Self Oscillating Class D

self oscillating class d

Here is the simplified schema of a self-oscillating class D amplifier.

Can anyone explain it to me how this circuit work.
Maybe some intuitive approach help me understand this circuit.

class d schematic

I can imagine it goes like this :

Assume the opamp will generate a variable duty cycle square wave (you mention class D), the square wave will be averaged out by the inductor/capacitor on the output side. This voltage is fed back to the input with the 56k resistor (the 51p capacitor creates a pole for higher frequencies). I assume the output will be amplified 56 times the input voltage.

The inner circuit makes the oscillator. This is a simple oscillator used a lot with opamps. A positive feed-back (resistors connectiong to the + input of the opamp)for setting switching thresholds, the 180k and 1.5nF at the - input set the time constant of the oscillator. (see page 11 of this datasheet **broken link removed**).

self oscillating amplifier

Thanks for help. I never thought about this circuit as a "simple" opamp oscillator.



Someone has anything else to say about this circuit

self oscillating pwm

Are the + and - inputs of the opamp correct?

self oscillate class d

Yes, the + and - inputs of the opamp are correct.
But I find another error. The M1 is upside down.
Here is the correct schema


I was force to switch the inputs of the opamp to achieve the negative feedback .
Because I added output stage (M1, M6), non-inverting input of a opamp works as a inverting input.
Increase in input voltage of the non-inverting input of opamp cause the output voltage of opamp increase to. This increased output voltage will open T1 and M6 is open to. Opening M6 connects the L1 to ground. Decrease the input voltage will cause the M1 to open and increased L1 voltage. (negative feedback)

Simple oscillator

class d schematics

Ok, that was my next question, I thought you used an emittor-follower for the output stage, but then you can not drive the output transistors into saturation due to the possible too highe saturation voltage of the opamp (or was it a comparator...).

The longer I look into it, the more interesting I find this system. I did some calculations on it, and I found a rather high switching frequency. The frequency chages with the input voltage.

Consider the system without the Feedback resistor of 56k. Then the open-loop gain of the PWM and power stage can be determined. It is supprisingly linear (except at very low or very high duty cycles). The open loop gain is about R3/R5 (refer to your last full-schematic) = ~-100.

Because the open loop gain is not as high as in normal opamps, you can not calculate the total gain just with R4/R5 (wich should be 56 in this case), but you need the full formule (refer to control theory --> forward path/(1+forwardpath*feedback)) where the forward path is the open loop gain. For your circuit it ends up with a closed loop gain of around 35.

At 50% duty cycle, (input voltage = 0) the switching frequency is about 1.6MHz, I think this is rather high for theoutput fets and the rather slow comparator.

class d amplifier schematics

Oh Yes this circuit is very interesting.
And I find it in a Philips service manual FM23AC plasma TV.

The switching frequency is about 333KHz÷350KHz (input voltage = 0)
But what is more interesting that the other audio channels (left low, Right High, RL) has a different value for R3411 (15K).
It's look like Philips match the R3411 to achieve the same switching frequency for all audio channels.

Can you explain the function of the Zener diodes (all of them).
And why do we need a AC-coupling capacitors ( C2430, C2455=100nF) in the input of the output stage?

Pspice calculate for my circuit about 380KHz (thanks to parasitic capacity I suppose), when I remove the C1 (1.5nF) spice show a pure sin wave just about 1.2Mhz. And the closed loop gain of around 32.

oscillating pwm

The output stage is fed with a higher voltage as the comparator, so some level shifters are needed. Also the gate- source voltage of the output fets must be limited to 20Volt!

With the Zener Diodes, the voltage at the bases of transistors 7430 and 7455 are kept between around +5.5 and -5.5V (zener voltage +1 forward drop). This sets also the level on the emittors of the transistors (because it is an emittor follower).

The pulse of +/-5V is transferred to the gates of the power mosfets through capacitors 2430 and 2455. They act as level-shifters. The value of the capacitors must be big with respect to the gate charge of the FETs. The zener diodes 6434 and 6435 limit the gate voltage.

This is a very cheap level-shifter, you need only the two transistors 7430/7455 for the two output fets.
An additional fact is that when there are no pulses at the base of 7430/7455, the power fets stop conducting, because the gates will discharge through the 4.7Meg resistors (3461/3436)

About the switching frequency, I did some calculations on the circuit you suggested, but with an ideal opamp (no switching delay, no input capacitance...). I also noticed that the ramp-threshold-setting resistor (R 3409) is 180 Ohm instead of 100 Ohm in your suggested schematic from before. This is very importand, because the ratio of R3408 and R3409 is in the equation of the switching frequency. (doubling R3409 (ok from 100 to 180 ) will result in half the frequency.

Are you planning to build a thing like this? Well maybe I will...

Stefaan

self-oscillating

I once used a similar "self-oscillating" PWM modulator for a controllable IR source, but it's properties are rather poor, compared to an usual constant frequency triangle oscillator PWM to my opinion.

The most serious disadvantage is in the fact, that the switching frequency is strongly output level dependant. May be, this is useful in some applications, but I dislike it.

Another point is the poor-mans gate drive circuit. I guess, it' working correct as shown in the TV schematic, but in your variant, apart from sourcing overvoltage to the gates, it also causes cross-current flow in the output stage. The CZ circuit provides an on-delay to prevent from this nasty effect. But it's far from a straightforward and easy manageable design, I think.

basic class d schematic

I'm not discussing the good or bad audio properties... I just think this is nice circuit for its principle of operation. Stealing or injecting current in the oscillator capacitor is a quite linear system.

jony130 : can you tell me how the output filter for this amplifier looks like?

PS. I give no guarantee that my calculations are correct.

simple self-oscillating class d amplifier with

Thanks for clearing up my confusion about this circuit.
And yes, maybe I will build this circuit in a couple of days.

Added after 3 minutes:

svhb said:

jony130 : can you tell me how the output filter for this amplifier looks like?

Click to expand...

Can you by more specified, because I don't understand.

PS I will use this cores TN17/11/6.4-3C20-A72
https://www.ferroxcube.com/prod/assets/tn171164g.pdf
https://www.ferroxcube.com/appl/info/gaptoroids.pdf

self oscillating d class amplifier

I mean is there something on the schematics you have from Philips?

self oscillator circuit

The output filter is only a LC filter show in schematic,
L=33uH C=440nF (5465, 2465, 2466)

class d schematic application philips

Ok, sorry, it was not on my printout (portrait instead of landscape). But indeed it's on your posted schematic...

self oscillating class d calculate frequency

I attach two more schematic of class D amplifier

First
SONY XM-D1000P5

I think sony works as is describe here
https://www.freepatentsonline.com/6753729.html

Second
Onkyo A9555

philips self oscillating class d amplifier

also nice document : **broken link removed**