. You have to include the drivers responce in the calculation of the filters. The midrange in your small enclosure probably has a roll off at 80 to 100 Hz with 12 dB/oct. To get a acousticaly 24dB/oct at 300Hz you only might need a 12 or 18 dB/oct electrical crossover, maybe with some correction at the low end. The same counts for the woofer. This one has probably a 3 to 6dB/oct high frequency rise that needs to be compensated in order to get a acoustically 24dB/oct at 300 Hz or else you will get only a crossover of somewhere between 18 and 20 dB/oct with all the phase differences.
Moving the top speaker to the back or front does not solve these problems.
For example. If your satelite has an acoustically roll off at 80Hz with 12 dB/oct, adding a second order butterworth filter at 80 Hz will give an acoustical fourth order Linkwitz filter at 80 Hz.
That's probably an accurate diagnosis- you don't actually have an acoustic LR4. My suspicion is that the problem is greater with the high-pass end of the mid.
But... just to be sure, double-check (maybe have someone else do it, sometimes I'm just blind to my own mistakes) all polarities to make sure something somewhere isn't backward. That's really easy to do when there's about two dozen wires to keep track of. Use impulses, test tones, AAA batteries, whatever you need to verify relative polarity mid to bass and channel to channel.
Moving the drivers won't help much because high-order slopes aren't super-sensitive to offset and the wavelenths at crossover are a few feet, not inches (sorry, "70 centimeters, not 70 millimeters").
But... just to be sure, double-check (maybe have someone else do it, sometimes I'm just blind to my own mistakes) all polarities to make sure something somewhere isn't backward. That's really easy to do when there's about two dozen wires to keep track of. Use impulses, test tones, AAA batteries, whatever you need to verify relative polarity mid to bass and channel to channel.
Moving the drivers won't help much because high-order slopes aren't super-sensitive to offset and the wavelenths at crossover are a few feet, not inches (sorry, "70 centimeters, not 70 millimeters").
Im actually very surprised at the results you describe. I had all passive on my 3way.
peerless 850146 sealed box bass
S Excel w15cy001 sealed box mid
SS D2905/950000 tweet er....sealed box
Anyway the xover between the full lot originally was all passive
with a single inductor on the bass and a single cap on the mid.
Running the mid/tweet full range sounded good (now i realise the xover isnt perfect but I sorted out the 8.2k peak in the mag cone) adding the bass bin and the xover didnt mess up the sound stage what-so-ever. It just added in some bloom that was quite unlpleasent, but the sound stage didnt really do bad. Oh and it did the bass MUCH better 🙂
Then I added active to the bass and mid. These were both at 12/db simple textbook xovers at 300hz. This would produce a sort of 24db on the mid and a 12db on the bass. This sounded better then the passive by quite a way, so I left it.
Then I rebuilt the xover with 24db on the bass and 12db on the mid and half optomised it using some free software. This was with LSPcad and their sample drivers. As both the bass they have and I have are roughly flat in the area I was crossing (now reduced to 150hz this sounded better as the w15 produced more) I simulated 150hz LWR 24db with it and then another one a 24db in a sealed box with the 12db active. This basically just made everything sound better intigrated.
Bottom line is you can refine the xover and you will end up with better soundquality.
But im surprised that it really messes with the SS, if this was a small degredation then I could understand it.
But then again these are completely different rooms diff setups diff speakers, so all of that will make a difference.
I use trimpots on all resistors to allow everything to be altered as desired.
I am making the full system active at the mo and am in the process of getting the PCB's etched, so Ill report how this goes when its done.
Hope you sort it Matt
One other thing to check for is the actual circuits are they working as you would expect them too. I have had problems in the past as all my things were done on vero strip board. This stuff is great for prototyping and stuff but it is easy to miss something. Like a track needs to be broken or some solder has joined some things together when it shouldnt have. Plus listen to each channel and see if it sounds the same, if it doesnt then worry and check everything is in order. You could have something really weird going on in the xover which is generating this god-awful sound.
I know full optomised and well integrated will sound better, but the wavelengths at 300hz are about 1 meter so you would need the drivers to be a long way away from each other to really mess up filter slopes that are say textbook and would sum (if on a woof and tweet at 3000hz) with a big peak. But here the effect is nowhere near as big.
Try reversing the phase of one of the drivers and see what effect that has on the sound
peerless 850146 sealed box bass
S Excel w15cy001 sealed box mid
SS D2905/950000 tweet er....sealed box
Anyway the xover between the full lot originally was all passive
with a single inductor on the bass and a single cap on the mid.
Running the mid/tweet full range sounded good (now i realise the xover isnt perfect but I sorted out the 8.2k peak in the mag cone) adding the bass bin and the xover didnt mess up the sound stage what-so-ever. It just added in some bloom that was quite unlpleasent, but the sound stage didnt really do bad. Oh and it did the bass MUCH better 🙂
Then I added active to the bass and mid. These were both at 12/db simple textbook xovers at 300hz. This would produce a sort of 24db on the mid and a 12db on the bass. This sounded better then the passive by quite a way, so I left it.
Then I rebuilt the xover with 24db on the bass and 12db on the mid and half optomised it using some free software. This was with LSPcad and their sample drivers. As both the bass they have and I have are roughly flat in the area I was crossing (now reduced to 150hz this sounded better as the w15 produced more) I simulated 150hz LWR 24db with it and then another one a 24db in a sealed box with the 12db active. This basically just made everything sound better intigrated.
Bottom line is you can refine the xover and you will end up with better soundquality.
But im surprised that it really messes with the SS, if this was a small degredation then I could understand it.
But then again these are completely different rooms diff setups diff speakers, so all of that will make a difference.
I use trimpots on all resistors to allow everything to be altered as desired.
I am making the full system active at the mo and am in the process of getting the PCB's etched, so Ill report how this goes when its done.
Hope you sort it Matt
One other thing to check for is the actual circuits are they working as you would expect them too. I have had problems in the past as all my things were done on vero strip board. This stuff is great for prototyping and stuff but it is easy to miss something. Like a track needs to be broken or some solder has joined some things together when it shouldnt have. Plus listen to each channel and see if it sounds the same, if it doesnt then worry and check everything is in order. You could have something really weird going on in the xover which is generating this god-awful sound.
I know full optomised and well integrated will sound better, but the wavelengths at 300hz are about 1 meter so you would need the drivers to be a long way away from each other to really mess up filter slopes that are say textbook and would sum (if on a woof and tweet at 3000hz) with a big peak. But here the effect is nowhere near as big.
Try reversing the phase of one of the drivers and see what effect that has on the sound
Looks like SY said what I said in a few sentences LOL I must learn to condence. I can write for ever and ever and ever and ever and ever and ever....... OK ill stop, on DIY audio. Anyway...Im leavin bzzzzz!!!
Nah, I missed the part about verifying that the circuits themselves were functioning properly. I need to practice more patience to get my replies longer 😉
Just one real simple listening test. Since you are active at the point the problems began. Just turn off one amp and listen to the low frequency section by itself. If you hear a lot of midrange, your are not crossing at 300 Hz at 24 db per octave.
The human ear/brain triggers on faster transients for spatial location. If the monitors sound fine by themselves, and adding the bass degrades, the two obvious causes would be midrange contamination from the bass section or a bass level much too high (a level imbalance can also produce more midrange contamination).
Either would be revealed by isolating each section and listening.
Good luck (maximum frustration always occurs during first implementation),
Mark
The human ear/brain triggers on faster transients for spatial location. If the monitors sound fine by themselves, and adding the bass degrades, the two obvious causes would be midrange contamination from the bass section or a bass level much too high (a level imbalance can also produce more midrange contamination).
Either would be revealed by isolating each section and listening.
Good luck (maximum frustration always occurs during first implementation),
Mark
EXCELLENT!
thanks for your help everyone, found it, the mid-trebble unit in the photo was wired wrong, found it by sweeping each channel through the X-over region with the sig' gen'. the right side sounded decidedly wierd, so i checked the wiring and hey presto, i have a sound stage.
I hear everyones comments about the X-over summing electrically, but not acoustically. The theory i have used is that the bass and mid drivers are mounted in the boxes of the 2 way kits which they are originally part of. Both kits sounded very good passive, so going active i should just have to alter the gains of the respective power amps to level match at the X-over point. There may be problems with this assumption, but the output is a lot more linear over the X-over region than in other parts of the driver range, which i was listening to in passive mode anyway. The problen is, the big Volt bass driver has a lot more linear output upto 3000Hz than the smaller monacor sph-135 mid, i can hear this easily on a frequency sweep. so really i should swap the bass and mid drivers, but the it would look even worse than it does now! Hope this all makes sense and Thanks again.
Mark
thanks for your help everyone, found it, the mid-trebble unit in the photo was wired wrong, found it by sweeping each channel through the X-over region with the sig' gen'. the right side sounded decidedly wierd, so i checked the wiring and hey presto, i have a sound stage.
I hear everyones comments about the X-over summing electrically, but not acoustically. The theory i have used is that the bass and mid drivers are mounted in the boxes of the 2 way kits which they are originally part of. Both kits sounded very good passive, so going active i should just have to alter the gains of the respective power amps to level match at the X-over point. There may be problems with this assumption, but the output is a lot more linear over the X-over region than in other parts of the driver range, which i was listening to in passive mode anyway. The problen is, the big Volt bass driver has a lot more linear output upto 3000Hz than the smaller monacor sph-135 mid, i can hear this easily on a frequency sweep. so really i should swap the bass and mid drivers, but the it would look even worse than it does now! Hope this all makes sense and Thanks again.
Mark
Just add a level match to turn down the bass and see what happens, if both kits sounded good passive then simply removing the bass from one and the treb from the other should result in a good match again. But the level of the two will need to be compensated for. Linkwitzlab.com has a useful gain application -+ 6db, I have used this and it works well. Alternatively you could add a few resistors to the bass to pad it down. I know this isnt ideal but for a quick fix would be better then the bass playing too loud. As it it active the xover wont change due to the extra resistance, but the box loading will.
If you have a pot handy put it in the bass out on the xover to pad it down, or if its not hard to do just build the little linkwitz circuit.
If you have a preamp or amp with preouts you could use that too.
Hope everything ends up as you would like it.
Congratz on the sound stage too, I cant stand a hifi with a bad SS it ruins otherwise good music.
Matt
If you have a pot handy put it in the bass out on the xover to pad it down, or if its not hard to do just build the little linkwitz circuit.
If you have a preamp or amp with preouts you could use that too.
Hope everything ends up as you would like it.
Congratz on the sound stage too, I cant stand a hifi with a bad SS it ruins otherwise good music.
Matt
- Status
- Not open for further replies.