SB13PFC25-8, SB19ST-C000-4 2 way. Crossover help needed.

kenrhodes

Here is what I have so far: SB 13 and SB 19 in a 2 way crossed at about 4k.(see measured response)
I think the crossover point for these drivers is too high at 4k and I think 2.5k is a better match for these two drivers. 
If I change the crossover to a lower point it should also help push down that 6.2k peak in the woofers response. 
would it be a good idea to filter the cone break up or would this be a waste of parts?
Thanks for the help.

dcibel

While you've got your computer out taking measurements, plug it in to your speakers with some EQ software and tame that lump between 4-7kHz, you may be surprised how much difference it makes to simply add some EQ. The Omnimic software can even calculate the EQ settings for you. Just click new average to save the curve, then open the equalizer and go to town.

Some okay EQ software for windows:

https://sourceforge.net/projects/equalizerapo/

4kHz is really pushing the limits of the SB13, you could experiment with simply increasing the filter slope (order) to better deal with the breakup. You may be able to make some improvements to your existing crossover with only buying a couple components. Failing that, if you want to rework the entire crossover, I'd aim for something in the 2-3kHz range for the SB13, and most 5 to 5.5" drivers for that matter.

kenrhodes

  I have all the parts to rebuild the crossover at 2.5k  I'll shoot for that in the morning. Hopefully that will work better. 
  I also need to reduce the padding on the tweeter. 
I think these will sound great when completed, even now they aren't as offensive as the measurements imply.
  Thanks for the help.

jhollander

It would be interesting to see the individual driver responses. A small cap across the woofer inductor would push down the break up

rjj45

I used those SB drivers in an MTM a few years ago. IIRC, pretty easy to cross, didn't need to suppress any breakup nodes. Here is what worked for about a 2Khz Fc. Double the coil and halve the cap for a TM. 

howard

Ken, this might be helpful:  http://ampslab.com/blog/2018/12/21/sparrow-19st-ver2/

Tom_S

I'm surprised to see a peak that large in the sb19. So much for the ultra-flat response they show in the manufacturer data. Seeing that the the Sparrow also shows a similar response, could it be a reflection from the woofer frame? I'm just guessing yours are surface mounted because those woofers would be a royal pain to flush mount.

S7910

The tweeter in the Sparrow looks to be surface mounted. Would that be a contributing factor as well?

kenrhodes

SB19 centered on a 7 inch baffle 10 uf cap installed to protect the tweeter. 

kenrhodes

sb13 woofer. LOTS of room issues... Woofer is flush mounted, 7 inch baffle.

dcibel

Based on your measured response vs the system response in your first post, it really appears like the woofer has no filter at all. Response from 3kHz - 8kHz is pretty much identical to the raw woofer response. Anyway, it sounds like you know what needs to be done to make improvements, but it would be interesting to see what the woofer response alone looks like with the filter in place, just for S&Gs.

kenrhodes

woofer with crossover5.6uf .7mh (1.5uf parallel).
 

kenrhodes

sb19 9.3uf .43mh

kenrhodes

combined response has a huge 3.5k suck
out. What can be done about that?

dcibel

Ok, I'm confused. What is the response in your first post?

For your combined response, you have some obvious driver interaction issues. While the woofer response with crossover in place isn't that bad, the tweeter is operating far too low in frequency to work well with the woofer filter. The tweeter is rolling off at like 1200Hz, not 3kHz. You can start by reversing the polarity of one of the drivers, might improve things, but that crossover needs to be reworked.

kenrhodes

Woops! I had the tweeter polarity wrong. This looks much better:

dcibel

Much better! If it were me, I'd drop the tweeter level about 3dB, but you can adjust to your own taste. I'd still have a closer look at the tweeter crossover myself as well, the on-axis response is much better, but if the response above with filter in place is accurate, I still think you are driving it too low.

rjj45

@kenrhodes
couple observations. I'm still learning after 10 years and many designs.
1 (deleted)
2 - (deleted)
3 - that SB19 is crossed too low. It may work, but might sound distressed if you push the volume.  Probably want 2.7Khz to 3.0Khz for that tweeter.
4. Look at the plot of the woofer with crossover - after the deep null, the level comes up to about 15 dB below the tweeter.
    you will get much cleaner upper mids if you can bring down the woofer there.  Recently I had some good results by using 
   a very small inductor (.03-.08) coil in series between the cap and ground. (Jeff Bagby tip)
5  That dip at 3.5Khz?  That's a beginning of a reverse null! I look for that, then know that I'm on the right track.
6. It's hard to tell, but you might want a larger coil on the woofer. You can do that (IME) most easily by measuring the woofer from the usual position (22-30 inches on the tweeter axis) without any crossover components. Save that response, then load it back into OM. Now pop a .60 - 1.0 mH coil on the woofer, and without touching anything measure that, save it, and load it back into OM.  You can read the amount of BSC you have with a particular inductor by looking at the FRs between 600- 1Khz.

Hope this helps

kenrhodes

OK, after ALOT more measurements and lots of plaing around with caps and coils here is where I am at:

dcibel

Not bad, but like I said before, if it were me I'd drop that tweeter level a good 3dB, bring the shelf at 4kHz down to the same level as the rest of the speaker and adjust from there.

You can widen your gate by a good 1ms as well, 1st reflection shown at 3ms.

Wolf

That 1.5uF tank-cap is too large. That is the reason your ripple above the null in the filtered woofer response is so high. Most of the time, if you go above 1uF for this part, your ripple will be too high, and you'll have other issues to resolve. I typically use 0.1-0.33uF here.

I'm with dB here- if the response is higher in the tweeter it will sound bright. 3 options exist:

- small series inductor to tilt down response. 0.05mH or so.

- CR across tweeter to tilt down response (better). About 2.5uF and 4 ohms to start.

- further attenuation. Resistor out front of xover or in parallel with tweeter do not tilt response. Series resistor after will tilt down response a bit.

kenrhodes

Im not sure how to take the 4k shelf down.

Wolf

The CR filter is likely your best bet here. Set the C to the point of required droop, and then set the resistor for the amount of loss. Or if the top-end droops too far, a standard mild notch will work.

If that doesn't work, you could always try a 'cross-pad' (Xpad) like I did in the Cecropia Supreme project. Essentially, it's a 6dB xover within 2 different attenuations. It was the only way to get it accomplished with the faceted-baffle induced rise, or 'anti-waveguide'. It involves 5 components; a series CR in parallel with a series LCR, placed in series with the driver. The LCR is the lower freq section, and the CR is the higher freq section. The C in the LCR sets the lower bandwidth limit of the circuit.

kenrhodes

kenrhodes said:

Im not sure how to take the 4k shelf down.

Sorry I did not see Ben's post when I said this. I will swap the tank cap to something smaller. And try the CR circuit to address the tweeter. 

Thanks for the help. 

squamishdroc

Here is my build with those 2 drivers Ken.

http://diy.midwestaudio.club/discussion/497/another-sb-build#latest

kenrhodes

Squamishdroc that makes me feel like giving up on what I have and taking the easy rout. Great looking build!
  What was the baffle width and the final tweeter crossover? My box tuning is identical but my baffle is 7 inches wide.
  Thanks for sharing.

rjj45

squamishdroc said:

Here is my build with those 2 drivers Ken.

http://diy.midwestaudio.club/discussion/497/another-sb-build#latest

That's some nice work, both in the box and crossover.

GeoffMillar

Those SBPFC drivers look really nice; I'm using the SB16-8 in a very nice two way with a VIFA tweeter, very happy with the sound.

A 3 way with the SB19, SB13 and SB20 sounds really nice, even if I really don't need any more speakers!

Geoff

squamishdroc

kenrhodes said:

Squamishdroc that makes me feel like giving up on what I have and taking the easy rout. Great looking build!
  What was the baffle width and the final tweeter crossover? My box tuning is identical but my baffle is 7 inches wide.
  Thanks for sharing.



Thanks - the baffles were 8" wide - I can't remember what the x-over point was - I want to say around 2200. 

squamishdroc

The SB13PFC and the SB16PFC are on sale at Solen right now for $15 and $17

GeoffMillar

Just for fun - and I'm very new at this - I modelled a three way in Xsim with the SB19, SB13PFC and SB 20PFC, using the squarmishdroc crossover above, I hope it's OK to borrow someone  else's hard work.  If not, my apologies but the 2 way model looked so good, I didn't see the point in changing it.

 I don't plan to build this as I have two projects on the go at the moment, but I know that three way crossovers are very hard to get right and would be interested to know what people think.  The SB20 zma and frd files were taken from fptrace  using SB's graphs and not actual measurements.

The project may happen if I can convince a friend that I can build him a better three way speaker than the mid 1980s generic three way he has at the moment. The total driver cost in Oz would be $240, not bad value.

Any thoughts welcome! 

Thank you

Geoff