Yes, and with an active amp emulating an inductor you can shape the response without needing to greatly increase headroom. But transitioning from voltage drive (>1ohm) at 100Hz to current drive (>20ohm) at 200Hz with a flat response is a bear of a problem to solve, and you don't get much benefit from inductance except well above Fs, at least for typical woofers like in my measurements.
You should emulate an inductor that is at least as large as Lvc. The greater the better. This would also mean that the reponse starts to drop off earlier than caused by Lvc and Re alone.. But because this is not caused by a physical series inductor it should be easy to equalise this out before the amp.
Regards
Charles
Regards
Charles
You mean by incorporating a filter, mimicking the inductor's transfer function, in de FB loop?
Well you take feedback from two things, the speaker voltage and the speaker current, and you transition from voltage feedback at LF to current feedback at HF and this naturally produces an output inductance characteristic. The amp gain at LF is set by voltage feedback and the gain at HF is set by current feedback so it is possible to flatten the response either by contouring the transition (contouring the output impedance), or adding input RC networks. You could set the current feedback so you will have high HF gain, and then you have more headroom to use RC networks at the input.
That's my recollection anyway from past experiments.
That's my recollection anyway from past experiments.
Last edited:
I've found in experiments that steeper than 1st order "crossover" slopes don't work well and are not worth the effort (for example, using a 2nd order split complicates the crossover pretty much as it needs to be linear phase, which means it has huge overlap and significant peaking in the individual voltage- and current-contributed paths).
Unfortunately not many class-d amps expose thr Feedback loop. LM3886 would be a good choice for a midrange amp with mixed feedback I think.
https://www.diyaudio.com/community/threads/lm3886-current-feedback.150767/
https://www.diyaudio.com/community/threads/lm3886-current-feedback.150767/
Thank you! I missed it last year :-(
List of lame excuses:
- occupied in another forum that time
- occupied with taming happy camping parrots on diyAudio
- lost in fruitless discussions about transducer distortion mechanisms, and shorting rings..."
- lost with only theoretical heroes here who type more or less gross nonsense with their keyboard.
With my DUT i use rockwool - this way there is no relevant CB resonance (fb).
And no "sound coloration" from the standing waves inside at higher frequencies either.
Last edited:
Dear >keantoken<
I really don't know how to come across as polite. So I write briefly and roughly, and beg for your indulgence. I only see "Girlanden" garnished with "Lametta" in your graphics. (Please use a translator of your choice to understand the “German” terms). Your graphics are enough to prompt long texts written by members who cannot have recognized anything. They just write what they always write, which is boring. Now it's out and I have questions.
Where in the setup did you measure the 2.83 Volts?
Where is 1 Watt?
I would like to see the associated SPL frequency responses, is that possible, please?
Best regards,
Bernd
What did you get?
I really don't know how to come across as polite. So I write briefly and roughly, and beg for your indulgence. I only see "Girlanden" garnished with "Lametta" in your graphics. (Please use a translator of your choice to understand the “German” terms). Your graphics are enough to prompt long texts written by members who cannot have recognized anything. They just write what they always write, which is boring. Now it's out and I have questions.
Where in the setup did you measure the 2.83 Volts?
Where is 1 Watt?
I would like to see the associated SPL frequency responses, is that possible, please?
Best regards,
Bernd
The SPL responses are in the REW files. You can open all 3 files in REW and compare them very quickly in a number of different views.
2.83V is across the speaker at all frequencies both for voltage drive and current drive. I did this by exporting the response difference as an impulse response (which is in the REW files) and using it with a convolver on my sound output. So on both graphs the speakers have the same SPL level at all frequencies. 2.83V is 1W into "8ohms", I chose this because you can at least roughly know where that is relative to your own listening levels or another measurement. I wanted to do 96db SPL measurements but I couldn't find my sound level meter and I don't think my amp has enough headroom.
2.83V is across the speaker at all frequencies both for voltage drive and current drive. I did this by exporting the response difference as an impulse response (which is in the REW files) and using it with a convolver on my sound output. So on both graphs the speakers have the same SPL level at all frequencies. 2.83V is 1W into "8ohms", I chose this because you can at least roughly know where that is relative to your own listening levels or another measurement. I wanted to do 96db SPL measurements but I couldn't find my sound level meter and I don't think my amp has enough headroom.
I did some tinkering and came to the conclusion that a differentiating current feedback path emulates the effect of an additional series output inductor (LS).
Here you can see how it can be used around an inverting amp:
Cshelf and Rshelf are there to compensate for the premature HF rolloff due to the inductor. Because the mutual influence of output inductor and driver impedance dips and peaks can be rather complex this is of course not sufficient by itself.
Differentiators are not per se stable in practice if they are designed to differentiate up to very high frequencies so it is advised to truncate its function up there by the insertion of a reasonably dimensioned resistor (the "dashed one").
Regards
Charles
Here you can see how it can be used around an inverting amp:
Cshelf and Rshelf are there to compensate for the premature HF rolloff due to the inductor. Because the mutual influence of output inductor and driver impedance dips and peaks can be rather complex this is of course not sufficient by itself.
Differentiators are not per se stable in practice if they are designed to differentiate up to very high frequencies so it is advised to truncate its function up there by the insertion of a reasonably dimensioned resistor (the "dashed one").
Regards
Charles
Last edited:
Well, there is the problem...
Dear >keantoken<
Thank you for your reply. Just let me tell what i did to prove my point. I do this in the form of an advice like in a manual. This way it fits a second purpose as an edit in post #1. It is a real chore to write sense in a foreign language ;-)
[1]
First (without the resistor) in ARTA-STEPS the generator was set to 63.17Hz. Then a multimeter was connected at the tranducer. 2.9V were set with the output volume knob at the interface. Then a SPL frequency response graph (including a step at 63.17Hz) was recorded with microphone. This graph was set as an overlay. Hence yellow.
[2]
Then the 20R resistor was put in series to the transducer. The overlay still on the screen, another measurement was performed. The volume knob was turned up until the reading on the actual graph (hence green) was the same SPL at the 63.17Hz step.
[3]
The generator was restarted at 63.17Hz. The multimeter at the transducer Terminals now reads 5.9V. At the same SPL as without the 20R resistor!
There is no need to EQ as long the frequency range of interest is roughly the same SPL. With the same SPL (eg. at a crossing) at the midpoint.
My advice is not to EQ and to run a true 2Ch measurement with a voltage probe at the transducer terminals. To get a valid reference for the Measurement system. This is not possible with EQ. This procedure excludes the always noisy distorting amplifier from the SPL (distortion) result. There is no need to own a lab grade amplifier. Sadly i bought a lab grade amplifier to get EQed responses. Nobody needs a lab grade amplifier to enjoy Speaker audio. Me neither. For sure any sane person will EQ their audio system. After evaluating the distortion of their proudly DIY made distortion free Loudspeakers.
Now >keantoken<, please show the proudly convolved fundamentals. And perform a valid 2Ch measurement. Clean up the display to D2 and D3. With 20R and without. Please!
Best regards,
Bernd
So far I didn't know you had a point to be proven. It still is quite unclear to me what it might be.
I could do the measurements again, but they would show the same result. I did them this way because the SPL response so close to the speaker is not realistic and will be different for every speaker. Normalizing shows what the motor is doing without the unrelated noise of the acoustic response.
The EQ was done in software with no distortion, and the amp I used has less than 0.01% distortion. The microphone SPL curves I got are the same for voltage drive and current drive. This could not be the case unless the voltage at the speaker terminals was the same at all frequencies.
I did the measurements the way I did so that the effect of 20 ohm drive would be completely isolated from any other variables.
The EQ was done in software with no distortion, and the amp I used has less than 0.01% distortion. The microphone SPL curves I got are the same for voltage drive and current drive. This could not be the case unless the voltage at the speaker terminals was the same at all frequencies.
I did the measurements the way I did so that the effect of 20 ohm drive would be completely isolated from any other variables.