Need help on Series XO

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all,

I need help from people who know.

I have been thinking of making a series cross-over between a ScanSpeak D2905/970k and a Peerless HDS-134. My plan was to make the XO at 2000 - 2500 Hz, but I need help on the math.

I suppose the drivers are connected in series, with an inductor across the tweeter and a capacitor across the HDS woofer.

Finally, I would make a 1st or 2nd order XO at around 250 Hz between the above system, and dual Peerless SLS213 bass woofers.

Can anyone assist with the math, especially on the series part?

Hoping for help,
Hi Jennice

See the drawing below for some good starting values for the series xo.

Download Andy Graddon's great series xo spreadsheet.

With your 2 bass woofers you have to make sure that the SPL is not greater than the tweeter mid combo, so I'd suggest that the you need to bi-amp with level controls.... 1 for the tweeter/mid and 1 for the woofers. You might have to crossover to these a bit lower depending on the 850488 F3.

For first order just use L=R/6.28f..... use the actual impedance for R (for 2 drivers in parallel will be halved) at the xo frequency, f.



  • d2905 850488.jpg
    d2905 850488.jpg
    20.9 KB · Views: 279
Hi Rabbitz,

Thanks a lot for your help :) :up:

There must be an error with the sheet, though (or your manual calculations). Surely R2 shouldn't be a short circuit !?

Also, R5 - C6: am I supposed to forget about R5, or what is the idea of this part (looks like impedance correction)?

EDIT: I got the sheet. Interesting tool, but what's the "Your aimed X-O zeta" all about?

Jennice said:

There must be an error with the sheet, though (or your manual calculations). Surely R2 shouldn't be a short circuit !?

Also, R5 - C6: am I supposed to forget about R5, or what is the idea of this part (looks like impedance correction)?


Suppose I better jump in, if no-one minds. :)

I suspect that Rabbitz has picked up the data from a data file, the zeros really should read, "n/a" or not applicable.

The use of zobel type impedance compensation, while in theory seems to be a 'must', can often be ignored. Depends on the woofer. It can be used to tame an overly aggressive upper mid response. You just have to remember that it does put a tad extra strain on the tweeter.
So if your woofer has a broard rise in the top of its response range, you may need one. If its reponse drops slightly through its upper range of use, you may not need one.

And of course , if you woofer has mountains etc.. you shouldn't be using a simple series !!

Think of a series x-o as a combining network rather than a controlling network (ie parallel) and you will understand their attraction, because that is what they sound like.
Hi Andy,

You're more than welcome to add your comments. I asked my questions for a reason. I want to learn.

The drivers (Peerless HDS-134 PPB and ScanSpeak D2905/970000) seem to have fairly nice responses --- at least it looks good to a beginner like me.

This project has taken a long time, and meanwhile I made myself a set of PC speakers with Vifa fullrange drivers. Obviously, there are limits as to how much air can be moved my these small drivers, and the top rolls off a little early, but I noticed something intersting:
Coherence! It all sounded clear and "right". This made me doubt my original 2nd order, and the suggested (from a friend) 4th order L-R filter. I am no longer convinced that a complex filter is always the correct answer. This is why I thought of the series- XO. It seems so beautifully simple. Also, the concept of constant voltage, going either to the tweeter or to the woofer, intuitively rang a bell of coherence in my head.

this is why I an considering this type of filter.

I am thinking of making a series XO for tweeter/midwoofer, and a "traditional" filter between the mid/high section and the bass woofer.

Sorry about the confusion Jennice but that little schematic is normally only used in my spreadsheet and I know that 0=not used.... I'll change it to a text note so if I do post one in the future it will be easier.

I don't use the R2 has that forms the part of a normal L-Pad and I only use the single resistor. Andy's spreadsheet show both options and I prefer the single resistor.

R5 and R6 are the Zobel values.

As far as Zeta goes just think of Z<1 (peak on the crossover area)is lively and forward and Z>1 (dip in the crossover area) is laid back. It's all to do with the overlap of the tweeter and woofer. Z=1 is basically your standard butterworth 1st order values.

Series are really nice.... just something about how they present music and so simple in construction (3 components in it's simplest variety) and Andy's spreadsheet is a blessing and a great starting point.

Then you listen and test and alter values. I have a test rig outside the speaker and I normally get the resistor value sorted out first so the top end is right, then play around with the cap value... as you increase the cap value the tweeter crosses over lower and vica versa. I don't tend to play around with inductor values too much as I've been through that learning process so have a good idea where I'm heading. My first series had about 20 variations to get the one I wanted and that changed again after living with them for a while. Now I normally jag it in about 2 variations. Each builder does it differently but that's how I approach it. The main thing is to use the impedance value at the crossover point for each driver.

The only thing is that series doesn't give you is tweeter protection of a cap, but I've never damaged one yet even with some stupid levels and a mistake once of crossing over at just above the tweeter's Fs.

I did try a few parallel crossovers on the last couple of speakers just to make sure and a double check..... even did a parallel 1st order with the same values.... tried them but I just went back to series again. I should have known but it was a good excercise.

Thanks for the exlanations. :)

I thought that the L-pad used tweo components for adjusting sensitivity AND linearizing the "apparent" impedance (the driver impedance seen by the XO-components. Doesn't the use of only the series resistor leave the impedance curve of the driver as non-linear as it originally is?

What's your experiences? Are 5W resistors OK for L-pads, or should I go higher, especially considering the resistor in parallel with the tweeter?

Most of the info that I have found on series state that they are better with the single series resistor and not an L-Pad and that's what I have found. Even in the parallel world, a lot use the single resistor including Vance Dickason. Yes L-pads do provide a constant load with leaving the tweeter impedance constant no matter what the attenuation is on a parellel network but I'm not sure if that is the case on series network.

The spreadsheet takes the resistor into consideration during the calculation and with your drivers with the series resistor (R1) only gives almost equal impedance on the tweeter leg and woofer leg of the network which is a good thing.

5W resistors are fine and most of mine are non inductive wirewound. Use R1 only (the one in series) as it will sound better but try the L-Pad with R1 and R2 if you like, but that will alter the cap value. As you reduce R1, C4 becomes larger and visa versa.... it's all linked together.

I'm eager to try it out when the boxes get back.
A friend of mine is doing the grinding / woodwork as the speakers will probably end up with him... but I have parts to make another pair. ;)

I use the meantime to think (or re-think) the X-over and get ideas....

... and the important help from fellow DIY'ers! :D
Hi Rabbitz,

I got a call last night: The boxes are sanded to a nice surface, and are ready for paint (we'll let a proff. car paint shop do it).

After that, it's out shopping for components.

Now it's all about getting rid of the flu *cough cough* :dead: to get started...

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.