IR2110 propagation delays, Cload = 1.7 nF, ns:
ON delay + rise time: 120 + ~45
OFF delay + fall time: 94 + ~30
IRF540Z delays, Rgate = 12 Ohm, ns:
turn ON delay + rise time: 11+35
turn OFF delay + fall time: 39+35
Common delays:
ON: 120 + 45 + 11 + 35 = 211 ns
OFF: 94 + 30 + 35 + 39 = 198 ns
So its much faster, which is good. So another 30secs I need, which the parallel diode will give with the 22ohms resistor.
ON delay + rise time: 120 + ~45
OFF delay + fall time: 94 + ~30
IRF540Z delays, Rgate = 12 Ohm, ns:
turn ON delay + rise time: 11+35
turn OFF delay + fall time: 39+35
Common delays:
ON: 120 + 45 + 11 + 35 = 211 ns
OFF: 94 + 30 + 35 + 39 = 198 ns
So its much faster, which is good. So another 30secs I need, which the parallel diode will give with the 22ohms resistor.
The output rise/fall time of the driver cannot be simply added here. 
I have done it for simplicity only.
Ideally, it is needed to take into account the mosfet's gate threshold voltage (3-4 volts approx) - this is the voltage, where mosfet starts to open. In this case, we should add not the whole output rise time of the driver, but only approx 1/5 of it: the driver's output rise time is given in datasheet for 15 volts, where the gate threshold is ~3v, so the mosfet will start to open much earlier (almost 5 times earlier), before the driver will charge its gate capacitance to 15 volts, so the 'effective' rise time of driver's output will be approx 1/5 of the time from datasheet. Also, using of the fall time from 15 volts to zero in our calculations is NOT fully correct too: mosfet will start to close at ~3 volts already. So we need to take approx 4/5 of the driver's output fall time (assuming we do not use the 'diode trick' here).
So the 'default' deadtime will be even not so much. Because of it I assumed to add another 30-40ns of deadtime.
I have done it for simplicity only. Ideally, it is needed to take into account the mosfet's gate threshold voltage (3-4 volts approx) - this is the voltage, where mosfet starts to open. In this case, we should add not the whole output rise time of the driver, but only approx 1/5 of it: the driver's output rise time is given in datasheet for 15 volts, where the gate threshold is ~3v, so the mosfet will start to open much earlier (almost 5 times earlier), before the driver will charge its gate capacitance to 15 volts, so the 'effective' rise time of driver's output will be approx 1/5 of the time from datasheet. Also, using of the fall time from 15 volts to zero in our calculations is NOT fully correct too: mosfet will start to close at ~3 volts already. So we need to take approx 4/5 of the driver's output fall time (assuming we do not use the 'diode trick' here
).So the 'default' deadtime will be even not so much. Because of it I assumed to add another 30-40ns of deadtime.
Last edited:
I think the best way to find out optimum "dead time" (actually drive pulse clearance or overlapping) is to look at the waveforms in detail (gate, drain, 20ns/div, >40Mhz scope) and adjust for little or no cross-conduction and the smallest possible delay error between sourcing and sinking current at the switching node.
IR2110 with gate turn off diodes is not likely to require any pulse clearance, in fact, in my current project the drive pulses require 15ns or so of overlap for optimum switching (the anti-cross-conduction gate snubbers that I use have to do with that, though, but it was already good without them).
IR2110 with gate turn off diodes is not likely to require any pulse clearance, in fact, in my current project the drive pulses require 15ns or so of overlap for optimum switching (the anti-cross-conduction gate snubbers that I use have to do with that, though, but it was already good without them).
This will reduce the current, but switching will be still hard
I would like to use here something between 1.5 - 2K, as a consesus 680 Ohm at 12 V will give approx 17mA. Whereas the recommended output current of tl494 is 200mA... Also, the output bipolar transistors of tl494 are saturating in common collector configuration. This will distort the duty cycle slightly. So 680 Ohms are even not so bad here
Also, how is the ringing? How does your ptototype sounds?
Last edited:
This will reduce the current, but switching will be still hard
680 Ohm at 12 V will give approx 17mA. Whereas the recommended output current of tl494 is 200mA... Also, the output bipolar transistors of tl494 are saturating in common collector configuration. This will distort the duty cycle slightly. So 680 Ohms are even not so bad here
Also, how is the ringing? How does your ptototype sounds?
ledmanias circuit worked fine.. yet to test mine
Upps, sorry, I meant open collector. This is how it is now configured in circuitcity's schematic.
http://img216.imageshack.us/img216/617/schematic.png
I built a 50W amplifier around the TI494 10 years ago and it sounds great. Sawtooth waveform works the same as a triangle waveform. If your using the TL494 you are not trying to achieve stellar sound anyway. With a good set of speakers you will be pleased. Keep in mind any class D amplifier that is driven with a clock signal like all switcher chips will emanate a small hiss (some more than others). That is why self oscillating is better because it acts like a sigma-delta converter and pushes the noise up and outside the audio bandwidth.
Well with ledmania's circuit I tested. The sound was good.
The speakers I was given are old but they look reasonable speakers to use. The Low pass filter characteristics in those speakers are great.
I switched at 196Khz and sound was good, but I had insane ringing.
I am yet to test the schematic I posted. Just the teacher got fed up with me playing around in breadboard (actually me too) and told me go directly to veroboard.
So that's why i am taking a bit longer now.
The speakers I was given are old but they look reasonable speakers to use. The Low pass filter characteristics in those speakers are great.
I switched at 196Khz and sound was good, but I had insane ringing.
I am yet to test the schematic I posted. Just the teacher got fed up with me playing around in breadboard (actually me too) and told me go directly to veroboard.
So that's why i am taking a bit longer now.
So you have connected the speaker with a long wires directly to the switching node..............
I do not wonder, why you had this "insane" ringing. You have connected the radiating antenna (speaker wires) directly to your power square wave generator. Had your teacher noticed about it? Had he recommended to make the deadtime bigger? Usually, peoples are trying to make the connections in switching node as short as possible (the connection between output mosfets and the trace to the output inductor), and are using the ground plane (one side of the PCB is a solid layer of copper, connected and used as Ground).
In your case it will be enough to simply add the output inductor (two for the full bridge). Use air gapped ferrite core or not gapped iron powder core (red or yellow). Connect it directly to the switching nodes.Good luck!
So you have connected the speaker with a long wires directly to the switching node..............
Usually, peoples are trying to make the connections in switching node as short as possible (the connection between output mosfets and the trace to the output inductor), and are using the ground plane (one side of the PCB is a solid layer of copper, connected and used as Ground).
Good luck!
Yeah he suggested both LC low pass filter and making the dead-time greater.
He also said about making the power flow loop as robust (short) as possible, just like what you mentioned. But I've had to take into account heatsinks so my loop is a bit big cause of it.
I'm using these inductors at the moment and yes I'm using two inductors and a single 100nF cap.
http://uk.farnell.com/epcos/b82111ec22/inductor-axial-22uh/dp/9753362
I just tested the power track today and it's ok. I regulate 20V down to 12V with the T0-220 I'm using..
Tomorrow I will solder up the PWM part and the XOR part.
This is my first time i'm desinging on veroboard, so I do have a lot of wires running everywhere. But I can't be bothered anymore in moving it around to minimize wires. It's almost time for me end this and write a report on it.
Last edited:
Ok did the PWM and XOR part.
PWM part is fine and I get 50% Duty Cycle (Approx).
But the XOR part concerns me. Though I get complimentary output PWM. With no input music signal, my output PWM out of the XOR is modulating, like there is a music signal.
I trigger my scope say rising edge and the falling edge is modulating.
I am Using CD4070BE XOR IC from Texas Instruments.
PWM part is fine and I get 50% Duty Cycle (Approx).
But the XOR part concerns me. Though I get complimentary output PWM. With no input music signal, my output PWM out of the XOR is modulating, like there is a music signal.
I trigger my scope say rising edge and the falling edge is modulating.
I am Using CD4070BE XOR IC from Texas Instruments.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.