Supertweeter experiment

[HighFidelityGuy]

I've been fancying doing some experimentation with supertweeters for a while and seen as I'm waiting for parts to arrive for my other projects I thought I'd make a start on another. ;)

Quite a few years ago I purchased some Vifa XT25TG30-04 tweeters second hand. They are a ring radiator design and supposedly have a useful frequency response up to 40KHz. They're probably not as smooth sounding as some ribbon drivers and fancy stuff like that but I'm hoping they will at least provide a taste of what a supertweeter can add. That is if I can actually detect any difference after their addition. I'm not entirely confident my hearing is up to the job but you don't know until you try this sort of thing. :mental:

To allow me to have extra flexibility over the supertweeter's control I plan on connecting them to my old Cambridge Audio A500 integrated amp which has remote volume control. So I should be able to gradually turn up the tweeter volume from my listening position. This will also prevent them from affecting my main amps and speakers as they won't be an extra load. To begin with I'll just prop up the tweeters on top of my main speakers using some blu-tack or something. I may then look at making some proper enclosures.

The tricky part is going to be finding the best crossover frequency to prevent the supertweeters from boosting too much trebble below 20KHz and adding harshness to the top end. For the sake of simplicity I decided to use a first order high pass design to begin with. Then if I prove the concept works I may look at building a second order. So to start with I've ordered a pair of 0.68uF SuperSound capacitors from Wilmslow Audio. I'm hoping that value should make the supertweeter just start to kick in at a reasonable level around 20KHz and gradually increase in level above that. That will be my starting point. I'll then decide whether I need a higher or lower crossover point from there.

The caps should be arriving before the weekend so I'll hopefully have some initial findings in a few days. :cool:

[John]

If it works should help those speakers you have

[HighFidelityGuy]

It'll be an interesting experiment. I'm not really sure what to expect and it's not like I'm not happy with the sound I'm getting, I just fancied giving it a try to see if I was missing anything. The frequency response on the Opus 3's is supposed to extend to around 40Khz. So based on that I shouldn't really "need" additional supertweeters but you don't know untill you try this sort of thing. So I thought that seen as I already had the tweeters and a spare amp I'd give it a go. :)

[HighFidelityGuy]

The crossover caps have just been delivered. So I'll hopfully be able to make a start on this tonight. :eyebrows:

[John]

A few people say they perfer a treble a bit more revealing on the Opus 3 The latest model has different caps to address that but for me I did mind the treble So be interesting to see how you get on with this.

[DSJR]

First order filters will bring about significant overlap, so I hope the existing tweeters will be in phase with the new ones.

I remember, when I owned Spendor BC1's, I asked Spencer hughes for some high output Coles super tweeters. Over the next few weeks, first one and then another appeared and I duly fitted them, to find the treble almost disappeared :scratch: it turned out that the Coles tweeters were wired out of phase with the Celestion HF1300's, as the latter have a 14KHz peak before their death-plunge. Matching the two together this way was an attempt at flattening the response - hmmm.....

[HighFidelityGuy]

A few people say they perfer a treble a bit more revealing on the Opus 3 The latest model has different caps to address that but for me I did mind the treble So be interesting to see how you get on with this.

I heard they had tweaked the tweeters in the newer model but I'm not sure what the end result was. It was supposedly an improvement but I guess that's subjective. I could probably cope with a slightly more detailed or revealing sound. So I guess that's what I'm hoping to achieve if anything. My worry is that I could end up with a more fatiguing sound. We'll see. :)


First order filters will bring about significant overlap, so I hope the existing tweeters will be in phase with the new ones.

I remember, when I owned Spendor BC1's, I asked Spencer hughes for some high output Coles super tweeters. Over the next few weeks, first one and then another appeared and I duly fitted them, to find the treble almost disappeared :scratch: it turned out that the Coles tweeters were wired out of phase with the Celestion HF1300's, as the latter have a 14KHz peak before their death-plunge. Matching the two together this way was an attempt at flattening the response - hmmm.....

I am a bit worried that I may end up with too much overlap. I initially wanted to use second order filters but I was stuggling to find inductors with low enough values and I didn't want the extra cost or complexity of using multiple inductors in parallel or getting them custom wound. So I decided to keep things simple for my initial trials. If I get some half decent results with first order filters I'll then look at second order.

You make a good point about phase. I've decided to just try the phase both ways and see if I can hear a difference. In theory I shouldn't be able to hear any difference above 20KHz but it's going to be the overlap area where I could get problems. I may end up overly boosting or attenuating between about 15-20Khz, either of which could be bad. I think there's going to be a lot of trial and error involved. :doh:

I'm first just going to run the supertweeters on their own with the 0.68uF caps and see how much frequency range I can actually hear. If I get tons of trebble I know the crossover point is too low straight away. Then I'll take it from there. :)

[Reid Malenfant]

I am a bit worried that I may end up with too much overlap. I initially wanted to use second order filters but I was stuggling to find inductors with low enough values and I didn't want the extra cost or complexity of using multiple inductors in parallel or getting them custom wound. So I decided to keep things simple for my initial trials. If I get some half decent results with first order filters I'll then look at second order.
Let me know what value inductors you are after & i'll wind you a couple to the exact value. I have loads of spare ferrites as i used to work on SMPS.

:cool:

[HighFidelityGuy]

Thanks Mark, I'll let you know. :cool:

You may be able to answer a question I have....

When choosing the component values to get a particular crossover frequency; do you use the nominal impedance rating, the DC resistance or the actual impedance at the crossover frequency? I'm guessing it's not the DC resistance but in my case the impedance varies between about 4 and 6 Ohms above 20KHz.

HERE's (http://www.tymphany.com/files/XT25TG30-04%20Rev1_0.pdf) the spec sheet if that helps.

Cheers.

[Reid Malenfant]

You'll need to take into account the impedance at the crossover frequency & to me it looks like 4ohms @ 20KHz ;)

[HighFidelityGuy]

Thanks Mark. I think at 20KHz the impedance is actually a bit closer to 5 ohms but it's a bit difficult to tell from the graph as there aren't enough scale divisions.

Anyway, when I chose the 0.68uF caps for the first order filters I decided to use the impedance at the crossover point, so I'm glad I got that right. More luck than judgement I think. ;) So in theory I should be getting a crossover point of approx 40KHz, which should mean that the supertweeters have a -6dB point at 20KHz. So hopefully that shouldn't allow too much audible trebble through below 20KHz. That's the theory anyway. I'll find out later tonight. :)

[Reid Malenfant]

Thanks Mark. I think at 20KHz the impedance is actually a bit closer to 5 ohms but it's a bit difficult to tell from the graph as there aren't enough scale divisions.
It's a PDF file, when looking at the graph go to the top & hit +% a few times & you'll soon increase the magnification & then you'll see much more clearly - 4 Ohms ;)

[HighFidelityGuy]

Well I've just got everything hooked up on my test bench and I've got sound from the tweeters. Unfortunately I'm getting rather too much sound. I seem to be getting lots of mid-range and trebble. :steam: Does anyone know what's going on? The caps definitely say 0.68 on them. Unless I've done something wrong it seems like I need much smaller caps or a much steeper filter. Any ideas? :(

[HighFidelityGuy]

I've just tried a pair of 0.05uF caps and I still get quite a lot of upper mid-range through them. So I think I definitely need to use at least second order filters.

Mark, would you be able to make a pair of 0.035mH inductors, or there abouts?
I reckon that a pair of those and some 1.5uF caps should give me better results. What do you think?

Cheers.

[Reid Malenfant]

I can make you whatever inductance you want ;) Let me do some calculations & see if i come up with the same figures as yourself. You want -6Db @ 20KHz with a second order filter & 4 ohm tweeter..

I'll be back :eyebrows:

[HighFidelityGuy]

Awesome, thanks Mark. I'll leave it in your capable hands as I'm sure you know much more about this stuff than me. I'm a crossover newbie. :lol:

I think standard cap values are going to dictate things quite a bit from what I was finding. I picked 1.5uF as it was in the middle but I guess 1uF and 2.2uF might be a possibility. I'll leave that with you though.

If it helps, this is the crossover design in my main speakers:

http://i2.photobucket.com/albums/y48/DTM2000/Hi-Fi/Opus3_Crossover.jpg

I'll be running the supertweeters off a seperate amp.

Cheers. :cool:

[Reid Malenfant]

Looks like you need a 0.995uf capacitor in series with the tweeter & an inductor of 0.064mH accross it.

I'm sure i have a load of 1uf film capacitors here, i can easily test them to get within 1%, in fact 0.995 is withing 1% anyway but the caps will probably be 5 or 10% tolerance as standard.

The chokes won't be a problem, i'll just nick some off of an unused secondhand crossover & unwind until i get to the required inductance :eyebrows:

E2A:- Why not go forth order? 24Db/octave rolloff in the sub 20KHz region will ensure you don't get any interference from the supertweeter :) Also because you mentioned that the standard tweeter goes higher than 20KHz it'd put both tweeters in phase so they wouldn't tend to cancel each others output..

It's just another cap & choke in each circuit :cool:

[HighFidelityGuy]

Awesome, thanks Mark.
If you think a forth order filter would allow better integration then I'll go along with that. The only thing I'm woried about with a forth order is that it's an aweful lot of components to change if I find I need to tweak the crossover frequency. ;)

Part of me keeps thinking that if the supertweeters really are only producing frequencies above 20KHz then in theory I won't be able to hear if they're working, which will make testing tricky. So that makes me think I'd like them to dip just below 20KHz so I can at least hear them a bit. However I don't know how close to 20KHz I can actually hear. So perhaps the next thing I should do is test my hearing. :scratch: At least I would then know what to aim for.

So I'll play some test tones tonight and see what I find. That should then give you a guideline to base the crossover frequency on. Does that sound sensible?

Cheers. :)

[lurcher]

I've just tried a pair of 0.05uF caps and I still get quite a lot of upper mid-range through them. So I think I definitely need to use at least second order filters.

Mark, would you be able to make a pair of 0.035mH inductors, or there abouts?
I reckon that a pair of those and some 1.5uF caps should give me better results. What do you think?

Cheers.

As you are using a seperate amp, why not add an CR filter before the amp to give you the extra order (at least for the experiment)

[HighFidelityGuy]

Hi Nick, that's an interesting idea. Would I need to know the input impedance of the amp to calculate the components needed? If so that may be tricky to find out due to the age of the amp. Cheers. :)

EDIT: It wasn't as difficult as I first thought. I found it listed as 47K ohms.

[Reid Malenfant]

Awesome, thanks Mark.
If you think a forth order filter would allow better integration then I'll go along with that. The only thing I'm woried about with a forth order is that it's an aweful lot of components to change if I find I need to tweak the crossover frequency. ;)

Part of me keeps thinking that if the supertweeters really are only producing frequencies above 20KHz then in theory I won't be able to hear if they're working, which will make testing tricky. So that makes me think I'd like them to dip just below 20KHz so I can at least hear them a bit. However I don't know how close to 20KHz I can actually hear. So perhaps the next thing I should do is test my hearing. :scratch: At least I would then know what to aim for.

So I'll play some test tones tonight and see what I find. That should then give you a guideline to base the crossover frequency on. Does that sound sensible?

Cheers. :)
Jeez, everyone is Dave on here :lolsign:

Yes a 24Db/octave 4th order filter will help it integrate much better. I doubt you'd need to worry about effects in the range you can hear because if you are about 40 or more like myself it'd be pretty uncommon for your hearing to extend beyond say 16KHz at a reasonable volume level, which might make the experiment appear to be a waste of time at first glance... However sounds that you can't hear will probably affect those that you actually can in the audible spectrum & this is where the extra "air" might come from.

My guess is that you'd not need to tweak the components ;) As long as you set it up correctly with a test CD or better still a function generator you'd be just about spot on, or lets say within a few DB, but as you won't be able to hear it directly anyway a few Db should be ok. If we were talking about where we can here then you'd want to match within 1Db ;)

[HighFidelityGuy]

Thanks Mark.
I decided to test my hearing anyway as I was interested to know how high I can hear.
I could definitely hear 18KHz. 19KHz was just about perceivable. I thought I could very slightly hear 20KHz but I think I was tricking myself to be honest. So there's hope for me yet. :lol:
I'm 29 so I think that's fairly good to say I've been to a fair few gigs and festivals without any hearing protection. I do look after my ears fairly well though.

So if the super tweeters kicked it at about 18KHz or above then that should be spot on.

I've been reading a bit more about crossovers and I'm starting to wonder if using an active crossover before my amp might be a more refined approach.
However that would require quite a bit more work so I think I'll stick with forth order passive crossovers at this stage and then take it from there.

So Mark, is that enough info for you to sort out the parts? :)

Cheers. :cool:

[Reid Malenfant]

So Mark, is that enough info for you to sort out the parts? :)
Damn i wish i was 29 again & knowing what'd happen on my 30th birthday party :lolsign: I'd have a sh*t eating grin thinking about it 3489

Yes, not a problem as i already know the components you'll need for a 4th order crossover, or should i say roll in filter for the supertweeter ;)

[HighFidelityGuy]

Lol, sounds like you had a good one. ;) There's a possibility I may celebrate my 30th in Vegas, so that should be fun. :eyebrows:

I'll leave the crossover in your capable hands.

Cheers. :)

[Reid Malenfant]

Hi Dave, i haven't managed to find time to dig out a load of capacitors & test the things to get you your 1% tolerance parts yet i'm afraid, too damn busy sorting things out this way.

However here is the circuit diagram...

3663

The component values are as follows:-

C1 = 1.055uf
C2 = 2.11uf
L1 = 0.02mH
L2 = 0.09mH


Now i reckon i can get to within 1% tolerance on those capacitors as the values are inside of the caps standard 10% tolerance, so it's just a case of testing a load until i find the magic numbers :eyebrows: The caps i'm thinking of are 100V MKT type - film caps..

[HighFidelityGuy]

Hi Mark, thanks for the update.
I'm not in a major rush. I've finally been sent some parts that I need for another project, so I've got other things to work on. So just concentrate on getting your room sorted out for now. After that once you've got the parts sorted just let me know how much you want for it all or what you drink and I'll make sure you're compensated. :cool:

[Reid Malenfant]

The only thing i'll add at the moment Dave is that you won't really be able to set it up by ear ;) As the filter is steep you'll need to set it up by measuring at the amplifier terminals (i recall you mentioning you'll use a seperate amp :scratch:).

What you need to do is find out the sensitivity of your Wharfedales & the sensitivity of the ring diaphram tweeters. At that point you can play say a 1KHz test tone into both amps & adjust the volume on the new amp to get the correct output voltage to match the sensitivities. So you'll need a multimeter & some kind of test record or CD or similar :scratch:

[HighFidelityGuy]

I am using a seperate amp so that will simplify things. I have access to the speaker specs, test tones and a Fluke multimeter, so I think I should be able to sort it. Once I have everything together I'll let you know so you can explain the steps.

Cheers. :)