Hi,

I'm building class d amplifier and 'm using lm319 as a comparator. My inputs are sine(150 mV p-p) and triangle(200mV p-p) waves. But I saw on output that -+ 150 mV p-p PWM.(I used each two outputs.)It's ok, real PWM but low voltage.

I want to see 0 V for low and 9-12 V for high on output. I tried 1k push-pull resistors for two outputs in LM319. Then I didn't see anything significant(there aren't any PWM). Are resistors values wrong?
I tried for two case: 1-) +V = +12, -V= -12 V, and ground = 0V
2-)+V= +12V, -V = ground= 0V
But it isn't solved I read datasheet again and again.)

What is the problem? (I'm working with physical components, not simulation.)And solution?

 

Does it work in your simulation?

I see you're using LTSpice. I downloaded occomp.lib from the Yahoo! LTSpice Users' Group, and set my simulation up very similarly to yours; it seems to work fine - but I don't think you got your LM319 model there.

Do you have bypass caps from +V to gnd, and -V to ground? I suggest 0.1uF ceramic or poly metal film, and 1uF or larger aluminum electrolytics (a total of 4 caps). If you don't have bypass caps, you'll have problems.

Check to make absolutely certain that your LM319 pins 3 and 8 are grounded.

You might have a "bad spot" on your breadboard. Try moving the circuit to another area.

Do you have any other load on the output besides the pull-up resistor?

250kHz isn't all that fast, but when your output is a square wave, you're requiring 10x or more bandwidth to get a decent-looking square wave out; as an ideal square wave is the sum of ALL of the odd harmonics of the fundamental frequency, which implies unlimited bandwidth. When you are putting out a non-square (not 50% duty cycle) rectangular wave, the bandwidth requirement increases.

 

Does it work in your simulation?

I see you're using LTSpice. I downloaded occomp.lib from the Yahoo! LTSpice Users' Group, and set my simulation up very similarly to yours; it seems to work fine - but I don't think you got your LM319 model there.

Do you have bypass caps from +V to gnd, and -V to ground? I suggest 0.1uF ceramic or poly metal film, and 1uF or larger aluminum electrolytics (a total of 4 caps). If you don't have bypass caps, you'll have problems.

Check to make absolutely certain that your LM319 pins 3 and 8 are grounded.

You might have a "bad spot" on your breadboard. Try moving the circuit to another area.

Do you have any other load on the output besides the pull-up resistor?

250kHz isn't all that fast, but when your output is a square wave, you're requiring 10x or more bandwidth to get a decent-looking square wave out; as an ideal square wave is the sum of ALL of the odd harmonics of the fundamental frequency, which implies unlimited bandwidth. When you are putting out a non-square (not 50% duty cycle) rectangular wave, the bandwidth requirement increases.

ANSWERS:

1-I used lt1016 for simulation. and my amplifier worked. Then I used LM319 for real amp. I didn't use breadboard. I used solder and perforated plate. But in first condition(low output voltage), my circuit worked.

2- I didn't use bypass cap. I'll try it. But this requirement couldn't valid for my first condition(low voltage output)? I didn't use any cap.
3- yes, pin 3 and 8 are grounded.

4-No , there aren't any resistor on output. only outputs now.

5- I didn't understand what do you want to say to me in last paragraph.

Special thanks for you, thank you.

 

The LT1016 is very different from the LM319. It's far faster, has complementary outputs, voltage supply range is different, and I could go on.

Proper bypass caps are mandatory. The inductance of the wiring between your power supply and your circuit will cause huge problems otherwise.

The LM319 has open-collector outputs. You must use a pull-up resistor to see any output, as the output can sink current, it cannot source it.

What I was trying to tell you at the end is your circuit's output waveform shape depends on the circuit being designed properly, with proper bypass capacitors across both supply rails to ground.

Here's the schematic of my simulation:

 

The LT1016 is very different from the LM319. It's far faster, has complementary outputs, voltage supply range is different, and I could go on.

Proper bypass caps are mandatory. The inductance of the wiring between your power supply and your circuit will cause huge problems otherwise.

The LM319 has open-collector outputs. You must use a pull-up resistor to see any output, as the output can sink current, it cannot source it.

What I was trying to tell you at the end is your circuit's output waveform shape depends on the circuit being designed properly, with proper bypass capacitors across both supply rails to ground.

Here's the schematic of my simulation:

But in this configuration, output voltages swing around + 12 V, -12 V?? Comparator basics : V+> V- == output ise about positive supply voltage

V+<V- == output is about negative supply voltage ??

I want 1> V for low output and 9< V for high output ... for IR2110 input

 

In the schematic I posted, the output varies between V+ and GND. It does not go to V-.

See the ground pin on the LM319?
The comparator won't sink the output lower than that ground pin. So, establish a solid reference somewhere between +V and -V that has the lowest voltage you want to output.

As far as the high voltage, you can establish another reference - say, using an LM7809 or 7808 or whatever you want - to pull the output of the LM319 to, using a resistor.

 

Ok, I'll try it tomorrow. I wish, my LM319 didn't break because of my configuration. Its output was absurd lastly.

I didn't understand one thing. Only interest. What does -V supply make? What is the difference -V= -12 and -V = ground?

 

Ok, I'll try it tomorrow. I wish, my LM319 didn't break because of my configuration.

I hope you didn't break it as well. The LM319 should be able to sink a fair amount of current; say up to 40-50mA without any adverse side effects.

Its output was absurd lastly.

If your output was hovering around ground, that makes me think that your output pull-up resistor was not connected to a positive voltage source (+V)

I didn't understand one thing. Only interest. What does -V supply make? What is the difference -V= -12 and -V = ground?

They are just convenient labels for "nodes" in the schematic.
I could have labeled them Vcc for V+ and Vee for V-, or +12v and -12v, or HI and LO, or many other possibilities; however V+ and V- simply mean that those labeled points are connected to the most positive and most negative voltages in the schematic.

I also included the voltage sources so that you can see what V+ and V- were in the simulation. Besides being meaningful, V+ and V- are more convenient than labeling something +12v; -12v, etc - as if you decided to change voltages, all you would have to do is change the numbers at the voltage source itself, instead of having to change all of the node names to keep them meaningful. After all, if you had a point labeled "+10V" but you actually had the voltage source set to 15v, the viewer of the schematic could get quite confused.

Vcc and Vee are common for transistorized circuits; Vdd and Vss are commonly used for MOS circuits. However, using V+ and V- leaves little doubt as to what is connected to what.

 

I have two 1k pull-up resistor and two of them connected to +12V.

I know -v is -12 V, but i didn't understand if output never goes negative side, what is -12 V's mission? sometimes ground and Vcc connects each other I see.

Last question , my triangle generator is XR2206 and in its circuit, I have 10uF cap between ground and -12 v and some caps. connected +12, ground and -12. Now, if I connect new caps as you said, any problem will occur?

 

Thank you SgtWookie, it works. PWM is generated 0V -12 V . But 12 V output , a bit high as IR2110 input? It mustn't be higher than Ir2110 supply, its supply is the same LM319's supply.

 

I have two 1k pull-up resistor and two of them connected to +12V.

OK.

I know -v is -12 V, but i didn't understand if output never goes negative side, what is -12 V's mission?

The comparator can only "see" input voltages that are from its' V- rail to the V+ rail less about 1.5v. If you had the comparators V- pin connected to ground instead, your output would not be correct when the inputs went below 0v/ground.

sometimes ground and Vcc connects each other I see.

That statement does not make sense. If ground and Vcc were connected together, the power supply would be shorted out.

Last question , my triangle generator is XR2206 and in its circuit, I have 10uF cap between ground and -12 v and some caps. connected +12, ground and -12. Now, if I connect new caps as you said, any problem will occur?

You need to post your existing schematic, and show where you are thinking of adding new capacitors.

 

Thank you SgtWookie, it works. PWM is generated 0V -12 V .

Actually, it should be between 0v and +12v, alternating between the two.

But 12 V output, a bit high as IR2110 input? It mustn't be higher than Ir2110 supply, its supply is the same LM319's supply.

You can connect the 1k output pull-up resistor to the IR2110's logic supply instead of V+. In the IR2110 datasheet, the IR2110's logic supply pin is labeled Vdd, and the logic supply ground is labeled Vss.

 

Actually, it should be between 0v and +12v, alternating between the two.



You can connect the 1k output pull-up resistor to the IR2110's logic supply instead of V+. In the IR2110 datasheet, the IR2110's logic supply pin is labeled Vdd, and the logic supply ground is labeled Vss.

No , I saw 0V and +12 V on the output of the comparator() as pWMs. But This values a bit high for IR2110 Hin and Lin inputs? (IR2110' supply is 12V)

 

The logic input voltage range is from Vss to Vdd.
If Vdd is at +12v and Vss is connected to ground, you can use the comparator configuration as it is now.

Read the datasheet.

 

The image you referred to in your PM:

looks pretty much OK, but you need 0.1uF ceramic or metal poly film caps across the Vdd and Vss pins, and if caps like C1 & C4 are not poly metal film or ceramic, then add a 0.1uF cap in parallel.

The gate resistors being 18 Ohms should be approximately correct, and the diode will help with dead time. You need to be certain that the high and low side MOSFETs on either half-H cannot be on at the same time. These MOSFETs turn on more quickly than they turn off. If you do not provide sufficient dead time, you will short the 12v or 15v power supply to ground, and it is very likely that one or both of the MOSFETs will be destroyed.

 

The image you referred to in your PM:

looks pretty much OK, but you need 0.1uF ceramic or metal poly film caps across the Vdd and Vss pins, and if caps like C1 & C4 are not poly metal film or ceramic, then add a 0.1uF cap in parallel.

The gate resistors being 18 Ohms should be approximately correct, and the diode will help with dead time. You need to be certain that the high and low side MOSFETs on either half-H cannot be on at the same time. These MOSFETs turn on more quickly than they turn off. If you do not provide sufficient dead time, you will short the 12v or 15v power supply to ground, and it is very likely that one or both of the MOSFETs will be destroyed.

C1 C2 C3 C4 are tantal caps.

0.1 uF cap for Vdd and vss is necessary? Because Vdd is equal to Vcc and , Vss is equal to ground. And there are many caps between power to ground.

I will try 18 ohm, then if mosfets destroy, I 'll use 33 ohm resistor and new mosfets.

 

Every IC you use needs a minimum of a 0.1uF cap across each power supply "rail". If it is a single supply like in this case, you need just 1.

The capacitor leads and trace need to be as short as possible, and the cap needs to be connected as close to the pins as possible.

 

Thanks for all.

I didn't use any charge pump circuit because of LC filter and its inductive effect.

Is there any problem??

 

C1 and C4 are boost caps.
The respective low-side MOSFETs must be turned ON before either C1 or C4 can charge.
You should see that in the simulation.

 

Hi again,

When comparator outputs connected to the IR2110 Hin/Lin, comparator outputs' voltage was ~2,5 V.

Then I cut this connection and looked comparator outputs, I saw 12 V PWM.

So, when connection was active, My circuit didn't work.

I need a buffer between comparator outputs and IR2110 inputs?? ( I have lm709 op-amp but it maybe slow, is it work as buffer?)