john curl said:
Don't throw the baby out with the bathwater, John. Bose or Sony is certainly not the only alternative to "cottage industry" guys who can't or won't do the most basic of testing of their product. Your own stuff is a good example - you DO measure.
I have patiently explained the pro's and cons of THD-20 and CCIF IM with spectral analysis. I am certainly not fixated on THD-20, as you wrongly assert. However, if all you did was test THD at 5 kHz and nothing else, I would certainly call you to task for it. If someone was only willing to give me one number as a distortion spec I certainly would pick full-power THD-20 over THD-1. If that is fixation, I guess I am guilty. But the real world is not that simple. As you say, and as I agree with, we want to know about the spectral makeup of the distortion. For that, as I have pointed out many times, given today's instrumentation available, full spectral analysis of CCIF 19 & 20 is probably best.
As far as TIM testing goes, you don't even spec it in your designs - all you say is "unmeasurable" - you don't say how low unmeasurable it is. TIM testing has virtually no value over THD-20 insofar as testing distortion that results from any SID mechanism. Period.
Bob
Dear Bob/Johan and other fanatics,
Although I am changing the subject from Transistors vs Fet, to Measurement and Analysis, this is a quick one that does not need to go to any extended discussion, just an opinion.
I have been messing around with my equipment this evening while my wife is playing bridge with some friends. This normally happens when I am bored.
Reading an article somewhere in the past in one of the diyAudio discussions someone mentioned placing a resistance in series with the loudspeakers.
I noticed a subtle difference in that the same song became more pleasant for the lack of a better word. The nominal speaker impedance (or rather DC resistance) is 3.2 Ohms and I placed a 1 ohm 30 watt in series.
I cannot imagine any real change in the performance of the amplifier and I do not listen at deafening volume settings either, but by-passing the resistor (and resetting the audio level) and then switching it back into circuit offers a slight but audible mellowing effect.
Any comments in this regard. Could it be audible or am I imagining it. It is not better or worse, only different.
Although I am changing the subject from Transistors vs Fet, to Measurement and Analysis, this is a quick one that does not need to go to any extended discussion, just an opinion.
I have been messing around with my equipment this evening while my wife is playing bridge with some friends. This normally happens when I am bored.
Reading an article somewhere in the past in one of the diyAudio discussions someone mentioned placing a resistance in series with the loudspeakers.
I noticed a subtle difference in that the same song became more pleasant for the lack of a better word. The nominal speaker impedance (or rather DC resistance) is 3.2 Ohms and I placed a 1 ohm 30 watt in series.
I cannot imagine any real change in the performance of the amplifier and I do not listen at deafening volume settings either, but by-passing the resistor (and resetting the audio level) and then switching it back into circuit offers a slight but audible mellowing effect.
Any comments in this regard. Could it be audible or am I imagining it. It is not better or worse, only different.
Nico Ras said:
Hi Nico,
Sorry, but I am not sure what you are saying here. Distortion does ALWAYS exist, no equipment is perfect. Taking into account the fantastic resolution capabilities of hi-end (
) test equipment, it would be safe to say that if the distortion is not visible on the test equipment it is not audible. We can measure THD-20 to less than 0.001 %, phase shift to less than a degree, freq response errors to perhaps less than 0.01 dB. NO BODY can hear that by several orders of magnitude.Jan Didden
Nico Ras said:
Assuming you started with a solid state amplifier with a damping factor of, say greater than 50, you just reduced your damping factor to about 8. Much like that of a tube amplifier. Maybe you heard what a lot of people may be hearing when they listen to a tube amp - the effects of a reduced damping factor. Depending on the impedance vs frequency characteristics of your speakers, it will cause a coloration.
Bob
Bob Cordell said:
Bob, I think you may have the right answerd. Although it sounded nice for a while with different material I started to note a sloppyness in bass passages, this I guess is from the amplifier asserting less control over the loudspeaker motion. This being quite a heavy bugger - Old B139 Mk1 (the 13 x 11 inch job).
I have removed the resistor though the high end sounded a bit smoother maybe, maybe more attenuated.
Thank anyway Bob.
Bob, I was 'teasing' you about the 20KHz fixation. However, it does cause a real problem with harmonic analysis. I realize that you like 19-20K difference IM, but I can't easily use that without a major test equipment change, and I don't think that it is worth it.
It is true that I don't normally measure TIM in my amps. They are fast enough to be TIM free.
It is true that I don't normally measure TIM in my amps. They are fast enough to be TIM free.
Hi Nico,
adding the 1r0 to a 4ohm speaker will change everything.
If you added 1r0 in series with the bass driver, then you would change the Q of the installation.
The lower Q of the bass unit will extend the bass, but also change all the other factors the designer selected for best overall response.
Putting the 1r0 in series with the crossover will be far more complex than just the bass driver alteration.
Try modelling the bass mod only.
adding the 1r0 to a 4ohm speaker will change everything.
If you added 1r0 in series with the bass driver, then you would change the Q of the installation.
The lower Q of the bass unit will extend the bass, but also change all the other factors the designer selected for best overall response.
Putting the 1r0 in series with the crossover will be far more complex than just the bass driver alteration.
Try modelling the bass mod only.
Hi Andrew,
The bass freuency response did not change audibly (sorry it is an 8 ohm driver). The Q may have changed somewhat, but it is front fired horn loaded so the horn really is predominant characteristic.
As I mention earlier, maybe I am imagining that I hear a difference. The experience I have had when you fiddle or as some refer to it, tweak, you always espect to hear a difference, and then you do.
Maybe a previous comment is valid that you are approaching the valve sound as the current source is now a voltage source.
The bass freuency response did not change audibly (sorry it is an 8 ohm driver). The Q may have changed somewhat, but it is front fired horn loaded so the horn really is predominant characteristic.
As I mention earlier, maybe I am imagining that I hear a difference. The experience I have had when you fiddle or as some refer to it, tweak, you always espect to hear a difference, and then you do.
Maybe a previous comment is valid that you are approaching the valve sound as the current source is now a voltage source.
People tend to worry too much about damping factor. Sometimes, a speaker will sound better with a series resistor. I did this about 43 years ago with an AR-1 (4ohm) loudspeaker by adding a 3.3 ohm resistor in series with it. It sounded better. Later, I found that AR had an additional 4 ohm resistor that was optional on the speaker itself. Today, I use a .3ohm series resistor in series with my WATT 1 loudspeakers, because the early Bybee network demanded a .3ohm resistor. Years, before I had to add a .5ohm resistor to the input of my first pair of WATT 1 loudspeakers in order to keep the Electrocopaniet amp that I was using from oscillating at 2KHz. It made little difference to the overall sound.
Carver made a powerful amp with an optional (current output) connection. It was an additional 1 ohm resistor that sometimes improved the overall sound.
The reason that we often think that a high damping factor is best is because many loudspeaker designers will 'voice' or 'fine tune' their design around a high damping factor, because that is what is normally available.
Carver made a powerful amp with an optional (current output) connection. It was an additional 1 ohm resistor that sometimes improved the overall sound.
The reason that we often think that a high damping factor is best is because many loudspeaker designers will 'voice' or 'fine tune' their design around a high damping factor, because that is what is normally available.
john curl said:
Good points, John. It is actually a little unfortunate that it is called Damping factor rather than just output impedance, because this causes people to associate its effects often mostly with the woofer. What often concerns me more is the coloration that can be introduced into some speakers in the midrange around the crossover points. Of course, the amount of midband impedance variation among different speakers is quite a bit. If a speaker goes from 2.5 ohms to, say, 15 ohms in the midband due to the crossover (Linkwitz-Riley XOs seem to do it the most), then there can be a moderate amount of midband coloration due to drive impedances on the order of an ohm. Other speakers have very tame midband impedance variations.
I also often wonder what kind of an amplifier the speaker manufacturers voiced their speakers with. The answer? Probably their favorite amplifier! A guy that likes tubes might voice his speakers with them, and then they might really benefit from a resistor in series with a solid state amp driving them.
I like that when they measure power amps in Stereophile they show a frequency response into a simulated speaker. It would be really great if they took a frequency response of the amplifier driving the actual speaker combination used in the subjective review of a given pair.
Bob
Bob Cordell said:
Anyway, "cross-over" frequencies of speakers impact on "cross-over" output emitter followers, so a series resistor helps to reduse this impact, right?
I don't see how inductive phase angles can increase distortions when they increase load impedance; it is obvious that higher load impedance reduces distortions of emitter followers; in turn, I already many times mentioned mechanism of dynamic distortions caused by emitter followers loaded by capacitive loads. It explains as well why your FET output amplifiers sound more natural than similarly measured BJT amps; the answer is simple: wider and smoother crossover region.
Dear Bob,
I knocked together a little amplifier this afternoon and this evening using some MOSFETs I had in my component boxes.
I also used full complementary symetry and a few other tricks. It was built from my simulation program (MicroCAP8) onto protoboard to see what happens.
Either I missed something but both the simulation program and my distortion analyser shows only reveals a second harmonic and nothing more. All higher order harmonics are way down near the noise floor.
It is lying in bits and pieces on the floor driving one speaker, while a KSA50 is driving the other speaker. The gains was adjusted so that they are identical and I am feeding both amps with L&R mono.
The MOSFETS sounds much nicer(for lack of a better description) than the transistor amp, eventhough they share similar topologies.
Any clues or is this what is typical of MOSFETS. I used 2SK1058/2SJ162 and bias current is about 100 mA.
Any comments?
Kind regards
Nico
I knocked together a little amplifier this afternoon and this evening using some MOSFETs I had in my component boxes.
I also used full complementary symetry and a few other tricks. It was built from my simulation program (MicroCAP8) onto protoboard to see what happens.
Either I missed something but both the simulation program and my distortion analyser shows only reveals a second harmonic and nothing more. All higher order harmonics are way down near the noise floor.
It is lying in bits and pieces on the floor driving one speaker, while a KSA50 is driving the other speaker. The gains was adjusted so that they are identical and I am feeding both amps with L&R mono.
The MOSFETS sounds much nicer(for lack of a better description) than the transistor amp, eventhough they share similar topologies.
Any clues or is this what is typical of MOSFETS. I used 2SK1058/2SJ162 and bias current is about 100 mA.
Any comments?
Kind regards
Nico
Another thing I would like to mention is that although the gains are set identical on both amplifiers using a sine wave, when music is played the KSA50 sounds amost twice as loud while the MOSAMP is tame and smooth.
The KSA50 bias currrent is 940mA per device while the MOSAMP is just on a 100 mA per device, which means the KSA50 is running a quasi class A mode.
I am not playing particulary loud as it is 02:30 in the morning and the rest of the family sleeping, so lest say an average of about 7 watts.
The KSA50 bias currrent is 940mA per device while the MOSAMP is just on a 100 mA per device, which means the KSA50 is running a quasi class A mode.
I am not playing particulary loud as it is 02:30 in the morning and the rest of the family sleeping, so lest say an average of about 7 watts.