Home Theater Forum and Systems banner

Speaker Polarity Check

16K views 21 replies 5 participants last post by  Vader 
#1 ·
Guys,

Fiance will be out of the house Wednesday night so this is my chance to run some sweeps and make some adjustments to my new mini statements (jim holtz diy design). The first thing I'd like to do is make sure all the drivers are correctly wired (in terms of polarity). I've done some searching it looks like this is possible using REW but I want to make sure that what I think is correct actually is.

The speakers are 3.5 ways so I will want to check each of the four drivers for polarity.

If I have REW run a sweep in each driver's "band" (I think I want to do this in the "middle" of the band so that the neighboring drivers will have minimal effect on the results).

I then look at the impulse graph and see if the highest peak is going upwards or downwards, I can then determine the polarity for each driver and make sure it is wired correctly.

Also, I was wondering since these speakers have an open baffle for the mid (the mid drive is in a tunnel which is open out the back of the speaker) if it is worth changing the gating for when I do full range sweeps?


Thanks,
Pete
 
See less See more
#2 ·
It is not necessarily that straightforward (or as complex as needing to measure each) as the "correct" polarity for as driver may depend on the crossover and the phase of the other drivers through the crossover region. A full range measurement should show a marked dip in the frequency response through the crossover region between drivers if one of them has the wrong polarity. You may find the folk in the DIY Speakers forum have more to suggest.
 
#3 ·
John,

I understand what you saying, I think. My plan once I have the impulse data was to go back to the crossover diagram and ensure the drivers are wired correctly in terms of polarity. I guess my real question is, does an upwards spike in the impulse graph correspond to a driver that has the positive lead going to the positive connection on the driver? I understand the positive "lead" from the crossover may actually be the negative lead, so when I say positive lead I mean like one coming from a batteries +.

Thanks,
Pete
 
#5 ·
John,

Two things, I noticed that the highest peak impulse flipped between when i ran w/o left channel loop back and when i ran with left channel loop back. I assume this means that the soundcard was flipping the polarity, but which impulse graph is showing me the absolute "+" or "-" correct?

When I was running REW to obtain the driver polarity I choose a sweep value above or below the cross over point to get the polarity of just one of the drivers. I noticed that all my impulse graphs showed a positive and negative peak that were nearly identical in amplitude and very close in time. What am I seeing here, the impulse for the mid and woofer (they are supposed to be out of phase)? I just took what ever peak was the greatest in amplitude.

Thanks,
Pete
 
#6 ·
Two things, I noticed that the highest peak impulse flipped between when i ran w/o left channel loop back and when i ran with left channel loop back. I assume this means that the soundcard was flipping the polarity, but which impulse graph is showing me the absolute "+" or "-" correct?
That's odd, you should get the same result with or without the loopback, just shifted along the time axis. Could you email me a measurement with and without loopback to look at?

When I was running REW to obtain the driver polarity I choose a sweep value above or below the cross over point to get the polarity of just one of the drivers. I noticed that all my impulse graphs showed a positive and negative peak that were nearly identical in amplitude and very close in time. What am I seeing here, the impulse for the mid and woofer (they are supposed to be out of phase)? I just took what ever peak was the greatest in amplitude.
Hard to say for sure. The actual sweep signal REW generates extends an octave below and an octave above the values you enter to prevent artefacts at the edges of the desired range, could try narrowing the sweep range to be wholly within the passband of just one of the drivers.
 
#7 ·
John, I should be able to get those to you when I get home (hoping I didn't delete them). If they shouldn't flip when i did the loopback, what did you mean by your earlier post:

"Yes, as long as your soundcard is non-inverting (which you can see from the loopback test)."

Yeah I can run a narrow sweep, or just unplug the woofer. I assume there is no downside to running a very narrow sweep if i'm just using it to determine polarity.
 
#8 ·
Using the timing reference loopback connection should not have any effect on the measurement, it may be inverted or not depending on the soundcard, but it should remain the same with or without the timing ref loopback connected. Measuring a loopback connection (e.g. as carried out during the soundcard cal) will show you whether the input is inverting.
 
#9 ·
John, what would I look for in the impulse of the loopback to indicate if the soundcard is inverting or not? I am trying to establish the relative phase between two non-colocated subs (easier said than done to move them). From what I have read, I can just look at the resultant phase chart of each sub, and if they are 180 degrees out of phase, the respective graph will be shifted along the time axis. I would also like to use the impulse, but I'm not sure if my card is inverting or not. Or, am I all wet in my understanding (pedestrian as it is), and there is an easier way from within REW? Thanks!
 
#10 ·
John, what would I look for in the impulse of the loopback to indicate if the soundcard is inverting or not?
Phase would be close to 180 degrees throughout the response, and if you look at the impulse on the %FS scale (not dB FS) the peak will go negative rather than positive.

I am trying to establish the relative phase between two non-colocated subs (easier said than done to move them).
If they are both the same type of sub their relative phase will vary according to the difference in their distances from the listening position. For example, if one sub is 15 feet away and the other is 10 feet away the difference in distance is 5 feet. The speed of sound is (approx) 1125 feet per second, so the sound from the sub that is further away will take an extra 5/1125 = 4.44 milliseconds to reach the listening position. What that means as a phase difference depends on the frequency. For a 100Hz tone a cycle is 1/100 = 10 milliseconds, so 4.44ms would represent 0.444 of a cycle which is 0.444*360 = 159.84 degrees. For a 50Hz tone it would be 0.222 cycles and 79.92 degrees. The difference would be exactly 180 degrees at 112.5Hz. If you have 2 subs and want their signals to be in phase across their frequency range at the listening position they either need to be the same distance away or the signal to the nearer sub needs to be delayed so that it will arrive at the same time as the signal from the further sub.
 
#12 · (Edited)
This discussion is very well timed, as I am embarking on the same task:

BACKGROUND:

In my case my mains are Infinity Prelude MTS towers. Each consists of a 6 driver passive mid/tweeter tower sitting on a powered bass/sub. I am tuning my system and my room and want to ensure my essentially 2.2 setup is in correct polarity at the individual driver level (I understand some multi-driver speakers have reversed polarity for a given driver, so i will want to check for consistency between the two towers).

I've done the 9v battery test and it "seems" my subs are ok, but its hard to tell out vs in, due to the 2nd reverse motion. Battery test is useless on mids/tweeters as I cant see the movement. As alternative to battery, Ive also tried positive/negative polarity test tones (clicks) from a testing disc. Ive also tried ios app called audiotools, which has a polarity checker. Also trying to use my Multi-meter to figure this out sending correct/reversed signals and seeing if voltage DC flips from positive to negative.

I get mixed and in consistent results from the above. Today I am trying to devise a conclusive method using REW (unless anyone can gave conclusive method using Multi-meter, etc).

REW METHOD:

If I understand the discussion ion this thread I would:

1) do sound card calibration and look at very 1st wave on the IR plot to see if it is up or down. This will tell me if sound card is reverse polarity.

2) Holding mic (calibrated behringer) a couple inches from a given speaker driver measure a sweep limited in frequency to the center frequency between the upper and lower published crossover frequency. Look at the IR plot to see if initial curve is up or down

Am I correct in my understanding?

Might I want to use "Polarity Clicks" as my signal vs sweeps? It strikes me that way I could measure a sequence of positive polarity clicks followed by negative polarity clicks and see that the IR plot is in fact showing these correctly and that they do reverse when the clicks reverse?? It would be a single measurement and I would see each click on the OR plot

Are there other REW tests for this or is the above the way to do it?

In the past I have looked at my prior measurements to see this and find I may have a some small IR initial blips and then a 1/2 height peak/drop and then a higher reverse peak/drop and maybe a few more more, before the ongoing small blips that follow. Some of these were measured outdoors to avoid reflection. This is to say, I am not quite sure which peak/dip to look at.

Thanks in advance for your help!!
 
#13 · (Edited)
Ive done the tests described above. It would be great if you could look at the IR pics below to confirm polarity. My basic question is: Many results show a 40%fs-70%fs peak or fall, followed by a 100%fs opposite peak/fall. Do I consider the 1st or 2nd as my true indicator of polarity. NOTE: all 50-70%fs were peaks (in polarity) and all %100%fs were falls (out of polarity). I also tested my soundcard and it was a single clear peak (in polarity).

I tested both subs and both sets of 6 mids/tweeters from about 2" from the driver. I set the sweep frequency to be in the center range between crossovers (avoided crossovers) with 512 length and 1 sweep.

Here are images: (need to figure out how to do this and will add tonight)

SOUNDCARD:

Technology Electronic device


Bass/Sun Connected Normally:

Text Line Plot Slope Diagram



Bass/Sub Connected Reversed

Text Line Purple Pink Plot


Midrange Connected Normally

Text Line Plot Slope Design



Low Tweeter Connected Normally

Text White Line Plot Font


High Tweeter Connected Normally

Text Green White Line Blue
 
#15 ·
My basic question is: Many results show a 40%fs-70%fs peak or fall, followed by a 100%fs opposite peak/fall. Do I consider the 1st or 2nd as my true indicator of polarity.
The largest peak (the one that reaches 100%). The test is difficult for subwoofers due to the limited frequency span, to get a clearer picture for the sub try sweeping to a higher frequency.
 
#16 · (Edited)
John,

I am not understanding what I am seeing (foregone conclusion, I'm afraid). After a few minor cosmetic changes in my HT, I re-measured everything, and while the response looks similar, the phase is messed up. The first measurement (made almost a year ago with dual subs placed symmetrically flanking the front channel) shows the periodic nature I would expect:

Text Pink White Line Green


whereas the one I did last night shows a similar response (different type of subs, but placed in the same positions), but the phase is completely(?) jacked:

Text Line Blue Green Wave


The subs are in phase with each other (which I determined brute force: I co-located them and verified the 6dB increase when using both as opposed to one). Note that the two graphs were done on different versions of REW. What am I missing? :scratch: :help:
 
#17 ·
.....
I re-measured everything, and while the response looks similar, the phase is messed up. The first measurement (made almost a year ago with dual subs placed symmetrically flanking the front channel) shows the periodic nature I would expect:...

What am I missing?...
The phase does not appear to be as different as you suggest. I would say they are very similar except that the polarity is inverted between the 2 sets of measurements and the older IR is offset further away from 0 ms than the new one above.

It is difficult to know for sure until you put the phase charts on an equal basis. To do that you may want to load the 2 measurements in the same .mdat file for convenience then:
> Select the Impulse tab and open the Graph Controls dialog. Invert one of the 2 IRs if necessary to have both large peaks pointing in the same direction (either up or down). [Since in your case each IR has 2 peaks reaching near 100%, pick the left one as the reference.]
> Select "Estimate IR Delay" and accept "Shift IR" if offered. The 2 IRs should now be aligned relative to 0 ms if viewed in the overlay/Impulse graph. If they are not aligned well together, then you will need to go back and manually offset one of them so that they overlay as closely as possible. Setting the peaks at 0 ms is a good practice for this type of comparison.

Once they are oriented in the same direction and overlaid well near 0 ms, the relative phase can be view. I suspect they will look almost identical.

I also suspect that you will have had to invert one of them to get the orientation similar. This just means that they were opposite in polarity. If so, and you remove the check in the "Invert Impulse" box, the overlaid phase will them show that the phase response is parallel, but 180 degrees offset from each other.

Note that the absolute polarity of the SWs is not really determined in this process. That is another question, and one that you asked earlier. It is not easy to say for sure and also you may be measuring the SWs with the XO filter in place. [If I were guessing though I would say your second chart, "Bass/Sun Connected Normally:", is a positive polarity if it was measured with no XO in place and negative polarity if it was measured with a 4th order XO filter applied.]

XO filters create phase rotation and thus impact the IR shape in a way that is proportional to the filter order/slope. In the end it is not important if the absolute phase is positive or negative. What is important is that the delays (relative speaker distance settings) are set correctly to align the phase of the SW to the main speakers at the XO freq. This can be done with either polarity. It is best however to have phase overlay well throughout the entire XO range and that may better result with either positive or negative polarity.
 
#18 ·
Hi John,

Thanks for the explanation, and I will try it tonight. I am piping REW through my AVR (Denon 2311CI), with a crossover of 250Hz (That's the best I can do to disable it), and this was the same for both measurements. A couple of questions, though. First, how would I go about loading the two measurements into the same MDAT file? Second, beyond the limits of the graphs I posted, the (new) phase curve really gets weird,with almost a square-wave shape. While the old phase shows multiple cyclical peaks, the new only shows the two at the edges of the new graph. Is this due to in-room reflections, and why would I have not gotten this in the old graph? Thanks for your help!
 
#19 ·
I notice now that I inadvertently referred to an IR graph posted by ccclapp and was thinking it was related to your question and your phase graph. I am getting very confused between the 3 different posters with data and questions and possibly different objectives. I do think most all my comments were still relevant to your situation and still may help your understanding.

Regarding your new questions:
1. To load 2 mdat files, first open one, then open the other using "File/Open Measurement" from the REW menu. Both will then be open in the same session.

2. I would have to see the graphs above the 150 Hz range to comment directly on what you see there. What you see will change when the IRs are properly aligned. The differences may disappear or they may not.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top