Celestion HF1000 replacement tweeter?

[DarrenHW]

I have 2 pairs of speakers (UL8 & Ditton 11) which use the 4ohm HF1000 tweeter. Two of the tweeters have been repaired so I'd like to replace the repaired tweeters with modern equivalents but I can't find any tweeters listed as replacements, does anyone know of a suitable replacement? Ideally I'd like to use the SEAS 19TFF1 I've used this to replace the HF2000's in my 44's.

[DarrenHW]

I'm thinking of using SEAS 27TFF http://www.falconacoustics.co.uk/downloads/Seas/h0831_datasheet.pdf

The HF1000's average 4.7ohm, the 27TFF (if I'm reading correctly) 4.8ohm.
The UL8 crossover frequency is 2000Hz and up to 28,000Hz, the recommended frequency range for the 27TFF 2000 - 25000Hz.
Size wise the 27TFF looks to be marginally smaller, not perfect but hopefully a drop in replacement.
I do not know what the sensitivity rating of the HF1000 is, the 27TFF is rated at 92db.

I'm hoping the 27TFF could be a drop in replacement with no need to modify the crossover, if anyone has any thoughts on this I'd be grateful for your input.

[tubehunter]

You could try the excellent Morel CAT 308 or CAT 328.

Both have low resonance with high power handling.

I'm listening to some 32's at the moment, very nice.

[Reffc]

You could try the SEAs units but you really need to know what the sensitivity is to match with the originals or at least ensure they're a good match with the woofer. There's no such thing really as a drop in replacement as the impedance with frequency will not be the same, no-matter what you choose to replace them with. Bear in mind also that the crossover frequency is usually the transfer function for the acoustic response and this isn't necessarily the same as the electrical crossover point (hence why crossovers cannot be "calculated"). For a crossover of 2KHz acoustic, you really need to check that any replacement tweeter has a self resonance of maximum, 600 to 700Hz. Less is better as distortion will be lower and power handling better.

The 27TFFC might be your best bet but it will require crossover tweaks. You will need to set the L-pad for the correct sensitivity match and also, you'll need to find the Lpad figures for the original unit so that total impedance seen by the filter can be checked against what you plan to do...it's a bit chicken and egg. Your main issue is the the HF1000 was supplied as either a 4 ohm or an 8 Ohm unit. The 27TFFC is a 6 Ohm unit. You can't match by looking at voicecoil DC resistance.

You might be better off looking for an Ebay pair of original replacement tweeters if you want a drop in replacement, otherwise you will have to rethink the crossover. That or just drop the SEAS unit in, and live with it, matching as close as you can, but you may loose phase accuracy and not achieve the crossover point you think it should be set at.

[DarrenHW]

Thanks for the response Paul, I feared it wouldn't be that simple. The only figure I can find relating to the sensitivity of the UL8's is 8.4w = 90db @ 1m taken from this leaflet:
15745
I think column 3 relates to this, based on the Ditton 66 brochure stating the sensitivity as "4.8 watts pink noise input produces 90dB SPL at 1 metre" the figure in column 3 for the 66's is 4.8.
I don't know whether this can be converted to the 2.83v/1m stated on the tweeter specs but I imagine looking at the figures the HD1000 is far less sensitive. I can't find any worthwhile information on the HD1000's. I'm reluctant to buy second hand replacements, I already have 2 repaired units so I'm guessing they're quite delicate and would ideally like to replace with a more robust tweeter.

[Reffc]

Depends on load Darren. 2.83V/1m equates to 1w/1m for an 8 Ohm load and 2w/1m for a 4 Ohm load. If your speakers are rated at 4 Ohms for the bass driver, then 4.8v/1m producing 90dB is close to 2w, so half that being 1w will produce 87dB/1w/1m (4 Ohm) or less (possibly 83 to 84dB) for an 8 ohm load. The reality is that it also depends on measured acoustic response in the cabinet not just theoretical driver specifications. If yours are 8 Ohm units (which will dip to 5 or 6 Ohms in the lower registers) then I'd expect more like 85 to 87dB 1w/1m, so the 19TFF would be suitable, possibly with very little damping. If yours are 4 Ohm units, then they'd still probably be ok with the 19TFF and without any damping needed.

[DarrenHW]

Thanks again Paul :thumbsup:. I like the idea that the 19TFF1 could be used although I am a little confused, the Free Air Resonance is quoted at 1700Hz, is this suitable for crossing at 2000Hz, earlier you recommended a resonance of 600-700Hz? I have 19TFF1's here so could mock this up easily but I have no idea how to establish how the crossover would need to be modified to accommodate this. I have used various online calculators to try and establish the existing configuration but I can't get any of the calculators to return results that match the values used. I had hoped if I could do this I would be able to use a calculator to spec component values for a replacement tweeter.

[Reffc]

Sorry Darren, yes if Fs is 1700 Hz they won't be suitable. The 27 TFFC would be as would the 27 TDC, 27TDFC or from Scanspeak, the Scanspeak D2604/830000, or Scanspeak R2604/832000. All are reasonably priced.

[DarrenHW]

Thanks again Paul, especially for taking the time to list suitable tweeters. I think I'm going to try the 27TDFC. The crossover for the tweeter appears to be Butterworth 3rd Order:
15766

According to the calculator if the 27TDFC measures 4.8ohm with a crossover point of 2000Hz C1=11.05, C2=33.16, L1=0.29, the current configuration is C1=16, C2=12 L1=Unknown. I realise this is only a ball park figure but ties in quite nicely with the spares I have so I won't need to order anything other than the tweeters, what do you think, do these values sound correct?

[Reffc]

It wont measure 4.8 Ohms Darren, as you're confusing voicecoil DC resistance with frequency dependant impedance. If you look at the factory spec sheet it's more like 5.5 to 5.7 ohms at 2KHz. The calculators only get you into a ballpark and wont be accurate, but if that's all you have, then use a figure of 5.5 Ohms and you'll be closer to the mark. Bear in mind that a calculated filter will not properly allow for the actual acoustic response of that tweeter and the fact you wont know what the original filter transfer function was designed around (it could have been skewed either way to account for the woofers rising response in order to generate a combined phase/acoustic accurate response). Still, have a bash and see what it sounds like if you have the parts.

[DarrenHW]

It wont measure 4.8 Ohms Darren, as you're confusing voicecoil DC resistance with frequency dependant impedance.

Yes I am :scratch: How's this for a simpler plan, record the frequency response of a stock UL8:
http://i1324.photobucket.com/albums/u606/drtwas/UL8%20150mm_zpsngt8bknw.png (http://s1324.photobucket.com/user/drtwas/media/UL8%20150mm_zpsngt8bknw.png.html)
install the new tweeter and manipulate the crossover values until the two charts match?

[Reffc]

Thats not a bad shout Darren but better if your did it for tweeter and woofer separately to measure individual response in room first, and that way you can see what the actual roll off slopes are, then try them together. You will of course be picking up all room reflections so this still wont achieve the accuracy needed but it's better than trying to calculate a filter. It won't tell you if correct phase response has been achieved but perhaps inverting the polarity of the woofer or the tweeter to see if you get a dip will be a help there. The bigger the dip with one out of phase, the closer the phase match when you re-invert.

[DarrenHW]

Thanks for sticking with me Paul!

Tweeter & woofer connected. Mic 150mm centred between tweeter and woofer:
http://i1324.photobucket.com/albums/u606/drtwas/UL8%20150mm%201_zpsv39puab3.png (http://s1324.photobucket.com/user/drtwas/media/UL8%20150mm%201_zpsv39puab3.png.html)

Tweeter only. Mic 150mm centred on tweeter:
http://i1324.photobucket.com/albums/u606/drtwas/UL8%20150mm%20Tweeter_zpsmjlv2gfd.png (http://s1324.photobucket.com/user/drtwas/media/UL8%20150mm%20Tweeter_zpsmjlv2gfd.png.html)

Woofer only. Mic 150mm centred on woofer:
http://i1324.photobucket.com/albums/u606/drtwas/UL8%20150mm%20Mid_zpskcsuqqkp.png (http://s1324.photobucket.com/user/drtwas/media/UL8%20150mm%20Mid_zpskcsuqqkp.png.html)

As stock the polarity of the tweeter is reversed, is it worth rewiring this and taking another measurement?

According to the Celestion documentation the crossover frequency is at 2k which I guess is the dip circled in blue, any idea what the dip circled in red is?
http://i1324.photobucket.com/albums/u606/drtwas/UL8%20150mm%201Exp_zpsjzx1321h.png (http://s1324.photobucket.com/user/drtwas/media/UL8%20150mm%201Exp_zpsjzx1321h.png.html)

[Reffc]

The best way to take the readings if in-room Darren is with one speaker positioned in the centre of the room and the mic at 1m from the speaker, level with the tweeter (which is where your ear level normally is when listening). I'd re-do the measurements for that. The dips could be due to room interaction or comb filtering caused by the cabinet design, so are likely null or cancellation points. However, the filter design might be partly responsible too and almost certainly is a significant factor for the first dip (read on...).

The tweeter's polarity being reversed as standard as with some 2 pole designs, indicates the designer thought it the best phase response using that polarity. What you can glean from the in room measurements is that the tweeter acoustic roll off with the standard filter is a steep 24dB/Octave between 1 and 2KHz and that the woofer starts rolling off at 1KHz with an approximate 18dB roll off (ie 3rd order acoustic) by 2KHz. This will give an uneven transfer function acoustically which is partly why you see that dip at approx 1.6KHz. The tweeter response is also far from smooth on axis with a notable rising output. I wont draw any firm conclusions from in room measurement though especially with the mic and speakers perhaps not ideally placed. Also, your measurements would have been taken slightly off axis. Whilst this wont have a significant impact on the result you see in your measurements, they would still benefit from being checked with just one speaker connected and situated as far from boundaries as you can get it, using a pink noise burst for measurement and gating the response if your software has that functionality.

[DarrenHW]

The best way to take the readings if in-room Darren is with one speaker positioned in the centre of the room and the mic at 1m from the speaker, level with the tweeter (which is where your ear level normally is when listening). I'd re-do the measurements for that. The dips could be due to room interaction or comb filtering caused by the cabinet design, so are likely null or cancellation points. However, the filter design might be partly responsible too and almost certainly is a significant factor for the first dip (read on...).

It will take me a couple of days but I can do that, I'll upload more charts once I have.



The tweeter's polarity being reversed as standard as with some 2 pole designs, indicates the designer thought it the best phase response using that polarity. What you can glean from the in room measurements is that the tweeter acoustic roll off with the standard filter is a steep 24dB/Octave between 1 and 2KHz and that the woofer starts rolling off at 1KHz with an approximate 18dB roll off (ie 3rd order acoustic) by 2KHz. This will give an uneven transfer function acoustically which is partly why you see that dip at approx 1.6KHz. The tweeter response is also far from smooth on axis with a notable rising output. I wont draw any firm conclusions from in room measurement though especially with the mic and speakers perhaps not ideally placed. Also, your measurements would have been taken slightly off axis. Whilst this wont have a significant impact on the result you see in your measurements, they would still benefit from being checked with just one speaker connected and situated as far from boundaries as you can get it, using a pink noise burst for measurement and gating the response if your software has that functionality.

I'm glad you agree the tweeter uses a 3rd order 18db crossover this was the only calculator that returned similar values to those used in the original crossover. Thanks again for all the help, I have more questions but I'll hold fire until I can upload more charts :).

[Reffc]

It will take me a couple of days but I can do that, I'll upload more charts once I have.




I'm glad you agree the tweeter uses a 3rd order 18db crossover this was the only calculator that returned similar values to those used in the original crossover. Thanks again for all the help, I have more questions but I'll hold fire until I can upload more charts :).

errrr...the tweeter response is acoustically 24dB/Octave (ie acoustic 4th order) judging by the in room response (the woofer looks like 3rd order acoustic). Once you can remeasure as suggested, and measure more on axis, that may well change the result and we can then better assess the filter requirements. Its nigh on impossible doing this by forum correspondence though! I'm really only trying to provide a few tips to point you in the right direction.

[tiguan]

Thread became a very usefull compact diy speaker guide and good lesson for me. Thank you Paul and Darren...

[DarrenHW]

errrr...the tweeter response is acoustically 24dB/Octave (ie acoustic 4th order) judging by the in room response (the woofer looks like 3rd order acoustic).

And there I was thinking I was getting my head around it :doh: The reason I thought the Tweeter was 3rd Order 18db is because it is wired like this:
15792
and the calculator puts the capacitor values in the same region as those fitted.
I thought the LF was 2ND order 12db because it is wired like this:
15793
and the calculator also puts the capacitor values in the same region as those fitted.
I guess it's not that simple?


Its nigh on impossible doing this by forum correspondence though! I'm really only trying to provide a few tips to point you in the right direction.

I appreciate that and also appreciate the help although I guess you may feel like your banging your head against a brick wall. I've taken some more measurements as you suggested, please don't feel obliged to reply if you don't want to :).

Some new measurements all taken at 1m in the most open space I have.

UL8 Tweeter & Woofer (Tweeter Reversed).
http://i1324.photobucket.com/albums/u606/drtwas/UL8%201m_zpsor5gd0mj.png (http://s1324.photobucket.com/user/drtwas/media/UL8%201m_zpsor5gd0mj.png.html)

UL8 Tweeter & Woofer (Tweeter not Reversed).
http://i1324.photobucket.com/albums/u606/drtwas/UL8%201m%20Rev_zps5qkjpjhb.png (http://s1324.photobucket.com/user/drtwas/media/UL8%201m%20Rev_zps5qkjpjhb.png.html)

UL8 Woofer.
http://i1324.photobucket.com/albums/u606/drtwas/UL8%201m%20Mid_zpspcmsvopl.png (http://s1324.photobucket.com/user/drtwas/media/UL8%201m%20Mid_zpspcmsvopl.png.html)

UL8 Tweeter.
http://i1324.photobucket.com/albums/u606/drtwas/UL8%201m%20Tweeter_zpsivh55uip.png (http://s1324.photobucket.com/user/drtwas/media/UL8%201m%20Tweeter_zpsivh55uip.png.html)

[Reffc]

The woofer measured from listening height exhibits closer to 2nd order acoustic (as expected)and the tweeter a steep 4th order still. As mentioned, 3rd order electrical does not always equate to the same acoustic...it depends on the drive unit and the crossover point. In this case it's because the tweeter is being crossed off low towards its natural roll off and that is steepening the filter transfer function, hence 4th order acoustic.

If you can super-impose the individual measurements on the same graph, that will tell you a little more, but it looks like no obvious 2KHz dip and it's difficult to establish if the crossover is phase accurate but I suspect not from what I see. (This is a crude attempt to look at what's going on when a more refined approach is really needed) and that is as a result of the individual driver phase at crossover point. You can also see the effects of comb filtering with the tweeter not reversed to its "normal polarity".

The easiest approach if you want to match how the speakers came from the factory is simply to get a replacement tweeter to match the tweeter curve you have, making very sure to level match, so take the tweeter measurement at a set level, leave the test gear in place and swap tweeters, tinkering with electrical HF filter until you match the "master" curve for the old tweeter at crossover. That assumes that there are no clapped out electrolytics in the original filter to start wih. Personally, I'd re-design it using a low self resonance tweeter of 500Hz or so and try to get a 2nd order acoustic at 2KHz for better phase matching. The filter values will not be calculable because if a flat baffle, time alignment within the filter is also required.

Just trying not to over complicate things for you, so have some fun with it and it would probably help you to do a little research to help you understand things like phase relationships.

This will be a useful link for you and I'd recommend it as good reading material:

http://sound.westhost.com/lr-passive.htm

Can't do too much more for you without doing the work myself (ie having the speakers here) but you'd not learn anything that way Darren, so once again, have a read of the above link and associated topics and just treat it as a bit of fun and learning and best of luck with it.

[DarrenHW]

The woofer measured from listening height exhibits closer to 2nd order acoustic (as expected)and the tweeter a steep 4th order still. As mentioned, 3rd order electrical does not always equate to the same acoustic...it depends on the drive unit and the crossover point. In this case it's because the tweeter is being crossed off low towards its natural roll off and that is steepening the filter transfer function, hence 4th order acoustic.

If you can super-impose the individual measurements on the same graph, that will tell you a little more, but it looks like no obvious 2KHz dip and it's difficult to establish if the crossover is phase accurate but I suspect not from what I see. (This is a crude attempt to look at what's going on when a more refined approach is really needed) and that is as a result of the individual driver phase at crossover point. You can also see the effects of comb filtering with the tweeter not reversed to its "normal polarity".

Well I'm glad it makes sense to you :scratch::scratch:! Unfortunately I cannot super-impose the individual measurements, I could manually enter the values and create a graph but this would be a very simplified representation.


The easiest approach if you want to match how the speakers came from the factory is simply to get a replacement tweeter to match the tweeter curve you have, making very sure to level match, so take the tweeter measurement at a set level, leave the test gear in place and swap tweeters, tinkering with electrical HF filter until you match the "master" curve for the old tweeter at crossover. That assumes that there are no clapped out electrolytics in the original filter to start wih.

Yes, I'm going to stick with this plan and no there are no clapped out lytics in the filter, I have recently recapped (like for like values) with Alcaps and have a stock of Alcaps to use modifying the crossover for the new tweeter.


Personally, I'd re-design it using a low self resonance tweeter of 500Hz or so and try to get a 2nd order acoustic at 2KHz for better phase matching. The filter values will not be calculable because if a flat baffle, time alignment within the filter is also required.

I'm pretty much set on the 27TDFC (Free Air Resonance 550 Hz, Recommended Frequency Range 1500Hz - 25000Hz) it seems to have a similar impedance to the HF1000 (based on assumption) and uses ferro fluid as does the HF1000, so (as read in Rod Elliott's article) negating the need for a Zobel Network, "Note that tweeters using ferro-fluid in the voicecoil gap will usually be very well damped, and may show little or no resonant peak. This means that no compensation circuit is needed."


Just trying not to over complicate things for you, so have some fun with it and it would probably help you to do a little research to help you understand things like phase relationships.

This will be a useful link for you and I'd recommend it as good reading material:

http://sound.westhost.com/lr-passive.htm

Can't do too much more for you without doing the work myself (ie having the speakers here) but you'd not learn anything that way Darren, so once again, have a read of the above link and associated topics and just treat it as a bit of fun and learning and best of luck with it.

Thanks for the link, I've read through it and although this has lead to further reading in an attempt to understand some of the principles covered it has confirmed some things I've heard but struggled to accept, things I've heard but seem to be juxtaposed with forum consensus and dispelled some forum based misinformation. Pretty much every project I undertake is a leap of faith based on internet reading so it's comforting to read articles in line with my own listening experience.

Yes, this project is intended to be an enjoyable educational experience, I'm a better (but a very long way from good) student with working examples. The UL8's are not my ultimate speaker choice but a speaker that fits into the space limitations I currently have to work with and integrate well as surround speakers with the Celestions I use as front speakers.

[DarrenHW]

2 months on and I think I can see the light at the end of the tunnel, I've not had much free time lately and this combined with the break-in time of the tweeters (100+ hours) has made this drag on a bit. I decided to go with the Seas 27TFF H0831-06 tweeter (http://www.falconacoustics.co.uk/downloads/Seas/h0831_datasheet.pdf) and stuck with the original 3rd order arrangement with amended values:

http://i1324.photobucket.com/albums/u606/drtwas/UL8%20xo_zpsid6ojcar.jpg (http://s1324.photobucket.com/user/drtwas/media/UL8%20xo_zpsid6ojcar.jpg.html)

C1: 10uF.
C2: 10uF + 16uF (+ 1uF Polypropylene).
L1: 0.30mH.
R1: 1R8.
R2: 27R.

I used calculators (using 5.6ohm for the tweeter) which suggested C1 = 9.47, C2 = 28.42, L1 = 0.33mH so I started at this point and fine tuned with the speakers in position and listened from my regular listening position. The only value (other than resistors) I have swapped is C2 and found 27uF to yield the best results, using a higher value the tweeter became tinny, lower the tweeter lost its sparkle. The resistor values have been established using resistors I had to hand swapping until I felt the balance between woofer and tweeter was right. I don't know whether this is the best arrangement for the resistors and would appreciate any feedback.

Initially I worked on only one speaker leaving the other stock so I could compare with the HF1000 to ensure I was satisfied the crossover was in the ballpark without detrimentally affecting the sound of the speaker. When I started to get close in terms of capacitor and resistor values it was becoming clear the 27TFF was far superior to the HF1000, not only does it have a more natural sound, it is more detailed and has a wider dispersal.

As speakers in their own right I think the 27TFF represents a noticeable upgrade, but in the AV configuration I have them in the upgrade is significant. The wider dispersal of the tweeters combined with the extra detail really extends and expands the surround illusion creating a far bigger sound stage with a better centre image. Integration with the main speakers is also very much improved I put this down to the tweeters being from the same family, the main speakers using Seas 19TFF. I'm hoping by using the same wire and capacitors in the UL8's as the main speakers will result in even better integration.

If anyone has any comments or suggestions please share them.

[Reffc]

Well done Darren. It seems you've come out the other side with a real sense of achievement, well earned!

[DarrenHW]

Thanks Paul and thanks for all your guidance, I'm happy with the way they're turning out. What do you think to the resistors, is there a better way to do this?

[Reffc]

You're probably not a million miles out Darren. The function of an L-pad is two-fold. Firstly it's to reduce tweeter sensitivity to match with the mid woofer, secondly, it's to maintain filter impedance loading as a constant. If the tweeter is around 5.6 Ohms and and you've added 1.8 series and 27R parallel, this equates to an HF load at crossover of 6.4 Ohms, so the filter should be designed around a load of 6.4 ohms at the desired crossover point. If this is adrift by half an ohm or so of the original crossover load at the acoustic crossover point, it will shift the crossover point and likely introduce phase inaccuracies.

The UL8 was meant to be around 90dB sensitive and that tweeter in raw form gives 91dB, so for a linear response you should only require a 1dB attenuation, perhaps 2 to 3dB maxdB for a gentle LF to HF slope. If the original tweeter measured between 5 and 5.5R at crossover, take the original L-pad values to work out total filter load at crossover (or supply me with the original L-pad values and I'll work up the nearest equivalent L-pad for the 27tFFC for you).

To maintain a constant filter load of say 5.6R, the required L-pad for each attenuation step is as follows (using nearest standard value resistors):

1dB = .56R/47R
2dB = 1R/22R
3dB = 1.6R/13R

[DarrenHW]

Thanks again for the input Paul :). There was no L-pad in the original UL8 crossover (thanks for the offer), I'll pick up some more resistors of the values you suggest and give them a go.