Djk (somewhat off subject),
Interesting. Have you any further details?
I tried (unsuccessfully) to find scematics for these amplifiers - I did not really expect to find any - but see that the Crest only offers 8 ohm output impedance. As one expects these to be used in sound systems, how do folks cope with loss on long cables? Or (not being in the business of P.A.) are there 100 ohm or other line options available? The effect of such transformation could have a substantial effect on the stability of a high NFB amplifier (which the Crest appears to be, with a DF of 400).
A friend of mine is inter alia connected with office block sound distribution systems. They have no end of problems there (and this guy knows what he is doing).
Interesting. Have you any further details?
I tried (unsuccessfully) to find scematics for these amplifiers - I did not really expect to find any - but see that the Crest only offers 8 ohm output impedance. As one expects these to be used in sound systems, how do folks cope with loss on long cables? Or (not being in the business of P.A.) are there 100 ohm or other line options available? The effect of such transformation could have a substantial effect on the stability of a high NFB amplifier (which the Crest appears to be, with a DF of 400).
A friend of mine is inter alia connected with office block sound distribution systems. They have no end of problems there (and this guy knows what he is doing).
I used the DO2000 for a number of years for sub duty in a small festival rig. It drove 4 of the CR "Earthquake" horns with no problem. They didn't make it big on the touring market because of their size and weight. Truck space is expensive. They were actually Gradient Drive amps for MRI machines that were modified for audio use.
However, the Crown Macro-Tech was the standard for many years on the touring market along with the Crest Pro Series and the Carver/Clair amps. The Crest amps were also one of the amps of choice on Broadway until they were discontinued and replaced with LabGruppen.
The Crest states a load impedance of 8 ohms, but most of them will drive 2 ohms or 4 ohms bridged with their eyes closed. I've actually seen a 9001 used for welding.
However, the Crown Macro-Tech was the standard for many years on the touring market along with the Crest Pro Series and the Carver/Clair amps. The Crest amps were also one of the amps of choice on Broadway until they were discontinued and replaced with LabGruppen.
The Crest states a load impedance of 8 ohms, but most of them will drive 2 ohms or 4 ohms bridged with their eyes closed. I've actually seen a 9001 used for welding.
1. True, the sound propagation isn't quite the same as the voice coil or former motion but it is a more satisfactory reference for feedback than anything else. Not a bad discipline for the driver designer to make the relationship closer (more piston-like) rather than sculpt the cone for some trick purposes.
The problem with cone drivers (even the most high-tech) is that we are trying to push air around with a heavy piston, vastly heavier than the air. Just not smart. Electrostatics, by contrast, use very thin Saran-Wrap and the air resists the motion and provides that feedback. Don't get me started on the subject because us ESL guys can be really swell-headed about our systems. I'm all ESL except for a single Klipschorn bass horn.
2. If I remember correctly, you want the amp output to the negative current image of the speaker, not low impedance or high impedance but negative impedance.
3. Motional feedback corrects the main resonance and ALL errors of cone motion. Because these systems are finicky, some systems in the past may have been configured to operate only around the low end. By "errors," we are also talking about cone output below resonance and MF can be used to drive the cone to the desired excursion (despite being held back by the box compliance). In other words, you can make the speaker work below resonance. Of course, you need to tailor the drive power down there because the cone output will require extra motion to generate equal room loudness. You even can make a little cone in a 50 Hz box play great bass.
4. Of course, bass reflex and other miserable resonant boxes can't use MF (or current output) because the speaker output is so badly uncorrelated with the cone motion around the three cookie resonances around the box resonance.
5. The Kenwood Basic Amp M1X Sigma Drive remote-sensing-type compensation for long speaker wires has to be the least of the benefits one would fuss over since the long wires don't distort, just harm the damping factor and power delivery and those kinds of losses doesn't hurt much.
The problem with cone drivers (even the most high-tech) is that we are trying to push air around with a heavy piston, vastly heavier than the air. Just not smart. Electrostatics, by contrast, use very thin Saran-Wrap and the air resists the motion and provides that feedback. Don't get me started on the subject because us ESL guys can be really swell-headed about our systems. I'm all ESL except for a single Klipschorn bass horn.
2. If I remember correctly, you want the amp output to the negative current image of the speaker, not low impedance or high impedance but negative impedance.
3. Motional feedback corrects the main resonance and ALL errors of cone motion. Because these systems are finicky, some systems in the past may have been configured to operate only around the low end. By "errors," we are also talking about cone output below resonance and MF can be used to drive the cone to the desired excursion (despite being held back by the box compliance). In other words, you can make the speaker work below resonance. Of course, you need to tailor the drive power down there because the cone output will require extra motion to generate equal room loudness. You even can make a little cone in a 50 Hz box play great bass.
4. Of course, bass reflex and other miserable resonant boxes can't use MF (or current output) because the speaker output is so badly uncorrelated with the cone motion around the three cookie resonances around the box resonance.
5. The Kenwood Basic Amp M1X Sigma Drive remote-sensing-type compensation for long speaker wires has to be the least of the benefits one would fuss over since the long wires don't distort, just harm the damping factor and power delivery and those kinds of losses doesn't hurt much.
Last edited:
"I tried (unsuccessfully) to find scematics for these amplifiers "
http://www.crownaudio.com/pdf/legacy/DO2000 Schematics.zip
"but see that the Crest only offers 8 ohm output impedance. "
The output impedance is less than 0.02R, and intended to drive loads of 2R or higher. The 10001 is rated for 7.5KW music program at 1R per channel, or 15KW at 2R in bridge.
http://www.crestaudio.com/media/schematics/10K1 Set.zip
"A friend of mine is inter alia connected with office block sound distribution systems. They have no end of problems there (and this guy knows what he is doing). "
Does he?
It's not that hard to build a bullet-proof 70V or 100V soundsystem.
A problem with good working existing system is people adding extra speakers, either loading the design down too much, or some idiot wiring up an 8R speaker with no transformer. Do a system impedance check (to see if the amplifier is loaded to hard). Hook a 12VAC transformer up to the line and walk the system. The speaker without a transformer will be trying to climb out of the ceiling.
http://www.crownaudio.com/pdf/legacy/DO2000 Schematics.zip
"but see that the Crest only offers 8 ohm output impedance. "
The output impedance is less than 0.02R, and intended to drive loads of 2R or higher. The 10001 is rated for 7.5KW music program at 1R per channel, or 15KW at 2R in bridge.
http://www.crestaudio.com/media/schematics/10K1 Set.zip
"A friend of mine is inter alia connected with office block sound distribution systems. They have no end of problems there (and this guy knows what he is doing). "
Does he?
It's not that hard to build a bullet-proof 70V or 100V soundsystem.
A problem with good working existing system is people adding extra speakers, either loading the design down too much, or some idiot wiring up an 8R speaker with no transformer. Do a system impedance check (to see if the amplifier is loaded to hard). Hook a 12VAC transformer up to the line and walk the system. The speaker without a transformer will be trying to climb out of the ceiling.
Last edited:
Djk,
Thanks for the references - mine would simply not open unless I join/pay for something blah blah blah
We seem to have one of those not uncommon duplicity of terms in electricity. What I usually understand under 'output impedance' and in accordance with most specs, is the impedance of the load a device caters for. The above figure is often called 'amplifier internal impedance'.
Sorry for that (I did notice what you are citing) - but I thought it would be clear from my comment, that my question had to do with the normal practice of pro-amplifiers providing higher - er - wel, if you insist, for higher load impedances than 8 ohm or such.
I cannot immediately notice why the Crown should give more problems that the Crest; almost to the contrary. One often gets more problems from quasi-complimentary output stages than full complementary. But such a conclusion here would be totally premature; there are obviously circumstances etc. connected with P.A. practices.
Yes. He is an E E Professor with two PhDs and designs/installs such equipment and systems, among his many other tasks.
Agreed, but irrelevant here. I was talking about an original installation. (I was further talking about a multi-storey office block with >100 offices, resulting in the use of several km of cable eventually, etc. etc.)
But I believe we are off-topic now, so back to the thread.
Thanks for the references - mine would simply not open unless I join/pay for something blah blah blah
We seem to have one of those not uncommon duplicity of terms in electricity. What I usually understand under 'output impedance' and in accordance with most specs, is the impedance of the load a device caters for. The above figure is often called 'amplifier internal impedance'.
Sorry for that (I did notice what you are citing) - but I thought it would be clear from my comment, that my question had to do with the normal practice of pro-amplifiers providing higher - er - wel, if you insist, for higher load impedances than 8 ohm or such.
I cannot immediately notice why the Crown should give more problems that the Crest; almost to the contrary. One often gets more problems from quasi-complimentary output stages than full complementary. But such a conclusion here would be totally premature; there are obviously circumstances etc. connected with P.A. practices.
Yes. He is an E E Professor with two PhDs and designs/installs such equipment and systems, among his many other tasks.
Agreed, but irrelevant here. I was talking about an original installation. (I was further talking about a multi-storey office block with >100 offices, resulting in the use of several km of cable eventually, etc. etc.)
But I believe we are off-topic now, so back to the thread.
Minimizing Speaker Cable Loss
Here is another approach. Place the amplifiers as close to the speakers as possible (works best with monoblocks). Next determine if the cable run from the preamp to the speaker can be treated as a lumped element. Typically this requires that the propagation time be less than fastest edge rate/20. (For most coax the propagation velocity is ~2.0 ns/foot). If it can be considered a lumped element then make sure your preamp is stable with the capacitive load. The capacitance of coax is ~30 pf/foot. Otherwise you will need to treat the coax as a transmission line and terminate it.
There are two types of terminations in common use: source terminated and load terminated. For the former, the load end is left unterminated, and the source impedance (Rdriver + Rterm) is set to match the coax's Zo (usually 50 or 75 ohms). A signal launched into a series terminated channel sees a 2/1 voltage divider at all points along its length except at the far end, where the signal reflects in phase with the incident wave, producing an amplitude equal to what the driver would generate into an open circuit. The reflected wave propagates back up the channel and is absorbed by the source termination. Obviously it is important that the driver be capable of absorbing this reflected energy without introduing distortion.
The second alternative: load termination uses the same source resistance as before, but also terminates the load to ground with Rterm=Zo. This introduces a 2:1 voltage divider compared to the source terminated case. Note that in either case the driver must be capable of supplying considerable current, since it is driving a 50-75 ohm load. Unless you are cabling the Boeing hangar you probably will not need to treat audio interconnect as a transmission line. Note, however, that a 30 foot coax run will yield almost 1000 pF of capacitive load.
Here is another approach. Place the amplifiers as close to the speakers as possible (works best with monoblocks). Next determine if the cable run from the preamp to the speaker can be treated as a lumped element. Typically this requires that the propagation time be less than fastest edge rate/20. (For most coax the propagation velocity is ~2.0 ns/foot). If it can be considered a lumped element then make sure your preamp is stable with the capacitive load. The capacitance of coax is ~30 pf/foot. Otherwise you will need to treat the coax as a transmission line and terminate it.
There are two types of terminations in common use: source terminated and load terminated. For the former, the load end is left unterminated, and the source impedance (Rdriver + Rterm) is set to match the coax's Zo (usually 50 or 75 ohms). A signal launched into a series terminated channel sees a 2/1 voltage divider at all points along its length except at the far end, where the signal reflects in phase with the incident wave, producing an amplitude equal to what the driver would generate into an open circuit. The reflected wave propagates back up the channel and is absorbed by the source termination. Obviously it is important that the driver be capable of absorbing this reflected energy without introduing distortion.
The second alternative: load termination uses the same source resistance as before, but also terminates the load to ground with Rterm=Zo. This introduces a 2:1 voltage divider compared to the source terminated case. Note that in either case the driver must be capable of supplying considerable current, since it is driving a 50-75 ohm load. Unless you are cabling the Boeing hangar you probably will not need to treat audio interconnect as a transmission line. Note, however, that a 30 foot coax run will yield almost 1000 pF of capacitive load.
How well would the below work for such circumstances, in terms of transmission lines/line delay? It essentially offloads the capacitive load to a separate buffer so that the main amplifier only sees a non-capacitive cable.
- keantoken
- keantoken
Analog_guy,
Stimulating and well-explained.
Considering all things (the buzz-word 'holographic' comes to mind), I am just wondering how to decide where the transitional region would lie. I think (hope!) that much of the worries about long loudspeaker cable has been put to rest here. Mostly the consequence below say 50m would be some attenuation, much of which can be taken care of by the power amplifier ability.
Before that time screened cable/transmission line concerns would already have entered the scene. The few times I was involved with this sort of thing we treated it as 600 ohm line business, but that was because the facilities were there, moreso than it being mandatory. But there was also a few occasions where we simply used rather thick loudspeaker cable in a 15 ohm system - no idea what the loss was, no measurement done.
Intuitively I still feel that a longer loudspeaker cable would be the better way to go in domestic systems. [Perhaps time one defines 'long' in the house scenario; for me that would be <10m. Perhaps I will have my horizons broadened here!]. After that a grey area ruled by other factors, and >100m I would say definitely transmission lineof the classic kind.
Please note that I am not minimising your viewpoint, but some input on what 'home' distances are applicable over the scope of 'poor people/affluent folks' would be informative.
Stimulating and well-explained.
Considering all things (the buzz-word 'holographic' comes to mind), I am just wondering how to decide where the transitional region would lie. I think (hope!) that much of the worries about long loudspeaker cable has been put to rest here. Mostly the consequence below say 50m would be some attenuation, much of which can be taken care of by the power amplifier ability.
Before that time screened cable/transmission line concerns would already have entered the scene. The few times I was involved with this sort of thing we treated it as 600 ohm line business, but that was because the facilities were there, moreso than it being mandatory. But there was also a few occasions where we simply used rather thick loudspeaker cable in a 15 ohm system - no idea what the loss was, no measurement done.
Intuitively I still feel that a longer loudspeaker cable would be the better way to go in domestic systems. [Perhaps time one defines 'long' in the house scenario; for me that would be <10m. Perhaps I will have my horizons broadened here!]. After that a grey area ruled by other factors, and >100m I would say definitely transmission lineof the classic kind.
Please note that I am not minimising your viewpoint, but some input on what 'home' distances are applicable over the scope of 'poor people/affluent folks' would be informative.
Hi Johan. I posted measurements in the speaker cable thread showing almost 3/4 dB high frequency droop @ 20 kHz from the inductive reactance of 25' of 18 gauge into a highly inductive speaker load. I recommend throwing out rule of thumb/short cut cogitating and breaking out the meters or Spice. 50 M of larger, even more inductive cable is sure to show significant FR variations into reactive loads.
Rdf,
So sorry, I vaguely recall that - I did not call up all that was said on the forum about speaker cable! I must go now; I did a quick search for that post but without result. If you would very kindly quote a reference to your post here?
I would however beg to say that I was not entirely guilty of 'rule of thumb/short cut cogitating'. My slight background here is also based on measurements, and I do know that substantially different findings can exist in that field, depending on the approach. Unless I missed something, nobody mentioned that here (as it then should have been).
But that by the way - if you would kindly point my nose (mouse) back to there I will be obliged.
So sorry, I vaguely recall that - I did not call up all that was said on the forum about speaker cable! I must go now; I did a quick search for that post but without result. If you would very kindly quote a reference to your post here?
I would however beg to say that I was not entirely guilty of 'rule of thumb/short cut cogitating'. My slight background here is also based on measurements, and I do know that substantially different findings can exist in that field, depending on the approach. Unless I missed something, nobody mentioned that here (as it then should have been).
But that by the way - if you would kindly point my nose (mouse) back to there I will be obliged.
The voice coil moves through a magnetic field that may or may not be constant, and may even be "altered" in order to cancel out nonlinear resistances of the surrounds, etc. A capacitive feedback mechanism turns motion straight into electrical energy linearly depending on the load at the other end, while an inductor does this indirectly through a magnetic field.
So I believe the capacitive method may be better, though I haven't done any tests.
How about covering the entire back of the cone with foil and making the other electrode be the speaker's metal housing?
- keantoken
Last edited:
Give the token a cigar! There are possible improvements such as a grill in front of the driver, which would increase capacitance & linearity and shield from noise, but I haven't the time to chase this idea any further. Have fun, people.
I never liked grills, really. Try blowing through one of them and you'll hear why. My Koss portapros sounded sooooo much better after I cut holes in the pads it's not funny (they were crappy foam pads anyways though). Bass should not buzz. (for the record, I like the idea of a woven mat type grill that is able to move with the pressure waves. A woven mesh of fine-gauge wire? on that note, the back of the speaker could be wrapped over with mesh for the electrode rather than using the grill as an electrode.)
Problem is EMI. Long cables could turn it into an RF resonator with a little inductance, and you most certainly don't want this in a feedback loop. RF goes straight through a wooden box, so just a front grill may not be enough...
We could solve half of the problem by attaching the charged part to ground through a capacitor, which would leak away RF, but how would be protect the moving side?
Alright, brainstorm report:
1: Front or back of cone is covered with a conductive coating, maybe foil.
2: Fine guage wire mesh on front and back of speaker, one side has to be elevated, preferably the side that no one will be touching.
3: Elevated side has a capacitor connecting to ground so as to filter away RF/EMI.
4: Elevated side is rigid, perhaps more coarse wire, so it doesn't move/resonate.
5: Unelevated mesh connected directly to ground.
6: Will any amount of shielding be enough? I suppose it depends on how sensitive/tight the feedback loop is.
- keantoken
Problem is EMI. Long cables could turn it into an RF resonator with a little inductance, and you most certainly don't want this in a feedback loop. RF goes straight through a wooden box, so just a front grill may not be enough...
We could solve half of the problem by attaching the charged part to ground through a capacitor, which would leak away RF, but how would be protect the moving side?
Alright, brainstorm report:
1: Front or back of cone is covered with a conductive coating, maybe foil.
2: Fine guage wire mesh on front and back of speaker, one side has to be elevated, preferably the side that no one will be touching.
3: Elevated side has a capacitor connecting to ground so as to filter away RF/EMI.
4: Elevated side is rigid, perhaps more coarse wire, so it doesn't move/resonate.
5: Unelevated mesh connected directly to ground.
6: Will any amount of shielding be enough? I suppose it depends on how sensitive/tight the feedback loop is.
- keantoken
Man, you have my sympathy!Following - briefly again.
Thanks a lot Rdf!
I did look; thought I would quickly read the rest of the thread before I reply ...... Permission to laugh! (For those not in on the joke - said contribution by Rdf was no. 987 or so in the thread ....)
I read somewhat in the surrounding pages. I now have more sympathy with you, if those were the 'trenches'. Rdf, again I must run (now to bed), and my subsequent questions would be off topic in this thread, eg. if you have the impedance graph of the loudspeaker you used (at those frequencies). But I do see what you mean. I will try to read/scan some of that monumental thread; measurements are always meaningful (provided one knows what one measured! - not aimed at you.)
If I may, in general: I have stated before that I have a problem with so many comments: They are qualitive and not quantative. "Item A is so much better than item B. Shame on item B!". But in reality both items constitute some 0,1% of the whole ....
Back to this topic.
Thanks a lot Rdf!
I did look; thought I would quickly read the rest of the thread before I reply ...... Permission to laugh! (For those not in on the joke - said contribution by Rdf was no. 987 or so in the thread ....)
I read somewhat in the surrounding pages. I now have more sympathy with you, if those were the 'trenches'. Rdf, again I must run (now to bed), and my subsequent questions would be off topic in this thread, eg. if you have the impedance graph of the loudspeaker you used (at those frequencies). But I do see what you mean. I will try to read/scan some of that monumental thread; measurements are always meaningful (provided one knows what one measured! - not aimed at you.)
If I may, in general: I have stated before that I have a problem with so many comments: They are qualitive and not quantative. "Item A is so much better than item B. Shame on item B!". But in reality both items constitute some 0,1% of the whole ....
Back to this topic.
i only got so far through this thread before i saw this, what about using dual coax with the sense wires down the center, and the speaker wires as the braid?. that would eliminate the capacitance issue, because the speaker wires would act as a "guard" shield (if you're not familiar with the term, it is a method used to eliminate the effect of capacitance when using coax cables at the input of an instrumentation amp where the shield is driven at the same potential as the center conductor). there will be some coupling between the speaker wires and the sense wires, but voltage errors from speaker wire resistance is likely to be greater. it would be an interesting experiment, and i would rather try it with a physical amp (either a "throw-away" or an "idiot-proof" one in case something goes terribly wrong) and speaker. theory is great for coming up with ideas, but they have to work in the real world. even an LM3886 amp would be a good choice, at least to experiment with, since it is a true "power op amp", and not prohibitively expensive as an experimental device.
i recently worked on a Bose receiver that had a sense wire for a speaker "servo". i don't know if the servo was actually a second voice coil or just a sense wire, but Bose had the problem of losing the feedback covered, as well as a switch to change the amp between servo mode and regular 2 terminal speaker driving mode. the amp was called the "Spatial Control Receiver" and was sold with 901 speakers. the primary idea behind extending the feedback loop is to place the low impedance afforded by negative feedback AT the speaker terminals.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.