John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
jneutron, your circuit can be solved to Re in series with 2 EMF sources opposing each other, one per coil. If one coil is 90% coupled to the other coil, the total EMF is 10% of the original EMF. So it's like 90% current drive, letting 10% of the EMF back in.You have a strange way of looking at it. ;) I understand what you mean however. That is why I would prefer bifilar.

If the coils are not perfectly coupled that implies they don't occupy the same physical space, which means the leakage flux paths will encounter different nonlinearities. So you could scale the output of one coil to match the EMFs, but there will be a residual which contains only distortion. So the distortion reduction when compared to equivalent impedance drive will be less than expected. Although with bifilar winding the coupling could be very high.we also agree entirely. If I nest the pickup in the interstitials of the top layer, even scaled one has to consider the pole piece may interact differently w/r to the front plate, and also a shorting ring may affect the nested more than a two layer.

Also, no rejection of thermal compression.
Well, that was one of the first things understood.
That is why my first postings on this , I mentioned EDM slit pole piece, laminated front plate, and no shorting ring..
Jn
 
Member
Joined 2002
Paid Member
No, it's not you.
At the top of Gallery's page, you will notice this:

Please note: This is a permanent archive. Due to this gallery module becoming incompatible with the latest version of the forum software, we have frozen/archived all old galleries (and their images) and made them read-only. No new images can be uploaded, all old images are still accessible, all images and links to those images will still work, and this archive will remain permanently. Once we move to our new forum platform, we will import these galleries to the new platform's gallery system, in whatever form that exists. If you would like to report a problem with any of the images stored in this archive please email contact@diyaudio.com. Thank you for your understanding.

George
 
If you look at Demians plot, fourth one, it shows that the electrical voltage on the second coil has harmonics that exactly match that of the acoustic distortion....
Is it also expected to get usable acoustic distortion component on VRS with current drive on the DVC?
 

Attachments

  • JN-Current Drive.gif
    JN-Current Drive.gif
    18.8 KB · Views: 239
When making thoughts about speaker distortion, it should be understood that there is not only one transformation in the speaker, but two. Speaker is an electro-mechanical transducer and also mechano-acoustical transducer. So it is not enough to consider speaker as an electro-mechanical transducer. Mechano-acoustic transformation is also non-linear, covering membrane modes, break-ups, etc. This is probably not expressed in the speaker current or in the second coil voltage. The possibilities to reduce speaker distortion by using second coil voltage are IMO limited and I would not expect improvement of 30dB, this is IMO overoptimistic. Maybe I am not getting something, but I would be convinced after successful experiment that would result in an usable controlled speaker, I think we have already had quite enough of mental exercises. At some point, they have to be converted into practice.
 
Any distortion reduction by either kind current drive is limited by the instabilties of the force factor BL(x) and BL(i), plus below resonance spring force instablility, and reluctance force (and thermals with the indirect scheme) aren't addressed as well. The BL instabilities also limit any effort of using the VC as its own sensor but full VC-based MFB can eliminate some other error terms.

10dB improvement with multitone IMD while still in the most linear part of the BL(x) curve would be quite a nice figure already. 30dB... no way

----------

Effect of BL fluctuation is tricky with VC-based MFB (or with very strong local feedback in the driver, the Qes<=0.2 league): Assume BL(X) has dropped to 0.5x, this reduces small signal sense voltage at that operating point also to 0.5x which means the control thinks the velocity is too low and ramps up the current until the measured value matches the input... but since the measured value is too low, velocity of the cone renders too large, overshooting. This stops only when BL has dropped so low that control runs out of corrective gain reserve or when output current finally limits.
That's why too tight a control loop control doesn't work with an unstable BL because it wrecks the sensor linearity. There should be a sweet-spot around resonance, though, were the overshoot tendency partly cancels the BL drop because it's just the right amount. Around resonance loop gain is highest when the amp is current output so mabye not the best choice to exploit that effect. It appears to work best when the open-loop behavior is most linear to begin with (sounds familiar), installed by proper degeneration of the current drive around resonance.
 
Last edited:
I have "reactivated" my current drive setup (it took all time from breakfast to lunch break). In principle, it works this way:

V_I_converter.GIF


Then I measured simultaneously tweeter (Beyma T2030, new) acoustical output and also the spectrum of the driving current. Please see it attached and you will see that acoustical distortion is much higher than driving current distortion. It is measured at 1Vrms across speaker terminals. Please note that acoustical distortion is about 30dB higher than current distortion.
 

Attachments

  • tweeter_current_drive.png
    tweeter_current_drive.png
    58.8 KB · Views: 205
Ah, I hadn't thought that would appear as an amplifier current distortion. I clearly need to go away and get my head around this.

I am not sure what you speak about. It is only in the acoustical output. Top trace is acoustical measurement, bottom trace is current spectrum. At some point, above certain level, acoustical output is messed with HF, though driving current spectrum remains unchanged.
I assume they are Al cone breakups.
 
Last edited:
The peak just above 10k in the bottom trace

It is catching that single peak from the air, it is there even without signal when measuring noise. The amp is not in the box, it is a "true DIY" , I think it catches some EMI, DC/DC chopper or so, maybe into measuring groundloop. It is unimportant, as it does not change with signal or level. I find much more interesting that the idea that acoustical distortion is mirrored in current is not general and may cover only certain cases. IME it is misleading to rely on such idea.
 

Attachments

  • noise.png
    noise.png
    35.5 KB · Views: 139
Last edited:
Status
Not open for further replies.