Doing a class D Amp project using TL494

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Such a high dead time results in too high distortion for audio. It may be converted into delay with compensation, but it's still a too long delay that limits maximum and minimum duty cycle. IR2111 is not in any way intended for audio, but for SMPS, it includes such a long dead time intentionally.

You should use a gate driver IC like IR2011, IR2010, IR2110, LM5100, etc... (Or a discrete driver with HCPL2611 style optocouplers and non-floating driver ICs, IR4427 like).

Dead time is not desirable in class D, I'm not using any at all in my amplifiers (although this makes them more sensitive to mistakes).

Actually, in some (quite optimized) circuits I have to add 10-20ns length to the pulses going to the driver in order to get things properly synchronized.
 
Take the output in wired-OR fashion and feed this to an inverter that is generating the opposite-polarity pulses for the driver IC. The latter can be achieved with equal delay for both pulse trains by using two XOR gates.

Depending on your circuit you might also need some voltage translation.

Regards

Charles
 
Does your teacher have experience developing high performance commercial products?

Some of us have. He may need some classes ;)

I use 74HC XOR gates and R-C-D networks to split PWM into hi/lo and generate dead time (or inverse dead time). To get matched propagation delays I use one XOR gate as an inverter and the other as non-inverting, both fed with PWM signal.

74HC gates (not HCT or schitt triggers) have quite consistent switching thresholds, so the dead times obtained are quite reliable, they don't vary much from one gate IC batch to another or with temperature. (Note that this alone is worth nothing if the gate driver IC does not provide consistent timing too... IR2110 at 15V is very consistent, even with wide temperature changes, but don't tell anyone :D )

Schmitt triggers have very useful hysteresis but the switching thresholds are not very consistent. The datasheets specify a wide variation margin. Same applies to 74HCT gates. I still use schmitt triggers for pulse narrowing and stretchign and missing pulse detectors, but bearing the higher timing tolerance in mind.

The inputs of some IR drivers like IR2110 include schmitt-triggers with quite consistent switching thresholds too. I tend to use these to actually generate the dead time (or pulse stretching), and the XOR gates only for high/low splitting.

Anyway, the first huge mistake of your teacher is to ask you to use TL494 for class D. No serious amplifier is done that way, and building the modulator with a couple of op-amps and comparators has far more educational value for you... (but does he know how to do that properly?)
 
That would mean another chip in between the TL494 and my current IR2110 gate driver.

He's not happy about it.

I agree with him. No active circuit is needed between TL494 and IR2110. I would give some hints if I saw a spark of understanding, but I see soooo many embarrassing mistakes, that make me say: circuitcity ought have to learn the basics of electronics before he should start doing this project. (And I don't wish to do his homework, because it's useless.)

I'm impressed very much by the fact that he has access for such an expensive oscilloscope.
 
Well it a project for second year. We get access to Electrical labs, that has the usual things, scope, soldering iron, multi-meter..etc..etc..
I've never worked with electronic circuits before, which is why its confusing why he would put up a complex amplifier as a project.

I liked my way better. I actually understood what was happening in my circuit.

Just a Op-amp + comparator gave me a input to Hin and Lin to the gate driver.

These chips are complex in circuitry and most of them i do not understand.
 
I'm sorry but I'm only asking for help because of the uncomfortable situation I'm in. They showed some fundamental theory and then expect us to build a class d amplifier.

If your saying that you tend not to use dead time Eva.

Can i Simply just feed my complimentary PWM that i get off my comparator into the IR2110, with the conscience of knowing it won't blow the MOSFETS (Which I doubt)?

I'll use the resistor-Diode network I read in a few IR application notes to get some dead time.

I'm waiting for the schmitt triggers to arrive through the school, so until then I'll be trying my own method until they arrive.
 
In my opinion you should drive IR2110 directly from outputs of TL494.
It has a lot of dead time, so on the one hand you may be sure to not blow your mosfets, but on the other it will produce distortion.
But hey, this is not cutting-edge hi-fi project!
Those who suggest ultra-low dead time, going triangle ramp etc. (technically very correct) either do not remamber what pointless nightmare studying on the university was, or haven't been there.
 
Eva wrote:

74HC gates (not HCT or schitt triggers) have quite consistent switching thresholds, so the dead times obtained are quite reliable, they don't vary much from one gate IC batch to another or with temperature. (Note that this alone is worth nothing if the gate driver IC does not provide consistent timing too... IR2110 at 15V is very consistent, even with wide temperature changes, but don't tell anyone )

Shouldn't the iput logic of the IR2110 be fed from a 5 V rail when the inputs are fed by the outputs of a 74HC... ?

Regards

Charles
 
darkfenriz said:
In my opinion you should drive IR2110 directly from outputs of TL494.
It has a lot of dead time, so on the one hand you may be sure to not blow your mosfets, but on the other it will produce distortion.
But hey, this is not cutting-edge hi-fi project!


Besides the gate drivers already mentioned here. Know any Gate drivers out there that have a IN login input and not HIN and LIN. Cause I'll let the gate driver do the work of getting complimentary pulses, instead of me running in a aimless loop.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.