Ah, are you thinking of the Nathan ? Unfortunately for me, $4600/pair. Oh well...
I did notice it has its crossover at 650 Hz. Requires a pretty special tweeter. Yes, I'll eat my words there.
Listening to just one driver with the crossover in place? Yes. Listened to the woofers through the crossover. I didn't hear anything horrible. I think you mean without the crossover in the way, though. I'll try that. That might tell us something about the upper end of the woofer's response, right?
How would I listen to the tweeter without the crossover? I wouldn't want to do that, I don't think.
--
The Nathan? Yes that's one example. 650Hz, are you sure? I'd have thought differently. Anyway, I am currently using a similar design. I use a 15" mid/woofer and I have run it as high as 1200Hz with impressive results.
I didn't expect you to run your drivers without their respective crossover. I'm sure you could glean more information out of the by doing so, but my point was simply to help you identify whether either one was giving problems on its own so you could target the biggest problem first. I feel that sometimes it can be more difficult to fix a system as a whole unless you have attended to the parts first.
I didn't expect you to run your drivers without their respective crossover. I'm sure you could glean more information out of the by doing so, but my point was simply to help you identify whether either one was giving problems on its own so you could target the biggest problem first. I feel that sometimes it can be more difficult to fix a system as a whole unless you have attended to the parts first.
try a series resistor between xo and driver(tweeter)
AllenB said:
Just did it. Put in 2 ohms.
It made the highs sound a bit softer, yes. I think there's a peak left in the 15kHz area, like how a kid pins the 15kHz band when they get their first graphic EQ to play with.
I'll try 1.5 ohms to see if I can hear a difference.
A 100 ohm resistor in parallel with the tweeter had a different effect. It made the highs sound softer, but also a bit rolled off. Probably means it was more accurate, but I'd need an accurate measuring microphone to know, which I definitely lack.
I wonder... L-pad? Rp of 100 ohms, Rs of 1.5 ohms?
Well, it's 9pm here, which means no more listening tests. The downstairs neighbors go to sleep early.
I'll put in the 1.5 ohm series resistors...
--
Last edited:
Rongon: have you tried a felt or fiberglass pad directly behind the diaphragm? Similar to the type commonly found in dome tweeters. I use such pads in all my comp drivers. I am going to make up some new ones using stacked layers of 1/16 inch wool felt. Give it a try and see what you think. While you are thinking is there any chance that you are actually hearing some component or cable in your system and not the speaker? Have you tried the speakers out in a known smooth system which does not have this issue? Are you sure this is the horn and not the woofer, your crossover caps are getting old and they are not sota parts by any stretch of the imagination something to consider. Just a thought. I have a number of different Klipsch speakers all with Ti diaphragms throughout and they are all smooth sounding to me. Best regards Moray James.
Last edited:
or bigger
now that I looked at it(xo), and thought about it
tweeter is not 18db, tho it appears so
two small and almost equal caps with a big inductor/resistor in between
I would consider it a 6db, with two caps in series, and a fiddly EQ in between, by the inductor/resistor
in light of this, I would suggest to try and reverse polarity on tweeter
Hi Moray,
Stacked layers of 1/16" wool felt behind the driver in the tweeter? That sounds interesting. Any place to point me to for more info? I'd certainly give them a try. Wouldn't that be funny if I've been hearing a sort of ringing in the tweeter all this time?
Still not sure if its the woofer or the tweeter. I suspect the tweeter, because putting a Zobel on the woofer only scooped out the mids somewhere kind of low (under 1kHz?). But putting resistors in various places on the tweeter makes noticeable changes in sibilance and whatnot.
The thought has occurred to me (and to tinitus) that my triode amps are not a good match for a Klipsch speaker. I wonder if they really like big, powerful solid state amps better? What are you using?
My system isn't SOTA by any means. But it's not basura either.
The CD player is a Pioneer Elite PD-D6J.
Some standard Belden/Switchcraft interconnects go from there to...
Niles Audio selector switch in a box, Intact Audio inductive attenuator (which is bright sounding, yes, but very smooth).
Old Radio Shack made in USA interconnects go from the line amp to the power amp. (Yes, I've tried other cables. These actually sound pretty smooth.)
The line amp is a common cathode 6N30P-EV. That drives a push-pull 2A3 amp, with split-choke loaded 6N6P LTP phase splitter/driver with CCS cathode load, DC-coupled to the finals. Tango OPT's (very deluxe). No global negative feedback, so I suspect the amp has a high-ish output resistance. But it's push-pull, so not like a little SE amp.
Speaker cables are doubled up 18 gauge solid core hookup wire, terminated in Johnson banana plugs.
I lugged my line and power amps over to a friend's place to play them through his Jordan arrays. The 2A3 amp held its own against his SE 50 amps with amorphous core OPT's. I'd say ever so marginally "harder" sounding than the SE 50 amps.
But no, I haven't lugged the speakers over to anybody's house. They're way too big and heavy for doing that on the spur of the moment.
BTW, the Crites titanium diaphragms are a big improvement over the original polymer diaphragms.
I have 2.2uF WIMA MKP-10 and a couple of 3.3uF metallized polypropylene caps, but they're huge compared to the stock parts. The stock crossover boards are impossibly tiny. I'd have to make a new crossover board to mount these larger parts. Maybe I could look up these values of Panasonic metallized mylar caps and see what their dimensions are. Maybe those would fit on the original board.
It's been suggested to me to bypass the bi-polar 33uF with a small value poly cap. I could try that too.
--
Which 2 ohm resistor? The one in series with the 33uF cap (in parallel with the woofer)? Or the 2 ohm series resistor I just put on the tweeter?
Thanks for your help with this. As I experiment, it might come to pass that your idea will sound the best. Hopefully, I'll know what's best when I hear it...
--
You mentioned a triode amp earlier. Not sure that it would be causing these issues. Considering that a higher output impedance will impress the shape of the impedance curve onto your frequency response, you might try something else.
You also mentioned the Williamson. I once built one, with 6SN7s and GEC KT66s. This was a very good amp, amongst the best.
You also mentioned the Williamson. I once built one, with 6SN7s and GEC KT66s. This was a very good amp, amongst the best.
The 2 ohm on the tweeter attenuates the tweeter. If it doesn't attenuate enough, put a 5 ohm, or 4 ohm. Listen again. A bigger resistor gives a more attenuated tweeter. Put the 10 ohm with the 33 uF for the woofer. Read my previous posts. All info is there to provide you with an excellent result.
I did a Spice simulation on the crossover just to have an idea of what the transfer function looks like. For convenience, both drivers are at 8 ohms.
Crossover is about 2KHz (12dB/oct on woofer, 18dB/oct on tweeter). We can see the effects of the "tuning" resistors. On the woofer, it created a bump at about 800Hz. The opposite is true for the tweeter. The 1mH+12 ohm pulled the tweeter down before roll-off.
My guess is acoustically, the woofer is rolling off early leading to the vocal receding into the background. The extra lift brings the voice slightly forward.
The horn response, on the other hand, is not flat. Perhaps like a CD type. By pulling it down, it flattens the response.
How they sum acoustically is impossible to tell unless a FR sweep is done. However, we know that it's difficult for a 10" to reach up to 2KHz. I would speculate that the vocals, particularly male voices, will have difficulty.
With regards to the sibilance and sizzle, that could be attributed to some peakiness in the tweeter response.
Again, I am simply thinking aloud. I could be completely off the mark.
Crossover is about 2KHz (12dB/oct on woofer, 18dB/oct on tweeter). We can see the effects of the "tuning" resistors. On the woofer, it created a bump at about 800Hz. The opposite is true for the tweeter. The 1mH+12 ohm pulled the tweeter down before roll-off.
My guess is acoustically, the woofer is rolling off early leading to the vocal receding into the background. The extra lift brings the voice slightly forward.
The horn response, on the other hand, is not flat. Perhaps like a CD type. By pulling it down, it flattens the response.
How they sum acoustically is impossible to tell unless a FR sweep is done. However, we know that it's difficult for a 10" to reach up to 2KHz. I would speculate that the vocals, particularly male voices, will have difficulty.
With regards to the sibilance and sizzle, that could be attributed to some peakiness in the tweeter response.
Again, I am simply thinking aloud. I could be completely off the mark.
Attachments
I've been thinking that too.
I have a different amp, built into an old Dyna ST70 chassis, with a simple 6SN7 LTP phase splitter/driver. It's running EL34's in triode now. I could up the global NFB in that one and use it with the Klipsch speakers, as a test. I have plenty of gain from the line stage to overcome the reduced sensitivity. I think I'll try that tomorrow. I'd like to know if the KG4.5's sound completely different when driven by an amp with a lower output Z, and I don't have a working solid state amp here at the moment.
Speaking of Williamson amps, I have a small stash of good transformers, including a pair of UTC LS-63, which were designed specifically for that kind of amp (10k anode-to-anode primary for triode-wired KT66's).
Did you build your Williamson with all that global negative feedback around it, like in the original? The one problem with that design is marginal stability, because of the global NFB around four stages, with the two RC couplings in there, plus the OPT.
But I digress...
--
Wow, Michael. Thank you. (I have to get motivated to learn LTSpice!) BTW, the Re on the woofer is 7.2 ohms and the tweeter is 8.1 ohms. Does that mean the modeling with 8 ohms should be pretty close?
I'll bet the tweeter is being used about as low as it can go before getting really honky. The tractrix horn is probably not designed to respond down to 800 Hz. Its opening is roughly 185mm W x 110mm H (7-1/4" x 4-3/8"), if that's any help.
So the woofer's upper end is boosted to make up for that. I can see why putting the Zobel on the woofer only made things worse. If you diminish that electrical 800 Hz boost, the mids recede right where we're very sensitive to them.
The speakers are kind of "scooped." It makes them sound "warm" except for the aggressive nature of the upper mids from the horn. Not unpleasant.
This speaker has a tractrix horn. I read up a little about CD horns and the need for response shaping (high freq boost necessary about 15kHz?). But couldn't find anything similar about tractrix horns.
Considering the approx. 2kHz crossover point, it sounds fairly acceptable on male voice. Those Klipsch engineers were pretty sly.
I'm sure there's peakiness in there. It's a really cheap looking, plastic horn.
I appreciate you sharing your thoughts on this. Fuel for more experimenting...
--
Had a feeling
I was even thinking you might add series resistance to exacerbate the problem and prove the point, but your idea sounds best. If you're keen you could always perform true Zobel correction to the speaker system as a whole (before the crossover) to overcome your issue, though maybe only above the lower midrange region?OK, I used a substitute transformer but I did conform to the common method of making that work.
By the way. I use a JLH solid state (Class A from 1969) using modern transistors, and the similarities to the Williamson are far greater than the differences.
What Michael has done here is the best estimative tool so far on the thread (and I was doing it behind the scenes as well), but a speaker is not resistive. This will be reasonably accurate in showing differences between two cases, but even less accurate at showing the absolute case, so the exact value is not really necessary.
By the way, the specs for this say the crossover is at 1k6Hz, for whatever that may mean.
The tractrix narrows its beam as frequency increases, so as the power in the highs drops, it is concentrated and appears correct on-axis.
The bump around 600 or 800 Hz has nothing to do with the 2 ohm resistor. The bump at 15 kHz has nothing to do with the 12 ohm resistor. The efficiency of the horn in the simulation is wrong. You guys are leading the thread owner off into confusion. You are just guessing and presenting it as facts. You are wrong.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.