Voice coil heating and the performance of passive XO, plus my funny experience at HTG

[Jay_WJ]

(This may not be relevant to the forum. If so, this can be removed without warning or notification. I understand that. Otherwise, the link below will be a good read about the issue of "the effect of voice coil heating on the performance of passive crossover.")

Hi,

I'm relatively new to DIY speaker building. I've been trying to be a serious learner, spending a lot of my leisure time these days reading useful information and asking questions by visiting several internet forums.

Perhaps I was TOO serious. I was finally expelled from HT Guide!! I think I acted too early before realizing what kind of forum the HTG IS---a very closed, narrow circle of people. Especially, it doesn't like a newcomer who SEEMS to be reluctant to join the adoration of a few important people.

In contrast, diyAudio is not partial, with various kinds of people contributing to valuable knowledge about DIY audio.

Anyway, at HTG I started a thread to ask a question about an issue, about which I really wished to have good knowledge. I hoped some experts could help me on this issue. Yes, I was very serious. Some people tried to answer my question. But unlike a usual newcomer, I didn't accept the answers as they were. I'm a scientist and not a simple person. When some statements or observations about a problem are given, I usually think of factors and variables that might not be taken into account in them. So, not being fully satisfied, I raised possibilities that could be considered additionally, hoping to get more detailed explanation, possibly from some experts.

Here's the thread if you want to take a look: http://www.htguide.com/forum/showthread.php4?t=26211

Sadly enough, it was only after I was expelled from the forum that I received some better explanation about the issue.

I admit that I talked in a bit "arrogant" manner there as a newcomer, but my only reason for doing that was that I really wanted more accurate information from more knowledgeable people. But that attitude was not accepted. My thread was immediately closed and after posting one more thread I was expelled. I understand that I was a bit too serious and my attitude could seem uncomfortable to people there. But I did not post "uncivily" at all. But my pure intention was totally ignored.

Anyway, I think this is a very good example that shows what KIND of forum the HTG is and what KIND of newcomer they would welcome. I find there are people of only two types there: 1) several very active ones who willingly join the adoration of the important person(s) and his designs; 2) occasional visitors who want to build their designs or simply do very casual talks about speaker building.

--Jay

[Lynn Olson]

I'm afraid I have to agree about "other" forums - what I like about diyAudio is the excellent moderation, a strong Euro-flavor that I like, and contributors who are knowledgeable yet respectful of some pretty wild ideas. (Talk about shameless buttering-up!)

Your point is valid in a broader context. VC heating in audiophile drivers is in fact significant even at 90~95 dB levels, and there's no clever techno-fix since it is result of the tempco of copper itself. Once you get used to the quite different sound of high-efficiency speakers, the time-delayed thermal compression of audiophile drivers becomes not only noticeable, but obvious. And I say this as someone who is not really a horn-fan ... but I have to admit what they do well, they do very well - dynamics and vivid tone colors.

As a scientist, you've probably done enough homework to know that high-order filters have greatly increased component sensitivity compared to low-order filters. Since the driver itself is an integral part of the overall acoustic filter, driver variations (with respect to level) degrade the operation of complex filters. A huge red flag is when values of crossover components have to be "trimmed" to match particular samples of drivers - it may match on a Tuesday at 7PM at 30% relative humidity at 80 dB SPL, but come next Thursday at 60% RH and 50 dB SPL, you can be pretty sure it won't be the same.

P.S. How real is power compression? Here are the specs for the 18Sound 15NMB420, a truly hard-core prosound 15" midbass woofer, rated at a continuous AES power of 350 watts. Click on the Technical Spec. tab, and you'll see that 0.7 dB of compression starts at 35 watts - even for this monster driver, designed for all-day-long full-power abuse in a disco.

If this driver starts to compress at 35 watts, what about a typical 6.5" audiophile driver? It certainly isn't going to be better, with a smaller voice coil and far less efficient cooling systems. The lifetime of a 6.5" audiophile driver at 350 watts continuous power would probably be measured in milliseconds - I'd guess about 50~200 milliseconds.

[augerpro]

Ironically of all the hobbies I have and forum memberships for them, this is the only forum that I ever had someone just come out of nowhere and try to belittle me. The mod had to delete all the posts it was so silly. I think it's more a matter of people in different "camps" tend to just stay out of the others thread. For example you'll never see Zaph post in one of Lynn Olson's threads. More civil perhaps, but most debates don't come to a conclusion either. Not that a lot of them ever can...eventually people just pick a side and ignore the other guy. But hey it works for me so whatever...

I WILL do that test I promised at htguide tonight! Give me a couple hours though. I'll probably try pink noise and actual music. I have my doubts that 35 watts at 5 continous minutes of pink noise (ala 18sound) is really representative of real music...

[David Gatti]

FWIW Jay, I dont know why you were expelled based on that thread. Seemed like a reasonable and civilised discussion to me. Unfortunately in writing, sometimes the tone is hard to get across and people take things in a negative way, which is a pity because it stifles passionate debate.

Lynn, you mention the 15NMB420, but that has 99dB sensitivity!
So 35 watts input = 114dB before compresson sets in. Should we be concerned about that?

My tests into my 90dB speakers, have shown that I listen at about 0.1 to 1W average power input giving about 85dB at the listening position from a stereo pair. Maybe I'm getting old, but I believe this is not atypical for most home users. How loud do you listen Jay?

Also lets say you have a 100W amp and assume that music has a dynamic range of 20dB. You should then only be listening at 1W. If you're listening at more than 1W average, you might start clipping your amplifier on the peaks - a situation far worse than thermal compression.

I'm sure its relevant in a disco, but in the typical home situation, and assuming you value your ears, power compression appears to be a non-issue. Still, if you want to be safe, build a line array ;-)

[Jay_WJ]

Originally posted by David Gatti
FWIW Jay, I dont know why you were expelled based on that thread. Seemed like a reasonable and civilised discussion to me.

I don't understand, either. The thread I linked above is the very one based on which I was expelled.

- Jay

[Lynn Olson]

Originally posted by David Gatti
Lynn, you mention the 15NMB420, but that has 99dB sensitivity!
So 35 watts input = 114dB before compresson sets in. Should we be concerned about that?

Yes. 35 watts @ 99 dB/metre is 114 dB, while 35 watts at 87 dB/metre is 102 dB. VC heating is roughly proportional to VC area, so the lower-efficiency driver is very likely starting to compress at 18 watts, which takes us down to 99 dB. (In practice, I'd expect it to be well under 18 watts, since pro drivers wind the VC on the inside and outside of the former, specifically to increase the thermal-emission area of the VC, so it can couple more efficiently to the thermal mass of the frame and magnet.)

In private communications with James Boyk at CalTech, he mentioned the typical peak-to-average ratio for classical music is 14 dB - sometimes more, sometimes less. So if we want uncompressed peaks at 99 dB, the average 1-metre distance playback level (as measured on a slow-responding SPL meter) shouldn't exceed 85 dB. That sounds about right, in my experience.

In a smaller listening room, listening to dynamically compressed jazz or rock, the requirements are more modest. In a larger room, listening to uncompressed acoustical music at live-concert levels, more is required.

[CeramicMan]

Hmmm!!! I had a quick look at the link and although I haven't looked at other HTG threads to form a more 'balanced' opinion, they sound like a bunch of downright snobs

Some of them seem to be forgetting that "common knowledge" is often wrong. Their unwillingness to discuss technical details "because it's been done to death and we already know everything about it" or whatever, sounds more arrogant to me than anything YOU said in that thread.

Anyway, back to the voice coils, I thought it was pretty well established that passive crossovers have lots of limitations because they go out of tune with temperature, humidity and things like that. Although a small dB change might seem negligible, but even for a 12dB/oct crossover, subtle changes near the crossover frequency might alter the lobing patterns (especially if the speakers are well matched), causing subjectively large changes in tonality.

Other things to consider:

The heat capacity of the metal. A voice coil isn't some magical homogenous unit, it's made up of thin strands of wire covered with an insulator. The 'average' temperature change over several seconds or minutes may bear little or no resemblance to localized swings in the actual metal.

Intermodulation distortion. Say for example, the speaker is playing two tones: a high amplitude 100Hz sine wave, and a softer 1.5kHz sine wave. Temperature oscillation due to the 100Hz tone modulates the amplitude of the 1.5kHz tone, creating side-lobes around 1.4kHz and 1.6kHz. Combine more complicated signals together with passive crossovers, and it could create a sort of 'network' of interdependent IM distorted tones that practically feed off themselves.

IMO, active crossovers make a huge difference and they also help to overcome limitations of the amplifiers – they distort too.

[herm]

Jay,

Do you think the rest of us don't read more than one forum?

The rest of the story...

If you come to learn, put your attitude away, Jay.

Both forums have many people willing to help, and many
people tired of attitude...

herm

[augerpro]

Ok Jay here are the test results! I'm using my center channel from my Bose Buster project: 3 way, Hivi M3N mid, Hivi M4N woofers, adn Dayton ND20 tweeter. In sealed box. Probably about the worst speaker to use for extended high SPL with all those little drivers and shielded motors to hold heat. Well the boxes are stuffed pretty well with acoustastuff so that should help. Anyhoo, the victim:






For music I used a 10 minute loop of Ministry- Filth Pig at about 99-102 dB. This is not easy listening music...control in red and after music in blue:





Hmmm...not bad. I'm happy to know this. But the next test isn't so easy. This right after the music test so no rest for the wicked little speaker. Five minutes of pink noise at 100-102 dB with 20v input...my poor speaker...again control is red and after pink noise is blue





Well there is a difference. I blew up the scale adn it's roughly 2/3 dB at the worst. And my little mid's cone was very warm. Like a cup with very hot coffee in it. A few resistors on the XO were hot and I could smell the hot glue I used on the components...I hope it was the hot glue...the only question I have now is how high of a level can the ECM8000 handle before the possibility of damage

[augerpro]

Also the XO points are 300 and 3000 Hz. The MLS gate was 6.7 seconds so 160 Hz or so is the cutoff and up to roughly 300 Hz is rolled of more than it really is due to the gate.

I think the Rod Elliot quote Jay used in that thread refers to higher than 2nd order networks, not acoustic slopes. An important distinction since it rarely (IME) requires a 4th order network to reach a 4th order acoustic slope. Usually a 2nd order network is sufficient, sometimes 3rd order.