mix current drive with voltage drive at LF?

Current drive has been discussed before here. The main advantages are reduced distortion through these mechanisms:

- force proportional to Bxl for current drive vs. (Bxl)² for voltage drive
- Le and Le modulation are rendered irrelevant because they will always be small compared to the impedance of a current drive amp
- dito for VC resistance variation due to self heating

The main disadvantage is that Q_e -> infinity, i.e. you loose electrical damping, resulting in a huge FR hump at resonance and possible uncontrolled overexcursion.

A Linkwitz transform circuit could flatten the FR, but due to the high Q involved, the compensation would be very sensitive to parameter variations, say over temperature. The problem of overexcursion is also not effectively dealt with.

Increased mechanical damping comes to mind, but the usual methods (lossy suspension, increased conductivity of VC former, air flow resistances, oil or fat damping) are usually very nonlinear in excursion or velocity or will vary with temperature.



Have there been attempts to combine voltage and current drive so that there is voltage drive around and below resonance and current drive above? In the simplest case the driver is only used above resonance, i.e. a midrange or tweeter. In this case, it might suffice to connect an inductor in parallel to the driver which would shunt voltages generated by resonant motion of the driver.


The other question I have been wondering about is, for an unmodified driver (as for the approach outlined here or the LT approach), there is only a relatively low mechanical loss above resonance. Will there be a significant difference in decay times between voltage and current drive above resonance?
 
Mixed feedback is alive and well in the Guitar World, >90% of SS Guitar amps use it.
I for one, have been using it since 1972, 50/50 mix so amp output impedance is 4 ohms for a 4 ohm load, or damping =1 , to mimic what I heard from my Fender tube amp clones which I had ceased making.
 
I honestly think this current versus voltage drive debate is one of the biggest time-wasters in audio. And I spent a lot of time building horribly unstable transconductance amplifiers in my greener days. :D

You can do it either way if you get the impedance flat. The maths says that they both then get to the same place. For sure, valves are fast devices, and I like them. But really it's all about feedback.

These days I get interested in flat impedance in a speaker, though we usually take a beating with the bass resonance, which is easiest with closed box. It's self-evident to me that it makes for an easy load for an amplifier. Which keeps the feedback in phase.

The BBC used impedance correction a lot:
Rogers Loudspeakers › LS5/9

I learned from this classic design. mh-audio.nl - Home

Not hard to estimate the inductance of a driver, or just lift the "Le" figure.

Then you can do nice things. An easy to drive speaker. What's not to like? :)
 

Attachments

  • Norton-Thevenin Equivalent Drive.png
    Norton-Thevenin Equivalent Drive.png
    20.1 KB · Views: 820
  • Marc Circuit.PNG
    Marc Circuit.PNG
    9.5 KB · Views: 811
  • Marc FR.PNG
    Marc FR.PNG
    19.1 KB · Views: 801
  • Marc Phase.PNG
    Marc Phase.PNG
    19.7 KB · Views: 798
Decades ago there was an AES article about current drive combined with motional feedback that only worked around the resonance frequency. They used an extra voice coil as the sensor for the motional feedback. A long time ago, I did some experiments with that concept using a double-coil woofer (meant to be used as a subwoofer). It seemed to work well. You need to include a filter that compensates for the magnetic coupling between the coils, though.
 
Current drive articles

Decades ago there was an AES article about current drive combined with motional feedback that only worked around the resonance frequency. They used an extra voice coil as the sensor for the motional feedback. A long time ago, I did some experiments with that concept using a double-coil woofer (meant to be used as a subwoofer). It seemed to work well. You need to include a filter that compensates for the magnetic coupling between the coils, though.
Hi Marcel,

Was that article by Mills and Hawksford? Distortion reduction of MC speaker using current drive March 1989, and Current drive power amp October 1989. I think the October article has the filter circuit to compensate for coupling between coils. A nifty circuit.

I concluded the 30dB of distortion reduction above resonance must have been due entirely to current drive since the voice coil feedback was rolled of at 500Hz, since there's not much feedback even at 200Hz for distortion reduction, but this was not mentioned in the papers.

The papers can now be downloaded from here as a compilation 38MB.

There are two other interesting papers I have seen. I have put them on myDrive in IansArticles here.
1. Catrysse, Loudspeaker Feedback (Current drive with Cap position sensor), JAES June 1985
https://drive.google.com/open?id=0B3i4bQpWjZBtZEFmeHZLbmhKUG8
2. Hsu, Current Driven Speaker MFB (2nd speaker), JAES Jan 1999
https://drive.google.com/open?id=0B3i4bQpWjZBtSlo1OUpKX0dKYUE

There's also Hans Klarskov, Speaker Builder, Jan 1990 here
1st page http://www.audiodesignguide.com/doc/Acceleration_Feedback_1.jpg

And if you are interested in parametric equalisation of speaker resonance using opamps then I also have available:
1. The Sokol parametric equaliser article 'Practical-Subwoofer-Design', Wireless World Dec-1983
https://drive.google.com/open?id=0B3i4bQpWjZBtT3FQdW9tbWo4dTQ
2. Jeff Macaulay, Variable Q speaker parametric equaliser, Electronics World, Jan-1997
https://drive.google.com/open?id=0B3i4bQpWjZBtdmp5cHBGdGw4WHM

If you are interested in a compendium of references on speaker feedback systems and current drive my Supplement to 'Speaker Feedback' EW May 1996 has about 50 refs with abstracts: https://drive.google.com/open?id=0B3i4bQpWjZBtQnJiQWx1NGpjbVk
and the EW 'Speaker Feedback' article is:https://drive.google.com/open?id=0B3i4bQpWjZBtbWhaWGMzd2VKbTQ

Last but not least, have you read Esa Merilainen's peer reviewed paper 'Comparative Measurements on Loudspeaker Distortion: Current vs. Voltage Control' under well controlled conditions and at moderate SPL?
It's here and download http://acousticsnew.ippt.gov.pl/index.php/aa/article/download/1780/pdf_255

It was a surprise to me that there was such a large difference in distortion readings with current drive at normal listening power levels of around 1W!! I had previously assumed that speaker distortion becomes inaudible with voltage drive when the cone movement is small, like normal listening level, but Esa shows there is still significant audible distortion at 1W with voltage drive. But at 1W and current drive the distortion reduces to around the threshold of audibility. That has renewed my interest in current drive of speakers (thanks Esa).

Cheers,
Ian Hegglun
 
Hi Marcel,

Was that article by Mills and Hawksford? Distortion reduction of MC speaker using current drive March 1989, and Current drive power amp October 1989. I think the October article has the filter circuit to compensate for coupling between coils. A nifty circuit.

(...)

Cheers,
Ian Hegglun

Hi Ian,

Yes, I think that was it. Thanks for all the literature references!

Regarding equalization of current-driven loudspeakers, in theory, at low frequencies where the loudspeaker can be considered a lumped system, the poles of the current to sound pressure transfer are the same as those of the current to voltage transfer - or impedance - of the loudspeaker. It should therefore be possible to make a self-adjusting equalizer that measures the impedance and adjusts itself accordingly.

With an equalization that is simply the reciprocal of the impedance graph, you get the same small-signal transfer as under voltage drive, but if you manage to accurately extract the poles it should be possible to do better than that and, for example, put the poles of a loudspeaker that even under voltage drive has a bumpy response precisely into Butterworth.

Best regards,
Marcel
 
Simulation example for Vdrive at LF & Idrive at MF+HF

Hi All,

Here's a simple PA to show Vdrive LF and Idrive mid+HF.

You can damp the bump at resonance with C4. You can trim the HF end 'lift' with R14 (an idea in my Speaker feedback article - see my previous post for link)

It includes a speaker and box model so you can get a rough idea of the SPL and LF and midF. The HF end of the speaker is not modeled (for the cone) but it still gives an idea of the HF lift R14. There's a second speaker, just uncomment it and comment the other to see it.

The idea of rolling off current feedback at LF has been mentioned in various threads eg http://www.diyaudio.com/forums/chip-amps/270956-lm1875-transconductance-amp-5.html#post4423509
and a similar circuit appeared in Lenard Audio article here http://copyright.lenardaudio.com/laidesign/images/a12/a12_cd-management.gif

I haven't built this amp, Pawel ('padamiecki') said the circuit is from Elektor. It needs heavy compensation to be stable, eg when overloaded. A second (universal) opamp is in my sim circuit so you to check the effect of a higher GBW opamp; swap the opamps then enable the .trans, then reduce the compensation cap C4 with clipping until ringing starts.

I hope this is helpful.
 

Attachments

  • diyAudio_transcond-SPL2.zip
    6.8 KB · Views: 93
  • diyAudio_transcond-SPL2.png
    diyAudio_transcond-SPL2.png
    67.8 KB · Views: 519
Is a voltage PA or a transconductance PA better for mixed V&I?

...Here's a simple PA to show Vdrive LF and Idrive mid+HF.
Did you notice the previous demo circuit shows a transconductance PA?

Other current drive circuits using clip amps, like the LM1875, are voltage output PA's; the current sensing resistor (in the speaker cold side) converts a voltage PA to current drive for the speaker.

So is a voltage PA or a transconductance PA better for a mixed V&I speaker driving? I didn't know the answer so I simulated the two types of PA's using the same demo circuit with the same open loop gains (where Av~gmRL for the transconductanve PA), both at once to plot the result. No compensation is used for this test. An Avol of 1000 (60dB) is used (most solid-state power amps can give this).

As you can see there is only 0.8dB difference at 10kHz -- so either a V or I power amp works much the same.
 

Attachments

  • diyAudio_transcond-SPL2-V-vs-I.png
    diyAudio_transcond-SPL2-V-vs-I.png
    132.4 KB · Views: 432
A variation on speaker transforms rely on state variables filters.

Albeit having a quite lot of op-amps, it has the advantage that the parameters can be set are independantly (initiel fo, Qo and target fp, Qp).

A circuit example has been shown by french excellent youtuber Jipihorn

Schematics :

https://jipihorn.files.wordpress.com/2013/06/linkwitz-variable-state.pdf

Film with full explanations in french.

https://jipihorn.wordpress.com/2013/06/03/linkwitz-et-filtre-a-variable-detat-mise-en-oeuvre/

I beardborded it only to check its noise, which I feared a bit because of the amount of op-amps. It is not audible.
 
Pole-shift equalisers

A variation on speaker transforms rely on state variables filters.
Albeit having a quite lot of op-amps, it has the advantage that the parameters can be set are independently (initial fo, Qo and target fp, Qp).
Hi Forr,

Thanks. I've been looking for anything on the Linkwitz pole-shift filter. Do you have any papers or articles (in English) you can share ? By Linkwitz?

I'd like to run the circuit you linked in LTspice when I get a chance. Does anyone already have it in LTspice to share?

I found an example of a pole-shift filter in use here
https://www.architonic.com/en/product/finite-elemente-modul-ls/1098995.

And Esa appears to use a pole-shift filter in his book or here http://www.current-drive.info/projects/17.
 
Linkwitz pole-shift filter

Re - https://jipihorn.files.wordpress.com/2013/06/linkwitz-variable-state.pdf

I had a quick look at the above circuit, & thought that the filters around U2B & U3A seemed odd. They are the same so i simmed one of them & it didn't work. On closer inspection they seem to be 9.87 Hz 6dB Oct LPF's = 23.7k + 690nf. So i added a 23.7k resistor in parallel with C2 & it now worked ! The same would apply to C1.

Anybody notice this, & have you found any other issues etc ?
 

Attachments

  • LPSF.png
    LPSF.png
    14.1 KB · Views: 411
Simulating voice coil compression with mixed V & I drive

Hi All,

With voltage drive voice coil resistance changes with temperature which causes volume compression, changing damping and crossover alignment.

With pure current drive all these unwanted effects are eliminated. But with current drive we get that undamped bump at resonance that needs equalisation or motional feedback to flatten it.

This thread discusses mixed voltage drive for damping resonance & current drive at HF. But with mixed drive (Vdrive at LF & Idrive at HF) we bring back volume compression, changing damping and crossover alignment when the voice coil temperature increases.

Here's some simulations with mixed drive to see how much voice coil temperature affects the frequency response.

The 1st attachment shows the circuit. It is the one in my previous post.

The 2nd plot shows standard voltage drive (jumper X3 open). SPL gain compression is 5dB from 0C to 200C. Damping changes by 3dB giving peaking at 60Hz at 200C. F-3dB stays about the same with temperature at 40Hz.

The 3rd attachment with mixed drive shows the frequency plots when Temp=0, 100 and 200. The voice coil resistance changes by 1.4 and 1.8 times, or 6 ohms at 0C, 8.4 ohms at 100C and 10.6 ohms at 200C. RV2 has been trimmed to get the HF gain the same as the LF gain. BTW this setting usually does not need to be changed when a different speaker is used.

With mixed drive the LF end is noticeably affected by voice coil resistance changes. There is about 4dB of gain compression at 200C (max at 150Hz). F-3dB SPL rolloff is 80Hz at 0C, rising to 150Hz at 200C. At 200C from 150Hz to 80Hz the roll off rate is gradual (-3dB/octave). At 200C this is probably less noticeable than voltage drive since peaking occurs (from 150Hz to 80Hz at +2dB/octave).

This shows mixed feedback has less negative impact on crossover alignment than voltage drive at high temperature. But the LF end rolls off sooner with mixed feedback when set for maximally flat at 0C (same as voltage drive), as mentioned 80Hz at 0C, rising to 150Hz at 200C.

The 4th attachment with mixed drive shows the frequency plots if we change the value of the V/I crossover capacitor (C4) from 10uF to 20uF giving flat at LF end at 200C and we get a consistent f-3dB of 55Hz, but now have 3dB peaking at 0C.

The 5th attachment shows the crossover capacitor changing from 10uF at 0C to 22uF at 200C, effectively reducing the V/I crossover pole frequency as the voice coil heats up. It stays almost flat but the low end f-3dB frequency still changes with temperature. It's shown here as a possibility -- it's not easily achieved in practice.

Finally, if you change the speaker you can see there is negligible difference in the plots. Fantastic! (BTW the simulation circuit attached zip includes parameters for a second speaker with a different cabinet.)

A simple mixed drive circuit like this would appear to allow plug-and-play of random speakers (within reason, nominal 8 ohm woofers and wide-range) and with minimal negatives. Fantastic! With this amp there's no need to tune an equaliser each time you change a speaker.:cool:

Also you can easily switching from standard voltage drive to mixed drive so your amp can do regular voltage drive speakers as well as current drive speakers.

With this amp current driven speakers could be passive, but it is far better to go active. And for an active system you don't need to mount the amp in the speaker enclosure because this amp can now drive 'random' speakers...

I know, all this is just simulations. But it's looks very promising so far. I'd like to make a system to see if it works as well as it simulates. Anyone else interested in making one?

Cheers,
Ian Hegglun
 

Attachments

  • diyAudio_mix-V-I-SPL-Temp4_20-Jun-2017.png
    diyAudio_mix-V-I-SPL-Temp4_20-Jun-2017.png
    15.6 KB · Views: 140
  • diyAudio_mix-V-I-SPL-Temp3_20-Jun-2017.png
    diyAudio_mix-V-I-SPL-Temp3_20-Jun-2017.png
    15.8 KB · Views: 134
  • diyAudio_mix-V-I-SPL-Temp2_20-Jun-2017.png
    diyAudio_mix-V-I-SPL-Temp2_20-Jun-2017.png
    16 KB · Views: 389
  • diyAudio_mix-V-I-SPL-Temp-cct1_20-Jun-2017.png
    diyAudio_mix-V-I-SPL-Temp-cct1_20-Jun-2017.png
    59.2 KB · Views: 407
  • diyAudio_mix-V-I-SPL-Temp5_20-Jun-2017.png
    diyAudio_mix-V-I-SPL-Temp5_20-Jun-2017.png
    15.2 KB · Views: 145
  • diyAudio_mix-V-I-SPL-Temp_20-Jun-2017.zip
    118.2 KB · Views: 62
...Anyone else interested in making one?

Cheers,
Ian Hegglun

Yes me. I have also played with mixed amps but unfortunately I can't hear a difference between current or voltage drive in mids or hights. The difference seems to be very small. So the most of manufactures takes the easily voltage method.
The mixed one sounds like a voltage driven.

I am very interested, particulary want to build current driven servo sub but I have no electronics skills at all... Maybe someone already did it and have something ready to go ;)
Maybe the ACE-Bass is interesting for you.
 
I have been playing with VFET single ended amp (output impedance = 7 ohm) for a month.

Attached 2 charts are measured with VFET SE and IcePower(Damping Factor=very high). The whole speaker is multi amp driven. The woofer is JBL 1400ND, crossover at 650Hz. The impedance peak is around 50Hz, and the peak is roughly notched out with a digital EQ when measure with VFET. The SPL number is wrong, since I don't have a SPL meter. They were measured at around 1 watt.

The third harmonics are much higher with IcePower above 70Hz. Second harmonics get high around peak resonance with VFET, and I guess it is mostly due to the 0.1H choke loaded output.

VFET sounds much cleaner, has much more resolution. Bass pitch is much easier to follow. However, modern studio recordings with loud and tight bass and kick apparently show the shortcomings of the low damping factor amp even after frequency compensation. They do still sound neutral and musical, but something is wrong or missing. We must remember that the today’s producers are making music with voltage driven speakers, and they sculpt the music for those speakers.

Well, I have no conclusion at this moment. I am still experimenting different amps. Currently, 1 ohm amp for 1st woofer, and Icepower driven 2nd woofer is staggered below 80Hz. It sounds OK at its best. If I could completely give up playing back the modern studio recordings, I would just go for VFET.
 

Attachments

  • VFET.jpg
    VFET.jpg
    125.2 KB · Views: 163
  • IcePower.jpg
    IcePower.jpg
    123.6 KB · Views: 150
Last edited: