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.
I'm building a class D amp using the following components
TL494, IR2110 Gate driver IRF540N mosfets.

I've never worked with this chip before and need some help. I've looked up on the "designing switching voltage regulators with the Tl494" guide and understand how to implement the oscillator components and also how to implement Dead-Time. But i do not understand how to actually implement the musical signal itself.

Do i just implement the musical signal into one of the error amplifiers (+'ve input of the error amplifier) and ground the other input (-'ve) ?

I couldn't find a application note on using this chip to create a class d amp, hence my confusion with implementing the music. I've also tried to follow how ledmania implemented the signal but don't understand why he has a reference voltage signal in the the negative input of the error amp with the music signal in the positive input.

http://www.diyaudio.com/forums/attachment.php?s=&postid=762246&stamp=1131481319

Here's a quick spice model of what i've done, though it doesn't run. Gives me errors.

http://img338.imageshack.us/img338/7121/attempt1w.png

In the schematic diagram above, I implemented the Fail-safe operation for the oscillator as specified in the design guide.

I wanted to simulate a 1Khz music freq and i'm running it on a single 20V power supply.

I will worry about Dead-time circuit and feedback and the other goodies later once i get the fundamental's corrected.


cheers
 
Far too many mistakes and misconceptions.

Have you ever heard about control loops, error amplifiers and frequency compensation? You seem to be overlooking everything related to these subjects.

Learn elementary electronics first ;)

Then do some discrete linear amplifiers, then some switching regulators and SMPS, then you would be ready to start with class D ;)
 
In further reading off ledmania's thread

"the data sheet says that "when the o/p of the error amp modulates from 0.5V to 3.5V, the o/p duty cycle will vary from 97% to 0%"in reverse order! So the proper bias point on this is to get the average voltage of 0.5V+3.5Vdevide by 2,you will get 2V. this will correspond to an o/p of 48.5%.bingo!.
So, 2V should be the quiescent bias voltage at the i/p..."


So now i understand why he's using voltage divider into the inputs of the error amplifier. But he's incorporating it as a feedback system. While i would like to initially test it as a open-loop system. So if i use Vref ( 5V ) and use voltage divider rule to get 2V from that, into one of the inputs of the error amplifier (Negative input - Pin 15). And in the other input (Pin 16) the music is fed through, do i have to make sure that the voltage is also 2V in this input (i.e Pin 16)?

If so i have to make sure the source music signal is 2V?

Also what's with the 3.3uF cap and 2.2K resistor which the music signal is put through?

What's the purpose of the other error amplifier (Pin 1 and 2)?
 
I have limited knowledge of control loop and error amplifier, but not for freq compensation.
The trouble is it's a actual school project that i have to complete by the end of the semester.

Any tips then on taking it little bits at a time with the TL494?

For example i was thinking i would just get the oscillator started and running and put a music signal in and use earphones like Ledmania used to check out initially to see if i've implemented it correctly?

Would this be a good starting point?
 
sorry should have added this as well to my last post.

I understand the basics of a class D amplifier..

The musical signal is compared with a high freq carrier (sawtooth or triangular) signal. This generates pulse width modulation, which is then fed into Gate drivers to operate the output MOSFETS. Harmonic distortion is introduced because of the switching action. The LC filter at the output filters the switched waveform from the MOSFETS to a sine wave (in a ideal world), before it's sent to a speaker.

A feedback is generally used to minimize errors at the output. Feedback is achieved usually by taking either the output right before it's filtered or after it's filtered, and filtering that by means of a passive RC filter and integrating that with the musical signal.

At the moment i'm not worried about introducing gate drivers, mosfets and such. I would like to just get the TL494 itself started and running.
I'm not too sure if my interpretation of Ledmania first post is correct, but it sounds like he just used earphone to listen to the output from the collector C1 from the Tl494, without introducing gate drivers and such into the circuit.
 
One of the biggest problems with using one of these switching regulator chips for an audio amp is the limited duty cycle. Ones designed for push-pull (or half bridge) have maximum duty cycle of just under 50%. You really want 50% quiescent (minus dead time), and to be able to swing both higher and lower, approaching 0% and 100%.

I once demonstarted to myself that it works, but what I did to start with was "regulate" 20 volts from an 80 volt supply. Inject audio at the reference voltage (2V reference, gain of 10X) and get a shade under 40V p-p. LC filter, cap coupled output. It actually sounded cleaner than I expected considering that it was thrown together for the heck of it.
 
TL494 is a bad choice, for example it uses sawtooth instead of a triangle wave. A simple class D amplifier with quite good performance can be built around just a fast comparator. Do some research on the UcD concept. It's a phase shift oscillator that amplifies audio and damps output filter resonance to flat response regardless of speaker impedance in the process...

Open loop amplifiers are a source of trouble, even getting no DC at the output is a challenge.
 
circuitcity!

Here's a quick spice model of what i've done, though it doesn't run. Gives me errors.

http://img338.imageshack.us/img338/7121/attempt1w.png

Yes, because it contains many errors! For example: what is R2 for?!? Why did you attach + and - inputs of OPAs together? You should never do this! Have you learned about OPAs? Etc...

Ledmania's circuit is faulty also.

try this one!

In the schematic diagram above, I implemented the Fail-safe operation for the oscillator as specified in the design guide.

Which is that design guide?

wg_ski!

TL494 is limited to 0...90 % if you connect pin13 to GND.

It's usable, but with care.
 

Attachments

  • 494.gif
    494.gif
    14.9 KB · Views: 3,302
Eva said:
TL494 is a bad choice, for example it uses sawtooth instead of a triangle wave. A simple class D amplifier with quite good performance can be built around just a fast comparator. Do some research on the UcD concept. It's a phase shift oscillator that amplifies audio and damps output filter resonance to flat response regardless of speaker impedance in the process...

Open loop amplifiers are a source of trouble, even getting no DC at the output is a challenge.


Yes, from the theory i read in books and IEEE, triangular wave carrier is the better option. But i'm just trying to implement a amplifier for a 25W speaker, purposely for academic reasons, not a high powered amplifier. Also i'm aiming for a oscillator freq of 200Khz.

Ok i'll look into the UCD concept as well.

Pafi said:

Yes, because it contains many errors! For example: what is R2 for?!? Why did you attach + and - inputs of OPAs together? You should never do this! Have you learned about OPAs? Etc...

try this one!

Ok i'll give the corrections you made, as a start to this project. I
ll try this configuration out.
Also your referring to Op Amp when you say OPA? We've covered op-amp theory in class.

Which is that design guide?

I used the "Designing switching voltage regulators" guide off ti.com

Also i'm thinking of using a function generator as my music signal for the moment. Is this ok?
 
Eva said:
A simple class D amplifier with quite good performance can be built around just a fast comparator. Do some research on the UcD concept. It's a phase shift oscillator that amplifies audio and damps output filter resonance to flat response regardless of speaker impedance in the process...

I'm sorry, but a quick google search shows me UCD is a pre-built amp.
Have i looked it up wrong?
Also can you give me example models of fast comparators. I'll look into them and some schematics perhaps as guide.

All I'm attempting to get out is a simple structured class D amplifier. I'm definitely not ready for complex structured class d amp.

A very simple class D amp that'll do the job at the end of day is enough for me.
 
initially i attempted to do this amplifier my way and did the following.

I used the TL072 op amp to create a triangular carrier. Which was then compared with a sine wave off a function generator using a LM361N comparator.

I introduced dead time to the output pulse width of the comparator, by first integrating that pulse width using a passive RC circuit, and then used a schmitt trigger to trigger off the thresholds i chose.

I would then use this dead time adjusted pulse width as my input to the IR2110 gate drivers and so on.

However i was told this concept would never fly and go look into PWM chips, and hence why I'm thinking of using the Tl494.
 
Pafi i setted it up from your edited schematic and my waveforms across C1 and C2 look like this now.

Output of C1

http://img339.imageshack.us/img339/1838/tek0005.jpg

output of C2

http://img383.imageshack.us/img383/5342/tek0004.jpg


The Blue signal is a sine wave coming from a function generator, with a 2v offset and with a 1v amplitude.

I'm providing a 5V reference into error amp 2 Pin (15), shouldn't it be 2v as well?


Any further suggestions as to what i can edit in that schematic and re-arrange?

I don't wanna move onto gate drivers and such until i get the circuit for the Tl494 circuit correct.
 
Eva said:
http://www.ciaudio.com/ucd_aes.pdf

;)

The triangle generator and comparator approach was far better, not only in terms of performance but in educational value too, than your current approach.

Self oscillating works in a not so evident way but it's also very interesting. The own carrier residual of the amplifier is used as a "triangle wave" :D:D:D


Yeah exactly, I actually understood what i was doing in my own circuit. I don't know what my supervisor saw that made him say go PWM chip.

I'll tell ya what then Eva, I'll post what i've done in breadboard of my own design and i'll also post a schematic of my design as well.

I'll post waveforms i've seen in the scope as well. In my own design i use the TL072CN op-amps to create my triangular waveform and i use the LM361N Comparators to create PWM.

Thinking of using LM360N comparator for schmitt trigger.

And thanks I'll read up on that document.

Pafi said:
Opps, I missed an other mistake: right side of R3, R4 should be connected to Vcc, instead of GND!

Have you heard about "open collector" output?

:whazzat:

Ok i had both C1 and C2 as open collector output. Thanks I'll make the modification and i'll post the images here.
 
I have 2 dumb questions to ask.

Q1: Is it ok to use a standard sine wave off a function generator as music input for now?

Q2: http://www.eetasia.com/ART_8800403692_499501_NP_988b9883.HTM


Eva and Pafi i was wondering if i can do what that link did. I know dead-time is important, but is it possible to get away with dead time by not building it and directly just connecting the PWM from the comparator to the gate driver, if it's only for a 25W speaker?

Wouldn't the propagation delays from the TL072 op-amp and the LM361N comparator be enough for delay?

Because inserting dead time is causing me problems at the moment. I'm having problems with my schmitt trigger circuit.
 
Is it ok to use a standard sine wave off a function generator as music input for now?

Yes, of course!

Q2: I'm not registered to that site.

Single ended vs. push-pull: it's related to basic SMPS topologies. Actually in his circuit the two modes are equivalent (because of the paralleled outputs).

Dead time is not a simple delay. Immediate turn off, delayed turn on of the MOSFETs. A resistor||diode on gate can do the job.
 
Hi circuitcity,
I missed a lot of nice thread in this forum due to the nature of my work and I’m surprised that there are still some guys interested in building tl494 as their basic class D amp for learning purposes. This is the reason why I open up that thread..
Upon careful scrutiny and testing on my breadboard (2 years ago) I found that pin 16 of the TL494 should be connected to voltage dividing potentiometer powered from pin 14 (Vref). It is important to bring the o/p to 50% duty cycle so that half of Vcc can be expected from the final output same as in conventional linear amps.

Here is the edited version of my old drawing but this time I used IRS20124 as the FET driver. The on board dead time generator makes this device very attractive and makes things less complicated for beginners. The values of those resistors in my schema have yet to be determined through their data sheet which I don’t have at the moment. I have not tried this whole circuit before but on the TL494 side alone, it works perfectly. You will be surprised the way how it sounds via headphones.

Take a look here
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.