[neolith]

Adding to the current conversation, PLLXO's have several positive features but they also have some drawbacks. The biggest hurdle, IMO, is that designing them beyond 1st order presents several problems with insertion losses and Q. The highest Q that can be obtained theoretically is 0.5 and designing for a specific Q is  difficult. The other problem is insertion loss which is difficult to control because it really depends on the preamp output and amplifier input impedances. Trying to get the right Q, crossover point and workable insertion losses is tedious. I have been trying to work out the math for a 2nd order LP filter and I am glad this is a just a hobby. FWIW, I did come up with a reasonable facsimile of the native speaker level XO for the MG 1.6 but it was more trial and error than calculation.  Here's is a graph of the results http://home.comcast.net/~neolith/Stereo/PLLXO16.jpg. It's a little messy but what it shows is the native xo (dotted lines) and a proposed PLLXO (solid lines). I also included a close approximation of the LP with a 1st order configuration (blue line) that is adjusted to the insertion loss so it can be compared to the 2nd order.  If anyone is interested in the details, just email me.
My point being, the PLLXO is great for 1st order filters but are not very flexible for higher order filters or where a Q >0.5 (Bessel or Butterworth) is needed and an ALLXO becomes more attractive.
 

[wushuliu]

Okay, I transformed the spreadsheet to Excel 2007.  The link is http://home.comcast.net/~neolith/Stereo/PLLXO_1st_v2.xlsm. In general, I try to stay away from Microsoft, I run my computer on Ubuntu. I only use Windows for a few programs that just won't run in Linux.

file extension is not valid, won't open in excel 2007

[neolith]

Works for me. Try again, also you need to enable macros and I think for MSExcel you need to have the analysis plug-in. You should look into LibreOffice, it's free.

[dawnrazor]

Hey Neo,

Thanks for doing this!!!

Davey I think you are right on a lot of accounts about the adoption of multiamping using line level crossovers (active or passive) amongst audiophiles.  I had a conversation with Al myself, but I heard a different story.  He basically said that magnepan knew what they were doing and that the passive high level crossovers were doing more than what a LL crossover could do.  Something about resonance IIRC.  Perhaps he just didnt want to go into it with me but that is what he was saying.

Anyhow I dont think neos spreadsheet will help many adopt.  The ones who can decipher the graphs and build the thing probably dont need the help anyhow.  For the rest of this those schematics are daunting.  Some pictures of actual crossover parts diagrammed so you can see how they go together would help tremendously.

I for one was super jazzed when I saw the balanced option.  this is exactly what I am looking for.  But it took me like 10 minutes to figure out what I was looking at.  How the F is this balanced??  Then I saw the 123 and figured that must be the pins for the XLRs.  makes sense.  But how does this do a whole speaker??  Must be one channel so you have to double everything, etc.??  I can make good assumptions but thats all they would be.  I guess I tap off of pin 1 in parallel for both high and low pass outputs?????  It is probably obvious but I am not 100% sure.  Some picts would help or a video of the actual creation.

Also I dont think the sheet works.  I click on the mmg tab and it just says its protected.  I put in values and I dont see anything chang on the balanced page.  How does it know which speaker I am using???? 

Also if you look at the data you have to input, some of it isnt clear to me.  What is the z out??  for instance my "pre-amp" lists the following specs:

• Input Impedance
     Balanced mode: 24 k , Unbalanced mode: 12 k
  
  Output Impedance
     Balanced mode: 100, Unbalanced mode: 50
  
  Output Drive Capability
     600 impedance, 0.2 µF capacitance

What do i use for zout????

My amps are supposedly 66k for the input impedance.   Do i enter 66000???  Or 66?

Anyhow while I am sold on the concept, it is a non starter at least for me and I imagine other audiophiles.  There needs to be a dummy version.

Sorry Neo, I do appreciate the effort and skill this doc took to create.  I just wish it was usable

Maybe Davey you can see now why I  use that "fancy" software for a 1st/1st crossover

[neolith]

I agree with some of your comments. Designing the crossover requires a re-iterative process that can seem a bit confusing. I included an instruction sheet but as often happens the instructions are only clear if you know what you are doing, but murky to the neophyte. Basically, you calculate the "ideal" cap and resistor values, then insert the closest E-12 value for the cap and recalculate a new resistor value to match. Then insert that value into the sheet and come up with the final product.   As an.  alternative, you can place a shunt resistor in parallel with the amplifier to change the load impedance. By doing this you can get the impedance to match a particular E-12 cap value rather than selecting the closest E-12 value. Again, it is a bit of trial and error to adjust the shunt resistor to get the desired result. It's messy but usually after a few stabs it is pretty easy.
The schematics are fairly straight-forward and they are what the are.  I will try to get some pictures up to make them appear more graphic.
If you want to look at the workings of the spreadsheet and one of  worksheets is "protected", you can unprotect them by using the password "Lolita".  I protected them solely to prevent unintentional changes -- somewhat paternalistic on my part.
The Zout (Zs on the balanced schematic) is the ouput impedance of the pre-amp. I have not seen the "drive impedance" before (perhaps Davey can enlighten) but I would use 100/50 (balanced/unbalanced) ohms.  All impedances are in ohms (not Kohms) and capacitances are in nf.
The schematic for the balanced xo is for both speakers.  Amp1 is being used for the two woofers (L/R) and Amp2 for the tweeters. Remember each channel feeds a driver and this not shown; perhaps, I should put in the speakers to make things a bit clearer.  Also if you look closely you will notice that Pin 2 is on the "upper" amplifier on the low pass but next to the "lower" amplifier on the high pass.  This is intentional so the the polarity is correct.

[Davey]

Davey I think you are right on a lot of accounts about the adoption of multiamping using line level crossovers (active or passive) amongst audiophiles.  I had a conversation with Al myself, but I heard a different story.  He basically said that magnepan knew what they were doing and that the passive high level crossovers were doing more than what a LL crossover could do.  Something about resonance IIRC.  Perhaps he just didnt want to go into it with me but that is what he was saying.

Nah, he was just speculating based on his lack of understanding of the crossover designs.  There's nothing that Magnepan is "doing" that is more than what a LL crossover could do.  Magnepan uses passive components to create a filter/EQ shape.  You can do the same thing at line level, but different components and a different configuration is required.  It's not magic. 

To successfully design a PLL you need to know a few things (minimum) about the upstream and downstream equipment:

The output resistance of the source (preamp usually) and whether it varies with volume control position.
The input resistance of both/all power amps.
Relative voltage gain difference between power amps (if they're different.)

And the final (most important) thing is the objective.  (The "target" electrical response you're going to realize.)

All of that is a mechanical process that can be completely designed/simulated on the computer before you ever mount RCA jacks or warm up the soldering iron.  There is no subjective evalution at all in the process......at least initially.

One of the first PLL's I designed many years ago was built by a fellow over in Europe.  He got back to me and reported he didn't like the way I "voiced" the crossover.    Come to find out, he had decided to use a different preamp than the one he gave me the specifications for.  It was an Audible Illusions pre with very high output resistance which skewed the result of the circuit.  After he switched back to a decent preamp everything worked great. 

Cheers,

Dave.

[neolith]

DR - My apologies and corrections to my last post.
My apology - in fact the schematics on the spreadsheet are for a single speaker and the crossover would need to be duplicated. 
I also re-did the schematics so it is a little clearer what was going on and in the process I found a minor error in the balanced xo which would have shorted out the output of the preamp - oops! 
To make it up, I plugged in the values for your preamp and amplifiers so if you go to the MMG column you will have your values already calculated. Also I fixed the "protected" problem so you should not have any trouble now with the spreadsheet.

[dawnrazor]

DR - My apologies and corrections to my last post.
My apology - in fact the schematics on the spreadsheet are for a single speaker and the crossover would need to be duplicated. 
I also re-did the schematics so it is a little clearer what was going on and in the process I found a minor error in the balanced xo which would have shorted out the output of the preamp - oops! 
To make it up, I plugged in the values for your preamp and amplifiers so if you go to the MMG column you will have your values already calculated. Also I fixed the "protected" problem so you should not have any trouble now with the spreadsheet.

No need for apologies.  You are doing us all a service. 

Your earlier reply threw me for a loop.  I couldnt figure out how it would work for 2 speakers...thanks for clearing it up, and thanks for putting in the values for my amps!

Though I think I need to change the 50 on the zout to 100 for the balanced figure right?

The schematic I think is less confusing the way you have it now.  I still need some picts.  Like if you follow the ground, do I really tap into an existing wire or is there some accepted way that one is supposed to wire that?  When I look at that I see myself removing a bit of insulation and soldering a wire there to make things like the diagram.  In actuality is that what people do?  Or is there some block that it gets wired too?

Oh and why the difference in insertion loss between the high pass and low pass???  Wont that mess things up?? 

And how do you INCREASE the insertion loss.  I know I know but with a digital vc it might be a good thing.

Again, thanks for doing this and taking the time to explain it to dawnrazor.

Why not post this in the PA?

[neolith]

You don't need to follow the schematic literally, you just have to have all the connections and components correct. For example, with the grounding, you could just tie all the #1 pins of the XLR's (input and output and both channels) to a common point (star ground) or solder them to the chassis. Likewise the shunt resistors could go directly to the #1 pins (output XLR).
With regard to the output impedance of the pre, the answer is "it depends." Some manufacturers will list the impedance across pin #2 and pin #3 and others list it as the impedance across #1 and #2 (and 1 and 3).
I assumed that it was 100 ohms across #2 and #3 so each leg is 50 ohms. Remember the spreadsheet designs a SE xo and then converts to a balanced xo.
The difference in the insertion loss is due to the use of EIA values for the resistors rather than the exact value. You could use three resistors in series to get the value closer. If you want to change the insertion loss just go back to the spreadsheet plug in a different value and get the new calculations.
I believe I did post the spreadsheet on PA several months ago.

[wushuliu]

ok, using the spreadsheet is over my head so I will just ask: is it possible to do a 100hz 6db highpass pllxo for an amp with 38k input impedance? preamp output impedance about 500 ohms.

[neolith]

If you just want a hp filter, just insert a  42 nf cap in series with the amp input. The formula is C=1/(2*pi()*F*Z).
You will not be able to buy a 42 nf cap but if you parallel a 15 and 27, you will have what you need.

[wushuliu]

If you just want a hp filter, just insert a  42 nf cap in series with the amp input. The formula is C=1/(2*pi()*F*Z).
You will not be able to buy a 42 nf cap but if you parallel a 15 and 27, you will have what you need.

Got it. Thanks!