[dpd123]

Hi all!

I came to this forum trying to learn more about passive line level XOs. Lots here, good site!

Unfortunately all links to Neolith's PLLXO spreadsheet are broken comcast links to an account that appears to be history.
Can anyone help me out with a URL or some other means to find this gem?

Thank you kindly.
David

[richidoo]

Never saw that spreadsheet, but it sounds like a good one. I made my own excel spreadsheet last year, since lost it, but I learned today that I must have forgotten the 2 in the denominator of the C1 formula because a filter I built that should have been 200Hz, turned out to be 400Hz in real life. DUH.  The math is easy enough that you don't really need a spreadsheet.

I use the article hosted by AC member 'planet10' on his t-linespeakers.org website for the math formulas.
http://www.t-linespeakers.org/tech/filters/passiveHLxo.html

There was another good article about PLLXO I can't find it at the moment, but I'll keep looking in my shortcuts. Says bascially the same thing in slightly different way.

The math is simple, especially for 1st order filters.

If you do 2nd order passive HP, then be sure to use a real resistor on R2, not rely on the amp alone for R2, contrary to suggestions in the notes. Some very popular modern ultra high performance opamps with ultra-high input impedance react badly to the filter without R2 and can output DC. Maybe some more normal amps would behave better.

To eliminate the math for 2nd order PPL filter, you can combine the passive design article above with the sallen key filter calculator application hosted by Rod Elliot at the bottom of this webpage: http://sound.westhost.com/project09.htm
Download and install it. It is intended for active filters, but they use the same math for R1 and C1, and you can derive R2 and C2 easily by multiplying R1 by 10 for R2, and by dividing C1 by 10 for C2.

Set your target XO freq, check off 12dB box, then enter your caps, usually this will be one of the following standard sizes:
.1uF, .01uF, .022uF, .22uF, .47uF, .047uF.  You have to enter the cap value in nF, so move the decimal to the right 3 places.
.1uF = 100nF   etc

Press Calculate. This gives you a precise resistor value for R1 that may or may not exist in a real part. Dale CMF55 resistors have many very close values and is a decent sounding resistor. Pick a real world resistor for R1 that is close to the calculated resistor. Mouser.com sells them and shows a list online.  If you are doing 2nd order filters, you will also need R2 resistor, such that R2 = R1 * 10. So when looking for R1, make sure R2 is also available. 1/8W, 1/4W, 1/2W sizes all OK, use only metal film or other high quality resistor. Juggle the values until you find parts that get you close to your target XO freq while obeying these other rules:

Choose caps which keep the resistor values on the low side, to reduce noise. But keep the R1 resistor in a range that your signal source buffer amp can handle, not too low. Rod Elliot offers more advice on choosing parts in the middle of that webpage above.

Example:
I was working on pllxo today. I wanted 200Hz. My choices are

1.)    R1 = 7.87k    R2 = 78.7k     C1=.1uF     C2=.01uF    = 202.2Hz
2.)    R1 = 3.65k    R2 = 36.5k     C1 = .22uF    C2 = .022uF     = 198.2Hz

As long as HP and LP are the same frequency, which they will be if using the same values, then 198 or 202 or 200 doesn't matter, HP and LP will be 198 with minimal phase error.

If I use .47uF for C1, then my R1 will be too low @ 1.69k being driven by THAT1200 which supposedly prefers load >3k.
If I use .1uF for C1, then my R2 is too noisy  @ 79k.
So .22uF is my best C1, and everything else falls in place according to the simple formulas.

Good luck
Rich

[SteveFord]

I sent him a PM so perhaps he'll chime in.

[dpd123]

Many thanks to both of you!

SteveFord I appreciate the contact. Neolith's calculator (from what I have read in discussions here) has all sorts of extra goodies, suggestions, analysis based on input impedance, load etc. I would really like to check it out so hopefully he can repost the spreadsheet somewhere or let us know a new URL if it has moved.

Richidoo WOW, thanks a ton for all these insights and tips. The kind of info I'm sure I'll look back to many times as gradually I understand more, and your info makes more sense Perhaps I can throw at you what I'm trying to accomplish? I'm building a portable boom box with Lithium batteries and a 2-way stereo+mono sub design. Sealed stereo chambers, MTM, each with:
2 * Tymphany TC9FD18-08 3.5" full range
1 * Tymphany BC25SC55-04 1" Square Frame Tweeter
Mono/centre sub: 1 * Dayton Audio SD270A-88 10" DVC Subwoofer
Amp: Sure Electronics AA-AB33184 4x100W TDA7498
Preamp/tone control: Yuan-Jing NE5532 Stereo Preamplifier Volume Control Board with Treble Midrange and Bass Tone Controls

My plan is to use 2*100w channels for stereo, with a 1st order PLLXO highpass at ~200hz for the full ranges, and a 1st order capacitor HPF at speaker level for the tweeters somewhere between 5,000hz - 10,000 hz for the tweeters since the TC9s have a very gradual roll off, way up high. (Cost is a big factor building these since I have an order for 5, and I'm doing them for cost, for some not-so-rich folks! So the dollars saved avoiding inductor designs for Subwoofer LPF is appreciated!)

For the design of this boom box I have been attracted to the PLLXO concept and doing channel attenuation and sensitivity balancing across the drivers, at line level, when it will be cheaper and easier than inductor LPF networks for speaker levels. I also hear efficiency, distortion, phase may have advantages with PLLXO. The subwoofer in this design is a bit more sensitive than the full ranges and tweeters so insertion loss and attenuation in the PLLXO circuit is interesting to me, I want to learn how I can deliberately factor this in to balance channels pre-amplifier. I have a UMIK-1 measurement mic and REW to test different configs, so I can take my time about this and get it right. I can also use hybrid PLLXO/XO setups, I'm not the slightest bit principled if there's a good technical justification.

My main questions right now are:
1. The Sure amplifier has an input impedance of 10kohms. The typical PLLXO calculator suggests 5k-10k R1. Is there any consideration here, for which end of the 5-10k scale to start from? Or even lower, since I have such a low input impedance for the Sure?

2. The dual-NE5532-based pre doesn't have any documentation about output impedance, really no info at all that I've been able to find. However the NE5532 datasheet states:
Zo Output impedance: AVD=30dB,RL=600Ω,f=10kHz...............0.3Ω
I'm not sure how to read this, since 0.3Ω actual output impedance seems really low and I don't know the impact of the rest of the circuit. Can anyone help me decipher this, and any idea how this might affect my PLLXO decisions?

3. I'm mixing the stereo channels into a mono bus for the sub, filtering that PLLXO/LPF (2nd order preferably), then splitting to 2 amp channels for the Sure amp blocks, which feed the 2 coils of the Dayton DVC driver. My summing circuit is simply 2 * 10kΩ resistors in series with input (+), tied together, into PLLXO/LPF. Any idea how I should START with a design like that, with line source splitting into 4 lines like this:

in 1 --------> Left main -----> PLLXO HPF, 200hz(?) 12dB/oct(?) -----> amp channel 1
          |
          -----> 10k R -----
                                      |-----> Center/Sub: PLLXO LPF, 200hz 12dB/oct(?) ----=====> amp channels 2,3
          -----> 10k R -----
          |
in 2---------> Right main ----> PLLXO HPF, 200hz(?) 12dB/oct(?)

I wired the sub already to a Harrison Labs FMOD PLLXO at 12dB/oct 200hz but I since learned my 10k input impedence on the amp will throw off the XO point, that it's a non-ideal design since it is built for input impedance variation over quality, and I'd be better off doing my own... which brings me here. Also they seem cool but overpriced compared to a simple circuit to DIY. On testing I heard a lot more high freq content than I'd expect with a traditional LPF.

My head is spinning a bit over:
- the effect of splitting the L and R to independent channels for 4 amp lines on PLLXO specs
- the effect of summing L and R for the sub pre-PLLXO for a LPF (should I build parallel PLLXO circuits, and sum them afterward maybe?)
- where to begin with that all-important initial resistor starting point for a 10k input impedence on the amp. If I could just get a hint there, I could at least start hacking & measuring with REW.

Thanks for reading!!!
David

[dpd123]

I'm also realizing that this is not a planar design so I might be in the wrong place. It seemed like this is where the PLLXO dudes all hang out so my apologies if it's inappropriate!

[dpd123]

Rich, at first your process you described above went over my head. But I downloaded Elliot's app and went through your steps and it's perfectly clear! Thanks so much for this straightforward and well explained guide.

My primary objective is actually to design a 200hz LPF and HPF at 12dB/oct so it's not only helpful theoretically but practically! Especially the part about solving for available cap types. It seems obvious now but was eluding me a bit until you explained it so well.

Now the only question I have for this specific PLLXO setup is whether the same solution you settled on, with 3.65k R1, do you think this would drive OK with the NE5532 based pre-amp I listed above, with 10k input impedance amplifier.

Also, if I were to mix 2 channels into 1, using 10k resistors in series on each stereo line's (+), how would I factor this in to the circuit? Or is there a better way to do A) PLLXO and B) stereo sum to mono, in one move? I bet there is...
Thanks again! David

[neolith]

Sorry for the delay. I haven't been visiting this site very often.
Here is a link to the spreadsheet (Google Docs).  It is written for LibreOffice (which can be dowloaded for free).
https://drive.google.com/file/d/0B6Rj4thFC-dnOWF6ZU9OeDhFQWs/view?usp=sharing

[SteveFord]

Thanks and don't be a stranger!