Design of output inductor for class D amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Ex-Moderator
Joined 2005
Re: Re: Design of output inductor for class D amplifier

Elvee said:
In addition, the "HF-bias" effect caused by the switching also complicates the matters as it tends to reduce strongly the non -linearities (but at least, it's also good news).

LV, do you have or know of any other sources of information on that topic (high frequency bias reducing nonlinearities)?
 
noblimps.jpg


This is the result with a rather simple inductor, on a RM12 core, the only special thing is, that it is wound first from down, and up, and then the second layer is also wound from down up (not from up till down).

As you can see there are hardly any 'blips'.

The bandwidth is 100 Mhz.
 
Lars you are probably cheating concerning the blips. If you measure when net output current is zero, the circuit will be performing resonant switching with gentle slopes.

Try the same measurement at 500mV/div but with a bank of power resistors connected between the output and one of the supply rails in order to make 5A or 10A flow to force hard switching (which is what will hapen most of the time when listening to loud music). Check also for EMI and ringing in these circumstances, as checking when idle is mostly useless.
 
IVX: I found the word 'Blips' in that material provided by Bruno, earlier in this thread.

Charles: Well now we use something with 3 mm airgap on both in- and outside. So it's halfway an aircoil ;)

EVA: Hi! I will try to do that test one day when i get a bit better time. You are right my reading only shows idle switching, but still it shows that almost no HF passes through from the switches to the output. The switching slopes are controlled not by the output current, but by the gate controller. So the flyback current of the coil really makes no difference here.
 
Lars Clausen said:
EVA: Hi! I will try to do that test one day when i get a bit better time. You are right my reading only shows idle switching, but still it shows that almost no HF passes through from the switches to the output. The switching slopes are controlled not by the output current, but by the gate controller. So the flyback current of the coil really makes no difference here.

Also it's going to be a very good static test in order to see if the slopes are really being controlled during hard switching or not. Testing with sine waves is a bit messy.
 
Re: Re: Re: Design of output inductor for class D amplifier

BWRX said:


LV, do you have or know of any other sources of information on that topic (high frequency bias reducing nonlinearities)?
Well, basically a class D amplifier is an excellent example: if you look at it in a conceptual form, it is a comparator followed by a power driver; the output can only assume two states, (+) or (-).
Without further addition, such an amplifier is hugely non-linear, but if you mix the input with a high-frequency bias (the triangular wave at the switching frequency), you create a number of intermodulation products, and some of these products conveniently combine to cancel the harmonics caused by the decision process in the lower end of the spectrum; the higher products and harmonics can then be eliminated by filtering.
Another example is the magnetic recording: without the ultrasonic bias, the hysteresis of the magnetic material on the tape causes severe non-linearities. The HF-bias smoothes out the "kink" around 0 in the transfer function.
LV
 
Ex-Moderator
Joined 2005
Re: Re: Re: Re: Design of output inductor for class D amplifier

Elvee said:
Another example is the magnetic recording: without the ultrasonic bias, the hysteresis of the magnetic material on the tape causes severe non-linearities. The HF-bias smoothes out the "kink" around 0 in the transfer function.
LV

Yes, I've actually likened class d to that phenomena before as well. I guess my question was meant more in terms of the high frequency residual left after the output filter reducing any nonlinearities (and thus reducing the effect of them) in something like the parasitic inductances of an actual circuit.
 
The HF-bias smoothes out the "kink" around 0 in the transfer function.

Could very well be the case but maybe it isn't. There are big differences between a recording head/tape and a class-d output filter:

The main lowpass function of the recording head/tape combination is given by the airgap and tape speed whereas the class-d output inductor is actually a part of the filter.
In case of recording RF bias the bias signal is of constant shape and amplitude and is combined with the signal by simple addition - while the RF signal in the class-d amp is actually the carrier of the information. So an in-depth analysis would be in place IMO to be sure.
I have to say that I once thought the same way as you do but I wouldn't bet on it for the aforementioned reasons.

Regards

Charles
 
phase_accurate said:


- while the RF signal in the class-d amp is actually the carrier of the information. So an in-depth analysis would be in place IMO to be sure.

Regards

Charles
You always have to be careful, of course, but basic modulation theory should apply here; this means that although the RF signal may be the "carrier of information", it can nevertheless be broken down into its components: average value, Fm, Fc, Fc+Fm, Fc-Fm and so forth. All these components are fed into a "modulator", a conceptual processing block incorporating all of the inductor's non-linearities. The result then goes to a perfectly linear (but not necessarily perfect) low-pass filter.
This may be somewhat counter-intuitive, but I'm convinced it's correct.
Similarly, I remember having discussed with an old "wireless enthusiast" who was convinced that an AM signal was a single "entity", and that trying to break it down into three spectral components was wrong; he wouldn't accept what he saw on the screen of a spectrum analyser.
LV
 
a conceptual processing block incorporating all of the inductor's non-linearities. The result then goes to a perfectly linear (but not necessarily perfect) low-pass filter.

I don't see how you come to the conclusion that the processing is incorporation the inductors non-linearities and how you come to the conclusion that the filter is linear ? Are you splitting the output filter into a linear and a nonlinear part ?
This alone doesn't confirm that the carrier has the same linearising function as the RF bias in a tape recorder, does it ?

It is IMO still advisible to achieve the least possible magnetic nonlinearitiy in the first place, independant of using pre- or post- filter feedback.

Regards

Charles
 
Classd4sure: Dear Chris. Try to be nice, talk a little more respectfully to your fellow DIY'ers. George (Soongsc)'s question is just as decent as anything else written in this thread. In any case i don't think this forum is only for the 'elite' in any particular subject. Anybody should have the right to participate on an equal basis, we are all experts on some subjects, and amateurs on other subjects. Doesn't give some (you) the right to deem other peoples rightful questions in-decent. I hope you agree.

Soongsc: The 'Blips' only have meaning in terms of RFI emittance. The sound is not affected by them. Whether this copper/ferrite coil sounds different than i.e. a foil coil or a silver / teflon coil is a good question. A couple of year ago i had a good discussion with Mr. Peter Quortrup of AudioNote in UK. He explained me that silver has a specially good coupling with a ferrite or iron core, which can be detected even on a molecular level. From an engineering viewpoint, i found this rather surprising, but on the other hand i think this guy knows what he is doing. Another thing is the teflon isolation.
Previously i would often use teflon isolation, but somebody demonstrated to me the difference between teflon and PE, and convinced me, that teflon can be a serious liability.

It should be noted that in the scope pic i posted some posts ago, no ceramic capacitors were used to dampen the RFI.
 
Lars Clausen said:
Classd4sure: Dear Chris. Try to be nice, talk a little more respectfully to your fellow DIY'ers. George (Soongsc)'s question is just as decent as anything else written in this thread. In any case i don't think this forum is only for the 'elite' in any particular subject. Anybody should have the right to participate on an equal basis, we are all experts on some subjects, and amateurs on other subjects. Doesn't give some (you) the right to deem other peoples rightful questions in-decent. I hope you agree.
..

Dearest Lars, in fact I strongly disagree. The topic of this thread is rather clear, seeking factual insight to filter inductor design, and "how does it sound" doesn't fit that description at all. Inductor design is not yet my forte, so I have remained a passive observer, very interested in learning and reading factual info or even the more hypothetical thoughts, and claim no elitness here whatsoever.

"how does it sound" only threatens the integrity of an otherwise excellent thread by firstly lowering the SNR. You'll find plenty of "how does it sound" discussions that don't concentrate on the design aspects already, a good number of which are asked by the "offender". This just isn't a thread for a question like that IMO.

I find such a question only leaves you, Lars of NewClassD, with the oppertunity to market further, instead of perhaps, taking Eva up on her insightful observations, which I happen to share, and which are likely to demonstrate things might not be as "pretty" as you tried to make it seem.

"EVA: Hi! I will try to do that test one day when i get a bit better time. You are right my reading only shows idle switching, but still it shows that almost no HF passes through from the switches to the output. The switching slopes are controlled not by the output current, but by the gate controller. So the flyback current of the coil really makes no difference here."

If you could stand by your marketing babble, you'd have done her test already, instead of avoiding it and babbling further.

I truly understand your wanting to defend the question, but I just can't bring myself to agree, or sit by while an otherwise very good and possibly educational thread, one of the few in a long time at that, gets dragged down into the obvlivion of subjective marketing babble. Had this been such a thread already, I would not have said anything.

I hope you can appreciate my position. Back to the actual topic please.

Best Regards,
Chris
______________________
www.diyaudio.com
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.