If you see a substantive difference from paralleled caps you'll be doing something wrong. Same with foil alignment. Sonically? Paralleled might have some advantages in terms of greater plate area, depending on cap construction, it's one of the things I like about "charge coupled" caps. Might be worth looking into that- some have even done charge-coupled balanced crossovers, though I've not tried the balanced crossover model- it doesn't make a lot of sense to me as a performance improvement option, where I can see the bias from charge coupling as having a logical MOA.
I will draw three mini conclusions from the above testing:
1) the outer foil of a cap should be at the driver end, this when using tube amplification. I will guess the outer foil should be at the amp end when using a solid state amp.
2) as far as the tweeter is concerned, capacitors below the optimal value, at the optimal value, or above the value, don't make a big difference. The 4.7uF trace is an inexpensive cap. Distortion was higher than expected.
3) multi strand wire is out. There was a uniform rise across the entire spectrum.
Tomorrow we test more variations.
1) the outer foil of a cap should be at the driver end, this when using tube amplification. I will guess the outer foil should be at the amp end when using a solid state amp.
2) as far as the tweeter is concerned, capacitors below the optimal value, at the optimal value, or above the value, don't make a big difference. The 4.7uF trace is an inexpensive cap. Distortion was higher than expected.
3) multi strand wire is out. There was a uniform rise across the entire spectrum.
Tomorrow we test more variations.
It is certainly sound practise to earth the outer layer of a capacitor, or the outer winding of a coil where possible. It reduces common mode noise. As does careful earthing. But more of a Radio Frequency thing TBH, where these effects can become significant, and given that you even know which is the outer layer. 
Now Bill, I have made some more progress with your circuit, and I am liking what I see. Start with small baby steps here, I present the 3 way version. Sorry the schematic could be better drawn. I actually amazed myself in correctly saying your circuit does some of the same things as the Fried type. Here's some hard data. The B80 is a good small fullranger.
The 4 way is done, but is trickier to align so far. Another day. Let's wrap our heads round this extremely elegant circuit first.
Now Bill, I have made some more progress with your circuit, and I am liking what I see. Start with small baby steps here, I present the 3 way version. Sorry the schematic could be better drawn. I actually amazed myself in correctly saying your circuit does some of the same things as the Fried type. Here's some hard data. The B80 is a good small fullranger.
The 4 way is done, but is trickier to align so far. Another day. Let's wrap our heads round this extremely elegant circuit first.
Attachments
Good morning Steve (system7)
That looks fabulous.
1) Regarding your phase graph. Could you explain to me what the highlights are. I'm one of those 'it needs to sound great, who cares how it measures' kinda people. You are my exact opposite, I like that.
2) This morning I will test many things, but more along the lines of what I did yesterday.
3) I have all the values on your schematic in house. If I can find the time this afternoon, I'm gonna try it out. If not, it'll be tomorrows project.
4) possible hiccups with this test;
a - my Acoustic Elegance TD15H+ 15" woofer isn't suited to play out to 1000Hz. For testing it'll be fine, just wouldn't use it in real life. Fkeller's TD12S may be the best driver I know of for your 3-way.
b - my Acoustic Elegance TD6M 6" is acoustically cut off at 150Hz, so it won't play low. For testing it could be used for the woofer with that one caveat.
c - my Accuton C50 will be tested for the midrange portion, but again, not in real life. Too much low frequency energy. Fkeller is looking for a midrange. I will be using the Morel MDM55 in one of my builds, with another using the Morel EM 1308. Either of these drivers will fit your schematic perfectly.
d - any tweeter on the planet will work.
Off to the basement for testing.
That looks fabulous.
1) Regarding your phase graph. Could you explain to me what the highlights are. I'm one of those 'it needs to sound great, who cares how it measures' kinda people. You are my exact opposite, I like that.
2) This morning I will test many things, but more along the lines of what I did yesterday.
3) I have all the values on your schematic in house. If I can find the time this afternoon, I'm gonna try it out. If not, it'll be tomorrows project.
4) possible hiccups with this test;
a - my Acoustic Elegance TD15H+ 15" woofer isn't suited to play out to 1000Hz. For testing it'll be fine, just wouldn't use it in real life. Fkeller's TD12S may be the best driver I know of for your 3-way.
b - my Acoustic Elegance TD6M 6" is acoustically cut off at 150Hz, so it won't play low. For testing it could be used for the woofer with that one caveat.
c - my Accuton C50 will be tested for the midrange portion, but again, not in real life. Too much low frequency energy. Fkeller is looking for a midrange. I will be using the Morel MDM55 in one of my builds, with another using the Morel EM 1308. Either of these drivers will fit your schematic perfectly.
d - any tweeter on the planet will work.
Off to the basement for testing.
Bill, I'm just playing with a new toy here, but I'm getting the hang of it now.
Power response is typical 3 or 4 way flat. Tweeter distortion might be improvable. There is no difficulty or any bad effect in adding a 10uF to the tweet to make it 3rd order. Takes the Fs down 6dB.
I don't know what to make of the phase yet, but it looks a bit 90 degree in places, which is OK too. Component values are not all that critical, it just seems to bend everything around changes.
Power response is typical 3 or 4 way flat. Tweeter distortion might be improvable. There is no difficulty or any bad effect in adding a 10uF to the tweet to make it 3rd order. Takes the Fs down 6dB.
I don't know what to make of the phase yet, but it looks a bit 90 degree in places, which is OK too. Component values are not all that critical, it just seems to bend everything around changes.
Attachments
Last edited:
Juhazi, I just saw your post.
I'm going to make a statement, and I'd like to be told if my thinking is incorrect.
Looking at system7's first picture, we notice a smallish midrange. Unless I'm mistaken, small diameter drivers have wayyyyyy better off axis response than larger drivers (as frequencies rise). The one thing you need to be mindful of is the center-to-center spacing. However, by offsetting the woofer from the mid and tweeter, we get a better 'sweet spot' response, at the expense of off axis response?.
I'm going to make a statement, and I'd like to be told if my thinking is incorrect.
Looking at system7's first picture, we notice a smallish midrange. Unless I'm mistaken, small diameter drivers have wayyyyyy better off axis response than larger drivers (as frequencies rise). The one thing you need to be mindful of is the center-to-center spacing. However, by offsetting the woofer from the mid and tweeter, we get a better 'sweet spot' response, at the expense of off axis response?.
I'm new to this thread as well, but I read something you wrote a few days back that told me you have the mind of a good DIY speaker builder.
Some of us have the experience to see why an idea won't be satisfactory, but let me tell you.. these conclusions aren't usually obvious until you've tried and tried. The most important reason for this IMO, is only becoming clear to the DIY community as of this decade.
I had to stop my testing and get to the computer.
badman made a comment in post #81. "Paralleled (caps) might have some advantages in terms of greater plate area".
I've seen Troel's do this many times. Love his website. Love Tony G's humble website as well.
FABULOUS (love that word)
Picture 1) New day, new start, new 6uF at C1 and C1B distortion measurement.
Picture 2) Can This Be!!!!!, distortion is down, and I mean across the board. I had to measure this three times to make sure. This with two caps of different values, 3.3uF and 2.7uF. Look how low D2 is in the crossover region (4200Hz)
Time to go measure what two identical caps in parallel look like.
I think I should listen to whatever badman says. Yes Steve, that includes you.
badman made a comment in post #81. "Paralleled (caps) might have some advantages in terms of greater plate area".
I've seen Troel's do this many times. Love his website. Love Tony G's humble website as well.
FABULOUS (love that word)
Picture 1) New day, new start, new 6uF at C1 and C1B distortion measurement.
Picture 2) Can This Be!!!!!, distortion is down, and I mean across the board. I had to measure this three times to make sure. This with two caps of different values, 3.3uF and 2.7uF. Look how low D2 is in the crossover region (4200Hz)
Time to go measure what two identical caps in parallel look like.
I think I should listen to whatever badman says. Yes Steve, that includes you.
Attachments
Opps, I don't have enough small value capacitors to test my 'identical' paralleled cap theory.
Midrange testing will use single caps, with the knowledge that final assembly will use paralleled caps.
Picture 1) Tweeter measurement with 6uF. This is for reference.
Picture 2) This is the midrange measurement. The tweeters capacitors are 6uF each.
D2; 40dB till 2000Hz. Drops till it crosses 30dB @ 5100Hz. ~24dB till 9500.
D3; Under 40 till 700Hz. Mid 30's till 3700. Dropping below 20dB
Picture 3) Just for fun, used paralleled caps (3.3 + 2.7). May as well have been the same measurement. Conclusion, the midrange is not affected by tweeter cap arrangement, only the value.
Picture 4) Dropping the tweeter cap values to 4.7uF. We didn't like this cap before. Let's see if foil orientation makes a difference.
D2; 40dB till 2200Hz. 30dB @ 5200Hz, 23dB till 9500.
D3; Under 40 till 700Hz. Mid 30's till 5200Hz (with some dips in between).
Picture 5) We just turned the 4.7uF cap around. Not a big difference from picture 4. Slightly worse than 6uF.
Picture 6) We drop to 3.9uF, and it's just plain worse.
Conclusions.
In Post # 5, sreten commented "capacitors affect the x-o point". 100% correct. The only thing we need to make sure of, the capacitors for the tweeter need to be higher than the normal series crossover ideal.
Our value of 6uF is more than twice what the parallel calculation would suggest.
In Post # 16, Bob Richard commented (near the end of his paragraph) "When you change one part's value, it throws off other parts".
Changing the caps for the tweeter affects the crossover point. Choosing a value too small adds distortion to the midrange.
Our new choice of 6uF now puts our crossover point below 4000Hz.
Not my ideal. I want the Accuton C50 (C51 in the future) to cross above 4000. Troels suggests the 2" Accuton can play easily till 5000Hz. He also calls it the best midrange he's ever heard.
I don't have any large guage (12awg) .25mH or .27mH lying around. Time to spend more money.
Midrange testing will use single caps, with the knowledge that final assembly will use paralleled caps.
Picture 1) Tweeter measurement with 6uF. This is for reference.
Picture 2) This is the midrange measurement. The tweeters capacitors are 6uF each.
D2; 40dB till 2000Hz. Drops till it crosses 30dB @ 5100Hz. ~24dB till 9500.
D3; Under 40 till 700Hz. Mid 30's till 3700. Dropping below 20dB
Picture 3) Just for fun, used paralleled caps (3.3 + 2.7). May as well have been the same measurement. Conclusion, the midrange is not affected by tweeter cap arrangement, only the value.
Picture 4) Dropping the tweeter cap values to 4.7uF. We didn't like this cap before. Let's see if foil orientation makes a difference.
D2; 40dB till 2200Hz. 30dB @ 5200Hz, 23dB till 9500.
D3; Under 40 till 700Hz. Mid 30's till 5200Hz (with some dips in between).
Picture 5) We just turned the 4.7uF cap around. Not a big difference from picture 4. Slightly worse than 6uF.
Picture 6) We drop to 3.9uF, and it's just plain worse.
Conclusions.
In Post # 5, sreten commented "capacitors affect the x-o point". 100% correct. The only thing we need to make sure of, the capacitors for the tweeter need to be higher than the normal series crossover ideal.
Our value of 6uF is more than twice what the parallel calculation would suggest.
In Post # 16, Bob Richard commented (near the end of his paragraph) "When you change one part's value, it throws off other parts".
Changing the caps for the tweeter affects the crossover point. Choosing a value too small adds distortion to the midrange.
Our new choice of 6uF now puts our crossover point below 4000Hz.
Not my ideal. I want the Accuton C50 (C51 in the future) to cross above 4000. Troels suggests the 2" Accuton can play easily till 5000Hz. He also calls it the best midrange he's ever heard.
I don't have any large guage (12awg) .25mH or .27mH lying around. Time to spend more money.
Attachments
My consern of distortion waas focuse on the mid driver blowing quite hard even below 300Hz and even lower. Acoustic measurement at 90dB and say 96dB at 1m should be run. The mid will propably exceed it's Xmax quite easily and even before that distortion is severe. Dedicated mids don't have much tolerance (Xmax), fullranges might be better.
Consern of off -axis was mainly for vertical, lots of overlapping W-M and M-T will cause lobing because of interference. How much this is a problem is questionable.
Consern of off -axis was mainly for vertical, lots of overlapping W-M and M-T will cause lobing because of interference. How much this is a problem is questionable.
system7's crossover
Picture 1) Midrange measurement. I left my own 4.7uF trace for reference.
Picture 2) I removed my red trace.
Picture 3) Tweeter measurement.
Picture 4) Tweeter with my trace removed.
Picture 5) Woofer measurement.
Picture 6) I decreased the range.
As expected, 1st order crossover have large overlaps.
I did a quick listen with music, at moderate levels, not bad at all.
I would cut the 3.5mH inductor in half. This will significantly raise the high pass for the midrange.
Picture 1) Midrange measurement. I left my own 4.7uF trace for reference.
Picture 2) I removed my red trace.
Picture 3) Tweeter measurement.
Picture 4) Tweeter with my trace removed.
Picture 5) Woofer measurement.
Picture 6) I decreased the range.
As expected, 1st order crossover have large overlaps.
I did a quick listen with music, at moderate levels, not bad at all.
I would cut the 3.5mH inductor in half. This will significantly raise the high pass for the midrange.
Attachments
I've made a few decisions.
I've decided to show you what my Wife goes through.
I've decided to show you what I think of the stupidest animal on planet Earth.
I'm content to be laughed at.
Picture 1) I've lowered the 3.5mH inductor from system7's crossover. The new value is 1.5mH.
This picture shows very little change in the woofer response. Certainly nothing near what one would expect. This shows me so much it's scary. I have no doubt very few of you will accept it.
At some point in our history, someone decided we needed to be bent over, pants pulled down, and you can guess the rest.
The theories used for series crossovers today are absolute crap.
There is no such thing as a first order series crossover. The slopes are not compatible with todays drivers.
Picture 2) The midrange. We now cross closer to 750Hz. This would be quite listenable at moderate levels. No way in hades you would turn up the volume.
Steve. Are you able to change your simulations code?. It's plain and simple wrong.
Solid State amplifier manufacturers paid off stereo magazines. This was a long time ago. We where told "distortion is 0.000000000000001%". What they didn't tell us was that is sounded like crap. Everything changed from there.
When was the last time you went to an audio show, and heard a speaker that floored you?. One that tricked your brain into thinking the performers where in the room with you?. Not for 30-40 years.
Like good cattle we followed the cow in front of us, got on the truck, and went to the slaughterhouse.
I swear to you, this will change.
Please do me a favor. Stop letting our newbies think about the latest abortion, the DSP.
I've decided to show you what my Wife goes through.
I've decided to show you what I think of the stupidest animal on planet Earth.
I'm content to be laughed at.
Picture 1) I've lowered the 3.5mH inductor from system7's crossover. The new value is 1.5mH.
This picture shows very little change in the woofer response. Certainly nothing near what one would expect. This shows me so much it's scary. I have no doubt very few of you will accept it.
At some point in our history, someone decided we needed to be bent over, pants pulled down, and you can guess the rest.
The theories used for series crossovers today are absolute crap.
There is no such thing as a first order series crossover. The slopes are not compatible with todays drivers.
Picture 2) The midrange. We now cross closer to 750Hz. This would be quite listenable at moderate levels. No way in hades you would turn up the volume.
Steve. Are you able to change your simulations code?. It's plain and simple wrong.
Solid State amplifier manufacturers paid off stereo magazines. This was a long time ago. We where told "distortion is 0.000000000000001%". What they didn't tell us was that is sounded like crap. Everything changed from there.
When was the last time you went to an audio show, and heard a speaker that floored you?. One that tricked your brain into thinking the performers where in the room with you?. Not for 30-40 years.
Like good cattle we followed the cow in front of us, got on the truck, and went to the slaughterhouse.
I swear to you, this will change.
Please do me a favor. Stop letting our newbies think about the latest abortion, the DSP.
Attachments
system7, how did I miss your schematic from post #87.
Your values are very close to my final values.
Only changes; we add the 2nd order components (greater slopes), we raise the cap values.
If I have time today I will test system7's crossover.
I will then add the 2nd order components, and measure.
Steve, are you able to write a new program?, or edit the one that you have?
Your values are very close to my final values.
Only changes; we add the 2nd order components (greater slopes), we raise the cap values.
If I have time today I will test system7's crossover.
I will then add the 2nd order components, and measure.
Steve, are you able to write a new program?, or edit the one that you have?
Cousin Billy, or Bill as I prefer to call you, since you are real human speaker enthusiast. You are getting bogged down in detail.
What you are doing is employing a crossover, the likes of which I have never seen before. You are a flippin' genius IMO. I'm loving what you are doing here. For all that, I need to think about what you are doing before committing myself entirely.
Naturally, you will encounter opposition from the slower intellects at this forum, and they are legion. Myself, I really don't care whether some capacitors are better than others, this is not the big deal in what you are doing. Or whether current-drive valve amplifiers are better than solid-state voltage-drive ones, because to my mind they deal with infinity in different ways, see Norton-Thevenin equivalents below.
Anyway, my friend, GREAT CIRCUIT! It's making me think.
What you are doing is employing a crossover, the likes of which I have never seen before. You are a flippin' genius IMO. I'm loving what you are doing here. For all that, I need to think about what you are doing before committing myself entirely.
Naturally, you will encounter opposition from the slower intellects at this forum, and they are legion. Myself, I really don't care whether some capacitors are better than others, this is not the big deal in what you are doing. Or whether current-drive valve amplifiers are better than solid-state voltage-drive ones, because to my mind they deal with infinity in different ways, see Norton-Thevenin equivalents below.
Anyway, my friend, GREAT CIRCUIT! It's making me think.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.