Big-t

Re: class d amps

djQUAN said:
I have thought about the diode trick but there is not much room to work with. :(

a small SMPS for amp9 is easy. there is a schematic in the power supply design forum somewhere. try to PM 'N-Channel' as he has made several of those converters already.
Looks liike you have your work cut out for you. :hot: I'm not really sure what half of the stuff you are doing means, but I'm enjoying the thread anyway. :D Keep up the good work, it's gonna work eventually.

I'll get hold of N-channel, I'd like to start off small (amp9 PSU), and work my way up, rather than go for something like a AMP8 PSU where I will be out of my depth.
Jamrak said:
guyz i'm new here in this forum i'd really like to learn more about amps thanks to all of you..

is there anyone could post for me a tutorial for a simple diagram amp,they are too complicated for me a step by step would do thanks in advance
This amp is a kit sold by 41hz.com.

They only supply the amp part, you still need to figure out some way of powering the amp. The easiest way if to use a huge mains powered toroid, teamed up with a bridge rectifier, and tank caps.

Once you get into SMPS it all starts to get rather complex I've found, I'm still trying to wrap my head around it all. :)
 
I'm sorry to say this but you have two critical mistakes in your PCB.

Think about the way you have wired the 12V input inductors, the filter capacitors and the center tap of transformer primaries. This explains the spikes and the diode failure very well.

Think also about the way you have wired the secondary windings and diodes, and about core balancing when the class D amplifier is drawing some current from one rail and returning half of this current to the other rail (this does happen during the tops of the cycles of bass notes) making one rail a few volts higher than the other.

EDTI: Wow! Your first mistake has gave me a good idea about how to implement regulation with the help of this primary side filter inductor and boost-mode operation (involving some overlap of MOSFET banks).
 
Eva said:
EDTI: Wow! Your first mistake has gave me a good idea about how to implement regulation with the help of this primary side filter inductor and boost-mode operation (involving some overlap of MOSFET banks).


that's exactly the reason why I did it that way (for PWM regulation) but without mosfet overlap.

the output is running bridged so current draw at both rails is equal most of the time.

to think, I have used this circuit before and it has worked without problems so there might be something else.

I'll try to bypass the input inductors and install them inline instead if all else fails.

since you mention the input L will naturally cause spikes on the primaries, I guess I have to add an aux winding for the gate supply instead.
 
The input inductor configuration *requires* overlap. The magnitude of the spikes is completely uncontrolled. The current through the inductor has no defined path to follow during dead time. The diodes blew because they were the easier path for this current. The next easier path are the MOSFETs but these are harder to blow. I expected you to have better understanding of the circuit.

You have to either connect the capacitors to the other side of the input coil or modify the control circuit.
 
TL494 datasheet mentions that max duty cycle at push pull is only 45% so I can't achieve overlap with a simple component change so what I did was bypass the inductors instead.

I replaced the inductors with pieces of 8ga wires so basically, all I have are four 6800uF and two 2200uF caps for input filtering. the hell with EMI filtering. I see car amps without input inductors! :clown:

but hey, as usual, Eva was right. :nod:

drain waveforms (this is identical to all four banks of fets)
CIMG4321.jpg


X10 of above waveform positive and negative slopes:
CIMG4322.jpg


CIMG4323.jpg


with moderate loading, the waveform barely changes. with no load, the waveform has straight angled sides but this one was with amplifier module connected at idle.

do you think I should lower the freq a bit more? this freq was 33kHz and was adjusted to have perfect square waves (with clean slanted sides) with no load connected. idle current with amp not connected is 270mA, and 2.6A with T amp connected and idle.

I have not yet connected the diodes for the gate supply but will still place them as before but will add ferrite beads.

I learned something new again. :D
 
This is looking quite good now. What you have to do, rather than just bypassing the filter inductors, is to connect the 12V capacitor bank to the transformer side of the inductors, and leve only a single low value capacitor at the input side (like 10uF with optional 100nF ceramic or 1uF film in parallel) for optimum EMI filtering.

You should allow the ramp to continue all the way down to the opposite state to avoid the turn-on EMI, which you have almost achieved. You can reduce the frequency even further (thus increasing crossover slope) if the core is up to the job (it depends on material, but exceeding 200mT is not usually recommended) or you can just add a bit more dead time (that won''t hurt at such a low frequency).

If you make it fully resonant and use the oscilloscope probe as a loop-antenna EMI sniffer, you will notice that the only EMI generated is a spike at each turn-off event.
 
while testing a while ago, I could turn the freq way low to around less than 20kHz before the supply starts drawing lots of current so I guess I could lower it a bit more.

I'll be adding an aux winding and tap the gate supply from there instead of the drains so that I lessen the risk of blowing diodes again.

will post again in a few minutes and we'll see what happens...
 
this thread seems to have turned into something that should be put in the powersupply design forum. :clown:

here's the latest waveforms.....

freq is dropped to 27.4kHz dead time is the same and gate drive supply is now driven from an aux winding.

edit: current draw at no load condition is now at around 400mA. that is the entire supply including a small fan, gate drive supply and PWM controller. only two LED's connected at the output via 10K resistors per rail.

CIMG4325.jpg


CIMG4326.jpg


CIMG4327.jpg


and here's a closeup of the gate supply:

CIMG4330.jpg


TO-220 device is a LM317 which provides regulated 17.5V to the totem poles. you can see the aux winding going to one ferrite bead and four 1N4937 diodes mounted vertically arranged as a bridge.

gate supply remains regulated down to 11V input.