The best affordable DIY open baffle loudspeaker I've listened to so far.

0 Members and 1 Guest are viewing this topic. Read 119343 times.

JBspeakerman


OK, here is my crude pencil sketch schematic for the Ultra OB.  I will try to talk Pano into posting something way more professional looking after he returns from his swamp trip!!

The configuration shown has been listened to extensively with a wide variety of music and amplifiers.  The amps included a 5 wpc KT88 SET, 15 wpc KT88 SEP, 30 wpc class A PP EL34, 30 wpc Virtue II chip, 38 wpc PE Dayton chip, Rat Shack 75 wpc receiver, Sherwood 110 wpc receiver and 200 wpc Musical Fidelity A308.  The stated out put power values are what I have measured into a static 8 ohm load and do not necessarily jive with advertised ratings.  After all of that, I arrived at what you see.  For most of the above, in a 20' X 25' x 8.5' room with carpet and furniture, the tonal balance is fairly neutral except with the PP transistor receivers.  The receivers are a bit dull and can benefit from a lower pad resistor, say 4.5 to 5.0 ohms.  One of the fun parts of this design is you can play all you want!  Some may want to experiment with attaching felt strips in various places, like on the inside of the wings adjacent to the Vifa and or the GRS.  Or on the baffle face.

This speaker has excellent and very full lower midrange.  Clarinets and such have a lot of body while maintaining surprising clarity.  Some may feel TOO much body.  To adjust this, a reduction in the high pass capacitor to around 5 MFD or so may be what you may find preferable.  I have made some previous comments concerning positioning distance from the rear wall.  I prefer about 36".  Some may find a bit of toe in toward the listener will have a positive effect imaging.

Make very sure your baffle structure is tight.  If you have a signal generator run it up and down until you eliminate all rattles.  Seems that the 45-70 Hz region is most prone to producing unwanted baffle noise.  Those with out a sig gen can use a test CD (warble tones and the like).

I will be interested in hearing back from those who build this project.  I truly feel most will be surprised at how good this speaker is with in it's out put and efficiency limits.  It is very revealing of amplifier and source quality.  What I found interesting during the audition process is that the Ultra easily demonstrated amplifier quality differences.  Very uncommon for a speaker of this simplicity and low cost.

Good luck and ENJOY!!

 




scorpion

And this is a MJK model simu with correct crossover values, not counting with the BSC-circuit and not corrected for the Vifas higer sensitivity:



/Erling
« Last Edit: 21 Dec 2011, 09:26 am by scorpion »

JBspeakerman

Scorpion....


Thanks for running the simulation!  Pretty ugly isn't it!!!  But the world of the OB can be surprising....

scorpion

To me you are wasting the Vifas capacity now and I would rather have that unit play than the probably 'sloppy' woofers like the high Qts Alpha15s. I could understand the coupling between the Peerless and the dome tweeter in the MK I because the tweeter could not be crossed lower. Now with other woofers and another fullranger I think the MJK type of construction is more rewarding also soundwise.

/Erling

JBspeakerman

I will see if I can put together some acoustic measurements from Arta & True RTA and post them.  Needless to say the spectrum at the listening position is radically different from the simulation. 

JBspeakerman

If I have done this correctly here is a snap of the Ultra Impedence and phase characteristics via WT3.





JBspeakerman



Here is a marginal resolution ARTA measurement taken on tweeter axis at approximately 8'.

I will work on a higher resolution snap tonight.  There is a bit of noise on both ends of the spectrum and anything above about 16K is not representive, same thing below 40 Hz.  You can see some baffle face reflection issues around and just above 2K.    And you see some wall reflection - room issues right at 50 Hz as well as some baffle buzz suck out at 60 Hz.  Will work on eliminating the Baffle noise before I take any more data.  But all and all flater over all than it has a right to be.  Will work on some Holm measurements next.






JBspeakerman



Here is the True Audio sweep.  1/3rd octave is what my version does and makes things look a bit better than they usually are.  But, I have found the rounded True Audio frequency curves tend to jive with what one hears in the listening chair.  True with mono or dipole sources.

Will be interested in what others measure in various venues.

Working on a couple of holm frequency and phase plots.





fastbike

  • Jr. Member
  • Posts: 8
Thanks for posting actual FR plots. The only thing I enjoy more than measurements matching listening impressions is when reality (i.e. measurements) disproves a bad simulation.



JBspeakerman

More observations.

Some may prefer the running the Vifa in reverse polarity.  This does provide a flatter on axis response.  However, it is a bit hotter to my ears on the upper end so for some listeners a little more pad may be in order, say 6.2 ohms or so.  There is some improvement in the area of midrange detail, but the mid bass seems to fatten up just a bit.  Need to experiment with Vifa polarity VS. rear wall distance and report back on that.

BTW, Jarome, thanks for sending your sim data via PM.

Each polarity has it's own sonic signature.  Both sound great which, givin the simulation differences is quite surprising.  It will be interesting to get feed back as to which polarity is most popular. 

Using the Holm measurements you can really see quite a difference.  Will post some snaps later as I have the time.

MJK

  • Full Member
  • Posts: 468
    • Quarter Wavelength Loudspeaker Design
Thanks for posting actual FR plots. The only thing I enjoy more than measurements matching listening impressions is when reality (i.e. measurements) disproves a bad simulation.

The ultimate is when a simulation matches the measurement and the listening impressions are favorable.



Æ

  • Full Member
  • Posts: 859
Hmmm, very interesting. How did you arrive at a different simulation than scorpion?

The ultimate is when a simulation matches the measurement and the listening impressions are favorable.





panomaniac

Martin. Thanks very much for the simulation.  It's looks very close to the measured responses.   What do you think accounts for the 20dB rise in the bass and mid bass your simulation?   Wall distance?

MJK

  • Full Member
  • Posts: 468
    • Quarter Wavelength Loudspeaker Design
Hmmm, very interesting. How did you arrive at a different simulation than scorpion?

I modeled the actual crossover layout as shown in the sketch and reset the SPL plot scale to match the measurement plot scale. The measurement plot scale and the smoothing applied compresses the peaks and dips so the response looks much flatter, Erling's plot looks much worse because of the scale and lack of smoothing. I set up the simulation to match John's description of his measurements.

I have reverse engineered hundreds (if not thousands) of speaker designs over the past 25 years and have usually been able to get my simulations to match the measurements if enough information is provided. You learn a lot by reverse engineering a speaker design.

MJK

  • Full Member
  • Posts: 468
    • Quarter Wavelength Loudspeaker Design
What do you think accounts for the 20dB rise in the bass and mid bass your simulation?   

The 20 dB difference between the bass and the mid bass is a result of driver selection and the crossover, nothing to do with the room. It appears to be inherent in the design. I have found that I like my OB's performance better if I boost the bass a couple of dBs but not as much as shown in the measurement plots.

panomaniac

Thanks Martin.  I 'm curios because your sim is very close to the measured results, except for the extra ~10dB simulated in the 40-50Hz region.  John's measurements show about 10dB of boost there, yours about 20dB.  John puts this down to noise, but I'm not sure it's all measurement noise, I think a lot of it is real content.  Measuring bass in a room isn't easy.

I've been helping John with measurement hardware and software and have seen many of these come across my desk.  It's nice to see the sim so close thru the midrange and above, just wonder what new info would useful to get sims and measurements closer in the bass.

MJK

  • Full Member
  • Posts: 468
    • Quarter Wavelength Loudspeaker Design
I 'm curios because your sim is very close to the measured results, except for the extra ~10dB simulated in the 40-50Hz region.  John's measurements show about 10dB of boost there, yours about 20dB.  John puts this down to noise, but I'm not sure it's all measurement noise, I think a lot of it is real content.  Measuring bass in a room isn't easy.

If I try and read the plot values at 50 Hz and 300 Hz and calculate the deltas I get the following.

ARTA = 17 dB difference
True RTA = 14 dB difference
MathCad = 18 dB difference

Measuring low frequencies in a room is difficult, but I have found that the problems start below 40 Hz. I have always had good results above 40 Hz if I am very careful.

One source of error in the simulations is created by my using manufacturer's data for the T/S parameters (we know that is a big assumption). If I were using measured values, I believe the simulation would be even closer to the measurements. If I were doing the design myself I would also use measured values for the crossover components.

fastbike

  • Jr. Member
  • Posts: 8
Truer words have not been spoken.



The ultimate is when a simulation matches the measurement and the listening impressions are favorable.

JBspeakerman

Thanks to Martin, Pano, Scorpion and AE for your comments and sims.

Nearly all the crossover layout decisions for the Ultra were made based on SQ.  Just wanted to see where my ears would take me.  Found out long ago using most of traditional monopole crossover layout theory with OB designs to be a waste of time.

As far as my measurements go, as we have come to know, getting OB in room measurements that reflect what one hears at the listening position can be problematical.  What I have posted thus far is more or less typical of what I am getting at 8'-12' of distance, on tweeter axis.  There does not seem to be much side wall reflection information getting to the mic at that distance. 

If I move closer or father away from the baffle, things, as you would expect, can change considerably.  The usual stuff, like the low end extending deeper with a more up shelf output as you move closer.  Mid room tends to have a bit of a mid bass null - lumpy-ness and way back the extreme bottom end comes back up and the bass overall smooths out.  Some of this is typical OB behavior and some is due to room interaction.

I think all will find the Holm measurements of interest.  I am optimizing a reverse Vifa polarity version based on listening.  The two polarities show very different measured behavior, but as I said before, not all that much SQ difference.  I will post the reverse polarity Crossover hopefully later today. 

JBspeakerman

Hello all

Here is the Ultra Holm combo plot of Vifa in phase (showing the classic crossover frequency summation dip) and out of phase exhibiting classic phase correct flat summation band pass behavior.  Looks like we have a classic 2nd order response.  There are of course room behavior artifacts, as well as compromised measurement technic issues.

But... I think it shows the crossover component values to be spot on.  How did we end up with what looks like 2rd or 4th order roll off symmetry on the woofer low pass with just a single inductor?  Same goes for the Vifa with it's single cap high pass element.

This data was taken at 9' prox on Vifa axis at a couple of volts input.  Baffles toed in about 5 deg. relative to rear wall with baffle to rear wall separation at 36".

Again, I must emphisize that both solutions sound quite nice and reveiling.  In phase, overall, especially in small rooms or at listening distances under 8' sound more coherennt.