Thanks for clarifying.
That's a very low impedance, which isn't ideal. However, it's at a frequency that normally doesn't have a lot of power, so it shouldn't stress an amplifier all that much. Also, the phase shift in that region is close to zero, so it's a quite resistive load, which probably makes it easier to handle.
6dB/octave produces a very wide range of interaction in the crossover region. 12dB/octave will provide better control and stop the higher-frequency driver from having higher cone excursions at lower frequencies within its operating band than it may be able to handle at higher input power levels.
The interaction of the low-pass crossover on the subwoofer with the driver's impedance curve may prove a little problematic. There's a lot of magnitude and phase shift happening around an impedance peak, and this can make it difficult to get a smooth filtered response curve. It will be interesting to see what happens with the particular subwoofer driver that you choose.
You're welcome.
I'm not sure that low of an impedance is correct and if not that might mean the zobel is wrong.
I might retest the Jensen tweeters just to make sure I calculated the zobel correctly unless sometimes it is normal for the impedance to drop that low. My concern is how that low of an impedance affects the crossover between the Jensen tweeters and the ribbon. My guess is given it's real close to the crossover frequency it may not affect it at all. Not sure if this is correct, but when I tested the Jensen tweeters separate from the crossover with the zobel connected it didn't appear to do anything at all.
Originally the woofer had no crossover on it either and I'm sort of surprised the woofer plays as good as it does crossed over as high as it is (close to stock crossover frequency based on tweeter impedance and the cap value attached to the tweeter), however it is a Jensen woofer which may be why it does sound so good playing that high.
Moving the speaker closer to a corner and angled about 45 degrees from the wall made it to where the bass sounds better and I no longer need a wider baffle or sub.
I'm not sure that low of an impedance is correct and if not that might mean the zobel is wrong.
I might retest the Jensen tweeters just to make sure I calculated the zobel correctly unless sometimes it is normal for the impedance to drop that low. My concern is how that low of an impedance affects the crossover between the Jensen tweeters and the ribbon. My guess is given it's real close to the crossover frequency it may not affect it at all. Not sure if this is correct, but when I tested the Jensen tweeters separate from the crossover with the zobel connected it didn't appear to do anything at all.
Originally the woofer had no crossover on it either and I'm sort of surprised the woofer plays as good as it does crossed over as high as it is (close to stock crossover frequency based on tweeter impedance and the cap value attached to the tweeter), however it is a Jensen woofer which may be why it does sound so good playing that high.
Moving the speaker closer to a corner and angled about 45 degrees from the wall made it to where the bass sounds better and I no longer need a wider baffle or sub.
There any grills I can get to put over the woofer and tweeters that will protect them but not affect the audio?
Another reason I use them is because they were made back when tubes were the only thing and later vintage speakers made when high power solid state was quite expensive. So therefore the speakers had to be relatively efficient and usually had light cones due to the lower power handling which makes the cone move quicker and easier which in my opinion would produce a more accurate sound.
Plus sometimes one can find the vintage drivers for a lower cost from someone who is just looking to get rid of them or a helpful member of a forum who doesn't need the drivers, isn't looking for top dollar and just wants to pass them on to a member who can use them.
Parts Express and Madisound seem to stock a variety of cloths that can be used for speaker grilles.
The impedance can drop on the low side, even if the crossover network is working properly. However, it is also often an indication that there is a problem.
I created a VituixCAD model of the circuit diagram and its components that you provided earlier. The circuit is shown below. The woofer used a 5.389 ohm resistance, each midrange a 2.108 ohm resistance, and the tweeter a 4.8 ohm resistance.
The impedance curve of the system is shown below.
The measured impedance curve of the complete system is shown below for ease of comparison.
It seems quite clear that assuming that the drivers are pure resistance loads in the VituixCAD simulation is not producing a particularly accurate model. If the impedance of each driver on its own was measured individually, then those measured results could be included in the VituixCAD model, making it much higher in overall fidelity.
The electrical responses of the low-pass woofer filter, the bandpass midrange filter, and the high-pass tweeter filter are shown below. It's apparent that the bandpass filter is not working very well, and the peak in its response occurs at the impedance minimum. Note that adding together the identical filtered output of the three midrange drivers would create a single overall output curve that is raised by 20*log10(3) = 9.54dB, so don't be alarmed by the lower level of the three green curves.
It seems highly likely that the filter working on the Jensen tweeters (midranges) is not producing a desirable filtered response function.
Thanks for that.
The green portion peak should be in the middle where the red and blue lines cross over, right?
When calculating the crossover I used a two way crossover calculator to calculate the values for the woofer and tweeter and the tweeter and super tweeter given that most three way crossover calculators I've found would only allow you to select the woofer/midrange or midrange/tweeter crossover frequency and it filled the other one in which in my opinion is rather limiting.
That said I did just find a three way crossover calculator that allows one to enter both crossover frequencies.
https://soundcertified.com/speaker-crossover-calculator/
Here's the result of that. Looks like the current bandpass crossover is just a little off.
So the crossover would then become this.
Now if I adjust the frequency to the values of tweeter and woofer crossover parts I get this. C3 I can parallel two .22uF film caps, C2 I can parallel two .22uF caps film with two 2.5uF film caps in parallel, L3 I can make by unwinding some of the current L3 and L2 I might can make by unwinding some of a 1mH inductor I have.
Will you plug those numbers in and see if the crossover is better? Before I go modifying the crossover I want to be sure the new values are right.
Perhaps the bandpass filter being off is causing that dip in impedance.
For the grills I was thinkling more of the waffle style grills.
I used these for a couple of 12" woofers in another open baffle build, but they don't have a 15" version nor a 5" version.
https://www.amazon.com/dp/B09SF3G3K1?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
I like those as the holes are larger which would affect the sound less.
The green portion peak should be in the middle where the red and blue lines cross over, right?
When calculating the crossover I used a two way crossover calculator to calculate the values for the woofer and tweeter and the tweeter and super tweeter given that most three way crossover calculators I've found would only allow you to select the woofer/midrange or midrange/tweeter crossover frequency and it filled the other one in which in my opinion is rather limiting.
That said I did just find a three way crossover calculator that allows one to enter both crossover frequencies.
https://soundcertified.com/speaker-crossover-calculator/
Here's the result of that. Looks like the current bandpass crossover is just a little off.
So the crossover would then become this.
Now if I adjust the frequency to the values of tweeter and woofer crossover parts I get this. C3 I can parallel two .22uF film caps, C2 I can parallel two .22uF caps film with two 2.5uF film caps in parallel, L3 I can make by unwinding some of the current L3 and L2 I might can make by unwinding some of a 1mH inductor I have.
Will you plug those numbers in and see if the crossover is better? Before I go modifying the crossover I want to be sure the new values are right.
Perhaps the bandpass filter being off is causing that dip in impedance.
For the grills I was thinkling more of the waffle style grills.
I used these for a couple of 12" woofers in another open baffle build, but they don't have a 15" version nor a 5" version.
https://www.amazon.com/dp/B09SF3G3K1?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
I like those as the holes are larger which would affect the sound less.
Last edited:
I went ahead and redid the crossover per the last posted schematic.
Soon as the glue dries I'll mount it and test it with the DATS V3 to see how the impedance looks now.
I suspect the dip in impedance might have been due to the tweeters and super tweeter overlapping in their reproduced frequencies.
Soon as the glue dries I'll mount it and test it with the DATS V3 to see how the impedance looks now.
I suspect the dip in impedance might have been due to the tweeters and super tweeter overlapping in their reproduced frequencies.
Here's the results using DATS V3 with the redone crossover. The impedance looks better and the phase shift is a bit smaller. Also just with a brief listening test I can already tell the speaker sounds better.
Interestingly enough now that the crossover is correct the bass sounds a little better and it's harder to tell that the speaker is three way versus full range.
That said this afternoon I did try the woofer by itself no crossover and it sounded fairly decent in the upper midrange lower treble region.
I know some crossover calculators are designed for the midrange to be reversed so I tried reversing the phase of the Jensen tweeters and it didn't sound as good.
Interestingly enough now that the crossover is correct the bass sounds a little better and it's harder to tell that the speaker is three way versus full range.
That said this afternoon I did try the woofer by itself no crossover and it sounded fairly decent in the upper midrange lower treble region.
I know some crossover calculators are designed for the midrange to be reversed so I tried reversing the phase of the Jensen tweeters and it didn't sound as good.
Last edited:
I think there are aspects of this speaker where it is likely that the crossover design is actually aided by 'sound & ear' rather than
trying to achieve some academic goal of impedance & phase response. Remember, the speaker is driven via an output transformer >
which makes impedance fluctuations take-on 'a different nature' compared to a high damping factor transistor amplifier.
trying to achieve some academic goal of impedance & phase response. Remember, the speaker is driven via an output transformer >
which makes impedance fluctuations take-on 'a different nature' compared to a high damping factor transistor amplifier.
Yes, most definitely. And the relatively sharp "peak" should be more of a broader plateau.
You hit the proverbial nail on the head. The bandpass filter actually serves to add a little bit of boost in the output, which is somewhat hidden by the method of presentation. That boost is what's leading to the dip in the impedance.
The crossover that you've measured appears to be largely resistive in nature above 100Hz. The impedance minimum also isn't as low as before, so that's a nice overall improvement.
I've gone and done a simulation of the crossover circuit with the component values that you provided. Note that I've disconnected the Zobel network, as this will upset the circuit as there's no inductance for it to compensate for. Also, I've amalgamated the three tweeter/midrange units into one driver with a composite resistance. This was done to better see the attenuation curve for the bandpass filter relative to the low-pass and high-pass filters.
The filter circuit as modelled is shown below:
The results that I obtained for the three filter transfer functions are as follows:
I won't provide the simulated impedance curve, as it will be nigh on meaningless.
My simulations would seem to indicate that you are in the right ballpark with your crossover. If you can be bothered, it would be interesting to obtain individually-measured files for the woofer, the three in-series midranges, and the tweeter. These would make the crossover filter simulation much more accurate, and allow the inclusion of the Zobel network in my modelling.
That's an interesting result. I would have expected the summed response to be better. However, there are still many unknowns regarding the true action of all those filters, so it's probably a bit unreasonable to make any comments, as they might be not much more than pure conjecture.
Last edited:
I've seen some crossover calculators that calculate for the midrange to be reversed polarity and some that calculate for the midrange to be the same polarity as the woofer and tweeter.
I suppose that I can feed a sinewave to the amp at the woofer/tweeter crossover frequency or tweeter/super tweeter crossover frequency and see if there is a drop in output and reverse the Jensen tweeters to see if there's an increase in output.
I can provide the DATS V3 data for the drivers when I get home today.
I suppose that I can feed a sinewave to the amp at the woofer/tweeter crossover frequency or tweeter/super tweeter crossover frequency and see if there is a drop in output and reverse the Jensen tweeters to see if there's an increase in output.
I can provide the DATS V3 data for the drivers when I get home today.
I'm a bit surprised by the impedance curve for the midrange drivers. There doesn't seem to be any resonance peak present.
Are you able to zip up the measured impedance files and post them here?
Are you able to zip up the measured impedance files and post them here?
You mean the files generated by DATS V3?
I'll have to look and see how to save what DATS puts out.
The cones of the Jensen tweeters don't move at all so there may not be a noticeable resonant peak like a typical driver has.
I can measure other tweeters to see if they have a resonant peak.
Also I see one why people like ribbons. The more constant impedance.
Will jump on this soon as I get home from dinner.
The program you use is it a reasonable cost or free and what's the learning curve for using it. If I'm going to keep building speakers it looks like something worthwhile for me to have.
I'll have to look and see how to save what DATS puts out.
The cones of the Jensen tweeters don't move at all so there may not be a noticeable resonant peak like a typical driver has.
I can measure other tweeters to see if they have a resonant peak.
Also I see one why people like ribbons. The more constant impedance.
Will jump on this soon as I get home from dinner.
The program you use is it a reasonable cost or free and what's the learning curve for using it. If I'm going to keep building speakers it looks like something worthwhile for me to have.
Last edited:
I can export the impedance data as a DATS formatted .txt or .zma file or a CLIO formatted .txt file.
Which do you prefer?
Which do you prefer?
Are those 3 tweets providing good midrange?. Maybe use 2 in series with a 4.7 ohm resistor ?
The woofer is crossed over to the tweeters at 6.305kHz, however that's fine as the woofer sounds good playing that high.
The stock crossover was a single 4uF cap so I took the impedance of the series connected tweeters and that cap value and figured the stock crossover frequency. I tried to keep the crossover close to that frequency with the three way crossover.
When testing the tweeters with the DATS V3 I tried one, two then all three. The output was louder as I added a tweeter in series so to use only two would make them unbalanced with the woofer and super tweeter.
My first choice would be the .zma files, as these should work well, as I've used them previously with VituixCAD. I'd anticipate that the CLIO file would also work well, as it's a well known audio measurement system.