6dB filter really bad for high Q tweeters

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I got some vintage Philips AD0160T4 and T8 and as I got them with 6 dB filter 5 & 12µF for 8 and 4 ohm respectively.
AD0160T.jpg


Impedance curves suggest good pair matching and high Qt.
Ad0160impedances.png


Frequency response is quite good with a gentle slope upward to 17 kHz and then a sharp drop. This is in agreement with WWDT
AD160T8abT4ab.jpg


The 6 dB filter looks like this
AD160T8-5micoCapBadIdea.jpg


Normalized it is - 6dB at 2.5 kHz and -12 dB at 1.2 kHz but then it falls of the track and is only about 3 dB down at 800 Hz. No surprise that it sound strained and harsh at moderatly loud levels. This suggests that high Q drivers really need higher order crossovers.

CapNormalised.jpg
 
Adding a 47 ohm or even a 10 ohm resistor in parallell does help some, but I think that even crossed over at 8-10 kHz, one would be better of with a higher order filter. An option would be to use hybrid filter with larger coil so it is like a first order filter for one or two octaves and then start to increase the slope.


normalised47Rand10R.jpg
 
speaking from my own impedance curves of woofers i have, or owned at some point, i wouldnt say those tweeters were a very good pair match.

Out of curiosity what were the Qt values? I guess even 10% tolerance is ok considering their age.

Another musing: would the addition of a notch to the 1st order xo help? Cauer filter? Or is that 2nd order and above? Ive never used a zobel, would that help?
 
Last edited:
If you look to the left on the impedance chart Qts= 1.5 and Qms=12 for one of them, the other ones has similar values with Qts in the 1.5 range. The two 8 ohms are almost overlapping, the 4 ohms are spread a bit. I have some Seas 10" woofers from the 70s and they overlapping all 8 of them. I ams still impressed that 30-40 years later that such complex mixes of many materials as plastics still work as specified.

I would use filters to keep the amplitude at resonanse way down. Traps and Eq can only tame amplitude errors not time domain errors.

A zobel will not help at all.
 
Member
Joined 2010
Paid Member
The parallel 1st order and the 3rd order crossovers have the same inherent problem;the series capacitor removes damping for the tweeter,where it is needed; unless you have a tweeter with excess efficiency and can fit a 6db or higher L attenuator.It is the "looking back" impedance that is important.For an accurate measurement,using swept frequency,the other leg of the crossover should be in cct loaded with a resistor,to simulate actual operating conditions.A better crossover,if you have suitable drivers is the series 1st order and its variant the quasi 2nd order.
 
Troels really does not think that the tweeter is bad despite the heading of the worst tweeter ever. Distortion is low and response up to 17 kHz is good.

Sure ferrofluids or large back chambers with resisitive damping makes the crossover work simpler. VaNarn I will try a even order crossover perhaps L-R second order as a start.
 
Member
Joined 2010
Paid Member
Considering their vintage in the light of what was available at the time,the Philips dome was a reasonable choice.Yes it has more colouration than the early cloth domes such as the Goodmans DT3 and the Peerless DT 10 HFC,but this should be kept in perspective considering its comparitive cost.It might be of some interest if the experiment could be continued in stages; with a 0.32mH coil and an 8 Ohm resistor for a l.f load in the sweep together with the 5 mFd cap and Philips dome.Then followed with the same components in a series arrangement.Both of these arrangements should give a xo freq of approx 3980 Hz.If you have surplus xover components handy, such as a 0.45mh inductor and a 3.5 mFd cap. then the quasi 2nd order xover could be part of the comparison.
 
I think the issue with this or any tweeter is that if you do not get the energy out of the tweeter below the lower rolloff point you end up with increasing IM and THD as power goes up due to the excess LF excursion.

If I remember properly (doubtful) the first order filter will merely flatten the excursion of the driver below the rolloff point, making it more or less constant as frequency goes lower. This is not a very good choice, since I don't think you want the little dome dishing in and out at bass or even midbass frequencies at all.

The Cauer or Eliptical filter suggestion has some merit, but the implementation will be tricky, in part due to the impedance rise that is there around LF rolloff and also the non flat passband of the filter. Carefully done, it might work fine. Of course, I'd not suggest a trying to get a first order out of such a filter. The dip is very high slope, and the remaining rolloff while down a reasonable number of dB, in a first order implementation (which is something of a kludge arrangement, I expect) it seems to me would potentially have some of the same problems as mentioned at LF with the standard first order filter, as the levels at very LF may not be all that much different... you'd have to run a simulation and see where you end up.

I used a number of these Phillips tweeters back in the day, and they were pretty nice when rolled off more like 2.5-3kHz... and the ones that were original (arguably better) did not have the little black "phase cap". The ones with the phase cap were later models, and I think they went heavier and cheaper with the domes... maybe even they were polycarbonate, and not mylar. Unsure.

Just guessing about the way to do it, based on only the curves, I think I'd try to adjust the Q of the xover for a slight bump and try to set the xover about 2kHz, this would flatten the passband a bit... the rise may be due to the phase cap... and I generally have not heard the phase cap sound quite right... but I think I'd opt for a very small low DCR inductor in series with the tweeter to flatten the rise, just enough... maybe make the coil start to have effect ~10kHz, and drop 3dB by 20kHz, that would give a flat response, you'd have to dork the coil value carefully, and actually a slightly lower freq start, larger coil is likely to have a better result to the ears...

A few cents worth of thoughts dredged off the top of the head...

_-_-

Btw... it's slightly unclear from the OP's discussion if the intent is to run the tweeter only from say 8kHz up??
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.