or Connect
AVS › AVS Forum › Audio › DIY Speakers and Subs › Hybird SEOS - Selenium 10PW3 and DNA350/SEOS12
New Posts  All Forums:Forum Nav:

Hybird SEOS - Selenium 10PW3 and DNA350/SEOS12

post #1 of 57
Thread Starter 
Well, it's about time for another mini-build at the Gorilla household. This time I thought I would try my hand at designing my own crossover - with some help from a very generous few on here of course. smile.gif A big thanks goes out to Ryan (Tux) for the help along the way. I wanted to start simple with a basic 2 way setup and a woofer that had a realtively easy going impedance curve. I had 4 of the Selenium 10PW3's laying around so I figured I would order some waveguides and CD's to pair them. In additon to the Matte SEOS 12s and DNA350 CD's, Erich came through as usual with the 10" SEOS baffles that helped me save some time. I'm very much looking forward to doing extensive comparisons between these and my CHT SHO-10s. tongue.gif

I'm planning on building 4 of these total to use as my surrounds with the Sentinels up front.











I had a few spare hours this weekend so I constructed the cabinets and started to take measurements outside. It was NOT warm out today, ugh. I stocked up on a number of popular crossover components from PE so I'll be starting my modeling with PCD shortly.

DNA350 and SEOS12, on axis


DNA350 and SEO12 ~20 degrees off axis


Selenium 10, on axis


Selenium 10, ~20 degrees off axis


Drivers combined in parallel, on axis


Selenium 10PW3 impedance


DNA350 Impedance


More to come including crossover modeling. biggrin.gif
Edited by Gorilla83 - 3/3/13 at 4:57pm
post #2 of 57
Thread Starter 
First shot at design. Still need to smooth out some areas. Any advice for tweaking or additional filters appreciated from the experienced. smile.gif

third order on the woof:
L2 - 1.32mH
C2 - 27uF
L3 - 0.44mH

second order on the tweet:
c9 - 12uF
L9 - 1.5mH

lpad:
R15 (series) 3.5 ohms
R16 (parallel) 3.5 ohms



Thoughts?
post #3 of 57
Beside the HF section needing more padding, we need to also see phase and what the individual driver responses look like with filters.

A R/C contour circuit may be needed to help flatten the top end, here's an explanation: http://www.diyaudioandvideo.com/Calculator/Contour/Help.aspx

You can also just use a series resistor instead of an lpad.
post #4 of 57
Oh Awesome. Glad to see this taking form. Nice measurements. wink.gif

To tweak a little bit more output out of the top octave and align the tweeter level with the woofer a little better, try puting a 5uF and 5ohm in parallel, and then put that in series with the tweeter circuit. So you'll have 5uF/5ohm + 12uF +1.5mH to ground + 3.5ohm + 3.5ohms to ground + tweeter.

EDIT - oops, I see that's what Face was basically saying to do as well. And I see an email from you I'll check it out now. Hopefully you sent me the files smile.gif
post #5 of 57
Thread Starter 
Quote:
Originally Posted by Face2 View Post

Beside the HF section needing more padding, we need to also see phase and what the individual driver responses look like with filters.

A R/C contour circuit may be needed to help flatten the top end, here's an explanation: http://www.diyaudioandvideo.com/Calculator/Contour/Help.aspx

You can also just use a series resistor instead of an lpad.

Quote:
Originally Posted by tuxedocivic View Post

Oh Awesome. Glad to see this taking form. Nice measurements. wink.gif

To tweak a little bit more output out of the top octave and align the tweeter level with the woofer a little better, try puting a 5uF and 5ohm in parallel, and then put that in series with the tweeter circuit. So you'll have 5uF/5ohm + 12uF +1.5mH to ground + 3.5ohm + 3.5ohms to ground + tweeter.

EDIT - oops, I see that's what Face was basically saying to do as well. And I see an email from you I'll check it out now. Hopefully you sent me the files smile.gif

Thanks guys for the help. I started played with adding what I thought were Contour filters (post HPF) and it seems I have to add REALLY high value resistors to help bring up that top end response, which doesn't seem correct. Here's what I've got on the T side now. I'm sure I misplaced a few things so don't be shy in scolding me. biggrin.gif



Tux- Yes, I sent you the full set of measurements. smile.gif

Also - I can attach the files here for you guys to peek at, but I'm limited to txt files it seems. Any harm in changing the extension?
Edited by Gorilla83 - 3/4/13 at 11:39am
post #6 of 57
I must have missed this thread yesterday. Diving right into crossover design, good for you. smile.gif

A lot can be about crossover design by playing around playing around in PCD. I find it just as fun as building the boxes.

What do your individual driver responses look like?



Edit:

The filter you are adding is parallel across the tweeter terminals which won't have the desired effect; you should be adding it in the top box (series after crossover)


Here is the circuit you should be looking at to adjust the HF response.


Use the series resistor to change the middle of your tweeter's response while the cap lets high frequencies pass through. A larger cap will let more high frequencies through though it it limited to how much padding you have applied with the series resistor. The parallel resistor has less effect on the response while allowing you adjust the level. Also make sure you select you HP box on the impedance graph to the right of your response graph that will let you view the impedance with the crossover applied. Make sure your impedance is not dipping too low there.


No harm in changing the zma and frd files to txt.
Edited by mtg90 - 3/4/13 at 12:03pm
post #7 of 57
Thread Starter 
Quote:
Originally Posted by tuxedocivic View Post

Oh Awesome. Glad to see this taking form. Nice measurements. wink.gif

To tweak a little bit more output out of the top octave and align the tweeter level with the woofer a little better, try puting a 5uF and 5ohm in parallel, and then put that in series with the tweeter circuit. So you'll have 5uF/5ohm + 12uF +1.5mH to ground + 3.5ohm + 3.5ohms to ground + tweeter.

EDIT - oops, I see that's what Face was basically saying to do as well. And I see an email from you I'll check it out now. Hopefully you sent me the files smile.gif

Tux - Here is my interpretation - let me know where I've jacked this up. smile.gif

post #8 of 57
You have them entered in C13 and R17. They need to be in C8 and R10. If you must leave that 10ohm resistor in R10, put it in R11 instead. Don't forget to flip to the PCD circuit layout every now and then to see what components you're actually changing.

Consider yourself scolded biggrin.gif
post #9 of 57
Thread Starter 
Quote:
Originally Posted by mtg90 View Post

I must have missed this thread yesterday. Diving right into crossover design, good for you. smile.gif

A lot can be about crossover design by playing around playing around in PCD. I find it just as fun as building the boxes.

What do your individual driver responses look like?

You know, I guess I'm not the type of person to just solder stuff together and not wonder what the heck is going on in there. smile.gifbiggrin.gif Driver responses:



post #10 of 57
Darn you Andrew thats not what I wanted. biggrin.gifbiggrin.gifbiggrin.giftongue.gif

Like this:
post #11 of 57
Thread Starter 
Quote:
Originally Posted by mtg90 View Post

Darn you Andrew thats not what I wanted. biggrin.gifbiggrin.gifbiggrin.giftongue.gif

Like this:

I'll be damned if I didn't check the wrong box. NEWB ALERT!!!

post #12 of 57
I also find using a parallel LCR notch in front of the cross over works well with the DNA-350/360. L8, C8, and R10. Something like 0.5mH, 7uF, and 8ohms.
post #13 of 57
Thread Starter 
Quote:
Originally Posted by tuxedocivic View Post

I also find using a parallel LCR notch in front of the cross over works well with the DNA-350/360. L8, C8, and R10. Something like 0.5mH, 7uF, and 8ohms.

OK getting closer, but still a bit rough and looks like I still need more padding on the tweet. I've been trying different filters all around. Also have a few spikes to work out:



post #14 of 57
Try this on the Tweeter, adjust parallel resistor value to move level up or down, The series RLC you see in there flattens the 350's impedance which makes things a little easier. The RC has the bypass cap which boosts the top end.
post #15 of 57
Did you gate your files Andrew? I can't remember if we went over this or not, because you're using omni mic. If you saved your measurements it's not to late. I've messed around with what you've given me, I'm gonna email you a few pieces of advice that would just bore this thread.
post #16 of 57
Thread Starter 
Quote:
Originally Posted by mtg90 View Post

Try this on the Tweeter, adjust parallel resistor value to move level up or down, The series RLC you see in there flattens the 350's impedance which makes things a little easier. The RC has the bypass cap which boosts the top end.

Very helpful, thanks Matt!

As a reference, when should one apply a RLC - when the T has a wild impedance curve? Is there a calculator that could be used as a starting point for the components?
Quote:
Originally Posted by tuxedocivic View Post

Did you gate your files Andrew? I can't remember if we went over this or not, because you're using omni mic. If you saved your measurements it's not to late. I've messed around with what you've given me, I'm gonna email you a few pieces of advice that would just bore this thread.

Email replied. I'm for sure going to need help with setting the 'gating' and help/explanation of what that does to the overall measurements.

OK - looking much better now! Here's what I'm working with after a bit of tweaking:



post #17 of 57
As for the RLC, it depends. I rarely use them, especially with compression drivers because the parallel resistor really flattens out the impedance that the rest of the network "sees". If you click on comp in the impedance window, you can see what the parallel R does. Start with nothing. Put in a 20ohm R. Then a 10, then 5, then 3. You'll see the impedance curve flattens right out as if you have an LCR on there.

Stop modelling, you're learning but wasting some time. Check out my email reply you need to work out z offset. Hint, it's over 0.1m which means the phase you're work with right now is WAY off. Let's sort that out then you can get back to modelling.
post #18 of 57
No problem Andrew.

When you have impedance spikes right around the crossover using an LCR to flatten them often makes things much easier vs fighting with them. It is not always needed and sometimes you can sometimes use those spikes to your advantage. I'll usually try to get the response I am after without one but if it does not look like it is working out I'll throw the LCR in.

Look on the bottom below the parallel leg 3. See the button that says Auto-Calc LCR Comp. wink.gif
Jeff made this really easy for us. smile.gif
post #19 of 57
Thread Starter 
OK I've been reading more regarding the Z-offset calculation and now my head is spinning. mad.gif

I understand this can be calculated both in omnimic (using the sum feature) as well as in PCD. As you can see above I've got the combined measurement of both drivers as well as the individual measurements. If I wanted to use PCD to calculate this, where do I start? So far I've imported the summed driver response (raw) using the 'import overlay' feature. What should I be matching up now?

On a related note - I've begun to map out where the components will lay out on the board. I'm going to go through my parts collection tonight and see if what I have matches up with what I have for my values in PCD right now. I figure I have nothing to lose by rigging up something and doing some listening! I'll start simple and swap one component at a time and note the differences.
Edited by Gorilla83 - 3/5/13 at 1:14pm
post #20 of 57
You're half way there. You have the woofer and tweeter loaded and NO filters turned on. You've overlaid the combined measurement. Now adjust the z offset (cell C29 on my spreadsheet) up and down until you get a nice overlay. Start with -0.02. Then -0.04 and so on. Like I said, it's around -0.1. Which, by the way, is quite a bit more than normal, but that's ok. Once everything lines up, you're good to go.

Basically, omnimic doesn't know when the sound leaves the driver, so you're telling the software where it is.

PS. If you google the subject, you'll find a lot about minimum phase and having to extract minimum phase. Ignore all that. Don't bother with x and y either. We just ignore that because it's not worth the hassle. As long as your mic was at least a meter away, you're fine.
post #21 of 57
This is what I do, in the driver offsets enter your vertical offset, usually I change the woofer as I measured from the tweeter axis. Then go to the z offset I have found moving the woofer makes the phase look better as you don't get all the wrapping of the tweeter phase when you move that. Since the woofer is in front of the CD start by moving the woofer forward in steps of 0.005m or 5mm as you start to move it forward the response in PCD should start to look more like the raw summed response. I would start at 0.050 becuase the woofer is going to me pretty far in front of the CD. Once the response in PCD matches your raw summed response you have your offsets entered correctly and you are ready to start crossover work again.

Ah Tux you too fast biggrin.gif
post #22 of 57
Thread Starter 
Quote:
Originally Posted by tuxedocivic View Post

You're half way there. You have the woofer and tweeter loaded and NO filters turned on. You've overlaid the combined measurement. Now adjust the z offset (cell C29 on my spreadsheet) up and down until you get a nice overlay. Start with -0.02. Then -0.04 and so on. Like I said, it's around -0.1. Which, by the way, is quite a bit more than normal, but that's ok. Once everything lines up, you're good to go.

Basically, omnimic doesn't know when the sound leaves the driver, so you're telling the software where it is.

PS. If you google the subject, you'll find a lot about minimum phase and having to extract minimum phase. Ignore all that. Don't bother with x and y either. We just ignore that because it's not worth the hassle. As long as your mic was at least a meter away, you're fine.

Quote:
Originally Posted by mtg90 View Post

This is what I do, in the driver offsets enter your vertical offset, usually I change the woofer as I measured from the tweeter axis. Then go to the z offset I have found moving the woofer makes the phase look better as you don't get all the wrapping of the tweeter phase when you move that. Since the woofer is in front of the CD start by moving the woofer forward in steps of 0.005m or 5mm as you start to move it forward the response in PCD should start to look more like the raw summed response. I would start at 0.050 becuase the woofer is going to me pretty far in front of the CD. Once the response in PCD matches your raw summed response you have your offsets entered correctly and you are ready to start crossover work again.

Ah Tux you too fast biggrin.gif

You guys rock! It looks like -0.09 is the best 'fit'. Please come visit sometime so I can buy you both drinks/food. It's not an exact match but it seems to be the best anywhere in that range of -0.01 to -0.3. Is something like this considered acceptable?

post #23 of 57
Gorilla, excellent thread, man. Really great stuff. I'll be doing the same soon. smile.gif

My head spins and then I go back down the rabbit hole. smile.gif These guys nailed it (of course) Use the "import overlay" to import your summed response. Leave the tweeter offsets at zero and enter distance of the tweeter center to woofer center for woofer Y offset. Enter the mic distance into PCD "Listening / Measuring Distance" (convert to meters of course - great smartphone app btw, "ConvertPad") Then begin manually entering values into woofer Driver Z Offset until the response converges.

Right here: https://www.box.com/shared/ouxjjsx0m8bs00cil5iq

I'll hang back and watch some magic happen here. cool.gif
post #24 of 57
Thread Starter 
RB - that's the exact same link that I read through about an hour ago. AWESOME information in there and definitely worthwhile reading for anyone out there who is curious.

You guys were right - had a big phase issue that I corrected by changing the crossover orders. With new values - F yes!



Anything look funky with the values? I like that it's a simple design too.


post #25 of 57
Hmm... they should be closer then that, almost if not identical.

Your not moving the wrong driver are you?

What does you driver phase look like when you select that?
post #26 of 57
Quote:
Originally Posted by Gorilla83 View Post


Is something like this considered acceptable?


You're close, and it probably won't change your current xo values, but no that's not acceptable. Try -0.107 on the tweeter or +.107 on the woofer. Don't bother with y or mic distance.
post #27 of 57
Thread Starter 
Quote:
Originally Posted by tuxedocivic View Post

You're close, and it probably won't change your current xo values, but no that's not acceptable. Try -0.107 on the tweeter or +.107 on the woofer. Don't bother with y or mic distance.

Doing so gets me close, but not exact:

:

I will check against my crossover to make sure phase is still in check.
post #28 of 57
?? Same files you sent me?

That does look closer though. What's your x and y? 0 I hope.
post #29 of 57
Thread Starter 
Quote:
Originally Posted by tuxedocivic View Post

?? Same files you sent me?

That does look closer though. What's your x and y? 0 I hope.

Same files, x and y are zero for all. I'm planning on rigging up some components and giving it a listen in the current configuration. Shoot me an email with what you came up with now and I will compare them. biggrin.gif

Just went through my inventory - might need to buy some additional smaller caps.
post #30 of 57
Thread Starter 
Update: Ordered some additional components from PE to build a few different crossovers. Parts are coming in Monday, but unfortunately have to travel for work all of next week and weekend. More to come in another week or so.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: DIY Speakers and Subs
AVS › AVS Forum › Audio › DIY Speakers and Subs › Hybird SEOS - Selenium 10PW3 and DNA350/SEOS12