Effect of driver resistance on crossovers

[Artmin]

When i bought my Celestion 33s, both tweeters were blown. Bass & mid are marked 4 ohms. Tweeter is unmarked but a spec sheet confirms 4 ohm too & I also managed to get a reading of 3.4 ohm from one of the voice coils.

To get the 33s up & running I used the 'same' tweeters from the Celestion 15 XR but the speakers were shouty at the top end. These XR units are reading 7.2 ohms. I have now rebuilt the 33 tweeters with new voice coils which give the dc reading of 6.3 ohm. When installed in the 33s, these too where shouty but are perfect when installed in 15 XR speakers.

Now I know that if I put an 8 ohm unit on a crossover designed for 4 ohms, I am going to change the crossover frequencies. My question is, to reduce the 6.3 ohm tweeter load to around 4 ohms, can I attach a ~10 ohm power resistor in parallel across the tweeter? Would this work or am I in cuckoo land? I also would have thought that an 8 ohm tweeter would have been quieter in a 4 ohm circuit, not the other way around.

I am currently running with them (@6.3 ohm) in the 33s with a 1 ohm resistor in series to attenuate the HF a bit, but I can hear a bit of a clash between the mid & the treble. I have also reversed the tweeter phase temporarily which is helping a bit, but I would love to get back the mellow Ditton sound.

Any advice would be greatly appreciated as these are doing my nut in :mental:

[walpurgis]

That's strange, a lower impedance will normally (but not always) equate to higher sensitivity, but in your case it sounds as though you have too much tweeter output with higher impedance. I'd recommend keeping an eye out for the correct Ditton 33 units really, at least the crossover will have the right component values and if you still find them a bit forward you can use a couple of resistors to form an 'L' pad and drop the output. Plenty of calculators and schematics on the net.

[Artmin]

That's strange, a lower impedance will normally (but not always) equate to higher sensitivity, but in your case it sounds as though you have too much tweeter output with higher impedance.

Thanks Geoff, that's what I thought. Yes, I need to get the crossover freqs. back to spec. I've just recapped them to original spec too. Any ideas on the 'resistor in parallel' theory. If that can correct the impedance then the battle is hopefully won & I can play with the attenuation.

Thank you

[walpurgis]

Although a resistor in parallel will alter the DC resistance reading of the driver and resistor pairing, it will be difficult to predict how it will affect the impedance as that is a dynamic value. The amplifier damping factor will probably be changed too. Unless you're sure of the effect, I'd say don't do it.

[system7]

Changing the tweeter impedance (all other things being equal) actually effects the DAMPING of the filter, not the crossover frequency.

So putting a nominal 8 ohm tweeter in a filter designed for 4 ohm nominal will create a sharp and peaky Chebysheff type response, rather than a gentler Linkwitz-Riley rolloff. This will interact with the midrange to produce a peak in response at crossover.

The filter will need a redesign. Smaller capacitors and bigger coil. Compare the below:

http://www.visaton.com/en/chassis_zubehoer/weichen/hw2_70ng_4.html

http://www.visaton.com/en/chassis_zubehoer/weichen/hw2_70ng_8.html

I'd really need to see the filter schematic and driver DC measurements to do better. No-one has posted a Ditton 33 crossover AFAIK. But I always enjoy modelling 3 way Celestion stuff, and sometimes it is easy enough to replace tweeters with modern ferrofluid types like the familiar 8 ohm soft dome SEAS 19TFF or perhaps a mylar Visaton DT94 http://www.visaton.com/en/chassis_zubehoer/ht_kalotten/dt94_4.html if the original really was 4 ohms.

It's always useful to measure the radius of the tweeter and the baffle cutout when seeking a slot in replacement too. Depending on the filter layout a resistor or two or a RC Zobel can then be applied to get the levels and slope right as mentioned.

[system7]

I'll just post this here as a a marker, since it's hard to track down.

The Celestion Ditton 33 crossover schematic:

http://img802.imageshack.us/img802/9499/ih33.jpg (http://imageshack.us/photo/my-images/802/ih33.jpg/)

Uploaded with ImageShack.us (http://imageshack.us)

The bass coils will be about 2 and 1 mH I'd reckon. The midrange shunt will be about 2mH too. I'd have to model it to see what phase is right.

Does look like a 4 ohm filter on the tweeter.

[Artmin]

Hi Steve, yes I think I am getting that peaky response that you describe where the mid meets the treble. I was punching the cap & ind values into a 2-way 3rd order calculator but I couldn't get the values to add up. I take it it is different for a 3-way 3rd order system but I can't find an online calculator for that.

[StanleyB]

Now I know that if I put an 8 ohm unit on a crossover designed for 4 ohms, I am going to change the crossover frequencies. My question is, to reduce the 6.3 ohm tweeter load to around 4 ohms, can I attach a ~10 ohm power resistor in parallel across the tweeter?

The resistance you read across the tweeter with a multimeter turns into a variable impedance during music. If you hook a resistor across it you are going to modify the frequency curve since the total resistance across resistor and tweeter is going to be of a different impedance value across the frequency spectrum compared to the tweeter on its own.
Also keep in mind that the resistor and tweeter are now sharing the load about 50/50. So if you got a 25W tweeter you are going to need a 25W resistor as well in theory.


I also would have thought that an 8 ohm tweeter would have been quieter in a 4 ohm circuit, not the other way around.

The efficiency of the two tweeters is very likely different, which is why the one sounds louder than the other.

[Artmin]

Might give the resistor a try, just to see what happens. If I don't I'll always wonder. Will let you know what happens.

[system7]

I use Visaton Boxsim for my simming. The trick is to use similar sort of Visaton drivers. I assume the physics is then similar, but it's always hard to estimate voicecoil inductance.
http://www.visaton.com/en/literature/software/downloads/

I modelled this Ditton 33 as a wallmounter and got some reasonable similarities. An 8 ohm nominal tweeter DID produce a 4-6dB peak around 3.5kHz. I won't bore you with the details, but replacing the 8 and 12uF capacitors with 4 and 6uF and an attenuating resistor around 2.2R tidied things up with an 88dB 8 ohm tweeter. The 7.5R plus 1uF Zobel suits most modern 0.05mH ferrofluid tweeters.

I would start there. Adjust the 2.2R for level. Won't be a million miles away. This speaker is actually not hugely efficient. It's is essentially the same tweeter circuit as my Ditton 44's with their HF2000 mylar jobbies.

[Artmin]

:wow: Steve, many thanks for that!

I have downloaded the software but I'm not sure where to start. I'd be very interested in having a play with that setup you have produced. Could I ask you please to email me the file at deletedforsafetyssake.com?

The voice coils in the tweeters that I have are actually around 6.3 so it'd be great if I could plug'n'play a bit with it.

Many thanks,

Gavin

[The Grand Wazoo]

Perhaps best to send Steve a PM with your email address on it Gavin - otherwise you might get some unwelcome interest from the many, many members of the Zwamgerian Royal family who need to find a place to park their untold millions for a day or two.
I've edited your post accordingly.
If you object, then let me know & I'll reinstate it.

[system7]

Gavin, I gotta tell you that Boxsim takes a while to learn. :)

You basically set up the cabinet, then fit some drivers and enclosures, then set up a reasonable crossover, then select some optimisations in red.

Then select the optimise button in the extras menu. Boxsim then gets the best possible values. It's magic. :eek:

But the skill is in giving it a good topology of filtering and matched drivers in the first place, so you'd best start with a two way system.

I've been doing this for years so I find it kinda easy. It's a sort of visual mathematics, and I often base it on classic designs that work well.

Nevertheless, I have a confidence that my tweeter circuit is about right here:

http://img27.imageshack.us/img27/5775/2w61.png (http://imageshack.us/photo/my-images/27/2w61.png/)

Uploaded with ImageShack.us (http://imageshack.us)

You'll want 7W wirewound resistors and some capacitors.

[Artmin]

Perhaps best to send Steve a PM with your email address on it Gavin - otherwise you might get some unwelcome interest from the many, many members of the Zwamgerian Royal family who need to find a place to park their untold millions for a day or two.

Chris, thanks for looking out for me. You get those emails too? Think of the bank interest I could be missing out on :doh::lol:

Thanks Steve, i can see that image much better now. So that's the original inductor just with different caps. Do you think I'd need the Zobel for the re-voice coiled HF1000? Would those cap values change with the 6.3 ohm impedance or would it not matter that much?

Sorry for all the questions. I used to think old Celestions were simple things :scratch:

[system7]

It's one of those internet rules never to post your email address. You just get sent a load of junk by the robots! ;)

The Zobel is kinda optional. You can either dump it or just fit a 22R resistor to damp the Fs voicecoil resonance a bit. Good point.

I'm not really sure what you have done to the tweeter TBH. But new smaller tweeter capacitors are compulsory here with the change from 4 to 8 ohms.

And an input resistor for attenuation of tweeter level.

[The Grand Wazoo]

Chris, thanks for looking out for me.
No trouble guv'nor - all part of the service.

[Artmin]

Here's what has happened. I had ordered a pair 10 ohm resistors a few days ago & they arrived today in the post. So out of curiosity, I soldered them onto the crossover between the HF out and -ve and added a 1 ohm in series to the tweeter after the crossover (I tried it before the caps but vocals sound fuller with it after). The result, well :wow:! The 33s have their voices back, sweet & mellow.

Now I know that in theory there are flaws in doing this, but then again with the rebuilt tweeters I don't know their response curve. But everything has clicked beautifully into place. The HF circuit is now seeing a dc resistance of 4.1 ohm. I plan to run them side by side with an unaltered pair of 33s in the next week or two, so until then I can't say if their voices have changed. But I'm well happy!

Thank you all for your help. I'm not ignoring what you say regarding changing values in the XO as I understand what you are saying, but this trial solution sounds great to my ears. :dance:

I will let you know how they go head to head.

[system7]

Nice work, Gavin. You have added about 3dB attenuation and treated the Fs resonance peak with the 10R damping resistor, which makes for less courseness and intermodulation in the sound.

This is a good thing to do with non-ferrofluid tweeters. Below the Vifa XT25 for comparison:

http://imageshack.us/a/img15/9313/oqd0.png

You've still got a bit of a bump at 3.5kHz. I found these values worked well to smooth it out without ruining the excellent phase alignment.

http://img856.imageshack.us/img856/3100/9k37.png (http://imageshack.us/photo/my-images/856/9k37.png/)

Uploaded with ImageShack.us (http://imageshack.us)

Here's a Troels Gravesen 3-way that does similar things to the Ditton 33 for interest. Troels is THE MAN with speakers...:)

http://www.troelsgravesen.dk/3WClassic.htm

[Artmin]

Thanks for that Steve, so I wasn't just imagining that it sounded good.

I'll have to have a play again with some of the those cap values. I should have enough cheapo ones about to mod those values (out of curiosity again, of course!).

I think you are THE MAN when it comes to crossover design. All I can do is plug & play, but that makes me :mental:, sometimes.

[Artmin]

I eventually got the opportunity to test the 33s side by side with an original, untouched pair. Mine still had the original crossover values, just with the shunt & attenuation resistor.

To our ears there was nothing in it. Both sets sounded great so I'm happy with the outcome, be it luck or not :scratch:.

Thank you for all your input!