Ok, maybe next time you'll be able to do this experiment..., the problem is that connect configuration shown in the application note seems to be wrong. I'll try to check with Purifi.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Purifi SPK5 Speaker Review (Prototype)
- Thread starter amirm
- Start date
I’m the builder and I am just finishing up the new sample. You are saying that the woofer should be wired out of phase to the tweeter? Or, both woofer and tweeter are to be reversed? Or, was this just done in testing to measure alignment?
SPK5 crossover
Would not recommend reversing the polarity. The crossover is acoustically equivalent to a fourth order LR filter. There is probably a typo in the SPK5 text (saying "2nd" instead of "4th").
Why?
1. The crossover shows a 3rd order electrical filter for the tweeter and a 2nd order electrical filter for the woofer (plus "stuff" that makes the filter slope steeper
).However, acoustically the (naturally) falling filter slopes of woofer and tweeter are added, so that the filter order increases even more and acoustically leads to a 4th order filter.
2. If the measurements in SPK5 are correct, they also show fourth order high pass and low pass.
I change polarity for one of the drivers (for example, tweeter) and get a smoother sound. Unfortunately, I cannot measure it properly
I did a simple simulation and it seems like the phase shift between LF and HF outputs of the crossover is around 218deg, which is much closer to 180deg compared to 0(360)deg. LF driver has impdance around 6ohm at 3.2kHz, so I used a 6R resistor in my model. Impedance of the tweeter is flat across the frequency range, 4R.SPK5 crossover
Would not recommend reversing the polarity. The crossover is acoustically equivalent to a fourth order LR filter. There is probably a typo in the SPK5 text (saying "2nd" instead of "4th").
Why?
1. The crossover shows a 3rd order electrical filter for the tweeter and a 2nd order electrical filter for the woofer (plus "stuff" that makes the filter slope steeper
However, acoustically the (naturally) falling filter slopes of woofer and tweeter are added, so that the filter order increases even more and acoustically leads to a 4th order filter.
View attachment 98944
2. If the measurements in SPK5 are correct, they also show fourth order high pass and low pass.
View attachment 98945
Unfortunately, that's not how crossover design works.
To put it bluntly, simplified and abbreviated, what happens electrically is "irrelevant". Ultimately, only the acoustic curve of the filter slopes counts.
When considering the electrical filter, the contribution of the chassis is completely missing. Even with an electrical equivalent circuit for the chassis, the reality is only partially represented.
Therefore, it is always the acoustically measured filter slopes that count. These are shown in SPK5 and can be evaluated.
It clearly shows that the filter slopes correspond quite well to a 4th order filter.
In order for the phase alignment of the drivers to fit around the crossover frequency, one has to slightly readjust the filters, since the "sound source plane" of the tweeter and woofer are different - the "sound source plane" of the woofer is offset by about 0.03m to the "rear".
That the alignment of the phases is correct is also shown in the SPK5 app note. The blue curve shows the frequency response when a driver polarity is reversed, so the phases should cause maximum cancellation.
The sample I'm preparing to ship tomorrow has dual binding posts so it is easy to test a reversed polarity on one of the drivers. The graph below is pretty primitive, using REW, an uncalibrated mic 30cm from a point between tweeter and woofer, 1/24th smoothing, no gating and decidedly echoic conditions. The blue line is normal polarity, the green line is reversed tweeter polarity.
Thanks for the measurements! If the initial polarity would be correct, reversing would create a dip, right? What is your personal preference? Maybe in my case it has do with an acoustics of the room...
As @Yevhen said, this does not look good. As he said, if the polarity of woofer and tweeter is reversed (to each other), a dip should result. The green line seems to be the "normal" one.
Try to measure woofer and tweeter separately (0.5-1m, with gate) to make sure that everything was done correctly when setting up the crossover.
UPDATE: The most important rule for building crossover is, check the correct connection of all cables and then do it again
The measurement as shown is not looking good. Maybe you should repeat it, at 1 m or at least 80 cm, at midpoint between MT (if that is the intended designing axis). Use gating, since the problematic area is high, no need for a resolution at lower range, ATM. As advised above, do recheck the wiring. Something is wrong.
To be clear, if wired to the schematic, all the drivers should use normal polarity.
now at 1m, gated at 5ms. Brown is normal polarity, green is reversed tweeter polarity only.
Measured separately, normal polarity, green is woofer, purple is tweeter
When I reverse polarities on individual drivers and run a plot, the overlay of normal to reversed shows no differences.
What does this all mean?
Measured separately, normal polarity, green is woofer, purple is tweeter
When I reverse polarities on individual drivers and run a plot, the overlay of normal to reversed shows no differences.
What does this all mean?
Attachments
Can you post the frequency responses of the individual drivers again with the common scaling of 50dB, rather than the 230dB shown.
When you reverse polarities on both drivers, you still maintain the same relative phase. If you care to show the phase as well, the two identical looking FRs will have a different phase response (or the absolute phase). Even better, show the step. And yes, please use reduced scale for the graphs: unofficial "rule" would be 50 dB range, 20 to 20KHz. If using gating, showing 200 to 20 KHz would be even better.
Edit: I just noticed, are you really "manufacturer"? Sorry, had to ask...
Last edited:
I think Manufacturer overstates it, but yes I put together and try to sell these copies of the SPK5 specification.
Here is an overlay of the SPL for individual woofer and tweeter. Mic at 1m aimed at centre spot between woofer and tweeter. No smoothing, Very short gating to try and capture direct sound. Scaling done to same limits as in Purifi App Note.
Same chart, but limited to 200-20kHz
Here is an overlay of the SPL for individual woofer and tweeter. Mic at 1m aimed at centre spot between woofer and tweeter. No smoothing, Very short gating to try and capture direct sound. Scaling done to same limits as in Purifi App Note.
Same chart, but limited to 200-20kHz
Without measuring (the raw drivers response in the box, impedance) it is difficult to say what is going on. Obviously, the tweeter is not attenuated enough (10 dB hotter relative to the bass-mid) and with a straight rising slope. Very odd. It does show a 3K aprx xo point. but other than that it should resemble what @ctrl showed above in # 186.
Because they aren't interacting?
So, here is a comparison of the measurements from the manufacturer and @sgoldwin.
This looks quite good. The crossover of the woofer is definitely set up correctly.
The tweeter curve looks a bit strange, but I think @sgoldwin said somewhere that his measurement microphone is not calibrated, which would explain the strong high frequency sound pressure increase.
According to this data, I would say that almost everything is connected correctly.
Only it should be so that if all chassis are connected with the correct polarity, then the green curve shown in post#192 should be measured and if the polarity is reversed for one chassis, the brown curve.
It is very important that it is done identically for both speakers so that the speakers do not radiate sound in opposing phase.
In the high frequency range, the deviation can be considerable with an uncalibrated microphone:
Similar threads
