How is the Barkhausen noise graphs in post 31 obtained?
Am I correct in understanding the assumption here is with current drive the Barkhausen noise is absent because the magnetic circuit is more stable than with voltage drive?
Am I correct in understanding the assumption here is with current drive the Barkhausen noise is absent because the magnetic circuit is more stable than with voltage drive?
The Barkhausen noise graph was obtained by using https://www.mathworks.com/matlabcen...peakers-for-aec-measurement-and-linearization. You can get the same with the latest beta of REW choosing FSAF as your measurement method.
My speculation on the absence of Barkhausen noise is that the domain flop ends up in EMF, i.e. a voltage spike. The current I=V/(Zspk+Zout), where Zout is the output impedance of an amplifier. For voltage drive, Zout->0, for current drive, Zout->infinity. Then F=I*Bl, and a=F/m, etc... A specialist such as Esa Merlinen would be the guy to confirm or discard my speculation.
My speculation on the absence of Barkhausen noise is that the domain flop ends up in EMF, i.e. a voltage spike. The current I=V/(Zspk+Zout), where Zout is the output impedance of an amplifier. For voltage drive, Zout->0, for current drive, Zout->infinity. Then F=I*Bl, and a=F/m, etc... A specialist such as Esa Merlinen would be the guy to confirm or discard my speculation.
Yeah, exactly, that's the logic that makes sense about, compared to some others floating in the forums.
btw. anyone interewted here is link to Esa's page, the book, and some articles that explain the stuff:
https://www.current-drive.info/
This one article of his is free to read and contains everything necessary to understand how this stuff works. Don't get stuck on voltage current drive amplifier debate, as amplifiers have very litte to do with any of this how acoustic distortion is reduced, because main difference is due to high circuit impedance between driver terminals and how driver can emit it's own distortion mechanisms from electrical to acoustic domain.
https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
btw. anyone interewted here is link to Esa's page, the book, and some articles that explain the stuff:
https://www.current-drive.info/
This one article of his is free to read and contains everything necessary to understand how this stuff works. Don't get stuck on voltage current drive amplifier debate, as amplifiers have very litte to do with any of this how acoustic distortion is reduced, because main difference is due to high circuit impedance between driver terminals and how driver can emit it's own distortion mechanisms from electrical to acoustic domain.
https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
Last edited:
Member
Joined 2003
Absolutely. However, BC or not, shipping my complete amp may have some cost due to the size / weight of the transformer and case. Two options, if you can provide some instruction to reproduce your measurements, perhaps I can complete them myself, unfortunately I don't have a SIG150 driver on hand. Other option is I could potentially disassemble the amp and just send you an amp board without case and transformer as a loaner. Send me a PM and we can try and work something out.
It's been many years since I followed the Mauro Penasa design information, I recall it was designed to follow some principals of Musical Fidelity A370. The thread is long, but I believe there is some good information burried in there. The LM3886 Mauro always describes as a "power current pump" in his circuit. I don't have a standard LM3886 circuit to compare, but I do like the amp for what it is. I also have a basic bottom of the line UcD180 with SMPS which I find provides a bit more "detail".
Dayton Audio DSC-165 subwoofer. Unfortunately, I did not have a good-sized box, and put it into ~6L, so the Fs went up.
Voltage Drive, H2 H3, for output varying from 60 to 80 dB SPL:
Same on the current drive:
Finally, comparing how distortions change depends on the series resistor 0/6/12/24 Ohm - and Current Drive.
Voltage Drive, H2 H3, for output varying from 60 to 80 dB SPL:
Same on the current drive:
Finally, comparing how distortions change depends on the series resistor 0/6/12/24 Ohm - and Current Drive.
HiVi D7500 midrange dome.
Voltage drive:
Current drive:
I also experimented with driving D7500 with series resistors 6/12/24 Ohm, current drive with LM3886 (GBP=2MHz, yellow curves) and LJM's MX50SE (GBP=7MHz, red curves). MX50 SE shall apply an additional 10dB of feedback loop gain, which could have lowered the distortions even further... but it did not. It appears that applying more feedback than LM3886 does not bring any lower distortions, you "run into a wall":
You can not observe any differences in objective exponential sine sweep measurement data of the acoustic output.
However, the differences are clearly audible when music (Mozart piano concerto) is used instead of sine sweep, and the residuals are objectively lower. If you can see it on spectrograms ... while ignoring different background noise... I am not so sure.
LM3886:
LJM's MX50SE:
LJM's MX50SE is certainly a better amplifier than LM3886, even if the sine sweep tests do not show it.
Voltage drive:
Current drive:
I also experimented with driving D7500 with series resistors 6/12/24 Ohm, current drive with LM3886 (GBP=2MHz, yellow curves) and LJM's MX50SE (GBP=7MHz, red curves). MX50 SE shall apply an additional 10dB of feedback loop gain, which could have lowered the distortions even further... but it did not. It appears that applying more feedback than LM3886 does not bring any lower distortions, you "run into a wall":
You can not observe any differences in objective exponential sine sweep measurement data of the acoustic output.
However, the differences are clearly audible when music (Mozart piano concerto) is used instead of sine sweep, and the residuals are objectively lower. If you can see it on spectrograms ... while ignoring different background noise... I am not so sure.
LM3886:
LJM's MX50SE:
LJM's MX50SE is certainly a better amplifier than LM3886, even if the sine sweep tests do not show it.
Great work! Could you specify the series resistor setup? Is it a current drive amp and resistor in series with the driver? Maybe a simple sketch would be the easiest way?
Thank you! The setup is a usual "voltage" amplifier, with resistors added in series to the driver, and the volume adjusted so that it produces the ~same acoustic output. Well, it can not be exactly the same due to calibration errors and the FR changes, but as good as possible.
BTW, the comparison of residuals between the voltage drive (red), current drive with LM3886, and current drive with LJM's MX50 SE (green):
BTW, the comparison of residuals between the voltage drive (red), current drive with LM3886, and current drive with LJM's MX50 SE (green):
@b_force in your reaction to this post by @tmuikku I believe are wrong (unless I am completely misunderstanding what you meant, in that case please ignore). You do not need to run simulations to see how frequency response will drastically change under current drive, you can just think about what is going on. I will explain:
The case is the same whether you use a physical series resistance or a feedback network and a sense resistor to simulate current drive. In both cases there is a resistance, R, in series with the driver in the circuit, which is the source resistance. The driver has its own impedance, Z, and so the voltage across the driver terminals varies as Vdriver = Z/(R+Z). But Z is frequency dependent for every moving coil driver, and therefore Vdriver also varies with Z. This is why, the more current drive versus voltage driver the more frequency response will undergo changes following the driver impedance. Simply put, current drive creates an impedance-based EQ network! For this reason, anyone performing comparisons between voltage and current drive, or any mix in between, must first EQ the responses to be identical in frequency response and SPL before making distortion measurements and subsequent comparisons.
I was just discussing current drive this past weekend with a guy in the speaker business. He claimed that he could get 10+dB (IIRC) using a physical resistor of 470 Ohm. Because of the losses he had to bridge the amplifier to have enough voltage and output to measure the effect. He only did it to test the effect and see what sort of distortion reduction could be achieved, but there are definitely measurable changes to distortion, mostly 3rd order.
When current drive is created using a feedback network and a sensing resistor, I think the situation is slightly different. This is because you are getting both some distortion reduction from the increase in apparent source impedance but possibly also through the feedback itself. When I say it is "simulating" a current source, this is because the feedback alters the gain of the amplifier following the driver impedance, just like with a physical series resistance. The frequency dependent impedance shows up across the sense resistor because it, too, creates a voltage divider with the driver impedance. The higher the impedance of the driver (e.g. at resonance and at high frequencies) the less the feedback. Because the feedback is returned to the negative input, less feedback means more output and vice versa. It is simulating the effect of a physical resistor and the gain of the amplifier is now impedance-dependent.
Here are links to a couple of good articles about using the feedback+sense resistor approach, authored by Rod Elliot:
https://sound-au.com/articles/cscaling.htm#s6
https://sound-au.com/project56.htm
As well as a post from a DIYaudio thread on the subject from 2006:
https://www.diyaudio.com/community/...rrent-source-amplification.91466/post-1073080
Also, you wrote:
This is correct (thermal runaway is technically possible) but in reality not a serious concern for DIY projects. Current drive tends to have less driver output SPL variation as its VC temp varies because it applies more current as the impedance magnitude rises with T. If you used a current source mode amplifier on a driver in a PA speaker and blasted low crest-factor music at maximum volume for awhile, then yes I think there would be a problem. For a home audio speaker and typical program material, I really doubt there will be any significant thermal "runaway" and the driver's "hot" SPL level will be more like the "cold" SPL level compared to 3dB or more compression for voltage drive and a "hot" driver.
Last edited:
Wow, that's a very interesting observation when using current drive. Thanks for posting your measurements on this.
SB12PFC25-4 is a typical corners-cut junk to get to the lowest price. Can the current drive make it usable?
Current drive (red) vs voltage drive with 24/12/6/0 series resistors:
At first glance, it looks promising. The LTI distortions on music (Queen's Another One Bites The Dust) filtered 300...3000Hz,
Red= Voltage drive, Yellow - current drive:
However, it is not as rosy as it may look. Alas, the sine wave measurements do not show the full picture. (Any) music shows stubborn spikes on 350Hz and 700Hz:
Where do they come from? I have no idea.
Current drive (red) vs voltage drive with 24/12/6/0 series resistors:
At first glance, it looks promising. The LTI distortions on music (Queen's Another One Bites The Dust) filtered 300...3000Hz,
Red= Voltage drive, Yellow - current drive:
However, it is not as rosy as it may look. Alas, the sine wave measurements do not show the full picture. (Any) music shows stubborn spikes on 350Hz and 700Hz:
Where do they come from? I have no idea.
How these sound like to ear? If you can could you post audio sample for voltage and current drive?
Here you go! Residuals amplified by 36dB
Try mounting it 'backwards' in the baffle and feeding sine wave to the driver at these frequencies, I suspect at least one of them is basket resonance.
Very nice info you're presenting! I found similar effects reducing H3 when adding series resistors or inductors. I like using multitone signals and looking at the 'grass' between the tones. I think it's a quick and easy visualisation.
That was a non-linear artifact, current drive only. You can't see it in the linear IR. Here is a wavelet-ish view of IR, with the window scaled up 1/f.
The box is 2' long, tapered at the end, and full of acoustic foam, least dense at the driver, and most dense at the end. You can speculate that 350Hz is about half of a standing wave lambda, but why is it only in the current drive?
The box is 2' long, tapered at the end, and full of acoustic foam, least dense at the driver, and most dense at the end. You can speculate that 350Hz is about half of a standing wave lambda, but why is it only in the current drive?
Because it is hidden by the incredibly brutal noise in the voltage drive?
Before I get lost in many words of a language I don't really speak:
Thank you!
No words can express how grateful i am.
Best regards
Bernd
My guess is that is around resonance (seeing now it's below 100Hz) of the driver in box, you have some ringing because the resonance has no electrical damping, only mechanical damping. Maybe you can check impedance curve to verify it?
IME once you really exactly(!) clone the driver terminal voltage of voltage drive those difference tend to go away completely.
The cloning method is to measure the impulse response of the driver current under voltage drive and then use that IR as the EQ (FIR convolution kernel) for the current under current drive. It contains all the warts seen in the current**) under voltage drive, including the effect of internal standing waves etc.
**) The linearized equivalent of the current, that is, precisely speaking. When the current is loaded with gross distortion then it could be the current steering may not fully/correctly EQ.
Many of the better softwares do a full Farina Log-Sweep where we can see the distortion in the current right away and know the problem zones.
When such a spike excites/stimulates a mechanical resonance the result is a non-linear artifact?
An artefact yes, but not necessarily a nonlinear one. But it might now be noticeable while it wasn't with voltage drive due to the better damping of the latter. So current drive is probably not a cure-all but another tool in the toolbox that has to be applied wisely in order to get the most out of it.