PERSPEX QUASAR ver 2.0 prototype build

Dedicated to those large boxes at one end of the room
JamesD
Old Hand
Posts: 784
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by JamesD » Sun Aug 30, 2020 1:46 pm

A few words on first order series filters

There is no one value of inductor and capacitor pair that sets the x-over frequency but rather a series of pair values that all yield the x-over frequency - these are, in effect, damped by the resistance of the drivers and hence have an associated 'quality factor'..

Its true that as a generic statement series x-overs tend to "sum flat" but it is not true that they are necessarily completly flat through the x-over region.

The "Q" value of the series filter indicates whether there is a bump or dip or completely flat region through the x-over. "Q" less than one produces a bump, Q greater than one produces a dip - these can be up to about +4dB or -6dB in practice. Doede Douma has produced a spreadsheet that helps explore this http://www.dddac.de/files/serial_filters_6dB.xls and a discussion of his experience with the x-over here http://www.dddac.de/op083b.htm

The sound of the series filter changes according to the chosen "Q" value :D

All the above is an simplification as it assumes the loudspeaker is dominated by its resistive component - this is not accurate - a loudspeaker electrically is a complex impedance so all the usual caveats regarding crossover design apply - the above does get you close to what sounds best

The AER/Supravox Quasar uses a 15mh low DCR choke and a 104uF capacitor comprising a 100uF PP with a 4uF PIO in parallel and moving to different values or types does change the sound - in my view for the worse.

Hope that helps!

JamesD
Old Hand
Posts: 784
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by JamesD » Sun Aug 30, 2020 2:01 pm

I should say that Greg has been utterly consistent in being able to hear the two drivers in the Quasar as separate locations from the first time he heard them at Eggborough and has always shared what he has heard with me. His observations and discussions have been and are valuable to me.

I look forward to hearing the results of Chris's further measurements. It does look to me like some driver interaction at the bottom end as well as room effects but that is pure guess work based on some of my previous observations.

I used to have discussion with Alan Shaw, the owner and designer of Harbeth loudspeakers, regarding how best to measure loudspeakers and what was the best mic to use... He found putting his loudspeakers on top of a 12' pole in the middle of a field helped him measure them accurately with the mic about 6' away and on axis with the mid range driver. Unfortunately the Quasar doesn't work away from the floor boundary so that doesn't help!!!

User avatar
Scottmoose
Needs to get out more
Posts: 1567
Joined: Sat Sep 08, 2007 11:03 am
Contact:

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by Scottmoose » Sun Aug 30, 2020 3:24 pm

Agreed, 1st order series will [electrically] sum flat in both electrical polarities only when Zeta = 1.0 (Butterworth). In theory, it should continue to sum flat with different Zeta (effective Q) functions, but due to the different phasing will only do so in one polarity, not both, or you'll end up with the results described. The actual acoustic behaviour is a different matter entirely of course. ;)

Likewise -I'll be very interested to see further measurements / data from Chris when he's able to take it too. Trying to get an accurate LF measurement is usually a pain; free air measures as Alan Shaw describes are about the best method assuming the speaker isn't intended for boundary loading (and providing the ambient noise is low enough). If they are, then you need to splice or blend gated far-field data for most of the range with nearfield for the LF, with a suitable level adjustment for the latter. Not a problem in itself, but if your crossover happens to be slap-bang in the same range as you're splicing / blending the two, you'll loose resolution. Better than nothing though.

As noted, I wouldn't take the filter Colin posted as anything other than what it was intended to be: a 'this will make noise' circuit based on a huge number of surmises because Supravox's data isn't great, Lii Audio's is bordering on non-existent, and unfortunately I'm in no position to take any measurements of them myself.
'"That'll do," comes the cry of the perfectionist down the ages.' (James May The Reassembler)
Website www.wodendesign.com
Community sites www.frugal-horn.com & www.frugal-phile.com

User avatar
IslandPink
Oooo, Roberts Radio's sound nice
Posts: 7992
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by IslandPink » Sun Aug 30, 2020 7:56 pm

Those are very interesting links, James, I need to read that in more detail.
I did put together a simple LT Spice model of the series cross when i got confused with the FX120 pair, that helped me get back on track, and is available to anyone who wants to send me a PM or email - but it's fairly quick to set up anyway.

ps. James , do you happen to know what the C79 and C12 units are, that the fella is working with on the series cross ?
"Peridots and periwinkle, blue medallions Gilded galleons spilled across the ocean floor" ( Mitchell )

JamesD
Old Hand
Posts: 784
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by JamesD » Sun Aug 30, 2020 9:01 pm

They are Accuton ceramic drivers replaced by the c25 and c90 drivers that are presently listed on their site. https://accuton.com/en-home/produkte/la ... er/keramik

Speaker is described here http://www.dddac.de/op08.htm with the units mentioned under the basic description.

Interestingly he changed the tweeter for the Diamond D20 for the final version...

chris661
Shed dweller
Posts: 2375
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by chris661 » Mon Aug 31, 2020 1:18 pm

I've been tinkering with the crossover simulator.

First up, I set off with the original drivers & crossover circuit, and it produced a familiar-looking response curve.

Image

Altering both the inductor and cap values changed the crossover frequency, as expected, but IMO we'd have to get heavy-handed with DSP to get that Fast-10 driver to be useful into the treble. A pair of notch filters on that one driver got things better, but obviously couldn't fill in the 3kHz dip. They'd still need a super-tweeter, too. IMO, no amount of break-in is going to fix that kHz range. The Fast-10 just isn't suited to this application, so I've simulated something else.

Here, I've replaced the Fast-10 with the compression driver & horn used in my big PA speakers: 18Sound ND1460 and RCF HF94. The HF drivers are a couple of hundred pounds each, and should be good company for the Supravox unit.

Image

Some notes on that:
- The impedance curve of the HF driver mandates the LCR network to flatten things out. Otherwise, the crossover goes to pot and it gets really peaky around 1kHz. This should be a consideration for all passive crossovers involving horns.
- The parallel resistor and cap shelve down the lower range of the HF driver. It's a constant-directivity horn, so this is to be expected. Varying that resistor allows some tuning of the overall balance.
- I'm a little concerned about running the HF driver this low. Chances are it'll be fine at sensible volumes, but cranking it up might make it struggle. Comes with the territory of first-order crossovers. If there seems to be promise here, though, the latest 4" beryllium compression drivers will offer more power handling at the bottom of the range. There are also coaxial compression drivers that might be worth consideration.

I'm not sure how legible the component values will be, so I'll list them below:
Crossover: 4.2mH, 33uF
RC shelving: 16ohm, 1.6uF - might need changing depending on how the sensitivities work out. I didn't measure the HF driver at the same drive level as the woofer (nor did I use the same mic gain), so I had to estimate that. I'm expecting the HF driver to have a 10dB+ advantage on the woofer, so there should be plenty of wriggle room.
LCR: 10ohm, 51uF (can be electrolytic), 1mH

It's more components than would be required with a theoretically ideal 500Hz-20kHz driver, but horns are difficult to work with and this is the best I could do while maintaining the original series crossover.


I'm expecting this to be interesting to listen to, and far less immediately objectionable than the Fast-10 driver. However, there are some things that could still trip things up:
- We've got a dipole covering half the range, and a monopole covering the other half. The room interaction will be interesting.
- Running a compression driver that low with a shallow filter could lead to distortion.
- The LF response isn't where I'd want it for a go-anywhere do-anything sort of speaker
- There's still the vertical spacing to consider

We shall see where the path leads.

Chris

JamesD
Old Hand
Posts: 784
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by JamesD » Mon Aug 31, 2020 2:22 pm

quick note - most full rangers are not dipole radiators after about 1500Hz but mono-poles so changing over to the compression driver won't cause much of a radiation pattern discontinuity compared to a twin cone driver - it may not be quite as gradual a change over but is a difference of degree not of type...

chris661
Shed dweller
Posts: 2375
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by chris661 » Mon Aug 31, 2020 2:45 pm

James,
If you take a look at the measurements initially posted, you'll note that by far the best measurement curve (and actually, the best sound) was from behind the driver. Apart from a dip at 4kHz, it was pretty flat out to 10kHz.

Granted, there may have been some reflections coming in - if the 9kHz Spike Of Death bounced around the room and back to the mic, it'd prop up the frequency response.

Chris

User avatar
Toppsy
Shed dweller
Posts: 2224
Joined: Mon Jul 23, 2007 12:10 pm
Location: red rose country
Contact:

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by Toppsy » Fri Sep 11, 2020 3:26 pm

After the measuring of the drivers and the particularly disappointing FR plot we got for the Lii Audio Fast-10 drivers I know that Nick sent Lii Audio an email enquiry highlighting the resultant measures we got and for their comments/feedback on the results being nowhere near their published data. Though I know the reply Nick got back it is not for me to comment on what they had to say other than the response prompting Nick to right off these drivers for this project and invest in a pair of Lowther PM6a silver wound drivers. These are expect sometime towards end of September, possibly early October.

When the new drivers arrive it is the intent for Nick and Chris to come over and we can spend some time to do some further testing/experimenting to find the best working combination of drivers and XO for these Quasar cabinets. As well as the new Lowther's to try out, Chris has a pair of 1.4" compression horn drivers he will be bringing to the party to try as an alternative. Should make for an interesting day.

This posses a few issues of how to modify the front baffles so we can quickly interchange from the compression horn to the Lowther's. The compression horn requires a 205mm cut-out hole. The Lowther requires a 190mm through hole and the existing cut-out for the Lii Audio is 216mm dia. Now the frame of the horn is 230mm square so this should fit straight onto the baffle to cover the hole and the bolt hole spacing give clearance to the cut-out hole. So an easy assemble on the day.

The mounting of the Lowther's however required the making of a couple of doughnut rings from 18mm chipboard with a 216mm outer diameter and 190mm inner diameter. These where simply made using the circular cutter jig on my router and a 6mm straight cutting bit. These were glued to 270mm sq. 6mm thick BB ply mounting frames. The end result looking this:

Image

Here is the existing cut-out with round-over profile:

Image

And here is the doughnut mounted:

Image

Whilst I was in modifying mode I decided to remount the Supravox bas units so these were flush mounted from the front of the baffle:

Image

If at the end of the day Nick decides to go with the Lowther coupled with the Supravox's then I can simply mount the doughnut ring from the rear and fill the round-over anulus on the front baffle with some wood filler and sand all smooth ready to paint the cabinets white if nothing else than to seal the chipboard.

Fun times ahead.

chris661
Shed dweller
Posts: 2375
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

Re: PERSPEX QUASAR ver 2.0 prototype build

Post by chris661 » Tue Sep 15, 2020 1:02 pm

Interesting times ahead, Colin, and the work looks great as ever.

I've tinkered with the crossover simulator a little further, and a cap inline with the compression driver (surprisingly small values, actually, 3u3 to 4u7) cuts down the <500Hz range from the HF driver, should mean it'll handle higher levels without struggling. I don't know how much of a problem it'll actually be, though, but it's something that'll help if the problem shows up.


I'm also curious about what the Supravox unit would do in a sensible-sized sealed box, but that's really moving away from the Quasar design.

Chris

Post Reply