Ergo's Wavecor Ardent Build

[Zvu]

Cool. It behaves quite well.

Thanks

[Juhazi]

My eyes focused at on-axis spl dip at 400-800Hz (a full octave), and increasing directivity too in that range. Could this be a measurement artefact?

I have had difficulties in measurements (or in real life?) in that range too, indoors. Anyway - rapid changes in directivity are beasts to compensate, in xo design shallower slopes usually help. But with a finished outcome/product the end user is in trouble, turning the speaker and head, never happy...

[ergo]

It can be an artefact indeed. I used a different window size for mid and woofer also and that allows more of reflections to creep in to woofer response. I concentrated the effort to optimize only to the mid/tweeter xover and left the low/mid ones untouched. Figured that these have been tried and are a good start point. At mid/tweeter crossing there was a phase matching opportunity and both CLIO and Fuzzmeasure measurements agreed with that, so I did few component optimizations. I'll be getting the parts next Wed/thu, so hopefully by end of next week I have them up and running. If I do not make that schedule then there will be another week of delay as I have a work trip to US coming up again.

[kimmosto]

^One problem with CLIO is limited IR window functions. Hanning/2 is not very accurate with rotation table and/or common window starting point for different drivers. Rectangular window function is stable ie insensitive for IR peak movement, but response suffers some artefacts due to sharp cutting of window.
I just sent feature request to Audiomatica: to add "Hanning/4" window function and short left window, adjustable in 'MLS Settings' window. Both ARTA and REW have few window functions with flat top but smoothed landing to zero which is stable with turning table. REW has also "left window" which would help with CLIO to adjust axial response close to minimum phase without post processing with negative delay in measurement program or XO simulator. ARTA has PreDelay which works too. But I'm not sure what are our chances to get improvements to existing versions. This is not big deal with flat hifi drivers but error could be significant with deep coaxials and horns when using Hanning/2.

[ergo]

Thats a good idea. Clio is lacking in these features indeed. I shall do the same. Getting more of the similar requests might make them raise this in priority on their backlog.

[ergo]

Finally some progress to report

Meanwhile the xover parts delivery was late, then I had a week of business trip to US and so the time flies.

I finally got all the parts and got the crossovers assembled. First tunes got played also and initial impressions are good. I'll try to find time tomorrow to take out the measurement gear and validate that the acoustic result is what it supposed to be. Plan is to do at least 0.. 30deg measurements and also 0..20 deg vertical. Then invert the mid and see if and how close the reverse null performance gets to the simulated responses.

PS. Most all of the capacitors are the Janzen Cross-Caps. I wanted to first make sure the measured result is correct as I did alter Jon's design a bit. Plan is to later upgrade to better caps....
I do also want to take it step by step and do the cap upgrades slowly. I'm even pondering on replicating the whole mid/tweeter xover plate with better caps, so that I could switch back and forth and listen. I would really like to have as objective way of hearing the difference as possible.

[sdl2112]

Nice work ergo. It's always good to check the design with less costly components. High end caps are pricey. I've bought the final caps for places I was pretty sure wouldn't change but I still have some to buy. I'll be very interested in your comparison if you make two versions.

[ergo]

Ardent xover ver1_EE measurements

Got some measurements done. I'll export/import these to VituixCAD later on but for now the images from Fuzzmeasure.

First the 1m on axis for 0, 10, 20, 30deg both with midrange in 'normal polarity' and 'reversed polarity'


0, 10, 20, 30, 40, 50, 60 deg horizontal at 1m


-20, -10, 0 , +10 , +20 vertical at 1m


Left and Right speaker at 2m (speaker placed mid room)


Distortion at 1m with 2.83V, 5.65V(+6dB), 11.27V(+12dB), 22.48V(+18dB) not sure if last one maybe hits the limits of CLIO amplifier already also.

[ergo]

Seems the moving GIF stops moving when uploaded to the forum..

2.83V, ca 84dBSPL @ 1m


5.65V(+6dB) , ca 90dBSPL @1m


11.27V(+12dB), ca 96dBSPL @1m


22.48V(+18dB) ca 102dBSPL@1m


Left and right speaker at intended positions - microphone at listening position

[kimmosto]

Seems the moving GIF stops moving when uploaded to the forum..

Forum converts (all) images to jpg to ruin quality, inflate size and stop possible animation.

[ergo]

Yeah, that would explain... I do not have a good place to upload/link the images to either. Anyhow, I guess it works works enough to convey the info

One image of the test setup itself also.

[Zvu]

I use https://postimages.org for images. It works ok for now and it's free.

[ergo]

Test to share from Onedrive



Hmm - this sort of works, but still not moving (at least with Opera browser) even though it appears to be GIF now

[Zvu]

I must ask, since your room seems big and i presume high enough, do you have anyone that can help you do the indoor quasi-anechoic measurements ?

In the past i've put the loudspeaker on the stand laying on the side (2 stands for floorstanding speakers), driver centers at 1/2 room height and then i'd made quasi-anechoic measurements on and off-axis.

I'm guessing that, doing that, you could get resolution down to 200Hz or lower so you'd be able to see what's really going on and are there diffraction products without smoothing other than what software does because of length of gate.

[Juhazi]

-Actually very good (looking) plots, considering that the room has nude walls without furniture or acoustic panels! I like to turn the speaker on a turntable instead of movin the mic, his way reflections stay constant.
-What is dB level in distortion measurements?
- 35Hz dip is obviously room mode, the speaker has acoustic suspension if I remember right.

[Zvu]

Agreed, good looking regarding the circumstances. Not as good as quasi-anechoic would be though, which should be the benchmark for us diy-ers without access to anechoic room. When possible/doable, of course.

[ergo]

Room height is the main challenge I have. The room is only 2.5m (8.2feet) as we renovated an old house and could not manage more. And thats as high as it gets in the house. Some rooms are lower than that.

Turntable in above Ardent case is a manual one below the speaker.

For Minerva project I did do the sideways measurement setup. Even built a mic rotation gadget for it...


The only problem is that these ones are at least a 1/3rd heavier it seems. So I have to think if I should try that or just drag them to work and use a weekend to do this in open office space around my lab there. Actually at work I even have an anechoic chamber, but it's for telecom type solutions - so physically small so the space between the bottom and top foam is perhaps some 1.5m, so these will height wise fill lot of it and again the weight is the main issue.

I guess the right way to build something like this would be to make a much lighter weight box for R&D phase and then the build the real one once necessary measurements are done. Will consider that for the next project that I'll probably try 'without the training wheels' - ie not copying one of Jon's designs again

[Zvu]

My friend and i made portable measuring equipment and planned to make indoor measurements in closed football or basketball courts. Battery powered D-class amp, notebook and external/usb soundcard with phantom power. We took one of those PA loudspeaker stands and modified it to suit our purpose. Lowest height is 2.5 meters which gave us 9ms of gate time. Not too bad and easily combined with groundplane to measure lower frequencies for merging.

As we failed to find someone to let us in a gym or closed court at night when the background noise is low, we did it outside. We measured R300's and his Q100's. Crickets at summer, believe it or not, were the biggest problem outside and rain, of course, that surprised us two nights in a row. We did that during workdays so i was pretty tired and sleepy through that week but the satisfaction when i saw the result made it totally worth the hassle.

Given how systematic your work is with Ardents and was with Minervas also, i'm quite eager to see what will be your next project and how will you make it happen

[ergo]

This is the SPL with 2.83V input @ 1m --> ca 84dB sensitivity in midrange. So the Distortion plots are ca 84dBSPL / 90dBSPL / 96dBSPL / 102dBSPL

[ergo]

I've been taking it slow and no changes to the Xover yet. I've learned that it's better to listen to the pair of speakers for extended period (at least 2..3 weeks) with a big variety of music, before deciding what are the aspect I really like or maybe dislike and only then go after changing something.

Below is a shot of my current evaluation lab - contenders from left are
Modula MT MKII EE
Wavecor Ardent EE
Minerva EE
Not sure even Jon himself has had a chance to listen to all these side by side

Anyhow. Final conclusions are yet to be derived, but as it stands the performance does follow the login. Meaning the Wavecor Ardents do excel as best sound and then followed by Minerva's and Modula MT MKII.

Modula's biggest shortcoming is the midbass and bass. Not sure if it is all due to inverter design or is it also the total cm2 of the driver and it being 2way, not 3way. In some sense it is not a fair fight either as the other two are much bigger for a box and driver countl. Aside from that, they do sound balanced but the midrange and high end clarity is mediocre if put side by side like this.

Minerva's take a step further in the aspects of bass / mid / high quality, but somehow they also do not reach quite the level of bass punch and speed of the Wavecor Ardents. What also kind of surprises me is that the Coax design should warrant a very good imaging and depth of sound stage, but again comparing to Wavecor Ardents it is not quite the same level of performance.

Which brings me to Wavecor Ardents.... these are for sure already a best sounding pair I've built to date. The quality of bass is good, midrange is very clean and clear and highs really do have brilliance when they play something that should have that.

I have yet to make in room measurements of all of these side by side, but it does sound a bit so that Wavecor Ardents might have a bit higher high mid / tweeter level than others. So that would maybe warrant some experiments.. but I might start with the line level filter Jon has been advocating as that has just about that much effect I believe.

Overall it has been a very interesting experience. One that is hard to put in words... If analogy from video is something to add then it is sort of like going from 720p to 1080p to 1080p with HDR, but where not only resolution changes put also the bitrate given to the stream. Moving from one speaker to the next in order of sound quality is like improving the SNR (and yet the electronics is the same, so there is not more or less noise from there). Just somehow there is more dynamic info there and this small level stuff adds so well to the experience of 'being there' and forming a good precise and deep soundfield.

[ergo]

Forgot to add the equipment list for reference
Cambride Audio Azur 851D as a DAC (Music from PC, either Spotify or local NAS + JRiver)
Bruno Putzeys Balanced Preamp -DIY kit with relay switched resistor ladder and remote upgrades.
ICEpower based Class-D amp DIY design

[sdl2112]

Nice family photo and thanks for the detailed description. I also have a Modula MT MKII. I did a real quick compare when I only had one up and running. I will need to do a more detailed review but the difference is substantial.

[Juhazi]

Excellent work and thank you for describing each speaker's sound. I haven't had the opportunity to listen to my speakers like that side by side, but I can easily familiarize with your impressions. I listened to 3-way speakers for 3 decades, then tried to learn to like 2-way+sub sound for 12 years - fail! Also a familiar finding that a "normal" dome tweeter sounds better than one in a waveguide (or is it the higher xo).

Still all these are great performers and someone else might value them differently. Are you planning to organize a group listening session with friends?

btw. nice clock on the wall, I have exactly the same!

edit - looking at response at listening spot, I think I see narrow baffle and midrange directivity making a recession 2-3kHz. Perhaps 1,5-2dB off from tweeter is needed...

[sdl2112]

Hey ergo, can you tell us your room dimensions and speaker locations at the intended listening location. Trying to correlate it to what I'm seeing.

thx

[ergo]

Juhazi, yeah this wall clock is bought from a store in Helsinki. So somewhat likely to have more of them on that side of the sea

sdl2112
I did a quick hack of an image. I could do a precise one, but as our family room is open to both kitchen and hallway it'll be hard to calculate the room modes with simple means.

[sdl2112]

That's a great sketch. My room opens to other rooms too. I was curious of the overall room size, as you show, speaker distance from the rear/side walls and to the listener. I've been playing around a lot and have made decent improvements. I planned on placing mine in the living room but the TV, fireplace and speakers don't want to co-exist...we'll see.

[Juhazi]

Yes, room size, shape and material are very important. Furniture, curtains and carpets have less effect. I have noticed that if the room has large openings, these make rt very short in upper bass, which is basically a good feature! But small speakers will be in trouble.

Here my dipoles measured in two totally different rooms. REW settings for waterfall shown too. My guess is that ergo's living room behaves pretty much like my living room.

[ergo]

I've done lots of listening as the Wavecore Ardents are now the main speaker set in family room for all music and movies. I've also done listening sessions with friends and we've been comparing amps and DAC-s to one another. I've never had this level of 'revealing of signal chain' and it is fun for sure to finally be able to make out differences in prior signal chain that easily.

Anyhow, few weeks ago when I still had all 3 speakers side by side I also did some measurements to all 3 at equal loudness.
All measured at same spot in room somewhat off the walls but not in middle of the room.
Microphone was at tweeter axis of each speaker and I attempted to set the SPL to 96dB around 1kHz.

This is the distortion result for each


I've added full size images here too

Since these measurements I've done a small tweak to the Ardents Xover... to be covered in next post.

[ergo]

The listening sessions comparing the 3 speaker did give an impression that Ardent had slightly more forward mid and more of top end. It was very clean sounding and not unpleasant, but in a way left me wonder if I would tune it more like Minervas... With Minervas I did use "Jon's voicing curve' to create a gentle shelving behavior from mid to top end. And I liked that result a lot back then. So some simulation turned out the following tuning which I've been listening to for about two weeks now.



At moment I'm very happy with all aspects of the sound and the low/mid/tweeter balance. For me this seems just right now - so I guess I should conclude that for me this voicing curve should be treated as a 'house curve' that I like

The ability for the speakers to reveal the quality of the sources is still very much there. Last weekend we did a listening session comparing 3 amplifiers - my IcePower (class-D) to Krell KSA based clone (class-A) to a Pavel Macura AB-1 (class AB). The winner was a bit of a tie between the first two, but more important was the finding by all three of us how easy it was to pinpoint the differences between the amps, while we are used to this being much harder with lesser speakers.

So as I continue listening I'm now pondering if it would be time to invest in better parts for the mid/tweet xover. I did build it based on Janzen entry level caps largely, so there might be more to gain there before I put the Xover into the boxes.

(Added these pics in full size to the link too)

[Zvu]

Nothing else to say about Ardents but - job well done.

I am a little confused by peaks of 3rd order harmonic in Minervas and that peak of 2nd HD in Modulas.

[ergo]

My assumption about the midrange distortion peak in Minervas is that the tweeter in the Seas Coax still struggles a bit as it still gets some energy in that range even though the xover frequency is higher. Also this is a relatively loud level. I wanted to make sure the noise in the room does not hide anything. In listening comparison there is bit more harshness in the midrange with the Minervas, so that seems to correlate with the result.
Hmm - I actually have Minerva individual driver responses as Fuzzmeasure files, but need a Mac at work to open. I'll get back to that as those should confirm if my above assumption is true.

The THD rise in modula is odd indeed. I did verify it several times as well as trying same with ARTA. Both REW and ARTA results agreed (same for Minerva). I'm not sure if a waveguide throat resonances could create a THD peak. Have to study that a bit more at some point. Or it's just that it has a big negative notch that creates a bad signal to distortion ratio at that frequency and the somewhat smoothed response does not show it.

The unsmoothed response does show a sharp dip there...

[ergo]

It's been a while since last changes to the Ardents - I've been listening to them daily and still happy, BUT I did have in plans to try to upgrade the tweeter and midrange Xover....

The original one was built with budget priced Janzen caps. Now I have a bundle of ClarityCap CSA and ESA series waiting and also Jantzen coils and resistors. Midrange caps will all be CSA-s, even the high value parallel ones. This way I'm hoping I get a maximum capacitor quality sound effect. As was the plan I will build another replica of same Xover with same values - just different parts.

Madisound had a sale for these Clarity Caps, so I hand carried them back from US two weeks ago from my business trip over there. It was a bit of funny though. These caps in a bad look a bit like a bundle of dynamite, so I did not dare to put them in luggage. I took them to carry on luggage, but then also removed these from bag before the X-ray. The border guard looked me a bit funny, but then agreed that it was probably smart to take these out to plain sight. With a snow storm at airport I did not want to add to the fuzz with bomb threat and all

Anyhow, hoping to find time this weekend to solder all this together too.

[Scottg]

As planned I ran a punch of measurements yesterday
[ATTACH=CONFIG]28459[/ATTACH]

Is that an RTX6001 I see in the back ground? 8)


Nice build-log!

[ergo]

Yes that is the RTX6001 - I was lucky to be in the Group buy over at the other forum. I like the audio test equipment too much, so I keep buying and improving on them. The problem is that by now I have too many and it creates it's own challenges as one has to make a choice which one to use each time So now I mostly switch between CLIO-12 and RTX + Fuzzmeasure on Mac or REW or ARTA on Windows.

Just before RTX6001 I also bought and assembled Jan Didden's Autoranger . With that a good pro sound card can also come close to the RTX in functionality, but yeah the RTX all in one is still a great tool. RTX is too good to let go but I have been thinking of letting go of the Autoranger.

[Scottg]

Yeah, there really isn't much that has better performance than the RTX6001. Even at $2500 US it's a good value (with Multi-Instrument), though Virtins really needs to have an annual sale for $1999 or less. If it wasn't such an "obscure" (poorly marketed) combination, Audio Precision might actually "take notice". 8O :W

Multi-Instrument software is a bit archaic in interface when compared to something like AverLAB though. ops: (..and AverLAB's Ethernet connectivity is pretty slick.)


I'd kind of like to see something like the RTX6001 for mic.s: a Class I capsule with wide bandwidth AND really low self-noise + good headroom. There is almost nothing like that period, let alone approaching "prosumer" pricing. :cry:

[JonMarsh]

Hello Ergo-

Just a reminder- when you were over here and hear the Ardent's at Jens place, as we mentioned then, the system was using one of my voicing filter prototypes. it remains installed at Jens home.

I am pondering whether I should make "flat" and "voiced" standard adjustable options in future crossover builds.

I will go back and read this thread in it's entirety as soon as I can find a little spare time (Hah! spare time? what is that? can you buy it on Amazon?).

I'm always interested to hear builders/listener's reactions- I have a lot of catching up to do for what's been posted on the board.

The listening sessions comparing the 3 speaker did give an impression that Ardent had slightly more forward mid and more of top end. It was very clean sounding and not unpleasant, but in a way left me wonder if I would tune it more like Minervas... With Minervas I did use "Jon's voicing curve' to create a gentle shelving behavior from mid to top end. And I liked that result a lot back then. So some simulation turned out the following tuning which I've been listening to for about two weeks now.

[ATTACH=CONFIG]28632[/ATTACH][ATTACH=CONFIG]28633[/ATTACH]

At moment I'm very happy with all aspects of the sound and the low/mid/tweeter balance. For me this seems just right now - so I guess I should conclude that for me this voicing curve should be treated as a 'house curve' that I like

The ability for the speakers to reveal the quality of the sources is still very much there. Last weekend we did a listening session comparing 3 amplifiers - my IcePower (class-D) to Krell KSA based clone (class-A) to a Pavel Macura AB-1 (class AB). The winner was a bit of a tie between the first two, but more important was the finding by all three of us how easy it was to pinpoint the differences between the amps, while we are used to this being much harder with lesser speakers.

So as I continue listening I'm now pondering if it would be time to invest in better parts for the mid/tweet xover. I did build it based on Janzen entry level caps largely, so there might be more to gain there before I put the Xover into the boxes.

(Added these pics in full size to the link too)

[JonMarsh]

Yes that is the RTX6001 - I was lucky to be in the Group buy over at the other forum. I like the audio test equipment too much, so I keep buying and improving on them. The problem is that by now I have too many and it creates it's own challenges as one has to make a choice which one to use each time So now I mostly switch between CLIO-12 and RTX + Fuzzmeasure on Mac or REW or ARTA on Windows.

Just before RTX6001 I also bought and assembled Jan Didden's Autoranger . With that a good pro sound card can also come close to the RTX in functionality, but yeah the RTX all in one is still a great tool. RTX is too good to let go but I have been thinking of letting go of the Autoranger.

Is it really possible to have TOO many good tools? :W

But I do have a sense of where you are coming from in this matter... different ones have somewhat different strengths they bring to the process... too much of a good thing is definitely a FWP- First World Problem.

[ergo]

Nice to see you back Jon!

My values in use have some tweaking in them already to basically follow your voicing curve. I had good success with it on Minervas and now on Ardents too. So I'b building it into the physical Xover.

I realised last weekend that I had ordered too few 15u and 8.2u Clarity caps due to rush before the trip, so could not start the assembly. But the missing parts are on their way in from UK and should arrive next week. Then I can make some progress again.... The whole Brexit saga rushed me on also - might be that we have to pay as much tax on stuff from UK soon as we do for US.

[JonMarsh]

well, one of our past top semi designers in Munich used to have this saying, "Slow work takes time..."

I'm sure you're looking forward to when those parts DO arrive!

Yeah, Brexit is pretty weird... still not clear to me if it will really happen. Fortunately stuff from the UK isn't taxed much here, as HiFi Collective is one of my favorite online shops, especially because of how fast FedEx delivery is by them, while being reasonably priced.

[ergo]

Finally some progress. Assembly almost all done. Now it will be 100% ClarityCap in Mid/Tweeter xover. As I want to compare the budget Janzen one to the ClarityCap one I decided to go all in and not parallel the higher values with budget caps.

It was sort of a challenge to fit it all on the board size to fit the base, so it ended up being sort of like 'double sided PCB' with components mounted to both sides.

[cochinada]

Never saw a double layer XO before. You will stack up this in a sort of silo I presume?

[ergo]

Yes, the current one looks like this and is made to fit the removable base of the Ardents (just has not got it's home there yet due to these experiments).

[TEK]

Based on the hight of the crossover I assume that your removable base is not a closed box, like the one Jon buildt, but a «lid» so that the crossover is going inside the speaker(?).
Hmm, you do have enough room below the vertical reinforcement? If I recall correctly I had to mount one part on each side of it...

[ergo]

Correct. My base is like a lid that goes over the opening at bottom of a speaker with a gasket in between. I also left a bit more room below the braces as I was expecting the xover to grow too high otherwise.... so hopefully I wont get a nasty surprise that it does not fit after all. But if that happens then I guess it'll just need another rework of it.

[cochinada]

.... so hopefully I wont get a nasty surprise that it does not fit after all. But if that happens then I guess it'll just need another rework of it.

Why don't you model your XO in 3D to avoid any surprises? It doesn't take much time or effort and this way you would know in advance if it fits or not.