Hi Jacco,
Glad you got it! It looks like they are getting out there ahead of expectations.
Thanks for the positive comment about the legibility of the figures. I used Microsoft VISIO for most of them, and McGraw-Hill was kind enough to devote the space to them that they needed for good legibility.
Cheers,
Bob
Hi Joachim,
Your observation about distortion in the current flowing in the loudspeaker is correct, I have also understood that there will be less distortion with an AlNiCo magnet, as that is a much higher quality magnet and may have a larger region of relatively constant B (I'm no expert here, it may also be that the voice coil overhang or underhang is different when using AlNiCo).
It is likely that both speakers showed low amounts of voltage distortion because of the very low output impedance of the driving amplifier.
Cheers,
Bob
Yes, the amplifier ( nearly ) shorts the speaker so what you see is basically the distortion of the amp. I can only speculate why the Alnico distorts ( much ) less in the curent domain. One explanation is that the Alnico magnet is electrically conductive and forms a kind of eddy current brake. I have the measurements somewhere and when i fid them i will post then.
Current distortion supression is posible when the amp has a high output impedance but then equalisation is necessary. Both Malkolm Hawksford and Nelson Pass have published work about this.
Current distortion supression is posible when the amp has a high output impedance but then equalisation is necessary. Both Malkolm Hawksford and Nelson Pass have published work about this.
Another similar way of viewing it is that if the feedback is sampling the output voltage it tends to stabilize it in front of the voltage perturbation created by the load current on its open loop output impedance ( this is equivalent to decreasing its output impedance). The feedback is correcting the distorted voltage but not the distortion of the current sunk or injected by the load. If the open loop output impedance is zero, there is no perturbation of the ouput voltage created by the load with or without feedback but the distorded current remains.
On the subject of intermodulation, it seems that if it exists, the injected current will be multipled by the output voltage through the amplifier non linearity.
If they are independent sources, they can have different frequencies and intermodulation exists, if the current source in the load is dependent on the output voltage, then there is no intermodulation but simply enhancement of harmonics and this with or without feedback and independent of output impedance. What is interesting is that the measurement showed no existance of an independent source in the loudspeaker even under heavy overload.
JPV
Of course if you sample the load current with feedback, you correct the distortion of that current, but you create a current source. This has to be equilazed because the ouput impedance influences the frequency response of the loudspeaker system.
Are current driven loudspeaker better sounding ?
JPV
In my time at Audio Physic we used higher output impedance in active subwoofers. As far as i can remember we made a compromise between current and voltage output because we had a thermal problem with pure current drive. Bernd Theis, the main designer left my company to be a journalist and unfortunately i found no measurements of distortion improvements in his files. I think the paper of Prof. Hawksford and Dr. Mills contains measurements of distortion. Current drive surpresses distortion that comes from the moving impedance and thermal compression, thats for sure. The subwoofer we build ( the Terra that was later renamed to Rhea ) was very good sounding and got high praise from Martin Colloms who called it "The most athmospheric
woofer an the market". If it had worse sounding using only voltage drive i can not say. To sell a current drive amp to the public is nearly impossible because the speaker and the amp have to be treated as a system. Some tube amps have a rather high output impedance of up to 3Ohm and when you connect such an amp to a wideband driver, bass and treble goes up. Such a wideband will sound dry and dull when connected to a conventional amp with high damping factor. Why amplifier designers and speaker builders do not work more closly together is a mystery to me.
woofer an the market". If it had worse sounding using only voltage drive i can not say. To sell a current drive amp to the public is nearly impossible because the speaker and the amp have to be treated as a system. Some tube amps have a rather high output impedance of up to 3Ohm and when you connect such an amp to a wideband driver, bass and treble goes up. Such a wideband will sound dry and dull when connected to a conventional amp with high damping factor. Why amplifier designers and speaker builders do not work more closly together is a mystery to me.
Hi Joachim,
I have never been able to get excited about current drive of loudspeakers because of the equalization nightmare. As we know, speakers are designed to be flat with a voltage drive, so the equalizer that would be used would have to accurately compensate the large variations in speaker load impedance caused by the LF resonances, crossover impedance variations, etc. Of course, these responses are not precisely controlled. Then the system would sound different with every different loudspeaker used.
Cheers,
Bob
Thats what i said. Commercially it whould be a desaster. In active systems this option is open but we experienced thermal runaway problems in our subwoofer design.
There are other ways to reduce distortion in a loudspeaker that is voltage driven. Copper rings or caps can reduce the rising impedance and when cleverly executed can also supress variation of inductance over level. BL curve has to be flat and symmetrical and the suspension design has to be symmetrical too. I do that by routine the last 10 years since i use the Klippel analyser. Modern speakers can have distortion as low as 0.5% second and 0.1% third at 95dB in 1m over 100Hz when voltage driven and that aproaches the performance of good tube amplifiers.
See my atattached measurement of a small 2 way speaker i design recently. It is a 10cm woofer and a 19mm tweeter with waveguide. The rising distortion under 60Hz comes from the fact that this speaker has restricted bass response. Note especially the low amount of 3rd order distortion. Second order distortion in the bass can be brought down by a factor of 3 by using two woofers in push-pull and i do that in my subwoofer designs.
There are other ways to reduce distortion in a loudspeaker that is voltage driven. Copper rings or caps can reduce the rising impedance and when cleverly executed can also supress variation of inductance over level. BL curve has to be flat and symmetrical and the suspension design has to be symmetrical too. I do that by routine the last 10 years since i use the Klippel analyser. Modern speakers can have distortion as low as 0.5% second and 0.1% third at 95dB in 1m over 100Hz when voltage driven and that aproaches the performance of good tube amplifiers.
See my atattached measurement of a small 2 way speaker i design recently. It is a 10cm woofer and a 19mm tweeter with waveguide. The rising distortion under 60Hz comes from the fact that this speaker has restricted bass response. Note especially the low amount of 3rd order distortion. Second order distortion in the bass can be brought down by a factor of 3 by using two woofers in push-pull and i do that in my subwoofer designs.
Hi,
You and many others quote at 95/96dB @ 1m.
Can you give an indication of how distortion varies with signal level?
What is typical at 65dB, 75dB, 85dB and 105dB @1m?
In a typical passive loudspeaker the distortion goes down monothonically with lovering level until you reach the rubb and buzz region and it goes up monothonically with raising level until you reach maximum linear throw. So it is at least in theory. I have to do some measurement anyway later in the day and i can include measurements of distortion at various levels. This is interesting for me to know too.
does this mean that at 75dB@1m (-20dB ref 95dB@1m) that distortion is reduced by a factor of ~ten? i.e. from 0.5% to 0.05% of H2 and 0.01% of H3 and almost no other harmonics?
In theory yes but at such low levels we have other artifacts that come from mechanical losses in the suspension and missalignement of parts. Hysteresis playes a role too.
At lower levels i expect a wide spread of harmonics though.
By the way, i got my book. I find the chapter about open loop and wide open loop bandwidth amplifiers particular interesting. Non NFB amplifiers with diamond output stage have been pioneered here in Germany by Audio Labor back in the 70th but of cause with much less sofitication.
At lower levels i expect a wide spread of harmonics though.
By the way, i got my book. I find the chapter about open loop and wide open loop bandwidth amplifiers particular interesting. Non NFB amplifiers with diamond output stage have been pioneered here in Germany by Audio Labor back in the 70th but of cause with much less sofitication.
Bruno Putzeys says his active speaker has partial current drive. The amp to his mid-bass unit starts off at low frequency as voltage drive and then moves over to current drive (higher output impedance) at higher frequencies - the mid-range distortion is reduced significantly, according to him. You can read a bit about it on this thread (assuming you're happy dodging the crossfire
) :PSW Recording Forums: Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Current-Driving of Loudspeakers
Successfully ran my first SPICE sim using the diff amp example. The current source wouldn't take "2 mA"--kept changing it to "2". That attempt to run a 2 A current source generated an interesting error log; .002 worked.
I've found a SPICE model for the Leach amplifier on his site; if I can get the hang of this software, it should be a valuable learning tool worth far more than the price of your book. Thanks for getting me motivated!
I've found a SPICE model for the Leach amplifier on his site; if I can get the hang of this software, it should be a valuable learning tool worth far more than the price of your book. Thanks for getting me motivated!
Something I played around with about ten years ago was a mixed mode voltage current feedback arrangement. This is a simple mod that can be done with any power amplifier.
I must add that it can cause stability problems and is not for the inexperienced hobbyist.
With this arrangement the gain of the amplifier is altered by the load impedance. The gain of the amplifier is reduced with an increase of load impedance say from 8 to 4 ohms and vice-versa. The gain is fixed to a maximum by the ratio of the feedback components and the minimum gain is unity. The gain of the amplifier and output power to the load follow the impedance curve of the speaker. I found that the output power of the amplifier very nearly matched that of a maximum power transfer curve.
It did cause some minor bursts of oscillation in the crests and troughs of a sine wave at higher output power levels but I think this could be over come by some minor tweaks to the amplifier's compensation.
I didn't pursue this any further because a lack of test equipment at the time, but the sonic results make it well worth looking into. There was both a high and low frequency extension and a thickening of the mid-range in a very pleasing way. It may be a reasonable alternative to a pure voltage or current drive.
Below is an illustration of the basic form.
David.
I must add that it can cause stability problems and is not for the inexperienced hobbyist.
With this arrangement the gain of the amplifier is altered by the load impedance. The gain of the amplifier is reduced with an increase of load impedance say from 8 to 4 ohms and vice-versa. The gain is fixed to a maximum by the ratio of the feedback components and the minimum gain is unity. The gain of the amplifier and output power to the load follow the impedance curve of the speaker. I found that the output power of the amplifier very nearly matched that of a maximum power transfer curve.
It did cause some minor bursts of oscillation in the crests and troughs of a sine wave at higher output power levels but I think this could be over come by some minor tweaks to the amplifier's compensation.
I didn't pursue this any further because a lack of test equipment at the time, but the sonic results make it well worth looking into. There was both a high and low frequency extension and a thickening of the mid-range in a very pleasing way. It may be a reasonable alternative to a pure voltage or current drive.
Below is an illustration of the basic form.
David.
Attachments
Shin Nakagawa amp => http://www.fidelix.jp/products/LB-4.pdf
Each monaural had two sets of terminals, an ESL63 connected to the terminals that included the load impedance NFB path made them oscillate (likely every electrostatic).
Each monaural had two sets of terminals, an ESL63 connected to the terminals that included the load impedance NFB path made them oscillate (likely every electrostatic).
Bruno Putzeys says his active speaker has partial current drive. The amp to his mid-bass unit starts off at low frequency as voltage drive and then moves over to current drive (higher output impedance) at higher frequencies - the mid-range distortion is reduced significantly, according to him. You can read a bit about it on this thread (assuming you're happy dodging the crossfire
PSW Recording Forums: Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Current-Driving of Loudspeakers
That actually may be convenient for Bruno if he is driving his speakers with a Class D amplifier, since the necessary output filter often kills damping factor (raises output impedance). Just speculating...
Cheers,
Bob