After many different loudspeakers built, my experience and more importantly, experience of many listeners is this:
If there is a big phase angle combined with lowish impedance at some frequency region, flattening the phase (by adding RLC compensation) around that region undoubtedly improves the sound. Consequence of this action is flattening the impedance, as a byproduct.
If there is acceptable (for the amplifier) phase swing, than adding RLC just to make impedance flat, in the most cases leads to mushy sound with less definition. How would I explain this? Easy - you are adding one unnecessary oscillator circuit (RLC circuit is exactly that) in the game.
Of course, there are a number of examples between these two extrema.
Last edited:
OK, noted. But they are still anecdotal, and I have shown previously that easily 100 million or more people can all believe the same thing yet can all be wrong.
Jan
You mean flattening the phase seen by the amp, right?
Then you must have one hell of an incompetently designed amp. Because the signal across the speaker doesn't change.
Or if you do something different, can you provide a schematic?
Jan
Yes, phase seen by the amp. It is not a problem of an incompetently designed amp, but a problem of an incompetently designed loudspeaker. Just see the measurements on many expensive "high end" loudspeakers in the Stereophile reviews. Phase angle of over 70 degrees?! No amplifier will be happy with that.
But hearing something is where potential audio improvements usually start. Imagine where we would be if people dissing early PCM hadn't led to detection and improvement of jitter.
dave
So it is the amp that acts differently whether the network is there or not. The cause may be the weird load of the speaker, but I maintain that a well-designed amp will not be impacted by this. There are exceptions - in another thread they discussed a Wilson speaker that can be considered an incompetent design, but those are exceptions.
What all this boils down to is that if Joe or anyone heard an objective difference with or without the network (and it's a big if) it should change with changing the amp.
Jan
Yes agree, but there are many cases where differences are imaginary and that lead to nothing more than a waste of time and money. Any half way serious researcher will first make sure that indeed there is a difference.
Jan
"The plural of anecdote is not data."
Just because "many around the world" claim to have been kidnapped by space aliens and anally probed doesn't make that true or even plausible.
I'll say it again: there is NO data to support this dubious claim. None. Zero. Zip. Zilch. Nichts. Rien. One more brick in the wall.
Just because "many around the world" claim to have been kidnapped by space aliens and anally probed doesn't make that true or even plausible.
I'll say it again: there is NO data to support this dubious claim. None. Zero. Zip. Zilch. Nichts. Rien. One more brick in the wall.
I've played with a compensation network on my crossover-less arrays. Just out of curiosity. I was warned before I started it wouldn't make a difference. I compensated for the impedance peak (I play the arrays trough that peak using EQ at the low end) and also the impedance rise at higher frequencies.
After several listening sessions I removed the compensation for the rising impedance. Without the compensation I experienced cleaner sound. Highly subjective I know. But somehow that compensation network for the impedance peak won me over. Even though I never found a reason for it in my plots. I correct the timing with FIR filters so that's not going to have an effect.
The one thing I did notice is when I measured the impedance in REW, and looked at the group delay tab of the impedance measurement plot there were noticeable differences.
Guess which one has the compensation, this was taken with the complete conjugate network vs no network. This was measured with a laptop soundcard and a resistor in REW, no separate amp.
A comparison of both impedance plots:
Making it easy to spot the compensated plot above.
To this day the impedance compensation for the peak has stayed inline, although I cannot explain why I like it, I just prefer it. No matter how hard I looked I could not find a clear answer in the raw measurements. Except for the above shown difference.
*The restlessness in one graph at 50 Hz is from the laptop power supply. That much I do now.*
Impedance looks like this with only the peak compensated:
After several listening sessions I removed the compensation for the rising impedance. Without the compensation I experienced cleaner sound. Highly subjective I know. But somehow that compensation network for the impedance peak won me over. Even though I never found a reason for it in my plots. I correct the timing with FIR filters so that's not going to have an effect.
The one thing I did notice is when I measured the impedance in REW, and looked at the group delay tab of the impedance measurement plot there were noticeable differences.
Guess which one has the compensation, this was taken with the complete conjugate network vs no network. This was measured with a laptop soundcard and a resistor in REW, no separate amp.
A comparison of both impedance plots:
Making it easy to spot the compensated plot above.
To this day the impedance compensation for the peak has stayed inline, although I cannot explain why I like it, I just prefer it. No matter how hard I looked I could not find a clear answer in the raw measurements. Except for the above shown difference.
*The restlessness in one graph at 50 Hz is from the laptop power supply. That much I do now.*
Impedance looks like this with only the peak compensated:
Last edited:
Just to show how much smoke is being blown up everyone's butt here, recognize that I showed exactly want is happening here back in my 2nd post, post 87 in this thread on page 9. http://www.diyaudio.com/forums/mult...rrent-compatible-crossover-9.html#post4581890 Now I've basically repeated that in post 472 on page 48.
So, 40 pages and 385 posts later and the dial hasn't moved. Very productive discussion. 😕
So, 40 pages and 385 posts later and the dial hasn't moved. Very productive discussion. 😕
As said, I correct with FIR filters after taking a raw measurement. Even the raw measurements did not tell me much, looking at more than FR alone.
The FR plot I do have comparing the 2 (made on different days) looks like this, prior to using FIR correction:
Again, this was taken with the full compensation network vs no network.
The FR plot I do have comparing the 2 (made on different days) looks like this, prior to using FIR correction:
Again, this was taken with the full compensation network vs no network.
Last edited:
That's one way to approach DIY audio, and a valid one.
Or...
It'd probably take less time to build Joe's example, have a listen, and see if it sounds any different than just debating for weeks on end from first principles to invalidate it.
I have just built a new amp with a couple of Hypex UCD180HG's. A lot of people have commented on the Internet that they sound quite good. It sounds much better to me than my old Pioneer A400X. Everyone else who's heard it in my system also prefers the little Hypex units. I have no data to prove it sounds better. However, I have actually built the amp, and listened to it, and to my ears it sounds clearly superior. I don't have any data to prove it, and to be honest I don't know where to even start to present data to prove to anyone on this forum that it sounds better than the old A400X.
I'm not against data. I use it extensively on my race car. The end goal is to win my class. Each upgrade should reduce my lap time. It's easily measurable and the average person understands and can relate to that measure. I care about the data for each and every upgrade I try.
The end goal with my hifi system is for it to 'sound great' to me. Each upgrade should 'sound better' than the last one. To me. It's a tad more subjective than my other hobby.
I'm surprised there aren't more responses to Joe along the lines of "I've tried this and I didn't find it an improvement... got any more data yet to convince me why I should spend any more time on this path?".
Shame there's no equivalent of a stopwatch for diyAudio. 🙂
I've played with a compensation network on my crossover-less arrays. Just out of curiosity. I was warned before I started it wouldn't make a difference. I compensated for the impedance peak (I play the arrays trough that peak using EQ at the low end) and also the impedance rise at higher frequencies.
After several listening sessions I removed the compensation for the rising impedance. Without the compensation I experienced cleaner sound. Highly subjective I know. But somehow that compensation network for the impedance peak won me over. Even though I never found a reason for it in my plots. I correct the timing with FIR filters so that's not going to have an effect.
The one thing I did notice is when I measured the impedance in REW, and looked at the group delay tab of the impedance measurement plot there were noticeable differences.
Guess which one has the compensation, this was taken with the complete conjugate network vs no network. This was measured with a laptop soundcard and a resistor in REW, no separate amp.
A comparison of both impedance plots:
Making it easy to spot the compensated plot above.
To this day the impedance compensation for the peak has stayed inline, although I cannot explain why I like it, I just prefer it. No matter how hard I looked I could not find a clear answer in the raw measurements. Except for the above shown difference.
*The restlessness in one graph at 50 Hz is from the laptop power supply. That much I do now.*
Impedance looks like this with only the peak compensated:
It's not a surprise at all. You are using a low Zout amplifier.
IMHO, the difference in sound Joe Rasmussen is talking about comes from the amplifier, not from the speaker. There is a compensation network to preserve the original FR (or the FR with low Zout amp, if you like). However a current amplifier still works in a different way, despite the fact it will not alter the FR.
I didn't want to intervene about the philosophical discussions on the Ohm's Law but in truth such law doesn't tell anything about what's really happening.
The difference between a low Zout source and high Zout source is that once the driver has been excited by means of a current circulating in the bobbin there will be a counter current (Len's law). The lower the source Z the higher the reverse current: it's a feedback (this is what is usually associated with the better capability of a low Zout amp to control the driver).
However the two main forces now present have different origins: the feedback current is simply proportional to the membrane velocity while the spring force due to suspensions is proportional to the shift from the rest point. As a result of this distortion is introduced. In this respect the higher Zout has an advantage, provided the drivers have enough quality.
Last edited:
Great if you are in a country where they sell these and have money to burn. Now if Joe offered 2 sets of speakers, one with all mods and one as new to be sent to burning amp this year for ears only demos we might get somewhere.
If people prefer something no one will argue. It's when they say 'it's better' you need to have some data. And you can only be subject if you listen ears only and don't know what has been changed.
Maybe the time has come to accept that some people probably had no
intent to discuss the topic. A lot of useful information is here and we
know a bit extra more about participants character by the way they
converse here.
I've just re-read the first page of this thread. My interpretation was the same as the first time I read it; that Joe has been playing with something on a different tangent to the norm, liked the results, and has shared with the diyAudio community. There was a clear explanation, lots of graphs (data?) and a how-to guide that even I can follow. There's a price to entry, as there is to just about everything in diyAudio short of speaker placement (assuming you already own some).
I'd expect to see lots of posts re-engineering the crossover in an endeavor to see whether it can be applied to any existing boxes on-hand and whether any benefit can actually be derived.
I'd expect to see posts suggesting the benefit could actually be from using higher quality crossover components, changing the stuffing, etc... Maybe even just from shifting the crossover outside the box (I've heard it helps but never tried it myself)?
Maybe even a few posts from people who were going to try it exactly as presented...
What happened to the "diy" part of this hobby? Build, evaluate, share views, seek further understanding/explanation etc... then collaborate and develop further or debunk and move on. Way too many threads in recent years just jump straight to debunk. If you think Joe's a nut, just jump straight to 'move on'.
The point in my previous email that I was attempting to convey was that in audio anyone can actually say they believe their crossover for a given speaker sounds better. There is no benchmark of what sounds "best". I don't believe there is any data that will substantiate Joe's view such that it satisfies your view that this approach is 'better'. We're mostly a bunch of technical people with a love of absolutes and data who are playing in the realm of the arts where the individuals personal opinion is the only thing that matters in the end.
Everything is there for anyone who wants to "do it yourself". If they like it better than the original crossover, then awesome! They got an upgrade! If they like it and want to delve into it deeper with Joe and others to understand further, this forum becomes more awesome! If they can't hear any benefit, didn't get an 'upgrade', chances are they had fun trying, learnt stuff and will re-use the parts for something else.
If you already knew this thread will be a dead-end and won't give you that next upgrade, move on and invest your efforts doing something that will. Then, please share it with others on this forum and make it awesome again. 🙂
"diy" does not mean chasing every dubious claim thrown out there without evidence, data, or analysis. If wasting your time that way is fun for you, have at it; lots of people try building perpetual motion machines, too. My lab time this week was spent on actual speaker design, not trying out implausible ideas.
To clear up my post;
The colour coding in my posted impedance examples may indicate something different, but the red trace in the group delay plot is the one without compensation. The blue trace is with conjugate network. Just to clear that post up. I did many measurements but could not find a true explanation for my preference.
Some of the tests can be found starting here: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-132.html#post4423888
The colour coding in my posted impedance examples may indicate something different, but the red trace in the group delay plot is the one without compensation. The blue trace is with conjugate network. Just to clear that post up. I did many measurements but could not find a true explanation for my preference.
Some of the tests can be found starting here: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-132.html#post4423888
Last edited:
- Status
- Not open for further replies.