UB class of crossovers

krzys
Posts: 1
Joined: 22 Dec 2019 14:45

UB class of crossovers

Post by krzys » 22 Dec 2019 17:18

Is there any information available for the different types of those crossovers? I see at least 6 choices in the drop menu.
User avatar
UliBru
Posts: 433
Joined: 19 Oct 2019 13:58

Re: UB class of crossovers

Post by UliBru » 23 Dec 2019 11:14

The new crossovers have been designed in consideration of the NT crossovers.
So the basic approach is:
- there is a passband which transfers all frequencies 1:1, no ripple, gain = 1
- there is a stopband which stops all frequencies, gain = 0
- there is a symmetric transition area between passband and stopband by definition gain highpass = 1 - gain lowpass
- at the corner frequency the gain is 0.5 for lowpass and highpass (-6 dB)

It is possible to find many solutions for a curve between passband and stopband which fulfills the requirements above.
A first approach is a curve like a cosine. This is the first selection in the combobox of the UB XO generator.
The other five curves are described by a polynomial function. Again there is a basic idea for the design. At the connection points between passband - transition area . stopband a kink should be avoided, the curve should be continuously smooth. This means that the overall curve should be continuous with continuous first, second, third, forth or fifth derivation. For the last case this e.g. leads to a polynomial function of 11th order.
So the other 5 choices in the combobox use the proper functions.

So why all these crossovers? For better understanding simply create e.g. a crossover with corner frequency 1 kHz. Create 6 crossovers according to the combobox and load XO1 in curve1 to curve6. Change the display in the time domain chart to RMSLog for all 6 curves and check the ringing time. You may even create a NT crossover and compare it to the UB crossovers.
Also compare the XOs in the frequency domain chart so get a feeling for the differences. The final selection must be by listening though.
Stay well tuned
Uli

Moderator
Acourate system: JRiver/Roon -> AcourateConvolver -> miniDSP U-DIO8 -> TacT M/S2150 amps -> DIY horn speakers
Ralf.Hoellmann
Posts: 8
Joined: 19 Oct 2019 19:07

Re: UB class of crossovers

Post by Ralf.Hoellmann » 08 Mar 2020 08:44

Dear Uli,

I have been using both, the UB cos as well as the UB 11. Sonically, I prefer the UB 11. However, have you, or has anybody else in the forum played around with the various crossover types and could share their (sonic) experience?

Kind regards, Ralf
User avatar
UliBru
Posts: 433
Joined: 19 Oct 2019 13:58

Re: UB class of crossovers

Post by UliBru » 10 Mar 2020 09:03

Ralf,

it seems that no one else like to answer. Probably there are not too many users right now trying the UB filters. Anyway I have not got bad comments at least. Have you tried these filters and compared them already?
Stay well tuned
Uli

Moderator
Acourate system: JRiver/Roon -> AcourateConvolver -> miniDSP U-DIO8 -> TacT M/S2150 amps -> DIY horn speakers
In-Ear
Posts: 47
Joined: 02 Jan 2020 18:41

Re: UB class of crossovers

Post by In-Ear » 10 Mar 2020 09:54

Hello,

I am using UB filters right now. I use UB 11, due to lowest pre-ringing, but I did not test or compare with the others. It works well, I used Bessel filters before and I hear a slight difference, most probably based on different directivity and lower distortion of the tweeter. Anyway, I think the impact of directivity and distortion behaviour of both speakers in the XO overlapp f-range might have much more impact than the "sound of a filter type" itselfe. Means it is pending on each individual situation which filter type sounds best - giving a general recommendation for any filter type might be very difficult.

Best regards,
Jörn
Ralf.Hoellmann
Posts: 8
Joined: 19 Oct 2019 19:07

Re: UB class of crossovers

Post by Ralf.Hoellmann » 10 Mar 2020 20:29

Thanks Uli, thanks Jörn. I am going to create a set of filters/ convolution files and want to compare them directly by uploading the ZIP into Roon.

I do agree, that many paramters have to be taken into account. I have compared the cos vs. the UB11 version on one speaker and the perceived (not the measured) sound quality seems to be better with the cos. This is subjective!

Regards, Ralf
Ralf.Hoellmann
Posts: 8
Joined: 19 Oct 2019 19:07

Re: UB class of crossovers

Post by Ralf.Hoellmann » 22 Mar 2020 19:09

In the end, five different combinations have been created and the corresponding ROON-Zips had been made available for quick comparison. Needless to say, that all CorFiles had been done with the same target function.

The ultimate audit was an extensive listening comparison, which is fairly easy in ROON switching the filters.

Without going through all the details, this have been my personal impressions:

- The close-up linearisation offers the best results (mic calibrated to free-field). The further away one goes, the more blurred it sounds. It appears, that the overall room correction of Acourate in combination with a partly correction of some of the effect doesn‘t work very well.
- The steeper the filter slope transition (e.g. j11) of the UB filter class the better it sounds – much more detail and depth of image.
- Even though 1st order filters are doing the thing, switching to 2nd order does provide a kick with more analytical details and clarity. Going beyond 2nd order creates visible artefact in the step response and sounds over-analytical.

Again, this has been a subjective assessment. I can only encourage all Acourate user with digital speaker set-ups to play more around with filters and share their experience.

Regards, Ralf
User avatar
Labdoc
Posts: 95
Joined: 19 Oct 2019 19:11

Re: UB class of crossovers

Post by Labdoc » 30 Mar 2020 16:34

@ Ralf.Hoellmann said- The close-up linearisation offers the best results (mic calibrated to free-field). The further away one goes, the more blurred it sounds. It appears, that the overall room correction of Acourate in combination with a partly correction of some of the effect doesn‘t work very well.

Hi Ralf, I am not sure I understand your set-up. Did you make Acourate measurements at the listening position? Or, are you "linearizing" the speakers using near-field measurements > convolving the correction into the XO and later, doing the room setup using these XOs in the room correction process?
Murray
Acourate/AConvolver on Win 11x64 > Thunderbolt > RME UFXplus (as DAC/preamp) > Balanced cable > Wyred4Sound SX1000 Icepower mono amps > ATC SCM50PSLT and stereo JL Audio F113 subs
User avatar
UliBru
Posts: 433
Joined: 19 Oct 2019 13:58

Re: UB class of crossovers

Post by UliBru » 30 Mar 2020 16:46

Hello Murray,

indeed I have also wondered a bit about Ralf's message. But in the meantime it has become clearer.
Ralf has measured his speaker drivers in various near-field distances. From very close up to about 40 cm distance. Based on these measurements he has prep'd driver linearisations. With an overall measurement and correction at the listening position he has recognized different results in the sound.

We have found that the very close-up measurements has led to irritating results because the measurement seems to show up break-up of the driver dustcap. So Ralf is now using a distance between 10 cm and 40 cm. He has reported that the 40 cm linearisation sounds best. The 10 cm linearisation sounds a bit more focused but a bit hollow.

IMO it is important to say that in this case the speaker drivers are mounted in closed boxes. Whereas a near-field linearisation of a dipole speaker often gives wrong results.
Stay well tuned
Uli

Moderator
Acourate system: JRiver/Roon -> AcourateConvolver -> miniDSP U-DIO8 -> TacT M/S2150 amps -> DIY horn speakers
User avatar
Labdoc
Posts: 95
Joined: 19 Oct 2019 19:11

Re: UB class of crossovers

Post by Labdoc » 30 Mar 2020 19:50

Thanks Uli.
Murray
Acourate/AConvolver on Win 11x64 > Thunderbolt > RME UFXplus (as DAC/preamp) > Balanced cable > Wyred4Sound SX1000 Icepower mono amps > ATC SCM50PSLT and stereo JL Audio F113 subs
Post Reply