So if I understand well, this is just with a series resistor or a series resistor inside the amplifier's feedback loop?
I am not really sure what you're trying to show in what context?
Because this only shows that a 10ohm resistor won't do an awful lot.
Which was to be expected already.
It does lower H3 a little, in the amount one would expect.
A 10ohm resistor in series is most certainly not enough to call the circuit current drive.
Because this only shows that a 10ohm resistor won't do an awful lot.
Which was to be expected already.
It does lower H3 a little, in the amount one would expect.
A 10ohm resistor in series is most certainly not enough to call the circuit current drive.
Thanks, I change my level to the level of the patient!
How am I doing?
I would never consider just putting a resistor a in series as current drive.
I am just a bit confused, because the TS mostly wanted to measure the distortion of current?
So anything that is big in value will influence that measurement.
On the other hand, just a resistor in series is not the most effective way to tackle distortion either.
So again, I am a bit confused that you're trying to show us?
Last edited:
Nice to see you cheer up
A 'current drive' amplifier is essentially a high impedance voltage source. Adding series resistance larger than the reactive load mostly defines the current delivered = current drive. It is not a practical way to build a high current source but it is an easy one for test purposes.
The increase in output impedance of the 2x 10R resistors is clearly large enough to have a significant impact on the low frequency range where the impedance swings about as well as at the inductive rise in the top end, so the behaviour of the drive current is certainly affected here.
I think my test is possibly flawed in two ways:
1) The LF distortion products are probably dominated by port design.
2) Output impedance is still small comapred to the inductive peaks in the impedance. Perhaps I need closer to 100R to see significant change in control of Lvc?
I could try again with a closed box and perhaps a different woofer if there is legitimate interest? The biggest labour cost is EQ'ing the response to match the voltage drive.
A 'current drive' amplifier is essentially a high impedance voltage source. Adding series resistance larger than the reactive load mostly defines the current delivered = current drive. It is not a practical way to build a high current source but it is an easy one for test purposes.
The increase in output impedance of the 2x 10R resistors is clearly large enough to have a significant impact on the low frequency range where the impedance swings about as well as at the inductive rise in the top end, so the behaviour of the drive current is certainly affected here.
I think my test is possibly flawed in two ways:
1) The LF distortion products are probably dominated by port design.
2) Output impedance is still small comapred to the inductive peaks in the impedance. Perhaps I need closer to 100R to see significant change in control of Lvc?
I could try again with a closed box and perhaps a different woofer if there is legitimate interest? The biggest labour cost is EQ'ing the response to match the voltage drive.
It does not help to produce graphs that are labeled D (distortion). If i am right you are using ARTA sine sweep farina method. This does not work. Do not ask me how i know.
01) Use STEPS (ARTA)
02) Bypass (omit) your equalizer
03) Set the STEPS - range to eg. 1000Hz-2000Hz (to save time)
04) Set the Steps to 1/6 oct (to save time)
05) Level match both measurements crudely with the output volume control of the interface/amplifier/software output mixer.
06) Use moderate Volume.
07) Publish the two fundamentals
08) Publish D3 and D3o in %
The distortion reduction by resistor is a given fact!
20R @4R nominal transducer is a guarantee for a significant difference. My measurements above are 20R @8R nominal transducer. To scale my measurements you could even use one 10R.
oh, and
09) ignore impolite useless members ;-)
Last edited:
I've got to ask 😋 Why not trust arta sine sweep method?
I'll try your method tomorrow.
I do know there was a problem of compatability between older and newer arta files that made the sine sweep distortion display wrong. It was corrected in thr latest version.
IMO we must match the frequency response with an equaliser so we have results that can be compared. If we only match the SPL (at 1khz?) using volume control we will see difference due to spl level difference.
I'll try your method tomorrow.
I do know there was a problem of compatability between older and newer arta files that made the sine sweep distortion display wrong. It was corrected in thr latest version.
IMO we must match the frequency response with an equaliser so we have results that can be compared. If we only match the SPL (at 1khz?) using volume control we will see difference due to spl level difference.
Welcome! Farina drove me nuts. Equalizing drove me nuts*. Then I read the ARTA manual. It says (analogously): Farina is a sensitive racehorse, STEPS also works in a coal mine. It is much less sensitive to noise, to reflections... . My living room is no coal mine, but STEPS works for me there. One point that may become important: With STEPS you get a chance to hear the amplifier clipping. Do not ask me how i know. With STEPS i can do an empty measurement for noisefloor by simply disconnecting the loudspeaker. I set this noisefloor to an overlay, so i see when a "D" sinks in it.
*) The demand in forums that a meticulous adjustment of the frequency response is absolutely necessary for a valid, meaningful measurement of "current drive" is gross nonsense. What needs to be equalized? It is only possible to equalize the excitation. However, a harmonic from non-linear distortion occurs in the drive current. Or in the material of the moving mass. Or both connected. An EQ in the excitation has no influence on either.
Just have a look at my microphone measurements in #157. There the fundamentals. It is not the perfect match - and more important: the "fine print" shows, that a perfect match is impossible. So if we want an engineering result for let us say 200Hz to 500Hz, we may do a simple level match for that range. The result will be much better than dialing on a parametric EQ for hours. Do not ask me how i know.
Than we might as well just close this entire forum lol.
Politeness is something that is extremely relative and subjective.
Something that might seem polite or impolite to you, could be very much to opposite for someone else.
The only thing we can do, is to ask each other for clarification.
You could for example also just totally misread someone.
Just calling people out because they don't respond exactly on the way you would like is pretty rude and impolite for a lot of people as well.
Last edited:
I was never cheered down to begin with?Nice to see you cheer up
A 'current drive' amplifier is essentially a high impedance voltage source. Adding series resistance larger than the reactive load mostly defines the current delivered = current drive. It is not a practical way to build a high current source but it is an easy one for test purposes.
The increase in output impedance of the 2x 10R resistors is clearly large enough to have a significant impact on the low frequency range where the impedance swings about as well as at the inductive rise in the top end, so the behaviour of the drive current is certainly affected here.
I think my test is possibly flawed in two ways:
1) The LF distortion products are probably dominated by port design.
2) Output impedance is still small comapred to the inductive peaks in the impedance. Perhaps I need closer to 100R to see significant change in control of Lvc?
I could try again with a closed box and perhaps a different woofer if there is legitimate interest? The biggest labour cost is EQ'ing the response to match the voltage drive.
Always cheery on this side!
That is a gross oversimplification of current drive.
But how effective the resistance/impedance is, depends on the impedance at a certain frequency of course.
I am mostly just confused what your goal is at the moment?
1 - Is it measuring the current distortion?
In that case the reference resistance needs to be as low as possible to not have a significant impact on the measurements.
2 - Lowering the distortion in general?
In that case there are better and more efficient ways, like adding an inductor or notch filter.
This time I closed the port of the loudspeaker box. The mid-range distortion result is a lot cleaner to analyse. I also switched to a ground-refenced amplifier.
Below: System noise floor.
Below: Voltage Drive EQ unit physically out of the signal chain.
Below: Voltage drive with EQ unit in the loop, but bypassed.
Below: Current Drive 10R level matched at 1KHz - 2KHz . EQ physically out of signal chain.
Below: Current Drive 10R with EQ to match the response to voltage drive.
Below: Current Drive 20R with EQ to match response to voltage drive.
Below: Current Drive 20R with EQ as distortion percentage.
Below: Voltage drive as percentage.
Conclusion: 20R output impedance provides a reduction mostly in 3rd harmonic. The most notable change is around 850Hz where voltage drive gives 1% D3 and 20R current drive gives 0.4%. Nice!
We see the biggest reduction in distortion where the woofer impedance is low. I think this is because the 20R series resistor will dominate current delivery most at that frequency.
If we have a very high output impedance such as from a current mode amplifier, will the distortion reduce that same amount (1% to 0.4%) across the full frequency range? Or are there many more factors?
If we have a very high output impedance such as from a current mode amplifier, will the distortion reduce that same amount (1% to 0.4%) across the full frequency range? Or are there many more factors?
I have experimented with distortion measurements and series resistors/coils to increase driving impedance on some drivers, and not all drivers behaved the same. I think even in Meriläinen's paper it was mentioned that drivers with alu voice coil formers were not responding as well as non conductive formers, and that seems to correspond with what I've seen too. They way I have done it is with multitone signals and measuring the acoustic output.
Hi, you are right, in general, better drivers get less benefit, while poor (motor) drivers more. Also, only some motor related problems get helped. If all motor issued were fixed, even still at some point some other distortion mechanisms start to dominate, like suspension.
Distortion wpuld reduce by reducing excursion as well, so effective distortion reduction mechanism is highpass filter, and likely electrical damping at the main resonance as well if it prevents un-necessary excursion, from xternal sources for example.
Although, it's quite complicated stuff as per in this very thread, jump resonance stuff and all. Everything depends Use good system design, best driver, and perhaps try some impedance manipulation. In the end, distortion doesn't seem to be much of a problem.
Distortion wpuld reduce by reducing excursion as well, so effective distortion reduction mechanism is highpass filter, and likely electrical damping at the main resonance as well if it prevents un-necessary excursion, from xternal sources for example.
Although, it's quite complicated stuff as per in this very thread, jump resonance stuff and all. Everything depends
Use good system design, best driver, and perhaps try some impedance manipulation. In the end, distortion doesn't seem to be much of a problem.
Last edited: