I have recently setup a test jig so that I can impedence sweep loudspeakers, and apply both RC compensation, and series RLC driver resonance and confirm the results.
I am finding the results to extremely worthwhile the effort on the tweeters of my old (ca1977) KEF 8" 2ways.
Sonic results are dramatically cleaner mids and highs, and extension of highs.
Instrument timbres and vocal timbres are much more accurate than before - I know the live performance and conversation spoken voices of some of the musos on my test records.
Anybody have similar experiences and results ?.
And what soundcard software are you using ?.
Eric.
I am finding the results to extremely worthwhile the effort on the tweeters of my old (ca1977) KEF 8" 2ways.
Sonic results are dramatically cleaner mids and highs, and extension of highs.
Instrument timbres and vocal timbres are much more accurate than before - I know the live performance and conversation spoken voices of some of the musos on my test records.
Anybody have similar experiences and results ?.
And what soundcard software are you using ?.
Eric.
Hi mrfeedback,
Are you talking about Zobel newtwork ?
http://users.ece.gatech.edu/~mleach/ece4445/downloads/zobel.pdf
http://www.trueaudio.com/st_zobel.htm
http://www.bonavolta.ch/hobby/en/audio/filters.htm
Rémy
Are you talking about Zobel newtwork ?
http://users.ece.gatech.edu/~mleach/ece4445/downloads/zobel.pdf
http://www.trueaudio.com/st_zobel.htm
http://www.bonavolta.ch/hobby/en/audio/filters.htm
Rémy
Hi Remy,
Yes I am talking about series LCR shunt network to cancel the mechanical resonance impedence peak, and also shunt RC network to cancel the inductive rising impedence characteristic, as discussed in the 'zobel.pdf' link that you give.
I had not considered that the RC network should be four components - thanks for the link.
I am getting a fairly close to flat impedence characteristic for the tweeter, and this is removes audible tweeter resonance effects very effectively.
Have you explored this /.
Eric.
Yes I am talking about series LCR shunt network to cancel the mechanical resonance impedence peak, and also shunt RC network to cancel the inductive rising impedence characteristic, as discussed in the 'zobel.pdf' link that you give.
I had not considered that the RC network should be four components - thanks for the link.
I am getting a fairly close to flat impedence characteristic for the tweeter, and this is removes audible tweeter resonance effects very effectively.
Have you explored this /.
Eric.
It´s a Bandaid. If you´re really doing a crossover optimization (and assuming you have a low-Z output amp), the parallel network may or may not prove useful. It´s most useful in cases where someone can measure Z, but doesn´t want to do anything with xover optimization and is trying a textbook xover. Real Men use target functions appropriate for the driver/geometry/polar pattern and use whatever topology it takes to hit them.
Polishing Turds.....
Hi Sy,
My purpose here is to improve existing speakers without major rework.
In this case, the result is very nicely successful, for minimum cost and time.
In this case I am sweeping the driver to determine its impedence characteristic, applying the networks and confirming the combined impedence curve - the result is measuring close to flat.
The sonic result of this is dramatically lowered overtones and hash from the tweeter.
Can you elaborate some on the 'real men' bit ?.
Eric.
Hi Sy,
My purpose here is to improve existing speakers without major rework.
In this case, the result is very nicely successful, for minimum cost and time.
In this case I am sweeping the driver to determine its impedence characteristic, applying the networks and confirming the combined impedence curve - the result is measuring close to flat.
The sonic result of this is dramatically lowered overtones and hash from the tweeter.
Can you elaborate some on the 'real men' bit ?.
Eric.
If you´re just adding them to an existing design, you´ll noticeably change the frequency response. Whether that´s a good thing or a bad thing is a random event, depending on the initial design of the speaker (if you did this to the woofer of of an NHT Super Zero, you´d definitely make the response more colored) and what you think is "better" (a lot of people really like really colored speakers).
Generalizing from the particular is a classic error. That´s the origin of the "real men" comment.
Generalizing from the particular is a classic error. That´s the origin of the "real men" comment.
SY said:
are you sure that zobels and RLCs change the frequency response?
I've been told that, since they are in parallel with the speaker, they both see the same voltage, so the frequency response is unchanged
the only thing that changes is the current drawn from the impedance compensation network, to maintain the total drawn current constant over the whole frequency range-> the amp sees a resistive load, and is happier
Rather Good Now.
I have added impedence comp and resonance comp across the tweeter.
FR is not particularly changed, but resonance behaviour of the tweeter is very noticeably changed.
Hard essing on vocals and cymbals is gone, and highs are much clearer and cleaner.
This KEF speaker is a 25yo design, and these tweeters have never been wonderful, but this mod improves them to being very acceptable, and well worth the effort and junkbox parts.
Eric.
I have not bothered to draw the schematic, but on first inspection looks like 12dB with impedence comp across the woofer.Bricolo said:
I have added impedence comp and resonance comp across the tweeter.
FR is not particularly changed, but resonance behaviour of the tweeter is very noticeably changed.
Hard essing on vocals and cymbals is gone, and highs are much clearer and cleaner.
This KEF speaker is a 25yo design, and these tweeters have never been wonderful, but this mod improves them to being very acceptable, and well worth the effort and junkbox parts.
Eric.
In A Nutshell
Hi Graham,
I have been using a program called RMAA, which is a stereo audio testing program.
In FR test mode, I feed soundcard line out into an amplifier.
The Active output of the amp drives the tweeter (woofer, box) under test, and 10 ohms to ground.
The other channel is two 10 ohms in series.
Soundcard line in is connected to the junction of the tweeter/10 ohms one channel, and the junction of the two resistors the other channel.
By applying the FR test, the driver under test can be compared to the reference 10 ohms resistor on the other channel.
This gives an upsidedown graph, but no matter.In A Nutshell
So far, this has worked well enough for me, and the acoustic results good.
Sorry that is all the time for now - I'll be back later tonight.
Regards, Eric.
Hi Graham,
I have been using a program called RMAA, which is a stereo audio testing program.
In FR test mode, I feed soundcard line out into an amplifier.
The Active output of the amp drives the tweeter (woofer, box) under test, and 10 ohms to ground.
The other channel is two 10 ohms in series.
Soundcard line in is connected to the junction of the tweeter/10 ohms one channel, and the junction of the two resistors the other channel.
By applying the FR test, the driver under test can be compared to the reference 10 ohms resistor on the other channel.
This gives an upsidedown graph, but no matter.In A Nutshell
So far, this has worked well enough for me, and the acoustic results good.
Sorry that is all the time for now - I'll be back later tonight.
Regards, Eric.
Sy is correct in that a crossover that has been optimised for non compensated drivers will need tweaking to bring the overall response back to correct.SY said:
I say that this ought to provide a smoother and nicer sonic result than using non-compensated drivers, and this is my experience.
Eric.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.