Modifications to Elac Uni-Fi Reference UBR62 Bookshelf Speakers

[terryforsythe]

What do you call that connector?

It is a circular connector with push-pull locking. The manufacturer is Weipu. It is in the WEIPU IP67 - Large SA28 series. There are different pin configurations available - see the section "Solder Spec".

SA28 - WEIPU

WEIPU Connector SA series: Large size SA28 with IP67 waterproof connector is made of anodized aluminum, specially designed for applications that require fast connection and are easy to connect. It has multi-color options with black, silver, red, blue, and green.

www.weipuconnector.com

They have a lot of different connectors. You may be able to use something smaller, like the SA20 or SA24, depending on the size of your cables.

 

[dagfinn]

Thanks. I use 2,5mm2 wires, so SA28 should be just right .

 

[somebodyelse]

Nice, fully active is another level . I'm converting my 4-way JBL250Ti's, one speaker is ready while I'm waiting for parts for the second (currently bi-amped). I'm considering 2x speakon, but one 8-pin is even cleaner. What do you call that connector?

Dagfinn

There's always the 8 pole speakon: https://www.neutrik.com/en/product/nl8fc

 

[dagfinn]

There's always the 8 pole speakon: https://www.neutrik.com/en/product/nl8fc

Thanks - I didn't know about that either, I thought 4 pole was max .

 

[restorer-john]

Why are you doing this?

Trying to make a silk purse out of a sows ear.

 

[terryforsythe]

Trying to make a silk purse out of a sows ear.

The results are impressive, significantly better than I expected when starting the project. The project was well worth the effort.

Also, I learned a lot which, in and of itself, made the project worthwhile to me. I have designed and built numerous passive speakers, albeit decades ago. I always used stout cabinets with a generous amount of internal bracing at various angles. This served both to reinforce the structure and to breakup internal parallel surfaces. I had never run into cabinet vibration issues to the level I did in this project, nor had I seen their impact on harmonic distortion (back then I did not have a means to accurately measure harmonic distortion). These things I have learned, and addressed in these speakers.

Also, I strongly suspected that eliminating passive crossovers would improve bass response. This project proved, at least to me, that this hypothesis was correct. Indeed, although I use to design passive filters as an occupation, I now am of the opinion that I no longer want them in my speakers.

The most significant issue with the speakers that remains are diffraction issues with the tweeter. Specifically, the diffraction at the tweeter/midrange interface and the diffraction at the midrange surround. The diffraction at the midrange surround has been mitigated a bit by adjusting the crossover frequency and slope. Placing the speakers slightly off-axis and using some DSP I also have mitigated the tweeter/midrange interface diffraction enough so that it is not audible to me.

 

[terryforsythe]

Here is an update. I did some tuning today, measured at my listening position with the speakers toed in a bit, probably 15 degrees or so. I found this provides the widest sound stage. The high frequencies do roll off above about 15 kHz, though, at that angle. But, at my age, 15 kHz is above my hearing range. The measurements are with 1/6 octave smoothing applied.

The left speaker is the green plot and the right speaker is the blue plot. There is a door opening behind the left speaker, which creates the dip at 85 Hz. I adjusted the output of the right speaker to compensate a bit at that frequency, which is fine since that frequency is non-directional.

The red line tracks the Harman curve above about 250 Hz. In the low bass region the red line is about 4 dB lower than the Harman curve. My low bass output is about 1 dB higher than the red line, so the bass is about 3 dB lower than the Harman curve. I like the bass of this preset better on music recorded over the last 30 years or so.

I have another preset with the bass closer to the Harman curve. I like that preset on older recordings, where the bass in the recordings typically is too low for my taste. I used Dirac Live on that preset after manually setting the EQ parameters, but Dirac Live actually made the frequency response a bit more choppy. I may or may not leave it on. I need to do more listening on that preset.

I am still awaiting the WEIPU connectors. They have black ones in stock, but I want silver. They are on back order until the next production run. WEIPU didn't say when that will be. Hopefully they will become available this summer.

 

[terryforsythe]

Recently, in another thread, the topic of damping factor came up. I modeled my system in Quc-s and used a spreadsheet to calculate the damping factor for my woofers connected to my amplifier, both with the passive crossover in the circuit and without the crossover. With the passive crossover removed I included the resistance, inductance and capacitance of my speaker cables (those values are negligible in the model using the passive crossovers). Here are the graphs (note the Y-axis scales):

 

[terryforsythe]

An update regarding the connectors. WEIPU still did not have the SA-28 connectors in silver the last time I checked. I ended up ordering the SA-20 connectors, for which they had silver in stock. Being smaller, soldering is going to be a bit of a pain, but they will look less bulky. I hope to have time to convert over to them this month.

 

[terryforsythe]

Here is my current speaker cable setup:

Here is the new connector plate:

Also, rather than using long cables as I do now, the cables on the speakers will be just long enough to plug into the wall. I am making extension cables to use when I pull the speakers out into the room when I sit down for listening. I have connectors for the extension cables as well.

 

[Rednaxela]

Gorgeous!

 

[dfuller]

Recently, in another thread, the topic of damping factor came up. I modeled my system in Quc-s and used a spreadsheet to calculate the damping factor for my woofers connected to my amplifier, both with the passive crossover in the circuit and without the crossover. With the passive crossover removed I included the resistance, inductance and capacitance of my speaker cables (those values are negligible in the model using the passive crossovers). Here are the graphs (note the Y-axis scales):

View attachment 441053View attachment 441054

That's a huge difference. What does it do sonically, though?

 

[terryforsythe]

That's a huge difference. What does it do sonically, though?

Tighter bass. More dynamic.

 

[terryforsythe]

Recently I have been doing more active crossover modelling in VituixCAD trying to improve the directivity index. But, I don't think my initial dataset is adequate. Also, I have added additional bracing and damping material to the cabinet since those measurements were made. So, I am preparing to do a more thorough set of frequency response measurements outdoors. To facilitate that, I have designed an automated spinorama turntable.

The turntable uses a controller and a stepper motor, and I created a GUI to interface with it. The way it works is that that the GUI commands REW to take a measurement sweep then, after it receives an event from REW indicating the measurement is complete, it commands the controller to rotate the stepper motor a set angle. It repeats the process a set number of times. Each sweep is named with "Full Range", "Woofer", "Midrange" or "Tweeter", depending on what is being tested, the orientation, and the degree angle, e.g., "tweeter_hor_+30".

The setup takes about 6 frequency response measurements per minute, including the time to rotate the speaker between each sweep. To do a full 360 set of measurements at 5 degree increments should take about 12 minutes. I will be doing that for each of the drivers both horizontally and quasi-vertically (with the speaker laid on its side). So, that will be 6 sets of measurements at 12 minutes per set, plus time moving the speaker cable between drivers. I estimate I should be able to do all of the measurements within 2 hours or so, though setup and breakdown themselves also may take that long.

Here is the prototype setup:

Here is the GUI:

The prototype works perfect without any weight on it, but when I placed the speaker on it, the teeth of the 6mm belt were slipping on the timing pullies. I have ordered a 15mm belt and timing pullies, which should arrive in a few weeks, and I will try those.

If the teeth of the 15mm belt still slip, I have a gear driven design, which eliminates the belt entirely, ready to print. I have been hesitant about 3d printing gears, the concern being printing tolerances on my 3d printer. I could use a 0.2mm nozzle, but it will take forever to print, but I may end up doing that. I'll first see if the 15mm belt and timing pullies cure the issue.

 

[kemmler3D]

Recently I have been doing more active crossover modelling in VituixCAD trying to improve the directivity index. But, I don't think my initial dataset is adequate. Also, I have added additional bracing and damping material to the cabinet since those measurements were made. So, I am preparing to do a more thorough set of frequency response measurements outdoors. To facilitate that, I have designed an automated spinorama turntable.

The turntable uses a controller and a stepper motor, and I created a GUI to interface with it. The way it works is that that the GUI commands REW to take a measurement sweep then, after it receives an event from REW indicating the measurement is complete, it commands the controller to rotate the stepper motor a set angle. It repeats the process a set number of times. Each sweep is named with "Full Range", "Woofer", "Midrange" or "Tweeter", depending on what is being tested, the orientation, and the degree angle, e.g., "tweeter_hor_+30".

The setup takes about 6 frequency response measurements per minute, including the time to rotate the speaker between each sweep. To do a full 360 set of measurements at 5 degree increments should take about 12 minites. I will be doing that for each of the drivers both horizontally and quasi-vertically (with the speaker laid on its side). So, that will be 6 sets of measurements at 12 minutes per set, plus time moving the speaker cable between drivers. I estimate I should be able to do all of the measurements within 2 hours or so, though setup and breakdown themselves also may take that long.

Here is the prototype setup:

View attachment 485563

Here is the GUI:

View attachment 485565

The prototype works perfect without any weight on it, but when I placed the speaker on it, the teeth of the 6mm belt were slipping on the timing pullies. I have ordered a 15mm belt and timing pullies, which should arrive in a few weeks, and I will try those.

If the teeth of the 15mm belt still slip, I have a gear driven design, which eliminates the belt entirely, ready to print. I have been hesitant about 3d printing gears, the concern being printing tolerances on my 3d printer. I could use a 0.2mm nozzle, but it will take forever to print, but I may end up doing that. I'll first see if the 15mm belt and timing pullies cure the issue.

This is pretty amazing work, you could call it a poor man's Klippel!

 

[rirelien]

Really interesting! How will you work with the resulting data, if the measurement system works? Re Klippel NFS, that’s where the heavy lifting/math happens, as I understand it.

 

[terryforsythe]

Really interesting! How will you work with the resulting data, if the measurement system works? Re Klippel NFS, that’s where the heavy lifting/math happens, as I understand it.

Import the frequency response measurements, along with the impedance measurement sweep for each driver, into VituixCAD. VituixCAD can handle up to 74 frequency response sweeps per driver.

I have thoroughly tested the measurement system with REW. It works, other than the slipping belt. That issue will be resolved, hopefully with the wider belt and timing pullies I have on order. As noted, if the wider belt doesn't fix it, then I'll go with gears - that definitely will work.

 

[Matias]

How about a vertical axis somehow? Thinking of the turntable mounted on a gas cylinder like the ones in office chairs on a tripod, controller raises and locks the latch so that is slowly goes down with each step, so that it can repeat the horizontal sweeps in a finite vertical steps. Poor man's NFS indeed hehehe.

 

[Sokel]

That's really nice, a hobby on its own, aside from audio, etc.
A silent night, outside, on a wide area and put somewhere fairly high can work its way out pretty good.

 

[terryforsythe]

How about a vertical axis somehow? Thinking of the turntable mounted on a gas cylinder like the ones in office chairs on a tripod, controller raises and locks the latch so that is slowly goes down with each step, so that it can repeat the horizontal sweeps in a finite vertical steps. Poor man's NFS indeed hehehe.

For the vertical measurements I will turn the speaker on its side. It is not a perfect solution, but it is practical. I am doing what I can to minimize diffraction from the setup.