QSC Waveguide project, XOs (active and passive)

[penngray]

I have been doing everything so far under the "learning HOLM" thread found here

https://www.htguide.com/forum/showthread.php?t=36100

Because I needed to learn how to measure. I think I have had some success so I want to focus on crossover design in this thread. Specifically my speaker build which started here.
Which are these

Image not available

The build is here..

http://www.avsforum.com/avs-vb/showthread.php?p=18338739&highlight=#post18338739

[penngray]

The project has One woofer (TD12M), 3 different CDs involved (I swap out the CDs I need to measure). I will build more boxes soon to add a center channel.

The CDs are Celestion 1425, Celestion 1745 and the BMS 4550

I have done full measurements of the 1425, Saurav has been extremely helpful in creating some XOs for me.

[penngray]

Here is the 1425 XO recommendation for the DCX

https://www.htguide.com/forum/showpost.php?p=496500&postcount=78

With this addition

https://www.htguide.com/forum/showpost.php?p=496699&postcount=141

Here is the 1425 XO (2Khz LR4)

https://www.htguide.com/forum/showpost.php?p=497301&postcount=187

Here is the BMS4550 XO (1500Hz)

https://www.htguide.com/forum/showpost.php?p=496699&postcount=141

BMS4550 XO (2KHz)

https://www.htguide.com/forum/showpost.php?p=497302&postcount=188

BMS4550 XO (1200Hz)

https://www.htguide.com/forum/showpost.php?p=498483&postcount=213

I have not run any of the BMS4550s yet. Im still working on the issues with the 1425.

[penngray]

1425 Measurements XO @1500Hz.

Gated Horizontal

Image not available


Gated Vertical

Image not available

On horn axis measurement

Image not available


so....I guess the fact that I did all measurements between the horn and the woofer then this is what measurement I get @ the horn axis (5" move vertically).

Furthermore, After spending the last week discussion the 1425 in another project I have concluded that its not wise to cross it over below 2KHz no matter what. I have found that the 1745 might be fine for crossing around 1500Hz or so.

1425 impedance

Image not available

1745 impedance

Image not available

BTW, the Lpad I have on the 1745 tames that very well. I will post that later.

[penngray]

I just spent 1 hour today getting measurements with the 1745.

This time I have an Lpad (5.1/4.7 ohm resistors) and I have some caps to try, after different caps (in parallel) I was able to get this measurement. Its 3.2 uF (2.2 uF+ 1 uF)

Image not available

Blue is no DCX,

Green is But18 @ 1500Hz
Red is But18 @ 1900Hz.

Those crossover points are guesses now, I forgot to write them down, its not a big deal though. They are not important to the discussion.
One thing that is interesting is the THD...

Image not available

The But6 alone (has much less THD @ 1K) but the THD increases if I add those 3 order slopes. Its something I never considered before.

I'm would love to just use a 1st order BUT high pass solution. What issue does that lead too though? The THD is lower with it.

[penngray]

Here is the woofer in the measurement. With the Lpad on the CD I had to drop the woofer by -4dB ( DCX setting for now).

The below is the TD12M with a But18 slope @ 1KHz.

Image not available

I summed the two to show the slope, looks okay...There is a peak around 1445Hz so I EQed that in the DCX

Image not available

here is the crossover region zoomed. This is zero DCX Eqing on the CD, just an Lpad and 3.2 uF cap. The woofer has a But18 slope @ 1KHz and EQing (-2dB @ 1440Hz).

Image not available

[Saurav]

It's difficult to tell how well your phase is tracking just by looking at the on-axis straight frequency response. What you need to do is invert one of the drivers (either in the DCX or in the physical wiring) and see what that looks like. With an LR4 crossover, you should get a deep null at the crossover frequency. With a B3 crossover, ideally you should see no change at all - inverting a driver affects the vertical polar response, and does not affect the on-axis response. In practice I've never achieved this ideal

You could also turn on the phase display in HOLM and see what that looks like.

[penngray]

Thanks, I know I have a lot more work.

I was happy to see just the response of the CD with that Lpad and 3.2 uF cap. its pretty flat from 2K Hz to 15Khz. Definitely works in my world.

Im still going to do the BMS4550, your hard work isnt being thrown at all either. Without your help I wouldn't be doing any of this!!

[Dennis H]

Doug, it's time to buy some software so you can save files. ARTA and LspCAD Std. are both pretty cheap compared to how much you've spent on the rest of this crazy hobby. In case you missed my mini LspCAD get-started in the other thread (there was a lot going on then) here's the link. I saw you asking about Speaker Workshop in another thread but you either have to be CJD or a real masochist to use that one IMO.

https://www.htguide.com/forum/showthread.php?p=496609#post496609

[penngray]

Yeah, I remember your post on LspCAD. Thanks, I was going to get too it again at some point.

I thought I was doing well with speaker workshop, I have drivers (imported FRD and impedance data), I have the network with the drivers added and I was just trying to figure out how to do the optimization stuff. Its asking for some predicted curve stuff....I do not have that though.

Btw, I do have Xover pro and I imported the data into that too. It doesnt do optimization though but it has its value though.

If I was to start over, yes I would have just purchased SoundEasy. Considering it has WT3 stuff and Xover pro covered.

Im still more confortable just changing values in the DCX to see what happens to my response.

[Saurav]

The optimizer in Speaker Workshop isn't really very useful, IMO. The one in LspCAD is much better. With SW you can only optimize one driver's response at a time, and to a pre-defined target curve. What you need is to optimize 2 or more drivers together, to a flat target. No idea about SE, the interface on that seemed like a steeper learning curve so I went with LspCAD.

your hard work isnt being thrown at all either

No worries I definitely understand wanting to experiment and learn. I'm also curious about how your hybrid active/passive solution's going to work out.

[penngray]

Thanks for the opinion on Speaker workshop, I won't bother with it if it isnt nearly as good as LspCAD.

I will starting figuring out the demo version of LspCAD. I do have a question about "target response". Do I need to create something (ie a flat response)?

[Saurav]

I can't really help you with LspCAD and passive crossovers, because there's a whole bag of tricks involved that I know nothing about. For designing with the DCX, this might help:

https://www.htguide.com/forum/showthread.php?t=35595

The basic idea is, you optimize each driver individually to its target acoustic response, and then you switch to the combined response, and optimize delay/level to get the phase to match. That's how I do it, anyway, because I (feel like I) understand what I'm doing

[soho54]

I wouldn't write SW off that fast.

It's true the the optimization routine is a waste of time, but that just means you have to learn to do it manually. Give yourself a few hours to really learn how each component effects one another and it will help out immensely with any program.

If you go back into the threads from around 05-06 here there are a lot of designs you can copy with the measurement files already online. It will really speed along the learning curve to try to recreate them yourself, and play around with them a little. The only way to learn it is to do it, and you might as well use the info on tried and true designs that are still available.

I know I still have crap from Jon M and cjd saved in the back of my old laptop. :B

It sounds like you have already figured out how to input your info, and get around in the program. These means you have the hard part figured out already. The SW GUI is it's biggest drawback.

[penngray]

Thanks!

I looked all last night for examples online. I only found the one Speaker workshop tutorial. Even the forum for it was a dead end.

Examples would be great! I would love to get buy without spending more $$$ on software.

[soho54]

Here is info for cjd's big RS WWMTM 3-ways, and the RS WWMT. Everything is there. I think v2 is what you want.

There are measurements for the modula MTs from two sources. You will have to get the xover from the thread as I can't find my SW version.

There are other things there, but as I was just playing with them when I was on the road to pass the time and didn't store them in an intelligent manor. :roll: Sorry. ops:

I have more on an old server drive, but it started producing errors with a random tick. I have it uninstalled right now until I buy another HD to try to save the data. I will never buy Maxtor drives for backup storage again. Every single one I have owned failed before 5 years were up. At least I caught this one before total failure. Now I have to decide on what order to pull the data off in case it dies half way through the 450gigs. :M

http://rapidshare.com/files/370584890/1_speaker_info.zip

[soho54]

Man, I just pulled up that old data in SW, and most of it is crap. ops: It was left over experimental stuff I was playing with while hotel hoping.

The Big 3-way data in 1 speaker info.zip\speaker\RSDriverData is ok.

All the SW data files in the main folder need a baffle compensation. I was playing with FRD programs at the time, and adding baffle effects afterward. :roll:

I get bored in hotel rooms. :B

Sorry, it wasn't as helpful as I though it could be. The good stuff is on the bad storage drive, of course.

[penngray]

Thats okay, I have started to figure it out...its not bad at all.

I still haven't built a target curve (too tedius).

but I have this so far.

Like you commented I used a network for each driver

TD12M

Image not available

BMS4550

Image not available

I then used the Combined feature to get this...

Image not available

Not bad for a noob so far, Im not sure its at all meaningful though, Are there phase issues?

I have not addressed any of the offset AC stuff at all and I still have no idea how to do that but Atleast I can get a relatively flat FR plot of two drivers with XOs.


and that peak pask 15K is really distrubing (it isnt in my Celestion measurements).

[penngray]

Do the vertical dotted lines around 1300 Hz mean we have a phase issue. Should that line match up perfectly. Is this part of the AC offset?

Do I figure out different crossovers to correct it. More or less, trial and error?


I also figured out how to create a target curve and then run optimize, it didnt do anything great for me.

Adding the BSC circuit drops my woofer response down more. I didnt know it did that, I was thinking in reverse and it would have raised my response below 250Hz.

[Saurav]

A passive circuit can't raise output, so it should drop the high-frequency output, which will bring it level with the lower frequencies.

When you run the SW optimizer, you need to check every component you want it to vary. You'll see L1, C1, R1, etc. in that window, check each one and then start the optimizer.

If you optimized to an LR4 target, you can check phase matching by inverting one of the drivers (right click the driver, it's in there somewhere), then seeing how the outputs combine. If you get a deep notch, then your phase matching is good.

I've never used that Combine feature, I usually create a new empty network, copy-paste both individual networks into it (so you end up with 2 sources, which is OK), then look at the output for this new combined network.

[soho54]

Adding the BSC circuit drops my woofer response down more. I didnt know it did that, I was thinking in reverse and it would have raised my response below 250Hz.

If you measured the speaker in the enclosure you don't need a dedicated BSC circuit. It was part of the frd, and when you flattened it you corrected for BS already. :T


Edited out a section due to not thinking. :roll:

[soho54]

I have not addressed any of the offset AC stuff at all and I still have no idea how to do that but Atleast I can get a relatively flat FR plot of two drivers with XOs.

---

Is this part of the AC offset?

It could be. Did you use minimum phase? If so...

Here is a link to help you out: http://www.rjbaudio.com/Audiofiles/d...alculator.html

[Dennis H]

Do the vertical dotted lines around 1300 Hz mean we have a phase issue.

More likely it's a graphing artifact. If the phase is -179.9, -180.1, -179.9, -180.1, you'll get a couple of vertical lines as it wraps from -180 to +180 on the graph but you wouldn't even be able to see that if the phase was unwrapped and allowed to plot <> -/+180.

[soho54]

I shouldn't post while watching TV. :B

[penngray]

If you measured the speaker in the enclosure you don't need a dedicated BSC circuit. It was part of the frd, and when you flattened it you corrected for BS already. :T


Edited out a section due to not thinking. :roll:

I didn't flatten it because I confused myself on how to handle BSC. I was thinking that I had to raise the response below 270Hz to compensate for the diffraction but that is because Im use to using an Active crossover.

I did figure out that this is the backwards appoarch and the idea is to flatten the curve on the woofer completely.

[penngray]

A passive circuit can't raise output, so it should drop the high-frequency output, which will bring it level with the lower frequencies.

When you run the SW optimizer, you need to check every component you want it to vary. You'll see L1, C1, R1, etc. in that window, check each one and then start the optimizer.

If you optimized to an LR4 target, you can check phase matching by inverting one of the drivers (right click the driver, it's in there somewhere), then seeing how the outputs combine. If you get a deep notch, then your phase matching is good.

I've never used that Combine feature, I usually create a new empty network, copy-paste both individual networks into it (so you end up with 2 sources, which is OK), then look at the output for this new combined network.

Thanks, raising the output was my noob thinking and that I have only used Active designs so its easy.

THe optimizer didnt really do much to my responses but it was cool to figure it out. I have some more flattening to do with my woofer so that I handle that drop @ 270Hz but right now I have a 3rd order buttworth slope on the woofer and a 1st order on the tweeter creating a 3rd order slope with the waveguide. I will check for nulls and validate my phase.

I used the combined feature because I didnt know you could copy the two sources onto one network (Great idea, dumb for me not to even try that easy copy)

The spike past 15K Hz on the BMS4550 is still an mystery, more measurements show its there but the Celestion 1745 doesnt have it.

[penngray]

More likely it's a graphing artifact. If the phase is -179.9, -180.1, -179.9, -180.1, you'll get a couple of vertical lines as it wraps from -180 to +180 on the graph but you wouldn't even be able to see that if the phase was unwrapped and allowed to plot <> -/+180.

Yes, I know its phase wrapping but the combined has the the vertical lines not completely together, they are side by side so I was curious if that matter.

[penngray]

Another question...

I found it interesting that placing the Cap after the Lpad on the Tweeter produce a slightly better response (in my noob eyes) but doing that is bad isnt it?

[penngray]

Another noob confusion this morning.

To get my response flat from 100Hz and beyond I have to use BSC circuit that drops my TD12M woofer response 10dB.

That doesnt seem right so I can probably assume that the measurement used isnt remotely accurate down below 300Hz. I know the Baffle diffraction will be 6dB below 276 Hz based on the formula but in room that drop should be less.

Do I just use the formula and add the circuit based on the 6dB loss @ 276Hz?

Or should I use my DCX to really find out what I need in room to get a flat response in room?

[penngray]

More crossover sims...

Image not available

I didn't like the drop at 2KHz so I moved the cap behind the Lpad on the tweeter schematic, I got this...

Image not available

Reversing polarity on the tweeter confirms a null

Image not available

I used the woofer crossover that Zilch and Pete used on their Eminence woofer, its simple and it seems to do a decent job for me too.

[penngray]

Not bad for 5 components so far. Im going to order some parts to just hear how it sounds and to get some other measurements done.

[Saurav]

That looks really good. Time to build and measure and see how well reality matches the simulations

The spike past 15K Hz on the BMS4550 is still an mystery

I saw that in my measurements too when I had a BMS tweeter. If you look at the raw tweeter measurements, you'll see a small spike. But once you flatten the CD WG's response, that tilts the whole response up in the HF (or down in the LF, depending on how you look at it), so that spike gets more pronounced / visible. Listen to it before you try to address it, I'm not sure how audible it'll really be.

[penngray]

Thx for the encouragement. I will order the parts.

btw, did you read the part about moving the 2.0 cap after the Lpad on the Tweeter? I have been told the cap should go before the Lpad but the measurement looks better with it afterwards.

[penngray]

OT: I just picked up a QSC DSP-30 for < $300, Im not done with active designs yet

[penngray]

I was playing around with the R2 value, 3 seems to be better then 4.7.

Image not available

[penngray]

Buying parts now....

[Saurav]

btw, did you read the part about moving the 2.0 cap after the Lpad on the Tweeter?

I did. If the responses looks better, it looks better. If you have raw driver impedance measurements, add those to the SW model, then right-click on the combined network and select 'calculate network impedance' or whatever it's called. Then you can see the impact of that change on the total impedance as well, that might shed more insight on which is 'better'.

FWIW, when I was experimenting with a WG tweeter, I had the cap for CD EQ after the L-pad:

https://www.htguide.com/forum/showthread.php?t=33242

See the first post there, that's roughly the crossover network I was experimenting with. If you read Wayne Parham's papers on XO networks for WG tweeters, he uses a similar topology - looking backwards from the tweeter, you get the CD compensation, then the L-pad, then the filter and any additional EQ. So I don't think what you're doing now is 'wrong'.

[penngray]

Thanks, another question about ordering caps..


What is the difference between the

Dayton PMPC-15 15uF 250V Precision Audio Capacitor

and

Dayton DMPC-15 15uF 250V Polypropylene Capacitor

besides double the price tag?


Also should I go through this sites link to Parts express to order this stuff?

[Saurav]

Also should I go through this sites link to Parts express to order this stuff?

Yes, that way you help keep HTGuide running. No idea on the cap differences, that's holy war territory

Edit: And I was going to suggest playing with the L-pad values to drop the tweeter level a little, but it looks like you already covered that.

[Paul W]

The difference between those caps is 1% vs 5% tolerance. Use the 1% where you want/need tighter tolerance or good repeatability...like matching tweeter XOs channel-channel.

[penngray]

Thank you! I will click on PE through this site. I know I should do that more often.

[penngray]

I just had a response on AVS explaining that having the cap could burn up if its after the Lpad so I will put it back in front and work on the X-over more.

[Dennis H]

Nonsense. Put it where it works best and use multiple resistors if you're worried about burning them up. Wire 4 of them like this to get the same R and 4x the power handling. They're only a buck a piece. Try it with one resistor for low-power prototyping and increase it to 4x once you decide on the final values. If you're prototyping with an external XO and clip leads, you can put a finger on the resistors to see if they're getting too hot -- they have to get hot enough to burn your finger before they are going to be damaged.

---|-R---R-|---
   |-R---R-|

That said, look at the impedance as well as the frequency response to see if it's dipping too low to be amp friendly.

[augerpro]

The shunt resistor is always a waste of power, that's how it works afterall, but...if it is after teh crossover filter the power it is shunting is only that of the passband of the driver, in this case the tweeter. If you put it before it will be shunting the full signal bandwidth, even bass, to ground. Not saying it is wrong, jsut something to consider. The actual resistor value used obviously has a big impact. How does the impedance look?

[Saurav]

I think that's why people mentioned the power thing - Doug's current XO has just an L-pad, and a single cap after the L-pad, so no filter ahead of the L-pad. Still, with drivers as efficient as what's he's using, I wonder if a 4 x 10W resistor group will be pushed hard.