605 time delay crossover

[donaldpatten]

Mike
Looking Good!
If you settle on 5 ohms, for the series R with the peaking coil, just use #30 awg wire , and omit the R.
For help on coil winding dimensions see Shavano Music On Line. The real key, is to make the coil form radius = the cross section height and width. This maximizes the Q with minimum wire. I can't wait until you listen and plot frequency curves.
Don

[Alien_Shore]

OK, continuing the testing for the HF filter. I think I understand what you are saying with the coils, Don. Matching the Q of the filter to the Q of the resonant peak seems to be very important for the filter to work correctly. By using smaller gauge wire on the coil, increases the coils Re, and can eliminate the need for a separate series resistor.

Here's what I have achieved so far (this is still with spare coils on hand) - I've not yet built a form and tried to wind my own!!



These two impedance curves are: Blue = barefoot HF tweeter, and Yellow = with Zobel and resonant-peak filter.

[donaldpatten]

Mike
Looks like you are almost there.
Re pots. Most of the ones I found were Junk
The coil , L6 and cap C8 form a peak to help lift the driver from 8 to 12k Hz You can leave them out if you want.
My drivers die up there, so they needed a little kick.
Don

[Alien_Shore]

Don,
I have a question for you - on the clone of the N-1500 crossover that you built for me, I want to adjust this for 8 ohm drivers. Would that imply to double C1 and halve L1? This would result in 0.75 mH value for L1 and 28 uF for C1... I don't have a circuit simulator that will let me do that for a series filter, but was going to try to prototype it this weekend.

[donaldpatten]

Mike
The Time Delay filter I sent is low pass only, with Zobel to match ( I hope) the speaker you ran all the curves on.This has 5 coils, about a dozen caps and one R for the Zobel. This is a parallel design.

The N-1500-A Clone was to be used with a " stock 604E " . This is a series design . The N-1500 has one coil, a HF Attn pot and is complete , no Zobel needed. It should work on 8 to 16 ohm drivers.
To complete the time delay XO, you need a parallel HF filter. I suggested a 4 uF cap and an L Pad, to start.

Don

- - - Updated - - -

Mike re changing impedance:

The XO freq is fixed at 1500 Hz so you can scale up and down the driver Zs. As I recall the 604 and 605s were designed around an old WECo standard of 12 Ohms.

ww =1/(LC)....where ww is : Omega squared , a constant. So if L increases 10% C decreases by the same %
Z= w L so if you want 10 % lower impedance . reduce L 10% and increase C the same %

Don

[Alien_Shore]

Here is the FR measurement of the 604E in the test cabinet I recently built.
This compares the FR using the N1500-A clone XO that Don built, versus the custom time delay XO:

[cradeldorf]

Hi Mike, Do you have a zobel on the HF? I noticed when modeling last night that it brings the HF down to match the LF pretty well, I just auto calc'd it and then fiddled with the values and it smoothed the sum line out pretty well. I'll leave that up to you and don on whether it's a good idea or not. My parts for the upgraded crossover will be here tomorrow and I'm trying one.

Isn't time delay a phase thing? Also do you have the tweeter polarity reversed in PCD? Or is it not reversed in this graph? I noticed when I used your .FRD files that it had a much smoother sum line if I didn't reverse the HF polarity as opposed to reversing them. It had a huge dip down at crossover with the HF reversed which seemed out of the ordinary...

[Alien_Shore]

Hi Mike, Do you have a zobel on the HF? I noticed when modeling last night that it brings the HF down to match the LF pretty well, I just auto calc'd it and then fiddled with the values and it smoothed the sum line out pretty well. I'll leave that up to you and don on whether it's a good idea or not. My parts for the upgraded crossover will be here tomorrow and I'm trying one.

This FR does NOT have the zobel on there - it's a simple 1st order HP with an 8-ohm L-Pad that I threw together. But yes, given the horrible rolloff @9K, I am trying to find my breadboard that has the HF zobel/notch filter I built. The FR curve looks awful at the HF resonant peak, so I intend to also put that notch filter in there and see if that helps to smooth things out. As you can see from my previous post, the zobel/notch filter has got some pretty big resistance in there from both the coil and the parallel resistor. I'll test both the zobel and notch separately. If the weather clears up, I might try to run the FR tests outside and see if I get better measures. I'd like to see your filter values you came up with in PCD.

Isn't time delay a phase thing? Also do you have the tweeter polarity reversed in PCD? Or is it not reversed in this graph? I noticed when I used your .FRD files that it had a much smoother sum line if I didn't reverse the HF polarity as opposed to reversing them. It had a huge dip down at crossover with the HF reversed which seemed out of the ordinary...

While I think phase is impacted, the bessel filter is more about aligning LF sound wave arrival times with the HF waves, since the voice coils aren't physically aligned in the 604. I didn't reverse the HF polarity. As you indicate, the PCD model shows it would be bad to do so, with both the series filter (N1500) and a parallel filter that I've modeled.

[cradeldorf]

whether it's right or wrong I have been getting the best sound with both the positives hooked to the left terminals and the grounds hooked to both the right terminals. Providing the speaker is oriented with the terminals upward. The crossover I came up with was conceived using my new measured .zma files and the blank frd files that come with PCD. I'm at work right now but tomorrow I can email you both the files I came up with, The one I'm building and the one using your .frd files. I kinda jumped the gun and ordered the parts the night before your files showed up. If I'm not totally satisfied I'll put together the ones I came up with using your .frd files. Mine are Linkwitz Rileys with zobels. I thought bessel was just a different crossover slope, like a butterworth or a linkwitz...

[Alien_Shore]

I thought bessel was just a different crossover slope, like a butterworth or a linkwitz...

The reason Don went with bessel is because each stage adds delay to the signal (like all passive filters do), but the key difference from those other filter designs is that the bessel delay is frequency independent, and it's linear phase response. Here's a technical paper that talks about the bessel in contrast to the other popular filter types. It was presented at AES, so it's brainy, but there are some good parts that I can understand...