I'd like to focus on the largest cause of difference between voltage and current drive to keep this manageable.
The whole thing is a bi-directional system. Everything interacts with everything. Eg: if the flux density changes with voice coil displacement, this will distort the back-EMF and it will also distort the force due to coil current. If the cone spring varies with cone displacement, this will affect both back EMF and net cone force due to coil current.
What is the largest non-linear effect that could show a difference in acoustic output distortion between V drive and I drive?
Also, if one does get a worthwhile reduction in the THD of the acoustic output by current driving, what about the elephant in the room of complete loss of electrical damping of the cone? Oops.
The tightness of control of the mechanical system could, in itself, account for THD differences couldn't it? Even if the electromagnet system were perfectly linear, one system with tighter velocity control than the other would show acoustic differences, especially with a music signal rather than sinewaves.
The tightness of control of the mechanical system could, in itself, account for THD differences couldn't it? Even if the electromagnet system were perfectly linear, one system with tighter velocity control than the other would show acoustic differences, especially with a music signal rather than sinewaves.
Last edited:
Joe is forwarding his own way of thinking in his way of presenting technical subjects which is deliberately complex and perplexed. Flaws also slip in.
Joe has invested into the subject he is pushing here (a site, some products and an image).
Please, keep this in mind when you volunteer to straighten what may look as technical irregularities. They are actually part of the packaging design.
George
Thanks Bill, another look at the feedback goes round and round I needed that.
Off on another tangent. I've been reading Bob Katz articles in innerfidelity since the beginning but had missed out on his latest that JC had been waxing about. Now I remain unconvinced that adding 0.1% 2HD can cause
But I was interested to read that, as the transparent bypass Bob uses THATcorp chips. If they are good enough for Bob they must be good enough for us?
I've thought that the diffamps they do should be useful in a number of applications as they have laser trimmed resistors in for a reasonable price in one offs from the usual suspects.
What's not to like?
But I was interested to read that, as the transparent bypass Bob uses THATcorp chips. If they are good enough for Bob they must be good enough for us?
I've thought that the diffamps they do should be useful in a number of applications as they have laser trimmed resistors in for a reasonable price in one offs from the usual suspects.
What's not to like?
I like this poster: http://www.klippel.de/fileadmin/kli...rature/Papers/Klippel_Nonlinearity_Poster.pdf
I couldn't find it because I kept searching for klipsch instead of klippel.
Can you explain the Bl(i) thing more and why current drive doesn't help with it? If the coil is held in place, current should magnetize the motor, and the resulting induced EMF should contain nonlinearities. If you use current drive, the nonlinear voltage won't affect coil current and thus force. So can I assume that the distortions you're speaking of occur only when the coil is moving?
Perhaps the magnet only wants to magnetize in one direction and in one direction Bl increases and in the other direction it doesn't? I guess I can see how that would change the force on the voicecoil while stationary. But if Bl changed linearly in response, would it still cause distortion or not? If so then the same would apply if the magnet were an air coil instead.
I was thinking about shorting rings. I think for Bl(i) you want a shorting sleeve more than a ring. The more copper the lower the frequency it will work at, so using only enough copper to level out the Le(x) doesn't fully address this. But using a shorting sleeve doesn't necessarily level off Le(x) and may (I think) even prevent you from being able to level it off, unless you don't use as large a sleeve as you could. Since current drive gets rid of Le(x) problems, it would seem a current drive driver would gain the full benefit of a shorting sleeve, the bigger the better. Another way to say it would be the optimal drive impedance would increase.
Voltage drive would only improve material noise within one or two octaves of Fs where the E-brake effect has control. This makes sense considering this is also where excursion is highest. I guess controlled breakup cones where there was some midrange impedance roughness might have a bit of a benefit in the midrange.
Based on this, if you want to design a cabinet for current drive then you don't want to use one that increases excursion, because you will be increasing material noise on two fronts.
I couldn't find it because I kept searching for klipsch instead of klippel.
Can you explain the Bl(i) thing more and why current drive doesn't help with it? If the coil is held in place, current should magnetize the motor, and the resulting induced EMF should contain nonlinearities. If you use current drive, the nonlinear voltage won't affect coil current and thus force. So can I assume that the distortions you're speaking of occur only when the coil is moving?
Perhaps the magnet only wants to magnetize in one direction and in one direction Bl increases and in the other direction it doesn't? I guess I can see how that would change the force on the voicecoil while stationary. But if Bl changed linearly in response, would it still cause distortion or not? If so then the same would apply if the magnet were an air coil instead.
I was thinking about shorting rings. I think for Bl(i) you want a shorting sleeve more than a ring. The more copper the lower the frequency it will work at, so using only enough copper to level out the Le(x) doesn't fully address this. But using a shorting sleeve doesn't necessarily level off Le(x) and may (I think) even prevent you from being able to level it off, unless you don't use as large a sleeve as you could. Since current drive gets rid of Le(x) problems, it would seem a current drive driver would gain the full benefit of a shorting sleeve, the bigger the better. Another way to say it would be the optimal drive impedance would increase.
Voltage drive would only improve material noise within one or two octaves of Fs where the E-brake effect has control. This makes sense considering this is also where excursion is highest. I guess controlled breakup cones where there was some midrange impedance roughness might have a bit of a benefit in the midrange.
Based on this, if you want to design a cabinet for current drive then you don't want to use one that increases excursion, because you will be increasing material noise on two fronts.
This seems relevant:
http://www.daytonaudio.com/media/resources/290-226-dayton-audio-nd91-8-klippel-test-results.pdf
Still not sure about the Bl(i) linearity of an air coil motor though... I thought Jneutron said something about that in the past?
http://www.daytonaudio.com/media/resources/290-226-dayton-audio-nd91-8-klippel-test-results.pdf
Still not sure about the Bl(i) linearity of an air coil motor though... I thought Jneutron said something about that in the past?
How would viewing the issue from the point of magnetic hysteresis help? ... hysteresis being the equal of DA in capacitors. As a factor in back emf.
What could you do to reduce the hysteresis affects? Would it reduce 2H?
-RNM
Last edited:
DA is not hysteresis
film cap DA is well modeled by the Linear RC ladder circuit - to the point that Pease was able to Linearly compensate a Mylar cap to equal a Teflon cap
Hysteresis is inherently Nonlinear - and can be seen in Ferroelectric caps like hi K ceramic caps
re the recent discussion of precison in engineering language the distinction is important for understanding
(looks like the wikipedea DA page needs an edit)
film cap DA is well modeled by the Linear RC ladder circuit - to the point that Pease was able to Linearly compensate a Mylar cap to equal a Teflon cap
Hysteresis is inherently Nonlinear - and can be seen in Ferroelectric caps like hi K ceramic caps
re the recent discussion of precison in engineering language the distinction is important for understanding
(looks like the wikipedea DA page needs an edit)
Destroyer, remember I don't sweat the audiophilia nervosa stuff (thanks, Bill for that one) and stick to more modest aims that at least I can justify with *some* logic (with the overarching goal of "having fun building stuff").
Of course, if you're trying to knock something up, then all this becomes moot. At a certain low price point you bemoan that you can't directly talk to a LSR305's dsp and deal with it.
I try to keep my enjoyment of music separate from my enjoyment of making things. My present design decisions are looking like an extremely well provisioned receiver in a single 4u case. (Albeit I'm using 2x TPA3255 EVM boards because cost:benefit was too favorable for me to not jump on 4 channels for $80 shipped)
Let's humor a 3-way arrangement with no subwoofer. This example would as well fit bookshelf speakers + stereo subs, obviously.
With that said, Graham (I think?) hits it pretty well on the head:
1.) Yes, I'm working from a premise of a digital source, and even more so a computer that can either locally store the files or stream them from other sources via wifi. Same cost regardless passive/active (RPI3 has more than enough oomph to do 6 channels of DSP).
2.) A quick look the now long-growing es9038q2m thread will tell you that getting superb DAC channels is not a terribly expensive proposition, especially if one sticks to the core essentials (it needs modest modifications, one of which is a buffed up PSU, which isn't going to cost much more to scale up to 3 boards). Similarly, Damian has shown measurements of a $60 (with case!) AK4490. Power it off a different power brick and drop a 49720 in and it's extremely good. We can argue esoterica this or that, but that's beyond my desires regardless. So in this vein, let's call it a $120 incremental cost.
3.) The PSU and chassis of an amplifier dominates its BOM. Followed by the output transistors. With multichannel and active, we're not dropping any power across the crossover AND we're presenting (typically) a kinder impedance to each respective amplifier (need look no further than the WATT puppies for a nightmarish low impedance). So can we at least agree that, all else equal, we don't need any more VA out of our PSU than a 2 channel amplifier capable of equal output nor need larger heatsinks? Even if we go with something like a boosted rail front-end on an AB amplifier, that rail can be shared. All the rails can realistically be shared. You need fewer output transistors per channel, although you'll end up with more total transistors. Bonus here is better heat transfer to the heatsinks, most beneficial to transient demands (i.e. mid/tweeter amps may have similar instantaneous power demands bu their continuous power needs falls off generally at 1/f). Pretty easy to get the extra boards at $25/channel (remember, nothing super fancy), so incremental cost is another $100.
4.) Dedicated DAC + amp means one can sensibly make a harness between chasses (if needed!) and bypass all the cable stupidity that sits around the topic by a nice STP control wire interconnect, like, you know the stuff for industrial applications where there's a shedload of RFI and crap floating around. Or all in one case, where twisted pair between boards is all you realistically need if you bond your grounds together correctly. (Run everything differential!!!!) Let's add $30 for the ostensible extra connections. Speaker wire, well: Speaker cables don't influence harmonic distortion!
A good 3-way crossover is going to cost how much? Let's be frugal and say $150 for the stereo pair.
Lot more labor, but realistically we're in the realm of a couple hundred bucks difference. This could be more or less depending on one's esoteric desires, but sets a ballpark.
I understand the difference quite well... it is an analogy only. Anyway.... how does hysteresis affect the distortion and how can amplifier help rid the speaker of its distortion?
This is in the context of the circuit I showed.
[BTW - in my frame of reference it is not films but bipolar electrolytcis I always refer to. And certain ceramics. I pointed out decades ago, the film types are more linear and preferred]
THx-RNMarsh
Last edited:
I certainly do think a bit different. My aim is to only attempt stuff that is competitive with commercial stars. So while I think there is more than enough silly fluff out there, I've also heard well made, extremely good, gear as well. And some of it was vastly far from being affordable. There tends to be a disconnect from serious high end and DIY idea of what that even actually happens to be. Every once in awhile I have some get some PCB's from me that has something like a Wilson clone speaker that's a DIY project with $7k in parts alone or whatever (and much better than the Wilson I imagine).
1. The newer ESS DAC's have a bunch of channels, so that's extremely helpful since they aren't total junk like some of the DSP things of yesterday.
2. I doubt I'd be satisfied with the classD amp. The only ones I have liked have been classA front ended Hypex (and yet still not my choice). So it is a matter of taste. On the same accord I'm pretty demanding on whether amps have really good dynamics (especially small volume changes) and texture while sound equivocally large. That is one of the lines that many DIY people never draw because they're fairly unfamiliar with broad room filling power, or erroneously presume it takes mega wattage they that isn't DIY friendly.
3. I'd have to check power usage vs. transformer sag, while using extra amplifiers. You should be using less power not having to drive so much reactance, but given the efficiency of say AB, I'm not sure off the top of my head what it all is going to look like on a LPSU.
4. Hmm I'd go with whatever cables allow identical overall gauge of conductors. I'm not actually a fan of shielded thus far... but that is mostly because BJC sucks, and maybe because balanced get double humped on whatever problems they do have; so if the line stage sounds a hair plastic it now sounds two hairs plastic.
Something that is really going to change pricing for me is that I'm going to use all sorts of noise reducing things for all the parts if I build what you describe. I wouldn't be using wall warts (at least not stock).
What I don't understand is your separation, as anyone should feel somewhat gratified if they put together what you describe, and the music should be a little sweeter for it. My respect for DIY is never ending despite all the different levels of understanding and crazy that come with it. I think it makes you a better human.
I have Merilainen's Current-Driving of Loudspeakers and although he's pro I, he does give some good summaries that would help get people on the same page:
I could post some summaries and equation/graph screenshots.
Last edited:
We have different objectives and that's all good. With regards to cabling, I come hard from the test and measurement world so go have a look at full foil shielded instrumentation cable. You can get multi-conductor cable. Perfect for schlepping multiple channels from one chassis to another rather than running 3x XLR or something. Nothing esoteric.
PLTC & PLTC-ER Cable by Belden
As far as mindset, there are those that are, at least in communication here on DIYaudio, extremely fussy about their gear, etc. Certainly listening on fancy gear (and especially my own stuff!) is more fun, but here I am with earbuds (decent ones at least) and whatever output from my work laptop. The music's good and I'm happy.
- Status
- Not open for further replies.