Help in Class D amplifier designing

[dileepkella85]

Yes, you are on the right track.

Yes, transducers can pass through 56KHz signal, which is demodulated in air and we get 17KHz audio hearable signal in air.

I am getting the sound at the output but with little power SPL.

[Pafi]

Originally Posted by dileepkella85

@Pafi, you were saying that it is frequency modulation with side effect of PWM, so please can you share this point in detail, means you are saying that this 555 timer IC is working in the fashion of frequency modulator!!!....please elaborate this.

To explain it in details takes too much time. Try to think it over what determines freq and what determines duty cycle ratio (normally PWM changes duty cycle ratio only) in that circuit! (The answer: difference between trigger-comparator reference voltage and VC is approximately inversely proportional to freq, and average of the previously mentioned voltages is proportional to 1-duty cycle). If you change CV, both of them will change.

If you want to know how much freq is dependent exactly, then simulate it!

[dileepkella85]

Yes you are right that if we change the CV of 555 both frequency and width will change, that is what I also observed in simlulation, but in hardware observing in oscilloscope is is observed that output rectangular pulses have exact 40KHz frequency whatever the input is.

Well at this moment I am looking to improve the power level then I will come to this problem of modulation.

I just want to have the results like the hardware which I had in PCB then I will remove 555 timer from the circuit and put my DSP over there, then there is issue of modulation.

[Pafi]

Originally Posted by dileepkella85

Yes, you are on the right track.

Yes, transducers can pass through 56KHz signal, which is demodulated in air and we get 17KHz audio hearable signal in air.

I am getting the sound at the output but with little power SPL.

If you want demodulation, first you have to modulate it (in the right way)! Now there is some amplitude modulation also, but I don't know if it is designed or just side-effect.

For this experiment you need only ultrasonic signal coming from the transducers, without any audible spectral components, otherwise you can misdirect yourself. PWM makes audio freq, so it should be avoided completely!

I'm sorry to tell this, but as soon as you eliminate audible components, the sound will disappear, or at least becomes very quiet. For good demodulation in air you need very high pressure, comparable to atmospheric pressure! I assume the lowest limit is around 0.001 atm= 100 Pa, which is 134 dB in SPL. Do your transducers able to produce this? Maybe on hair or similar lightly connected small particles a little lower pressure can be demodulated also, but don't expect more than a whispering!

What you hear now, is probably simply the normal sound coming directly from the transducers. If you want to know if it is demodulated signal, or direct sound, you have to determine spectra and angular dependence. If you hear the sound only in a narrow beam, and you can hear not only high pitched tones, but middle also, then you really hear the demodulated signal, otherwise not.

To get clean signal (without audio) the best would be to modulate the power supply of the output stage, and drive it with a simple square signal.

On an off-tuned resonant circuit frequency modulation also turns to additional amplitude modulation, but to use this effectively first you have to measure impedance curve.

[ChocoHolic]

Originally Posted by Pafi

56 uH and 142 nF results a 56 kHz resonant frequency.

...and the circuit runs with 40kHz, means below the filter resonance.
Means hard switching, even during idle.
Do you get much heat in the MosFets?
If yes:
Better increase L to 100uH.
Or keep 56uH and increase switching frequency to 70kHz.

I am not sure which kind of signal mixture you are feeding to the system.
But theoretically low output vs. high output in your experiment could also be depending on the relation between resonance frequency of the filter and the switching frequency of the amp and Q of the transducer-filter-circuit.
I guess that the transducer and filter is the same, no matter if you use your breadboard or the circuit on the PCB.

Does the circuit on the PCB run with exactly the same frequency like your breadboard - or does it run with a frequency more close to 56kHz?
Basically you should get more and more ultrasonic output (and with this more gain) when you move the switching frequency towards the resonance frequency of the transducer-filter-circuit.
Avoid full resonance, it may lead to destructive currents.
For low switching losses it is better to operate some kHz above resonance rather than some kHz below.

[dileepkella85]

@Pafi
First of all, I clear one point that these ultrasonic transducers cannot out audio frequency because, these ultrasonic transducers have resonance frequency of 40KHz and lower cutoff frequency is 35K and upper is at higher frequency of 100K (around), so it is confirmed that transducers cannot output audio frequency.

We put inductor (choke) at the output which is acting as low pass filter, with capacitive load and its cutoff frequency is around 57KHz. if we see the spectrum of PWM there are difference frequencies and higher harmonics, and we want to remove higher harmonics above than 60KHz, and at the output there is 40K (carrier) and demodulated difference frequency (audio) which we hear.

@ChocoHolic
on working PCB there is also 40KHz.

Well, at this moment I am not concerned what is happening there in the circuit, I just want to replicate the circuit with the same SPL so that I can make changes in that in the future.

As I told that I have working PCB of that circuit, I just wanna replicate that now with discrete components, schematic is also posted in very first post, circuit is now in working condition but with low SPL as the same circuit on PCB giving high SPL, I need to improve that....please suggest on this, possible problems....

[Pafi]

Originally Posted by dileepkella85

@Pafi
First of all, I clear one point that these ultrasonic transducers cannot out audio frequency because, these ultrasonic transducers have resonance frequency of 40KHz and lower cutoff frequency is 35K and upper is at higher frequency of 100K (around), so it is confirmed that transducers cannot output audio frequency.

Cutoff frequency doesn't mean it absolutely doesn't work under that.
It means only that it's sensitivity drops below a specified value, and falls gradually, but never reaches zero (only at f=0).

if we see the spectrum of PWM there are difference frequencies and higher harmonics, and we want to remove higher harmonics above than 60KHz, and at the output there is 40K (carrier) and demodulated difference frequency (audio) which we hear.

Have you done any analysis about demodulating PWM signal?

As I told that I have working PCB of that circuit, I just wanna replicate that now

Then it is not designing despite of what you stated in the title of the topic.

Do whatever you want!

[dileepkella85]

Yes you are right that it will not be zero, but it should significantly be reduced, in my case it is not decreasing that much, sound level is significant at output.

No, not particularly analysed PWM signal, but will analyse in soon.

About last comments.....hahahahahah,
yes for this part you can say that, major part is remaining, trying to tackle the issue in parts.

[ChocoHolic]

Originally Posted by dileepkella85

@Pafi
Well, at this moment I am not concerned what is happening there in the circuit, I just want to replicate the circuit with the same SPL so that I can make changes in that in the future.

TI am afraid you have to be concerned about what is happening there in the circuit.
The circuit on the bread board does not act like the circuit on the PCB, so the issue is probably caused by what is happening in the bread board circuit.

Now after you have reached a status, where the bread board does not burn anymore, it should be possible to find the reason by a straight forward comparison.
A) Measure the values of the unknown components.
B) Check all voltages and signals in idle state and full power - node by node.
C) Make sure that in your comparison measurements and in the listening tests the signal before the bread board is identical to the signal on the working PCB.
D) If necessary double check comparison of all components (also the known and boring ones...)
These actions do not need any special circuit know how, but should lead you to the reason for the difference. You just need to do it.

[dileepkella85]

Very good suggestion.... of comparison...
will come with the results in next post

Thank you very much ChocoHolic.

Regards
Dileep