Op-amp/transistor/transformer connection problem in inverter circuit

[xeno]

@crutschow: Thank you. I wanted to have 1 amp @ around 10-12 Volt going through the primary.
If the calculation is not more complex for that than ohm's law then i would need 12 Ohm resistance/impedance for the coil.

My real-life op-amp is a TS912, i have not been able to correctly include the spice model data for that model, so i chose a different op-amp for spice.
I will also experiment with higher inductance coils after your suggestion.

 

[crutschow]

If you want 1A then that should be a resistive load, not inductive, as a transformer just reflects the secondary output load (which I assume is resistive) back to the primary.

 

[xeno]

Ah okay, so using a 10-20 Watt 10 Ohm resistor could work i guess.
Signal looks better when uping the inductance.
Not sure if the connectivity of smaller transistors to bigger transistors
is correct like that.
Ideally i had hoped that i would find an op-amp that would amplify the
weak XR2206 signal enough by itself to drive the big transistors, but the smaller transistors should do
the job fine.

**broken link removed**

 

[MrAl]

Hi again,

The feedback moved again

The feedback resistor R2 needs to connect to the very output.

A larger inductor simulates a transformer better, but you also eventually need to add some parallel resistance to simulate the load.

 

[xeno]

MrAl, i moved the feedback to the output emitter before the cap.
I increased the cap to 2200 uF, inserted a 10 Ohm resistor in series to the coil.
In parallel i cant get a reading for some reason.
I also added small transistors to boost the OP amp output.
Now my waveform across the coil looks like this:
**broken link removed**

**broken link removed**

 

[MrAl]

Hi again,

Did you set R2 to 100k again? Set that to 20k like R1 and try again.

 

[xeno]

Yes i set it to 100k again, because i have a hard time to get 10 Volt out of the XR2206 output in the end, but i will do what you say and compare )

Edit: Much better ! Tops are smoother, but that stepping is still there. What causes it and how can i improve it?

**broken link removed**

Would i need a different op-amp than the TS912?

EDIT2: If i leave the inductor out and just have the 10 Ohm resistor as a load after the transistors, then the waveform looks more like the intended waveform:

**broken link removed**

EDIT3: My learning curve concerning this is not as steep as it should be, i apologize.
I have come to the conclusion that the deformed waveform is a consequence of a magnetic problem probably related to saturation of the core.
When i put a microwave oven fan coil in place for my toroidal inductor then i get a wonderful sinusoid waveform. As has been already pointed out, the low number of turns (100) on my inductor in comparison to the many thousands (0.775 H) of the oven fan coil might not be enough.
If someone knows exactly what is going on here, i would very much like to hear an explanation, so i can
understand the phenomenon better. Thank you.

 

[MrAl]

Hi again,

Since you were having so much trouble with this amplifier i decided to sim it myself too. The results are shown in the diagram attached.
This is with two 25k resistors for R1 and R2 and on the output i have 1 ohm in series with 10uf in series with 10mH.
The input is a 1v peak sine 50Hz. The blue wave is the junction of the two 100k resistors, and the red wave is the output at the emitters of the two transistors. The two transistors are 2N4400 and 2N4403, and that is all that drives the output. At this somewhat lower voltage level the sine output looks quite good as expected with the type of feedback we are using.

You can see some initial spikes on the output, but they go away as the caps charge up to their nominal values. Toward the end of the diagram you can see the sine start to look very good.

If you are getting saturation in the real world, you wont be able to use that particular coil. That's the way it goes. If the coil primary can not sustain the voltage presented to it at the required frequency then all you can do is get another coil that can handle the higher current levels.

How are you getting these waveforms BTW, are you using a scope? If so, try looking at the current wave into the coil and post that too. It would also help for you to post the pics with a little bigger text so we can read the text too.

 

[xeno]

Thank you for the simulation and the explanation.
I had used 0.8mm wire for winding the primary, i will try a much smaller diameter wire which
will have more resistance.
The amplifier part should work fine now as your simulation showed.
All that is left for me is to educate myself more on how to design the transformer
appropriately to avoid saturation.

Here the current measured across a 1 ohm resistor in series with the coil (GREEN)
I hope that is the correct way to measure the current in that context.
The voltage across the coil is the yellow curve.

**broken link removed**

It was good that i finally measured the current, because it seems to show that the current is far from
1 ampere like i wanted it to be.

 

[MrAl]

Hi again,

For a sine wave, here is a little formula that is handy to have around...

Bmax=E*1e8 / (4.44*Ae*N*F) in Gauss

E = applied RMS volts
Ae = cross-sectional area of core (cm^2)
N = number of wire turns
F = frequency (Hz)

B is the flux density of the core. You have to look up your core data or use a typical max value for B to find what Bmax should be for your core. Once you know that, you can use that formula to figure out the minimum number of turns for a given applied voltage (sine). Typical values for EI steel laminations commonly used for 50 or 60Hz operation are 10000 to 15000 Gauss. The formula assumes no average DC current is present in the winding.

 

[xeno]

MrAl, thanks for the formula.

Here is the exact schematic of my built circuit (The XR2206 part was not included)

**broken link removed**

 

[MrAl]

Hi again,

You're welcome.

The feedback resistor R2 is still not connected to the very output emitters, that should be done first.

What kind of core are you using for the coil?

 

[xeno]

My bad, i have it correctly connected in the real build, but forgot it in the schematic.
I updated above schematic.

At the moment i am using a ferrite toroid, since there was nothing better at hand
and i did not anticipate the core material to influence the circuit functionality at an early stage.

For a later experiment i want to go higher in frequency and then a ferrite would be suitable, but maybe not for 50 Hz )

I just measured the output of the XR2206 again. Through a 1 ohm resistor the current is 4.8 mA
and the voltage 0.8 V RMS. I will look into uping that a bit inside the actual XR2206 circuit.
But the output current seems to be normal for the IC.

 

[MrAl]

Hi again,

The output of the XR chip is only 0.8 volts?

Depending on the toroid, it could saturate quite easily. It depends highly on the core material and if there is any gap and how big the gap is. Do you know what kind of material the toroid is made up of?

Perhaps you should start testing with just a resistive load and get that working properly, then go to another type of load later.

 

[xeno]

I have the output of the XR2206 now at 1.5 Volt RMS.
A 2.5 Volt can be reached but then the waveform is not clean sine, so
i will try to work on that a bit more, maybe squeeze a few millivolt plus.
You are right about working just with resistive load at the moment.
I will try now to measure the currents between the Op-Amp and the first
transistor stage. I have tried already a 1 ohm resistor to ground and scope across, but
that killed the signal (on second thought, it is obvious why hehe). Probably the TS912 Op-Amp load impedance requirement of 100/600 Ohm must be met by the resistor or better expressed there must be a load resistance + the current measuring resistor.
To have full control over the current amplification of the stages i need to know
the current out of the Op-Amp with the transistors connected. The output short circuit current as printed in the data sheet is about 70 mA.

About the toroid i just know it is ferrite (coloured in grass green)

 

[xeno]

Okay, i have made the scope measurement behind the Op-amp.
I used a 100 Ohm resistor and got this:
GREEN : Voltage
YELLOW : CURRENT

**broken link removed**

Then i remembered the coupling with the capacitor hinted by eric and got this:

**broken link removed**

So the measurement sub circuit looks like :

**broken link removed**

If the method to measure the current is correct, this tells me that
there is 2.81V RMS / 100 Ohm = 0.0281 A = 28.1 mA current
delivered by the Op-Amp to a 100 Ohm load.

 

[ericgibbs]

hi xeno,
Is that R5 value supposed to be 100K.?

For a 50Hz sine wave input I would use a 470uF or 1000uF for C3.

 

[xeno]

@ericgibbs: Thank you, i used the copy function in spice and forgot to change the value of R5, it is 100 Ohm.
Depending on what load a pair of TIP120/TIP125 would present to the OpAmp.
it seems that it would be theoretically possible to drive them from the OpAmp only.
The darlington transistors have a hfe of 1000.
To get about 3.5 Watt (10 Volt p2p @ 1 Amp RMS) in the primary,
i would need a base current of 1 mA [1A/1000(hfe)] only.
But my calculation could be wrong.

In my initial test i just attached the transistors to the Op-Amp, but that looks like it is not
presenting an impedance to the op-Amp, so maybe a 1000 Ohm base resistor could do that which should still give 2.8 mA of base current and with full current amplification even 2.8 A.
I lack the experience in setting up complex amplifier transistor circuits.
Something like this:

**broken link removed**

If this approach is flawed, then i need advice on how to drive via smaller transistors, what resistances to use.

 

[MrAl]

Hello again,


You keep moving that feedback resistor R2 to some other point. The right side of it has to be connected to the two emitters or the distortion will be way too high. ALWAYS connect it to the output emitters unless there is some REALLY good reason for not doing so.

The reason this is so important is because the two emitter base voltage drops mean the bases require a non sinusoidal voltage in order to get a sinusoidal voltage at the emitters. The op amp takes care of this as it slews up (or down) to match the emitter voltage. If the feedback comes from the output of the op amp instead of from the emitters, the op amp puts out a sinusoidal voltage which is not the correct signal for the two bases. This doesnt matter as much when the sine voltage is a lot higher than the base emitter voltages, but when it's like like 10v or even 20v it's going to make a huge difference in the distortion.

 

[xeno]

It has not really moved, while adding more components like the transistors in Spice i forgot to attach the feedback to the output.
I understand the importance of putting it to the output. Thank you for your elaboration, it helps.
The correct driving of the transistors is the main concern now.
I must read up on classical amplifier design to tackle this and appreciate any
suggestions.

EDIT: The Op-amp to transistor circuit might need to be like this : https://redcircuits.com/Page1.htm

**broken link removed**

If i interpret this circuit correctly, the resistors between op-amp and first transistor stage just set the voltage offset, so there is no actual base resistor in the classical sense.