605 time delay crossover

This is for you lucky 605 Duplex owners
It describes a crossover that is time delay corrected for the offset caused by the woofer / HF horn VC magnet placements.
It turns out that fixing the delay also takes care of the phasing problems too,

Also included is a test signal you can play to verify the time delay on any two way system.
This would help you set the delay for any active XO.

This information was lost in a move by Apple/MAC to close the URLs.
The new URL is:

www.donpatten.com

Click on X-Over or DIY

Sorry for any inconvenience this caused.

Don

Don,
This is an impressive body of work you've done! Much of it is still over my head a bit, but I'm studying up...
I'd like to try and use this for my 604E, so I may hit you up for some tutoring if you're open to that.

Mike
I would love to.
Frist: Tell me about your self, your 604E and what test equipment, cpu software you have ( or can beg bos ) . Can you do freq response, impedance curves, look at scope traces of pulses, etc.

Don

hah! Can I call ya? My fingers would hurt typing all that!

Mike
We need the woofer to be a flat impedance from 400 Hz to 4000 Hz. Do have Fluke 87 DMM?
Add a Zobel to your woofer . This is a series R & C across the woofer . Pick the values that give a flat Z curve. Use an L Pad to find the proper R (10 ?).The C ( 30 uF ? ) can be several caps in parallel

Make a CD of the test signal on my web site. and have it repeat 10x or more

To get the delay time we need to play the test CD , with the woofer on, and look at the acoustic signal 12 inches on axis with a scope and almost any omni microphone. Connect the HF horn in series with a 4uFd cap.
This gives you a bare foot woofer and a single pole HF high pass crossover.

Sync the scope with the amplifier output and turn the HF on and off.
Note the LF peak and the HF peak. Estimate the time between the peaks (300 to 500 micro seconds )

Don

Quote:

---

Originally Posted by donaldpatten

Mike
We need the woofer to be a flat impedance from 400 Hz to 4000 Hz. Do have Fluke 87 DMM?
Add a Zobel to your woofer . This is a series R & C across the woofer . Pick the values that give a flat Z curve. Use an L Pad to find the proper R (10 ?).The C ( 30 uF ? ) can be several caps in parallel

Make a CD of the test signal on my web site. and have it repeat 10x or more

To get the delay time we need to play the test CD , with the woofer on, and look at the acoustic signal 12 inches on axis with a scope and almost any omni microphone. Connect the HF horn in series with a 4uFd cap.
This gives you a bare foot woofer and a single pole HF high pass crossover.

Sync the scope with the amplifier output and turn the HF on and off.
Note the LF peak and the HF peak. Estimate the time between the peaks (300 to 500 micro seconds )

Don

---

Very nice explanation Don ! Thanks for helping out Mike.... :smile2:

Quote:

---

Originally Posted by donaldpatten

Do have Fluke 87 DMM?

---

Not the fluke, but I do have a good DMM - Extech 22

Quote:

---

Originally Posted by donaldpatten

We need the woofer to be a flat impedance from 400 Hz to 4000 Hz.
Add a Zobel to your woofer . This is a series R & C across the woofer . Pick the values that give a flat Z curve. Use an L Pad to find the proper R (10 ?).The C ( 30 uF ? ) can be several caps in parallel

---

I have not yet built a test enclosure for this speaker, so I assume that you are speaking about all free-air values.
I installed DATS on my desktop PC; it is much more modern than my laptop, so I was thinking I might get more accurate results. Here is the free air test result on the woofer:

https://www.hostboard.com/forums/hbm.../2012/09/1.jpg

Quote:

---

Originally Posted by donaldpatten

Make a CD of the test signal on my web site. and have it repeat 10x or more
To get the delay time we need to play the test CD , with the woofer on, and look at the acoustic signal 12 inches on axis with a scope and almost any omni microphone. Connect the HF horn in series with a 4uFd cap.
This gives you a bare foot woofer and a single pole HF high pass crossover.
Sync the scope with the amplifier output and turn the HF on and off.
Note the LF peak and the HF peak. Estimate the time between the peaks (300 to 500 micro seconds )

---

OK, I can use DATS for my 'scope. I have a calibrated condenser mic that I've been using to make FR measurements. I'll start now on the zobel and take a few measurements and report back.

Thanks for the help, Don.

Some quick bread-boarding and I managed to derive a Zobel that looks pretty good against your design parameters:

https://live.staticflickr.com/65535/...e649dae2_b.jpg

The textbook calculator suggested C1=6 uF and R1=9.2 ohms.
With experimentation I finally arrived at real-world values of C1=18.33 uF and R1=10.14 ohms.

Mike
The curve shows 7.5 ohms 500 Hz to 5k Hz. This is a critical number, and is the design load impedance for the low pass filter.
C1 to C5 also L1 to L5 are calculated from this value along with the delay time. The next step is to build a lows pass filter and check the delay.
I have enough spares to bread board so you can test it with your 604.
PM me your shippiing address full name and phone number.
Please verify your test set up with an 8 or 10 ohm resistor.

Don

Don,
Here's my first attempt at measuring the woofer using the new zobel and your test chirp:

https://live.staticflickr.com/65535/...7f8153ea_b.jpg

I'm a bit lost on how to compare the HF...when I isolated just the HF, the pulse intervals seem to line up to the LF pulses (when each is measured independently). I don't know how to discern the difference in the delay between the two...

Oh, and as far as testing the DATS rig, yes, I did that thoroughly using precision resistors and caps I have on hand. All those calibration tests were looking good.

L1 1.4605425
C2 17.41598667
L3 0.80942775
C4 12.58409333
L5 0.6199425
C6 9.364906667
L7 0.42345225
C8 5.516826667
L9 0.18948525
C10 1.133253333
Mike

Look at the fourth image a pr of red & blue freq resp curves, and the text with them. Can you trigger the scope (dual ch'l) on the amp out and look at the mike preamp out on the other ch'l ? Your traces show a large LF pulse followed by a smaller HF pulse, repeated 6 times.
Try to reduce the time base by 5 or 6, to show only one pr of pulses. Now turn off the HF driver. Turning the HF on/off will help iD the two pulses.

www.donpatten.com

New coil & cap values See my web site


L1 1.4605425
C2 17.41598667
L3 0.80942775
C4 12.58409333
L5 0.6199425
C6 9.364906667
L7 0.42345225
C8 5.516826667
L9 0.18948525
C10 1.133253333

Mike
The CPU list of coils & caps is 1 to 10
The schematic is C1 to C5. Sorry about that
C1 = 12 + 5.6
C2 = 12 + 0.33 + 0.22
C3 = 4.7 + 4.7
C4 = 3.3 + 2.2
C5 = 0.33+ 0.33+ 0.47

For the schematic and text see:

www.donpatten.com

Great - that clarification helps Don.
I'll get the other breadboard out and start putting the test rig together.

I also have an active XO which provides the ability to play with time delay - I might look into that to do some prototyping.

Mike
When you tune L1 & C1:
Connect 12uF and the 5.6 in parallel to make C1 . Connect C1, L1 coil, and a. resistor all in series, Try the DAT by measuring a sweep impedance. To tune a coil add or remove turns of wire.L1 & C1 should bottom out at 998 Hz.

Don

Mike
I suggest to test the breadboard board low pass filter you run a sweep freq resp.with the woofer and Zobel connected as a load. Do not connect the HF driver. The sweep tone is connected to the power amp input. The measurement connection is to the woofer terminals ..Take care as not to short the speaker line, since both the tone source and measurement connection are probably common thru your PC.
When your make a measurement like this be very Quiet as your Super Duplex is a super microphone.
Don

Hmmm, ok so I don't use a mic for the measurement? I attach both the amp out and the PC soundcard input to the woofer terminals?!?

Yes Mike
This is a test of the freq resp of just the coils & caps. WE will include the drivers next. To keep the filter happy we need the Zobel and woofer as the load. Try a sweep curve to see how we are progressing
Don

OK, well, I did an inventory of my caps and coils and I don't have what I need for the LP filter. I will order some wire to wind some coils.
Do you agree with the general tips provided on this webpage?
Also, where is a good supplier of magnet wire?
For 200' of #18:
Powerwerkz $27
Amazon Small Parts $16

Mike


The # 26 AWG Magnet wire is available from Mouser or any Belden Distributor: part # 8052 is a 1/2 pound, 800 foot spool. I got large spools with long lengths of #18 and #15 AWG from BAE Wire ( 941/309-0900 ) on eBay.
The price of copper has gone crazy.
See

www.donpatten.com

Mike
You will want to listen to your speaker, so connect the Lowpass filter to your power amp and speaker. The woofer Zobel (10R+18F) is all ready on the circuit board.
To connect the horn you will need a 4uF cap in series and an L Pad . Connect the driver leads to the L Pad terminals 1 and 2 . Connect terminal 1 to the power amp common .Connect L Pad terml 3 to one wire of the 4uF cap and the other wire to power amp hot .

Lets check the new prototype LP filter time delay, to see what correction is needed.
Play the test signal through the filter and woofer.Trigger the scope on the input signal to the filter. Connect your test mike to pick up the on axis sound about 12 inches in front of your 604. Play the woofer . I use a dual channel in the chopped mode, sync to the power amp . Adjust the display to show the power amp output as a single pulse. The other channel shows the woofer acoustic output. Now turn on/off the horn and see any changes in the combined woofer/horn acoustic signal. When the pulse gets taller and narrower we are done. If adding the horn output causes the pulse to jump sideways ... The amount of jump is the amount of timing error.

Don

Don, I received the woofer TD filter today - thanks again for putting that together for me.

I've gotta run to practice, but I thought I would at least post a quick imped sweep of the woofer with the filter:

https://live.staticflickr.com/65535/...ea623d6c_b.jpg

I'll do the pulse testing later tonight if I have time.

What's a TD filter?

A gi-normous chunk of copper wire and little plastic cylinders :biggrin2:


Time Delay

cradeldorf

Most XO filters , Butterworth , etc have time delays that vary with freq. this is not what we want.
The Bessel filters have constant delay and are what the Dr ordered.
We hope to delay the low freqs , all of them, by the same delay to match the high freq delay caused by the driver being behind the woofer.
And if we are lucky, this filter will not only provide the proper amount of delay, it will have a frequency response that rolls off the high freqs to match our hi freq filter.

Don

Pano -
I swore that you posted recently about using Holm to detect timing delay of different drivers in a system.
The board's search indexes apparently haven't been updated since JUNE so the search function isn't working worth a crap!
Can you tell me how you use Holm to do that? I don't have a 'scope, and the DATS scope is rather limited in functionality - I can't split the input between devices, so the method that Don is describing and asking me to do I'm at a loss.

I still can't get the DATS scope to do what I want it to, but I went ahead and took some test screen shots anyway for you all to look at.

First is the scope of the woofer only (through the TD filter) with the test chirp signal:

https://live.staticflickr.com/65535/...ca6393de_c.jpg

Second is the HF horn only with the same test chirp signal:

https://live.staticflickr.com/65535/...1649fb0b_c.jpg

Last is the combined woofer and HF horn:

https://live.staticflickr.com/65535/...5c3afdcf_c.jpg

The amplitude seems to only increase when the woofer and HF play simultaneously, versus getting "messy".
But I'm not sure what else I can deduce from this without being able to trigger directly from the amp signal and compare...

As a layperson who's been watching this thread with interest, I'd say it looks like you're there.

It would be interesting to compare those shots with similar shots using a standard n-1200 crossover without the TD circuit.

eso

Mike
I think eso is right.
Could you change the scope horiz time base by 6 or 10 to blow up the image ?
Re splitting the input... I use a small A to D converter to get my analog L&R to USB. It looks like a double stuff Oreo with a tail. The tail is a USB cable and the cookie has two 1/8" stereo jacks one for Line and one for Mic. I got one from the Apple Store and another from Goodwill.
Don

Quote:

---

Originally Posted by donaldpatten

Re splitting the input... I use a small A to D converter to get my analog L&R to USB. It looks like a double stuff Oreo with a tail. The tail is a USB cable and the cookie has two 1/8" stereo jacks one for Line and one for Mic.

---

Like one of these?

http://www.hostboard.com/forums/hbmc...2012/10/25.jpg

Griffin Technology iMic USB Audio Interface GC16035 B&H Photo

eso

Thanks guys - I agree that we are just about done here; what remains is more academic in terms of precision of measurement.

Here's the scoop on the DATS oscilloscope. It is "dual channel" but limited. One of the channels is fixed on the internal "reference signal" (signal generator), while the other is selectable (you can use the DATS test leads to probe an external signal, or you can point to the software audio device selected for use as input to DATS - normally the sound card or mic pre-amp). Since my crappy laptop only has two USB 1.0 ports, which are both filled by the USB mic/preamp, and the USB sound card, I didn't have an empty port to use the DATS test leads. One last thing I can try is to find a USB bus and get another port going, or try to use the laptop internal sound chip and free the USB port up.

And yes Don, I can adjust the scope base time and take a few more screen shots. Will try that this evening.

Eso - the 604E runs stock on the N1500 XO. Unfortunately, the speaker I bought didn't come with the XO, so I can't test it with the N1500, unless someone wants to mail one over. :) I have the schematic (it's a very simple circuit) and could slap one together...

Great

Send the schematic . I never turn down one.

Don

N-1500-A Schematic

https://live.staticflickr.com/65535/...336a67b44b.jpg

Quote:

---

Originally Posted by Alien_Shore

the 604E runs stock on the N1500 XO. Unfortunately, the speaker I bought didn't come with the XO, so I can't test it with the N1500

---

Bummer. It would be fun to the the difference. I believe the compression driver in the 604 is based on the 802 drivers, and those cross at 700Hz in an A7 system. Would it be possible to give both drivers a 1K test tone with no filtering to illustrate the time smear?

For some background, I'm working on re-creating the Urei 801B duplex which was developed by United Recording Electronics after they ceased using the 604 based diver due to QA issues in the early '80s. The picture in my avatar are some of the early horns for my project based on the Urei horn. Their Time Aligned? crossover is quite similar to the filter Mr. Patton is using, although I think with modern testing Don may be attaining higher accuracy that the production Urei units.

eso

Mike
I can put one together sunday , you would have it next week .
Don

Here are the same measurements, with a bit more sensitivity on the scope:

LF only:

https://www.hostboard.com/forums/hbm.../2012/10/1.jpg

HF only:



Combined:

https://live.staticflickr.com/65535/...32a063fd_c.jpg

No joy on the dual channel triggering. I did manage to get all three devices going with a USB bus, but the scope software still only will look at the internal signal generator and the probe. So with that experiment I was at least able to see the electronic trace of the test chirp as it came out of the amp into the filter, but just cant put anything else on the display with it. I plan on building another tube amp this winter, so I'll be on the lookout for a standalone scope or some better software.

Quote:

---

Originally Posted by donaldpatten

Mike
I can put one together sunday , you would have it next week .
Don

---

Thanks Don - I received the N1500-A clone in the mail today.
I'll work on putting together some tests...

Do you have your E in a cabinet yet?

oh hell no. I've been busier than a one-armed paper hanger.
Although my woodworking shop construction project is moving along! The concrete guys came today and poured the slab.

Ran a quick impedance sweep comparison of the N1500A clone that Don built along with the Time Delay filter.
These are sweeps with the 604E woofer attached only. Remember that my 604E is reconed at GPA with 8 ohm parts.
In this graph, the red/blue lines are the time delay low-pass filter, the green lines are the N1500A clone with the HF shelving control set at mid-point. The cursor is set at the crossover point of 1500 Hz.

https://live.staticflickr.com/65535/...798f6e80_b.jpg

The shelving control on the N1500A clone works as you'd expect from looking at the schematic. As you decrease the HF attenuation, the woofer rolls off more quickly, and vice-verse.

Here are a couple more impedance sweep graphs.
This is the HF tweeter barefoot (green), along with the zobel (magenta) I'm working on:

https://live.staticflickr.com/65535/...01517819_c.jpg

I need some smaller resistors and caps to fine tune this a bit further, but it's getting there. I had to do some un-natural series/parallel joins of caps to get at this arrangement since all I had on hand were 1 and 2 uF caps. This is using 12 ohms R and 1.5 uF C. I think I need a little bit more R and C to get this thing fairly flat at about 8 ohms. Does anyone know if it is possible to smooth out all those crazy ripples after Fs?