Hello,
The downside of a current transformer is as analog was saying, changing wire resistance with temp.
I think it could be worth trying for lower powered amps though.
As far as the hall effect sensors are concerned their linearity and insensitivity to drift much over wide variations in temperature seem to make them an excellent candidate, the closed loop linear current sensors as Jaka has shown are said to be among the best type for linearity of ~0.1%, and bandwidths of 100Khz.
They appear to be nothing more than a very fancy current transformer.
I think both methods are worth trying, but for sensing actual output capacitor current, the hall sensor seems king.
Regards,
Chris
The downside of a current transformer is as analog was saying, changing wire resistance with temp.
I think it could be worth trying for lower powered amps though.
As far as the hall effect sensors are concerned their linearity and insensitivity to drift much over wide variations in temperature seem to make them an excellent candidate, the closed loop linear current sensors as Jaka has shown are said to be among the best type for linearity of ~0.1%, and bandwidths of 100Khz.
They appear to be nothing more than a very fancy current transformer.
I think both methods are worth trying, but for sensing actual output capacitor current, the hall sensor seems king.
Regards,
Chris
Hi Chris,
tha fancy part of LEM current transducer is the Hall sensor and a little PCB containing opamp with discrete buffer. Current transformer itself is wound on laminated iron core, not even ferrite is used.
Problem with small ferrite current transformers is low frequency response. While 400kHz is reproduced faithfully, 50Hz does not go through and can easily saturate transformer. Temperature of winding is no problem since current transformer is a current source. But secondary winding must be designed for expected Volt second product and this is where core becomes large for low frequency response.
One possible technique I have never seen used is integrating inductor voltage. Differential integrator would have to be used and that is probably too much for opamp.
For capacitor current sensing I think that Fig. 19 from Carver patent that you linked is the way to go.
Best regards,
Jaka Racman
tha fancy part of LEM current transducer is the Hall sensor and a little PCB containing opamp with discrete buffer. Current transformer itself is wound on laminated iron core, not even ferrite is used.
Problem with small ferrite current transformers is low frequency response. While 400kHz is reproduced faithfully, 50Hz does not go through and can easily saturate transformer. Temperature of winding is no problem since current transformer is a current source. But secondary winding must be designed for expected Volt second product and this is where core becomes large for low frequency response.
One possible technique I have never seen used is integrating inductor voltage. Differential integrator would have to be used and that is probably too much for opamp.
For capacitor current sensing I think that Fig. 19 from Carver patent that you linked is the way to go.
Best regards,
Jaka Racman
classd4sure said:
Rob Williamson (the inventor listed on the Carver patent) was a long time friend of mine. Sadly, he died suddenly and unexpectedly several years ago. He, I and several other friends had many a pleasant engineering discussion around my kitchen table. He was a valued technical sounding board and provided much encouragement that enabled me to work out my leapfrog design technique.
Active damping as outlined in the Carver patent does not look at or feed back dc current, but it is more-or-less a first order subset of the general leapfrog method (which provides the optimum control solution for any number of low pass ladder filter elements). I can't remember if it is written up in the patent, but I think Rob was the one who recognized that it is possible to look at output capacitor current by steering a small fraction through a much smaller parallel RC sense network.
For producing high fidelity audio, feedback signals must be virtually distortion free (after all, they are the standard to which feedback aspires). Quality voltage feedback is transparently obtained at the cost of a few resistors or maybe differential opamp. Too bad equally good current feedback (especially down to dc) is so problematic.
Regards -- analogspiceman
Hello, a couple of years ago I also had to build some circuits with a variant of current mode control.
Both Fets opened synchronously, the current was measured across R14 and R15. The resistors R16 through R19 removed common mode swing and doubled the measured output. Maybe, this is an idea for full bridges.
I also used current transformers for C-leg current sensing in a piezo-motor application at around 80kHz with very small ferrite cores, diameter was about 6mm from Epcos. Have to check the correct data, if interested.
Regards, Timo
Both Fets opened synchronously, the current was measured across R14 and R15. The resistors R16 through R19 removed common mode swing and doubled the measured output. Maybe, this is an idea for full bridges.
I also used current transformers for C-leg current sensing in a piezo-motor application at around 80kHz with very small ferrite cores, diameter was about 6mm from Epcos. Have to check the correct data, if interested.
Regards, Timo
Attachments
analogspiceman said:
I would be very cautious about trusting the integrity of a so called ground referenced current sense. My experience is that it is much better to sense "ground" referenced current differentially rather than try to star signal ground from a high current sense resistor. Your layout person will thank you.
Hello
Could you please give some hints about that?
I am planning the design with ground reference current sensing.
I thought to use a ground plane for power things and speaker return and take the ground directly out of current ense resistor (befor it connects to the plane) for a signal ground (input and modulator).
Is it the no-no situation you've mentioned here? Needless to say I am going to be my layout person.
Ragards,
Adam
PS. don't correct my misuse of articles please
Hello,
That subject seem to be an old inactive one.
I always thought that current was not usable for class D amplifiers for the reason that it carries the distorsion due to the varying output impedance. introducing the current in the feedback loop meant to introduce distorsion in the loop.
I have missed something ?
That subject seem to be an old inactive one.
I always thought that current was not usable for class D amplifiers for the reason that it carries the distorsion due to the varying output impedance. introducing the current in the feedback loop meant to introduce distorsion in the loop.
I have missed something ?
The biggest contributors to class D distortion are:
1. dead time distortion
2. inductor nonlinearities
which are both phenomena of distorted voltage caused by changing output current.
With current mode these two are virtually vanished even before you apply global negative voltage feedback.
Another thing is getting rid of a second order filter in a voltage loop and getting two first order loops: one current and one volatge loop.
1. dead time distortion
2. inductor nonlinearities
which are both phenomena of distorted voltage caused by changing output current.
With current mode these two are virtually vanished even before you apply global negative voltage feedback.
Another thing is getting rid of a second order filter in a voltage loop and getting two first order loops: one current and one volatge loop.
Oh yes, simple measuring of the high frequency current (current around and after the filter cutoff frequency) is the key of close-loop amplifier. That's is for sure the reason why many used it (band pass current mode control, capacitive current Feedback, ouput voltage derivative...).
On the other hand, the low frequency current (before the filter cutoff frequency) namely the "audio band" current is much harder to measure.
But the point is: Since Load is unknown, audio band current is unknown.
To correct those low frequency distorsions, one might know the audio band current set point, no ?
On the other hand, the low frequency current (before the filter cutoff frequency) namely the "audio band" current is much harder to measure.
But the point is: Since Load is unknown, audio band current is unknown.
To correct those low frequency distorsions, one might know the audio band current set point, no ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.