Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
131.137.245.208
I know this subject has been touched on before, and I read some previous posts, but I have a question:
I recently got my 1.6QR drivers back from the factory for repair and I was rewiring them. I use an active line-level XO. Long story short, attached is a schematic for how I have them wired, and appears consistent to me with Magnepan's schematics
But looking at Magnepan's schematics, it appears the positive terminal of the amplifier goes into the positive terminal of the woofer and the negative terminal of the tweeter.
I realize that doesn't necessarily make them out of phase since the crossover shifts the phase. But what I'm unsure of if my XO shifts the phase the same way as the factory XO. I have the same slope settings as the factory XO, but the -3db points are slightly off.
I figure the best thing to do is
1) play a 600hz tone
2) mute the tweeters and make sure the tone sounds centered and in phase
3) mute the woofers and make sure the tone sounds centered and in phase
4) unmute and reverse the polarity of both tweeters and see which sounds better.
Follow Ups:
So I took out the safety capacitor I had in line with the tweeters. The capacitor was there to protect the tweeters from any DC signal (power surge or amp failure), but as a result of this discussion I realized signal to the tweeters was 90 degrees out of phase.
I figured it wouldn't make much of a difference but I took it out today anyways. Well, either it made a HUGE difference, or I'm a poster boy for placebo effect. I didn't know my speakers could sound like that...
Songs like Stairway to Heaven (I took the elevator today ;) ) and the Hell Freezes over version of Hotel California absolutely sprung to life. Generically, the only way to describe it was more coherent. But specifically, both songs have multiple guitars playing similar sounding chord progressions. Before, when one guitar was playing, it was crystal clear, dynamic, and coherent. But the other guitar would come in and I wasn't sure where it was coming from or exactly what was happening. I've have that sense at blues clubs, so I figured it was normal, but today was a different story. The differences between the two guitars were much more resolved. I could hear they were playing different chords at different times, and had a much better directional sense.
Again, either that 90 flip from the cap was super bad, or I'm Mr. Placebo-man.
Sounds like you were out of phase and now are in phase - so that is good. If you want to have a protective cap on the tweeter use a big one a couple of octaves below the XO.
Yes that's exactly what I had. But it doesn't matter the value of the cap, it still shifts the phase 90 degrees. Perhaps two caps in series and wire the tweeter backwards?
It should be 53 degrees at one octave into the passband and 28 deg at the second octave up and then 14 degrees into the third octave, 11 degrees at the decade. It is still a phase shift but is probably in the range of adjustment you could address with toe in.
I think you're confused, but maybe I'm confused. The phase shift for a capacitor is 90 degrees, that is, voltage lags current by 90 degrees. Are you saying the whole signal (voltage plus current) is shifted depending on the frequency?
180 - arctan(f/fo) for HP filterfo is the elbow freq, -3db
as Davey noted from o in the pass band to 45 deg at f0 to 90 deg well below.
Edits: 05/06/14 05/06/14
I whipped out my college text book and am promptly reminded why I got a C in the course. I am pretty sure the phase shift in your equation above is the phase angle for impedance, i.e. the angle between 0 and the impedance vector for real and imaginary impedance.
I was speaking to the phase relationship between current and voltage (ELI the ICE man, anyone?), which I couldn't find in my textbook, but the internet tells me it's 90 degrees current leading voltage (or, rather, voltage lagging current, since capacitors 'resist' changes in voltage).
Those two are completely separate concepts, right? I thought the point of matching phase was to make sure the tweeters and woofers both fire the same way for the same signal, e.g. a solid tone at the crossover frequency, and I thought the ICE man phase shift was the important part for that (which is why Magnepan's tweeters are wired 'backwards'). Is it that impedance angle that is important to match?
Problem with this stuff is no one has translated it into straight English, lol!
The IV phase difference is not what we are looking at, it is the phase angle in the signals to the drivers that need to be matched. The number of LC active stages does change IV phase but only in the speaker level does that matter, since the amps don't care about the IV phase upstream, they are driven by the incoming V. While it is true that you have a better utilization of the amp's IV envelope with speaker level XOs with odd orders, that is not the phase of the acoustic output.
OK well I think I'd need to stand next to you at a whiteboard with dry erase markers before I really understood why.
What we have is an electromagnet (the drivers) in a static magnetic field (from the pole piece). The fluxating magnetic field from the driver interacts with the the static field and we get motion and sound waves. As far as I know, that fluxating magnetic field is based solely on current, not impedance or voltage.
I guess what I'm saying here is there's a piece of the puzzle missing in my brain, and if you're laying it down, I'm not picking it up. Do you have any links to a good primer on the subject?
OK - 2 circumstances
speaker level XO -
it has a Z composed of reactive and resistive components, it causes I to differ from V in phase because of the reactive portion - which changes with frequency. In matching the phase of the output from the drivers (in our case the drivers are themselves non reactive - or very nearly so), so the XO will do all the phase variance. The amp delivers a voltage modulated by the incoming signal and that pulls current through the output devices.
The phase in the two (or more) pass bands of the XO is near 0 once we are an octave away from the XO freq, while that in the filtered band is 90*Norder, Norder being the order of the filter. The amp sees both the filtered portions from each half of the XO, hence the amp always sees 90*Norder, however, the signal going to the driver has only the phase characteristics provided by the filter. That is given by the arctan formula a couple of posts above. So the phase that needs to be matched is that in the signal reaching the driver past the filter.
circumstance 2 - line level XO
The amp sees only the residual reactive impedance of the drivers (very little in our case) and their resistance. The line level XO provides the filtering before the amp and induces phase changes with frequency - in the voltage, but since the amp is directly connected to the driver it essentially has no IV phase differences, all the phase change is nearly identical in both I and V.
Hope that helps.
Anyway, there are numerous coursework websites from EE departments around the world, just pick one. None of them are stellar in clarity.
I think what you're telling me, is the internets were wrong about the VI phase angle always being 90 degrees. Either way, it's time for me to hit the button.
The main point is that what the amp sees when driving the XO of the speaker is BOTH sections of the XO - both the LP and the HP, but the driver sees only its own filter. While the phase in each filter is near 0 in its pass band (highs above HP, lows below LP) it is always not 0 in the other filter - which is in the stop band.
Say the XO is at 1khz, the impedance the amp sees at
2 khz is that of the HP (near 0) and that of the low pass (phase = -arctan(f/fo))
500 hz is that of the HP (phase = 180 -arctan(f/fo)) and the near 0 impedance of the LP.
In the normal parallel XO the amp sees both the impedances, and are like parallel resistors and combine 1/(1/Zhp + 1/Zlp). The impedances are complex numbers with the resistive (R) + reactive (i*X), where the freq variable portion is the reactance.
No, it is 90 degrees. :) However, I think you're sort of conflating two things here and just confusing yourself. Don't worry about it too much, but if you want to perform further reading try an outline of capacitive and inductive reactance. Here's a good one:
http://en.wikipedia.org/wiki/Electrical_reactance
Dave.
I think you both might be confused. :)If you're in the passband there is no phase shift imparted by the series capacitor but once you go down in frequency then you start inducing phase shift that will max out at 90 degrees when you get about a decade below the cutoff frequency.
Maybe I missed it, but you didn't mention what value this protection capacitor is. It's easy to plot the phase response if you know that.
You mentioned 10Hz cutoff in your initial pictorial, but I'm afraid that's not possible. (The capacitor value would be VERY large.)Dave.
Edits: 05/06/14
I wrote 10 hz because I couldn't recall the capacitance and was too lazy to look. It's 2200 uF, so by my calculations 18hz cutoff, close enough for government work.
Regard phase shift, see reply to Satie...
Yeah, a 2200uF capacitor has no business being in series with any transducer in any speaker system. It's no wonder your system sounded better with those removed. (Many other issues with this sort of capacitor beyond phase shift.) :) I'm wondering what possessed you to install those to begin with.Those might be suitable as power supply caps in your amplifier but not in a speaker system.
The 1.6's are quasi-ribbon anyway, so a "protection" capacitor is completely unneeded.
Cheers,
Dave.
Edits: 05/07/14
I figured being such high capacitance they would be transparent to any frequences coming from the high pass of the XO, and therefore inaudible. Obviously I was wrong.
I just got back from a trip and perhaps read this all above in haste. It does look as if you know what you are doing and the gang has also helped.
One thing did come to mind that you may want to explore later. I am listening to music as I write this. My heavily modded MMGs are playing alone in stereo, no other speaekrs or subs. The "mode" they are playing in is their best (overall). It is my favorite for a number of reasons, too many to cover now. Significantly, this mode has both the tweeter and the mid/bass drivers wired in "same-polartity".
This is something that I've always liked more than the typical "inverted polarity" since the day I began to bi-amp (PLLXO) many years ago. The xover points & slopes that I now use are very similar to what the factory used as speaker-level xovers. In those eaarly days my MMG version had factory xover slopes similar to the 1.6 (1st order for tweeter & 2nd order for the mid/bass).
I've always wondered if other Maggies could benefit from this. To be sure, "same-polarity" is a bit more challenging to optimize. I reckon that I've been lucky here and other factors, like the room and is configuration, may have helped.
I lost the schematics I made of my MMGs wiring the day I first altered them. Therefore I do not know what polarity they had then. However, the last time I measured the factory speaker-level xover, I only had the factory original parts for one channel. With these FACTORY parts the best response curve came if I used "same-polarity". Curiously "inverted-polarity" looked just as usable.
To this day -- in PLLXO bi-amp mode now -- I can switch to "inverted-polarity" in a minute. I do so at times for some recordings (bad recordings are more easily forgiven, for example). Even with good & great recordings, "inverted-polarity" mode does not disappoint.
Yet, in this room "same polarity" can be much more addictive. The better the texture captured in a recording of a voice or instrument are, the more its richness comes across. While there are imaging & staging attributes that I can approximate to "equal" in both modes (making adjustments), NOTHING has gotten me close enough in terms of texture richness and natural timbres as "same-polarity".
Were it not for how meaningful the experience may be I would not encourage the heavy tweakers among us to explore it. It does require a little extra tweak work. Yet, when it works...even the Mylar may disappear, if you know what I mean.
you may want to spend sometime with these graphics to verify the wiring and colors of wire now vs OEM stock.
http://www.integracoustics.com/MUG/MUG/tweaks/green_lantern/SCHEMATIC.jpg
http://web.archive.org/web/20081026005801/http://www.precisioncomputer.com/ocean/MagUp/Label.htm
When I did my ALLXO and PLLXO updates, if I get the phase wiring wrong it created a very noticeable midrange suck out in the sound.
So what I didn't mention in the original post was that when I rewired them yesterday, I looked at the schematics I make before I unwired, and it looked out of phase! I was convinced when I originally bi amped the speakers I messed up and they've been out of phase all along!
So when I had everything hooked up last night, they sounded good but something was off. The factory did a full new Mylar job, so I was thinking break in, but in the back of my mind was phase.
We'll turns out my original wiring job and schematics was correct, but damn I keep looking at it physically and it looks wrong. On one speaker, the positive terminal (from the amp) go up the voice coil wiring on the outsides, and one the other it goes on the inside and the ground from the outsides. Maybe in that speaker they put the magnets in backwards!! :D
Anyways I did the test above and also this helpful page helped me.
http://www.richardfarrar.com/are-your-speakers-wired-correctly/
Soundin' perty magnificent right now!
Also, listening to a straight 600hz tone sucks
So you figured it out? right?
Yep, all's well that ends well.
The wiring on the left and right speakers are mirror images of each driver other so +- -+ on one side becomes -+ +- on the other. It confuses everyone but Magnepan :). Enjoy!
Edits: 04/24/14 04/24/14
No, actually that's still not correct. It does really get confusing, but it depends upon whether you're considering the transducer polarities only or the related wiring as well.
If you're just looking at the bottom of the transducers themselves and you have no wiring/components connected then it looks like this:
+ - + - + - + -
You can confirm this visually by looking at the traces as they lead out of the connectors and head up to the mylar. None of them cross over each other and both tweeter and woofer "circulate" in the same direction. So, tweeter and woofer would both move the same direction if voltages connected the way I represented.
When you include the related wiring (this is where the tweeter polarity reversal happens) then it gets confusing.
In that case you'd end up with something that looks like this: (As you noted.)
+ - - + - + + -
I labeled pluses and minuses here arbitrarily. They could be represented (all eight connections) in opposite polarity as well.
Cheers,
Dave.
lol I'm still confused. I look at the driver and can't fathom why wiring like that results in movement in phase.
It operates on essentially the same principles as electric motors. Since I've always done poorly academically on motor theory, and could never master the right hand rule, perhaps it's not shocking I don't get it :D
I had problems with the right-hand rule when I was a freshman in college too. I was busy using my right hand for things other than physics :).
It just confused you too. :) Take another look please.
Cheers,
Dave.
I must have caught my error at the same time you did, hence the edit. :)
As Davey says the phasing should be correct provided you used the same filter orders (2nd order LP and 1st order HP). If you reversed the polarity of the drivers in relation to each other, you will hear a large "hole" around the crossover frequency.
If your line-level crossovers have the same/similar slope characteristics as stock, then they will have the same phase characteristics as well. So, you'd want to maintain the same wiring polarity within the speakers you had originally. As you noticed, in the case of the 1.6, the tweeter is wired with opposite relative polarity from the woofer.
Some users are bi-amping with active crossovers that have much steeper slopes than stock. That's a whole different can of worms. :)
Your step #4 is fine, but it's possible the preferable sounding configuration might be different than you expect since these are broad overlap filters. Probably not likely though. I think you'll find the correct wiring configuration to sound preferable.
Cheers,
Dave.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: