Butterworth vs L-R question

[Earl K]

Brad & O.G.

Thanks for the support , most appreciated .

- it's always good to have others paraphrase .

[Earl K]

JSP,

- Should you decide to build Jeffs' 605a network / to compensate for the fact that it is designed for the small horn found in the Altec duplex ( instead of the 811 ) reduce the value of C1 from 5 uF to 3 uF ( while leaving the coil value as is ) .

- Doing so, starts this cap filtering at a higher frequency ( @ 3315 hz instead of @ 1990 hz ).
- This newer ( higher filter ) point, in combination with the original coils filtering action, moves the combined F3 point ( of the 2 poles ) to @ 1838. hz / up from @ 1424 hz .

- If this opens up too much of a hole in the response between the woofer & horn / start adding back in some capacitance at C1 ( in .5 uF increments onto the 3 uF value ) .



FWIW to this discussion ; changing the cap value as directed , alters the HiPass Filters' classification for an electrical transform , from being almost pure ButterWorth to somewhere in between Bessel & LR ( a bit closer to Bessel )

<font color="#FFFFFF" size="1">[ October 26, 2007 05:53 PM: Message edited by: Earl K ]</font>

[Earl K]

edit #1

The attenuation curves for a 806/811 combo on a stock N800F ( Valencia ) crossover .
( Curves courtesy of Zilch at LHF )



<font color="#FFFFFF" size="1">[ October 26, 2007 06:08 PM: Message edited by: Earl K ]</font>

[Earl K]

edit #2

- From LHF ; the schematic for the N800F



- Here's the AC impedance plot for the Hipass portion of this N800F ( run by Zilch on Woofer Tester II ) .



The N800F was the recommended network for the "16 ohm" components that you have .

- You can see this here near the bottom ( of the following pic from LHF )





<font color="#FFFFFF" size="1">[ November 01, 2007 10:04 AM: Message edited by: Earl K ]</font>

[bowtie427ss]

Thank you Earl K, JSP's situation sparked my interest as i'd like to build new networks for my 9844's. I'm thinking i'd like to move the crossover point up into the neighborhood of 1200, but it's not exactly a straightforward proposition as the HF presents a 16 ohm load and the LF an 8 ohm load. I'm hoping at the same time to employ some "smoothing" similar to what was incorporated in the 19 network.

Somewhat OT, can anyone tell me if it'd benefit my efforts to swap out the 806 bodies for 802's?

[JSP]

Earl,

OK, I'm going to build the 605 plan posted. Ordered the components from PE and will begin the process of putting it all together soon. My plan is build new walnut cabinets first and then install one set of drivers. I think I will assemble one speaker first and then I'll be able to compare the sound of the new unit with the orginial flamenco.

I did change the value of C1 to 4 - as a drop to 3 seemed a tad much, but this can be adjusted if needed.

I'll report the results when I get the new speakers running.
JSP

[Earl K]

I did change the value of C1 to 4 - as a drop to 3 seemed a tad much, but this can be adjusted if needed.

<font size="2" face="Verdana, Helvetica, sans-serif">- 4 uF might just be perfect for you. It's certainly a good value to start with. I used it this morning & it worked well in this circuit .

- FWIW, this morning I wired up this circuit using what was on hand in the parts bin .
- I substituted a JBL 2410 ( le175 ) for the 806 since it was the closest driver that I had on hand ( that was 16 ohm or thereabouts ). All my 802s & 902s are 8 ohms .
- Since I didn't have an 811 ( or 511 ) I used a cheap plastic exponential horn from a Korean company called "Sammi ".
- The 16 ohm variable Lpad was dialed down @ 8 db to approximate the act of balancing the 806 to the 416Z ( 803 ) woofer .

- Anyways, as expected , the 5 uF cap allowed too much of the lowest octave to remain in the horn .
- The 3 uF cap started filtering early ( as expected ) and gave a sloppy banana looking curve ( which is to be expected from a filter that somewhat resembles a 2 pole LR ).
- The 4 uF was right on the money for the horn used ( which is smaller than an 811 ).
- This 4 uF gave me an acoustic F3 of 1250 hz ( the 3 db down point ) / while the electrical F3 was higher , up at 1565 hz ( measured electrically ) . Acoustically, I used an RTA to measure the resulting F3 point of the network/driver/horn combination .
- I listened to all three types with pink noise running / the 4 uF sounded ( & looked ) the most balanced by quite a margin ( again, I stress , on the horn that I used which was not an 811 )
- FWIW, ( since you seem to be interested in these matters ) / the resulting "electrical filter transform" is evenly split between being classified as Bessel or Butterworth .

- This acoustic F3 of 1250 hz ( in my case / yours may be different ) doesn't give the desired 15% spread that Jeff M. built into his design / this may or may not actually be a problem when you wire up your crossover.
- If a problem arises , it's most likely going to manifest itself as an acoustic "bump" ( of a couple db around the crossover region ) / when the woofer is blended in . ( Remember to flip the polarity around on either the woofer or horn crcuit , when you wire them up / you can see this flip in the wiring directive in Jeffs' schematic )
- To deal with a "bump" / some crossover values could be adjusted ( this is easiest to do in the horn circuit ) . Alternatively, a series LCR notch ( strapped across the transducer ) can be designed to attenuate only the offending area ( this LCR can be in either the Hipass or Lowpass circuit ) .

I'll report the results when I get the new speakers running.

<font size="2" face="Verdana, Helvetica, sans-serif">- I'll be interested to hear about your results.
- I'd also suggest that you "BookMark" this thread and then add onto it at the appropriate time ( it'll pop back up to the top of the queue .)



<font color="#FFFFFF" size="1">[ October 31, 2007 05:04 PM: Message edited by: Earl K ]</font>

[Earl K]

FWIW, here's a link into a schematic for a Tannoy duplex ( the 15" Monitor Red ).

It's a 15 ohm network. The web site ( where this .gif resides ) states that this is a 1000 hz crossover .



- Note, the recommended ( un-Official ) modification to the original circuit. This is the series LCR notch filter( to be strapped across the HF driver ) I spoke of earlier .

- As noted in the diagram , this LCR is centered around 3K ( and a passing analysis shows that it's of fairly low Q ) .



- The schematic above shows that Tannoy officially adopted the use of the LCR notch-filter when they released their "Monitor Gold" generation of duplexes . This LCR notch ( the 3.3 uF cap & the 0.82 mH coil , with R coming from the Re of the coil ) is centered at 3060 hz , & has a higher Q than the "un-Official" LCR modification .

- BTW; remember, you may not require a notch-filter ( I'm just loading into this thread info, that I consider might be relevant at a later date )

What is the difference (to the listener) between speakers built with a butterworth vs Linkwitz-Riley xover? (second order?)

<font size="2" face="Verdana, Helvetica, sans-serif">- Referring back to this portion in your first post ;

- An analysis of the values chosen to form the lowpass filter ( in Jeff Ms' basic 605 network ) , reveals that the LC values ( working into a "flat-line" 16 ohm load ) will create a 2 pole L-R transform ( or extremely close to the L-R transform ) .
- One may need to add a Zobel circuit to actually achieve this transform / if the impedance curve of the 416z deviates from being a straight 16 ohm load ( till at least somewhere past 2500 hz ) .



<font color="#FFFFFF" size="1">[ October 31, 2007 04:25 PM: Message edited by: Earl K ]</font>