Trillium build thread

[Dennis H]

The response looks OK to me, but what's up with the phase?

You need to remove the time of flight on the impulse window. It will guestimate it pretty close by setting the marker (right click) at the peak of the impulse and then clicking 'get'. Then move the marker to where you want your gating to be.

[Dennis H]

LSPCad manual shows how to do that... although you might be ok with the ARTA pink noise and not have to splice data like MLS.

He isn't using random 'pink noise'. It's a periodic signal with a pink spectrum that in many ways is superior to MLS or chirp signals for measuring the impulse response. Floor bounce will still be alive and well but ARTA has the tools to splice in the nearfield. The manual is your friend.

[Jed]

The manual is your friend.

In either case, if using justMLS or ARTA, I recommend he splice in data from nearfield if measuring in a room where boundaries effect the data below 500hz or so. He didn't mention that in his list of procedures, and I find it an important step, especially for a 3 way.

Here's my list that I follow, but I'm not listing the process I go through to merge/splice data here:

1) Measure the drivers for harmonic distortion and impedance
2) Measure on axis responses, and fixed mic responses on tweeter axis
3) Evaluate the data to determine best acoustic slope and crossover point
4) Import data into LSPcad and simulate the design with the different
sets of measurements to verify the final response results are the same
for each set of measurements
5) Order the parts and assemble the crossover
6) Measure the speaker and individual drivers with the crossover in place to verify final response matches the LSPcad model
7) Listen, listen, listen!!!
8 ) Measure off axis response to make sure there are no major dips and peaks
9) Tweak the crossover and adjust for taste and the flattest response as possible
10) Finalize design

[Paul Ebert]

This is why when I measure a speaker, I set it up, set the mic up, and move NOTHING till I've measured everything.

Phase can look like crap but the key here is it's ALL EXACTLY RELATIVE so it lines up right, thus leading to the filters all being good.

There are all sorts of other options for extracting minimum phase and such to work with.

There is an option for displaying minimum phase along with the frequency response. I assume that if I then export this to the frd file, the file will contain the minimum phase information. Is this what I want? I have to admit that I've never understood the concept of minimum phase.

And yes, I will be studying the manual.

I really appreciate the help. I figured this part would be a challenge.

[Jed]

There is an option for displaying minimum phase along with the frequency response. I assume that if I then export this to the frd file, the file will contain the minimum phase information. Is this what I want? I have to admit that I've never understood the concept of minimum phase.

And yes, I will be studying the manual.

I really appreciate the help. I figured this part would be a challenge.

If you use minimum phase you have to set the driver coordinates of each measured driver in your crossover modeling program to represent the location on the baffle. You would also want to measure on axis of each driver at a set distance for each driver.

The nice thing about the fixed mic measuring is you set all the driver locations to 0; however, you don't get the added benefit of modeling the power response.

That's why it's good to do it both ways. Fixed mic, and on axis- minimum phase and compare the results.

As Dennis mentioned, correcting the excessive phase rotations can be done by adjusting the time of flight.

[JonW]

The response looks OK to me, but what's up with the phase?

Hi Paul,

Looks good. :T I found getting good measurements to be the toughest part of plodding through my first design project. You've got good help here from the resident experts. In case it clarifies things any more...

The phase of the driver is the phase of the driver. It's pretty simple. What you've got to do now is have the computer see it in it's simplest form. I'm using LspCAD and in there, there is a distance box where you type in how far the microphone is from the driver. Start with 100 cm. If you see wildly wrapping phase like you have, just try typing in other numbers close by. 98, 103, etc. Get it to the point that the phase looks the smoothest, with the flattest plot and the least wrapping around. (This will not change the frequency response.) Then you're good.

Now move to the mid driver. Do not touch the microphone. Do not change any parameters in the software. Just change the wire connections so you can measure the mid driver. Take the measurements. Then move on to the woofers. The woofers will be more complex because, as Jed says, you may need to merge near field data. But start by taking a regular measurement without changing anything. Once you move the microphone even 1 cm from it's original position used for the tweeter measurement, everything will be off (with regard to phase).

Essentially, what you want to do is get pretty looking phase for the tweeter. Then take all measurements referenced to the tweeter. That way you have a complete set of measurements and when you change things in your crossover model, the phases will all move together or apart as they should.

I hope that helps. Good luck. If you can get by these measurements... and you've already built your complex cabinets... then you're on to smooth sailing.

[TacoD]

I like the dimensions and proportions of the loudspeaker. It is a real looker. The x-over will need some work, those Dayton RS woofers are not the easiest to work with. The resonances may take a notch or two .

[cjd]

I like the dimensions and proportions of the loudspeaker. It is a real looker. The x-over will need some work, those Dayton RS woofers are not the easiest to work with. The resonances may take a notch or two .

Sure they are.

[Paul Ebert]

OK, while I'm reading the manual (I'm on page 45 of 146), I'll experiment with the time of flight and mic distance tweaking as suggested by Dennis and Jon.

TacoD, thanks for the compliment. I'm hoping that the RS225s won't be too difficult given the anticipated crossover frequency of around 750 - 800 hz, but, yes, I could have picked easier drivers, not to mention an easier design, all around. But I do my best work when I'm in over my head .

The ARTA calibration process calls for using a sound calibrator for determining the sensitivity of the microphone. Has anyone actually done this? Where does one get access to a sound calibrator? I'm not inclined to buy one as it looks like they cost around $300. Is there another way to do this part of the calibration? I can get access to a good RMS voltmeter and a signal source at work, so calibrating the soundcard should not be too difficult.

[JonW]

Hi Paul,

A couple more things to suggest…

If you cannot figure out the mic distance/time of flight thing (they’re related or the same thing) it could help to download the LspCAD tutorial for the measurement program within LspCAD called “JustMLS.” It has a walk through of how measurements were taken for one system (“ugly ducking.”) I forget if it’s in the manual or a separate file or what. It’s not very explicit and, in general, a lot more for details could have been included. But if you get really stuck it might be worth a read.

On the topic of calibrations, you must get your microphone calibrated. Typically costs $40-50 or something like that. Send it out to someone who will send it back with a calibration file. My ECM8000 had an 8 db rise by 10,000 Hz. Some people have dips with the same microphone. I’d say you cannot cook up a decent crossover without measurements from a calibrated microphone.

[Paul Ebert]

I have had my microphone calibrated and am using the frequency compensation in ARTA. But, that gives me a relative measurement (deviation from 0.0 dB vs. frequency). As part of the calibration, ARTA expects a sensitivity of the mic in mV/Pa. The calibration report I received does not provide that. It does give the sensitivity of the reference mic in "dB re 1 mV/Pa". That value is -32. Doesn't this come out to around 0.025 mV/Pa? If so, the sensitivity of my mic would have this value as well since they are basically the same except above 3k or so. That value, however, strikes me as quite small.

ARTA also wants the preamp gain, but I can set that to 1.

[Dennis H]

Yeah, that mic sensitivity doesn't sound right. I've seen numbers in the 8-30 mV/Pa range depending on the brand of mic. Can you check with the guy who did the cal to see what the deal is? To check if it's in the ballpark, you can compare Arta's SPL meter to an RS meter at 1kHz. The RS meter may be off a couple of dB but you can at least see if they are somewhat close.

[Paul Ebert]

I've been using PCD to do some modeling using the measurements I've taken. One concern I have is that the system impedance seems weird (see attached). Will the peaking above 400 or so be a problem? I can get rid of it by deleting the HP on the tweeter. Nothing else I've experimented with has impacted it.

If this is problematic, does anyone have any suggestions?

[Paul Ebert]

Trillium Tuneage!

So, after several weekends of measuring and modeling, I decided it was time to build a prototype crossover. It took most of the weekend putting it together, but I got to try it out this evening and ... it's surprisingly good! No real nasties. Perhaps, a bit of muddiness in the lower midrange and a bit of reticence in the highest octave.

But, the bass!!!

Despite having most of the bottom open, it sounds deep, strong, and clean. I can't wait to hear what it will sound like when the bottom is properly attached (either sealed or ported - I plan to try both). I think I'm going to be glad I went with the dual RS225s. For my room, a single one would probably suffice, but having two allowed me to raise the sensitivity a few dBs and have some more flexibility with baffle step compensation.

Anyway, here's a photo of the prototype crossover:



I used an idea from Roman Bednarek to use binding posts to connect the parts. This allows one to put parts in parallel by adding a dual banana jack with the additional part(s) into the binding post. For now, I'm using el cheapo non-polar electrolytics. When I've got the values nailed down I'll go for better caps (and, maybe, Mills resistors instead of the Daytons and an air core inductor in place of the P-core).

So, I guess the next step will be to put in the bottom (sealed) and see how that sounds. And, I'm sure I've got more tweaking to do.

This sure is fun!!!

arty:

[Paul Ebert]

Well, I'm in the middle of crossover development (The changes I'm making are too big to call it tweaking, at this point). I'm finding this stage to be quite difficult. I've gone back and forth quite a bit between crossovers that measure like this:



or, like this:



So, it seems the choice is between reasonably flat above 500 hz or bass, though the bass does not sound as emasculated as the first graph seems to indicate. You'll notice, from the line at the bottom of these graphs that the gating is set to some point above 300 hz. What exactly, does this mean? Does it mean that the graph below that point should be completely discounted? I'd like to find a way to get the bass of the second graph (or better) with the flatness of the first (again, or better), but changing the components hasn't gotten me there, so far.

So, a question: how do you measure and tune bass, when you design? Any options beyond finding a huge room (or anechoic chamber) or going solely by ear?

[pedroskova]

Although I've yet to use it, ARTA allows you to "splice" in nearfield response with farfield. Measuring just inches away from the driver allows for a larger window with fewer room artifacts. ...see the help files.

You can also just increase your windowing, and turn a blind eye to the upper bass / midrange. The overall response gets pretty peaky, but you can get some idea of what's going on.

[JonMarsh]

:T Bravo, Paul. Had to follow up to find out what the Trillium is, with the reference from Craig J. A man after my own heart, as one item on my "to-do list" is to convert my M8ta to three way with an RS52 midrange. :T :T :T

[JonMarsh]

Oh, I forgot to explain the name. To me, 'Trillium' sounds like some sort of unstable element or some other cool thing befitting such a large and ambitious three-way design, but actually, it's a small three petaled flower. No matter, I like the sound of it.

I do to... M8ta is so "dry".

[Paul Ebert]

:T Bravo, Paul. Had to follow up to find out what the Trillium is, with the reference from Craig J. A man after my own heart, as one item on my "to-do list" is to convert my M8ta to three way with an RS52 midrange. :T :T :T

Thanks, Jon! Glad to see you can come up for air every once in a while. You certainly were the main inspiration and mentor for this project (with a measure contributed by Zaph and others). Monday I was finally able to temporarily mount the drivers on the second cabinet and hear them in stereo. I say 'temporarily' because I haven't cut the facets on the second one, yet. It looks like that will have to wait a week and a half as I'm accompanying my son on a college visit this weekend. I'm so glad I was able to sneak these in before the tuition bills pile up even more than they have been :W .

I'm still working on the bass - it can be a bit boomy. Also, if I go by my cruddy, and highly suspect, measurements, they seem to plummet at around 65 hz. I'm still trying to figure that out. I may have to wait until the weather is good enough to measure outside.

I will be drawing up and posting the crossover soon. It's not final - I plan to take the pair to InDIYana and, based on the reactions they get, I may make more modifications (I am, after all, a newbie at this).

Sure has been a learning experience and a lot of fun. Even in their current state, I'm very happy with them.

[jquin]

:T Bravo, Paul. Had to follow up to find out what the Trillium is, with the reference from Craig J. A man after my own heart, as one item on my "to-do list" is to convert my M8ta to three way with an RS52 midrange. :T :T :T

Oh don't do that. I've only just finished my M8ta.
It better not sound nicer than the current M8ta!!
I don't want to replace all those x-over components! Sniff

[TacoD]

Oh don't do that. I've only just finished my M8ta.
It better not sound nicer than the current M8ta!!
I don't want to replace all those x-over components! Sniff

I think you have some time to enjoy, it will take some time to finish the design, before we know it Jon is occupied with work again

[JonMarsh]

I think you have some time to enjoy, it will take some time to finish the design, before we know it Jon is occupied with work again

you know me and my wretched lifestyle far too well! :rofl:

I was really hoping for a long break from business travel, but the last few weeks have demolished that idea... but I do have a week off every month for the next three months.

[Dennis H]

but I do have a week off every month for the next three months.

Sounds like you have a clever boss..... give you 4 weeks work and 3 weeks to do it in. :W

[JonMarsh]

Why, yes, he is pretty clever that way.... as long as I rise to the challenge. But at some point, there's only so many hours, and once you run out of them, you're out of them. We won't mention that I've spent this morning on my time off catching up with what I couldn't finish last week....

[---k---]

My company had a Management Consultant/Motivation Speaker come in last year and give a 2 day seminar. This page is straight from his book:


Unfortunately, management is in denial that this page exists. We're not quite in the same boat as you, but during the summer when construction is in full swing, they expect us to be 110% utilized to make up for November and December. It can be insane at times.

I used to be an over-achiever with a yearly average billable utilization of 80%+ (there is also a lot of required non-billable crap). But now, I'm what another part of the book calls refers to as a "disillusioned learner".

[JonMarsh]

Yeah, that's a kind of familiar issue; the guy who was my manager until 2001, and who retired the end of 2007 (laid off, but with 24 years with the company, got a large severance package (2 weeks pay + 1 week pay for every year of service), was in a 11-12 hour a day mode most of the last three years of his employment. Then someone decided we didn't need all this conformance tracking and quality motivation stuff-

The only thing working in my favor is that with attrition and retirement they've lost so many of the long experience high knowledge people, that I can get more done in 7-10 hours than some guys can in two or three days... this works in my favor, as regards trying to rationalize my life. I have more leverage, but am using it sparingly so far.

[pedroskova]

I still haven't gotten over the news that National Semi laid off Bob Pease. Who's next...the Pope? I'm just a simple tube cretin (even opamps shiver at the sight of me), but even I know who Mr. Pease is.

[JonMarsh]

Yeah, I was just talking about that to a young German engineer at work- he didn't know who Bob Pease was; I explained, it was about like laying off Jim Williams. But he didn't know who Jim Williams was. When I explained it was like laying off Dilbert (who he's a fan of), then it got home.

[jquin]

I still haven't gotten over the news that National Semi laid off Bob Pease. Who's next...the Pope? I'm just a simple tube cretin (even opamps shiver at the sight of me), but even I know who Mr. Pease is.

My old boss had the excellent book
Troubleshooting Analog Circuits by Bob Pease.

I must admit I was never taken with analogue. RF yes, digital yes, analogue no.

Still the world I live in now you just buy boards from OEMs and integrate them.
Unless you work for an OEM there isn't much call for expertise in engineering anymore, just integration.
Having said that you can never have too much background. It's invaluable when things go wrong.
It's just a shame that it's almost impossible to gain that level of experience on the job nowadays.

[pedroskova]

My old boss had the excellent book
Troubleshooting Analog Circuits by Bob Pease.

I must admit I was never taken with analogue. RF yes, digital yes, analogue no.

Everything electronic is analog in the end, unless I missed the part where transistors, capacitors, etc., aren't needed inside those gray square things with all those legs stickin' out of them.

EDIT: sorry for taking this thread OT. We will now return to our regularly scheduled thread.

[Paul Ebert]

I got the facets cut on my second speaker today and its crossover mounted inside (rather than on the floor). Good progress.

Cutting those facets by hand sure takes a lot out of this 50 year old. I think, if I were doing it over, I'd make the sled that cuts them on a table saw. The second one turned out much better than the first. The first will need some bondo to fill in some gouges and slightly rounded over corners.

Tomorrow, I hope to do final assembly of the second crossover (it's currently on the jig previously shown) and get that mounted. Then I'll be pretty much ready for InDIYana!

[jquin]

Everything electronic is analog in the end, unless I missed the part where transistors, capacitors, etc., aren't needed inside those gray square things with all those legs stickin' out of them.

EDIT: sorry for taking this thread OT. We will now return to our regularly scheduled thread.

Actually everything is quantum!

[Paul Ebert]

Tomorrow, I hope to do final assembly of the second crossover (it's currently on the jig previously shown) and get that mounted. Then I'll be pretty much ready for InDIYana!

No such luck. Soon after getting out of bed this morning, I noticed that the sump pump was not turning off. This meant that its pipe had frozen and I was seconds away from a flooded basement. And, since the sump is in my workshop, it would be a flood with a good amount of mdf dust mixed into the water.

Well, after much panic and barking out orders, we managed to avert the flood and get the pipe cleared and the pump working again.

I spent the rest of the day sweeping up the shop, just in case...

I think I'll hold off building any more cabinets until I have a router with dust collection and an air filter.

[brianpowers27]

These sounded great at InDIYana. Where are the crossovers currently set? Do you think you are going to keep them as they are now?

[viki_v2]

any link to crossover