First project - FAST / WAW like

[Stal]

OK. Looking at the 3rd screen.

Take away the resistor from the SPH completely.

Change the resistor across the FF85 to 24 ohms.

Change both inductors to 4.8mH.

Change the woofer cap to 33uF also.

Do not worry about the impedance peak below 100Hz this is expected.

 

[Stanislav]

Elmans,

I would strongly recommend to try a 1st order Series Xover at the BSC of your intended box, which might happen to be around 300 - 400 Hz.

1st order series will take care of the peak of the mid tweeter rather nicely and also sums flat. On top of it it is self adjusting, you change one value somewhere and it just shifts the xover point up or down and also varies the "zeta" of the filter.
1st order series is very forgiving and in a FAST application uses only the minimum number of components. You can also add a zobel for the woofer later, but try it first without one.

Have look at this link, you might find it helpful:
Audio Dandy - How to enjoy High Resolution Audio

P.S. Below you'll see an example of a 1st order series xo.

 

[Stal]

Stal, I'm not sure I properly integrated the values you gave me, because the response looked weird when I tried.


Attached are my tries at :

1. 1st order on both drivers : I think response is good, but afraid it's not safe enough for the 85
2. 1st order on woofer, 2nd ordrer on full range
3. 2nd order on both : I like that phases seem to match all along, and that impedance has no more peaks

Dou you have any recommendation as to which solution would be "best", and whether it would need further tweaking ?

By the way the Fs of the 220HQ should be down near 30Hz but your graphs show it about 60Hz. Are you sure you've got the correct drive unit data and not some other drive unit that you've labelled as a SPH-220HQ.

 

[Stal]

Elmans,

I would strongly recommend to try a 1st order Series Xover at the BSC of your intended box, which might happen to be around 300 - 400 Hz.

1st order series will take care of the peak of the mid tweeter rather nicely and also sums flat. On top of it it is self adjusting, you change one value somewhere and it just shifts the xover point up or down and also varies the "zeta" of the filter.
1st order series is very forgiving and in a FAST application uses only the minimum number of components. You can also add a zobel for the woofer later, but try it first without one.

Have look at this link, you might find it helpful:
Audio Dandy - How to enjoy High Resolution Audio

P.S. Below you'll see an example of a 1st order series xo.

I agree a 1st order series is worth exploring, I currently have it on my set up.
I tried modelling it at home with these drive units using the same software that elmans is using but it didn't look quite right.
I think an issue with the FF85WK is its quite high Fs and an xover target of around 400Hz as you mention.
I think 2nd order will show what is possible by taking away the issue with the 115Hz Fs and a xover point of 400Hz. We could then re-explore 1st order either series or parallel.
Here's one I drew earlier. The parallel 27R is just to make the FF85WK to look like 6 ohms so that both drive units are 6 ohms at the xover freq. This helps choose our filter values easier using conventional calculators.

 

[elmans]

Thank you both.

I have tried both options as recommended in pictures attached. I thought the XO freq could be anywhere between 400-600Hz ?



About the 220 Stal, I have not found a way to check. I have redone the SPL and Z calculations using the "enclosure" tool, and making sur to check the SPH-220HQ. These 2 quantities are the only things that I feed to the main program and are used for XO simulations.

 

[Stanislav]

I rarely use software to simulate xovers.
Usually calculate values for L and C, wire them up and start measuring amplitude and impedance.
XSim is the exception.
I have to admit I'm a great fan of 1st order series for 2 and 3 ways and have not been disappointed after many projects.
But I trust my ears more than I trust sims.

2nd order is not a good candidate for a series xo, introduces a lot of anomalies in the response and to compensate you start adding components which defeats the idea of simplicity, makes everything more complex than a parallel xover, so one might just stay with parallel then.
Also bad transient response.

That 115Hz Fs is of no consequence in a 1st order series, it is dampened by L, which has to be of very low Rs, so one needs a cored inductor.

This article by Rod Elliott is a good read:
Series vs. Parallel Crossover Networks

 

[Stal]

@elmans:
I've actually traced in the drive unit frequency response for both these drivers. I'll have a closer look at home when I get a chance. I think there's something up with the 220HQ data you have. The impedance peak should definitely be lower. The data sheets show an Fs of 30Hz.

 

[Stal]

Thank you both.

I have tried both options as recommended in pictures attached. I thought the XO freq could be anywhere between 400-600Hz ?



About the 220 Stal, I have not found a way to check. I have redone the SPL and Z calculations using the "enclosure" tool, and making sur to check the SPH-220HQ. These 2 quantities are the only things that I feed to the main program and are used for XO simulations.

Just as a quick check, take away the parallel R and see what happens.

 

[elmans]

@Stanislav : as much as I agree with you, I don't have any components at hand, so I have to resort to sim to know what to order, and given the cost of components I'd rather adjust as much as I can beforehand. I think in the pictures I posted, first is parallel 2nd order, and second is series 1st order ?


@Stal : if I sim the 220 on infinite baffle, I do get the impedance resonance at 30Hz (like spec sheet). However adding 0.707 sealed moves it to about 58Hz. Otherwise, driver parameters seem consistent between spec sheet and VituixCAD file.

 

[Stal]

I rarely use software to simulate xovers.
Usually calculate values for L and C, wire them up and start measuring amplitude and impedance.
XSim is the exception.
I have to admit I'm a great fan of 1st order series for 2 and 3 ways and have not been disappointed after many projects.
But I trust my ears more than I trust sims.

2nd order is not a good candidate for a series xo, introduces a lot of anomalies in the response and to compensate you start adding components which defeats the idea of simplicity, makes everything more complex than a parallel xover, so one might just stay with parallel then.
Also bad transient response.

That 115Hz Fs is of no consequence in a 1st order series, it is dampened by L, which has to be of very low Rs, so one needs a cored inductor.

This article by Rod Elliott is a good read:
Series vs. Parallel Crossover Networks

I mostly agree about 1st order series xovers and can clearly see the merits.
This is the first time I have been using a simulator in order to help elmans along.
elmans has not bought all his drive units yet never mind built the boxes, so we are at quite the preliminary/feasibility stage. I think it helps to see what the merits of various solutions are.
2nd order parallel also has an inductor to shunt the FR at low frequencies and a capacitor to shunt the woofer at higher frequencies.
The issue with the FF85WK is that it is a small drive unit with a low xmax and a 1st order with a low xover point may still be a bit much for it at higher volumes.
Also I'm not convinced that the shunting of the 115Hz Fs is quite enough. At 115Hz the XL of a 2.4mH inductor for example is 1.7 ohms add the series R and you have nearly 2 ohms. You won't get the exact 6dB roll off in this region although admittedly it may not be an issue from freq response point of view but maybe from a xmax point of view.

 

[elmans]

Just as a quick check, take away the parallel R and see what happens.

I did in the picture attached


Edit : attached the modified series 1st order

 

[Stal]

@Stanislav : as much as I agree with you, I don't have any components at hand, so I have to resort to sim to know what to order, and given the cost of components I'd rather adjust as much as I can beforehand. I think in the pictures I posted, first is parallel 2nd order, and second is series 1st order ?


@Stal : if I sim the 220 on infinite baffle, I do get the impedance resonance at 30Hz (like spec sheet). However adding 0.707 sealed moves it to about 58Hz. Otherwise, driver parameters seem consistent between spec sheet and VituixCAD file.

OK that's good to know.

 

[Stal]

I did in the picture attached

Sorry I meant on the series 1st order version.

 

[Stal]

Try also a different box response say about 0.5 or 0.6 for both boxes.

 

[elmans]

Here is the 2nd order on both drivers, with 0.577 boxes.


I don't wish to go that way because it makes the enclosures too big, I have to keep them about (ideally) 30~32 total inner volumes, then add stuffing.


Edit I have updated my previous post #71 to attached the modified series 1st order you suggested

 

[Stal]

I appreciate you don't want a big box. It was more an exercise to see what happens to the resonant peak, it has moved down in frequency. I didn't realise a 0.707 alignment moved it up by that much.

 

[Stal]

With the 2nd order xover could you remove the parallel R to see what happens.

 

[Stal]

I did in the picture attached


Edit : attached the modified series 1st order

With the 1st order series could you change the box alignment of the FF85WK as low as possible, its only a small box anyway.

 

[elmans]

Here are both 2nd order versions (0.707 and 0.577) without 24R parallel resitor.


As for FF85, it seems 0.577 is as low as it can go, beyond that increasing volume does not change Qtc