Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

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

Status
Not open for further replies.
Okay i got it all connected now, but must "debug" the connections probably, because something "short circuits" the XR2206 signal completely to flatline when i scope the XR2206 output.
Just to make sure i understand the symmetry of the bridge arrangement, are the 2 amplifiers identical or does the lower one need to have the transistors swapped?
I have also 2 questions concerning your diagram in post +47 and the load connection.
Is the 1000uF cap an AC cap, and if not what is the polarity?
You symbolized a GND symbol close to the output.
As i understand it the load is completely floating between the 2 amplifier outputs or does it need to be grounded and if so then at what side?

Spice Simulation of the circuit only shows 4 V Peak voltage at the output:

**broken link removed**

The circuit :

**broken link removed**
 
Last edited:
As far as i got was to have 2 relatively nice 4.7V RMS signals that are out of phase to each other at the 2 emitters of the FIRST transistor stage with the 2nd transistor stage emitters not connected to VCC and GND.
For some reason as soon as i connect VCC or GND to them the signal dies.
I will try to replace the TIP120/125 tomorrow, to see if they are broken and sotosay short the signal constantly.

Here the first transistor stage signals:

**broken link removed**

Here the circuit how i made the measurement:

**broken link removed**
 
Last edited:
You don't need C3 since the output of each of the two bridged amplifiers is +6VDC.

With the emitters of the power darlingtons disconnected then the little transistors do not have a collector voltage so they act like simple base-emitter diodes.

With a +12.0V supply the max output level at clipping is about 10.6V p-p (and less when loaded) which is only 3.75V RMS, not 4.7V RMS.
 
Okay i got it all connected now, but must "debug" the connections probably, because something "short circuits" the XR2206 signal completely to flatline when i scope the XR2206 output.
Just to make sure i understand the symmetry of the bridge arrangement, are the 2 amplifiers identical or does the lower one need to have the transistors swapped?
I have also 2 questions concerning your diagram in post +47 and the load connection.
Is the 1000uF cap an AC cap, and if not what is the polarity?
You symbolized a GND symbol close to the output.
As i understand it the load is completely floating between the 2 amplifier outputs or does it need to be grounded and if so then at what side?

Spice Simulation of the circuit only shows 4 V Peak voltage at the output:

**broken link removed**

The circuit :

**broken link removed**


Hi again,


You dont have to ground anything as the outputs are floating. In fact, you can *not* ground anything or you wont get the full output voltage. Also, no transistors have to be swapped in the lower section. This is a bridge arrangement where we build two amplifiers each that can put out some maximum voltage Vp but one output is inverted so that when we take the output across the two amp outputs we get twice the output Vp as that of a single amp. The lower amp is already inverted so there's no need to swap any transistors. The result of using two amps like this is we get more output peak voltage which results in more output power into a given load.

As Audioguru says, you might not need C3, but you do have to be careful about this in case there is any output DC offset. Output DC offset can cause a high level DC current to flow in a low DC resistance device like an inductor or speaker, so depending on what the final load is going to be and what the DC offset is (and drift) you may or may not need C3 in the final application. For a 10 ohm resistor load, a 0.1v DC offset only means 0.1/10=10ma DC offset current which isnt bad. In an 8 ohm speaker it wont be too bad either. For a low resistance inductor however it could be quite high, depending on the effective series resistance.

You'll also want to note that by using Darlington Compound transistors on the output you'll be losing more peak voltage. That's because of the extra emitter drop. Using a single transistor instead will mean less voltage drop. If you see clipping with the Darlington you might not see that with a single transistor. If you do see clipping and you want to get rid of it without changing transistors you would have to lower the gain. To lower (or raise) the gain you would change the two feedback resistors, making the lower one 25k higher than the upper one (assuming the two other resistors are both 25k like they are in you diagram).
 
Last edited:
Thank you Audioguru and MrAl for the detailed explanation of the DC offset.
Indeed the waveform is clipped.
It did not change with a non-darlington pair (TIP3055/2955)
I will try now different resistors now.

EDIT: Hmm, even with both resistors being 25k the waveform still looks like this.
I only check the upper amplifier for now with the 10 Ohm load to ground as
a collector load and the scope at the collector.
Somewhere this cut in the sine top is being introduced.
Since i should be at unity gain with both feedback-resistors equal, it can't be
the op-amp.
Disconnecting the load resistor makes it less severe, but still there.

**broken link removed**

**broken link removed**
 
Last edited:
You must not connect the load to ground when you test one amplifier of a bridged amplifier arrangementWhen the power sdupply is a single polarity. Because then the lower darlington does nothing.
Connect a 1000uF capacitor to the output of the amplifier and the load to ground at the negative end of the capacitor.

Your 'scope is a nightmare because it shows RMS voltages but a real 'scope shows true peak voltages with graticle lines showing peak volts.

I would expect the output of one amplifier with a output coupling capacitor feeding the 10 ohms load to symmetrically clip with a peak-to-peak voltage of about 8V.
 
@Audioguru: Thank you for your patience. Your suggestion was most helpful.
The introduction of the coupling capacitor smoothed the signal.
The peak to peak voltage is in the range you have anticipated (9.3 V at unity).

At unity gain:

**broken link removed**

With the original resistor values 25k/50k:

**broken link removed**
 
Last edited:
The slanted clipped portion of the sine-wave is caused by the phase shift of the input and output coupling capacitors and the feedback to ground capacitor because their values are too small for a frequency as low as 50Hz.
 
@audioguru: This helped. I understand now how the component values influence
the waveform.
I raised the capacitance of the output coupling capacitor to 4700uF and the "ripple"
close to the peaks of the waveform is gone now.
The slightly flattened parts of the sine peaks could be compensated with actually lowering the feedback to ground capacitor value to 100 nF.
That also lowered the output voltage peak to peak value to 7.8V.
I assumed it's a trade-off situation between waveform shape and output voltage.
Maybe i can make up for this by raising the ratio of the feedback loop resistors above unity.
 
Last edited:
After equalizing the voltage peak to peak of both amplifiers to about 7.8V, i bridged the amplifiers and get 14.8V p2p across the 10 Ohm resistor.
This is exactly in the range that i wanted to reach.
The little ripple can't be removed with a DC cap or even 2 DC caps from both sides like i did with a single amplifier.

**broken link removed**

I am wondering if lowpass-filtering the signal would clean that up?
If i assume a first-order low-pass RC filter, then
1/2*pi*10 Ohm * 50 Hz=0.00031830988618 F = 318 uF (quite big :/ )

EDIT: I get the best looking signal with a 47uF AC cap:


**broken link removed**

Bottom sine peak looks only truncated due to oscilloscope scaling.

And with the transformer primary coil in series (just out of curiosity):

**broken link removed**
 
Last edited:
Reducing the value of the feedback to ground capacitor value to 100nF reduces the amplification of low frequencies (50Hz) but the distortion from the XR2206 at 100Hz and 150Hz has the original amplification so there is more distortion at the output.
Your 'scope shows severe distortion at the output. The amplifier should show no distortion so something is wrong.
 
If you refer to the small ripple waves on the fundamental wave, then this is most likely coming from my USB scope. I am a bit punished with it, because it has this ripple even with everything disconnected and after calibration.
I have no clue how to fix this, actually it probably can't as manufacturer states 3 % noise as normal
http://www.hantek.org/asken/iaskdetail.aspx?id=2009100405494952


Number 1 on my To-Get-List is a better scope. *sigh*

If you refer to a different aspect of distortion than that ripple, then i can try to fix it, if you have an idea
what to change to improve it.
 
Last edited:
This explains why your circuit or oscilloscope is garbage:
 

Attachments

  • 'scope photo.PNG
    'scope photo.PNG
    80.5 KB · Views: 155
Thank you. My oscilloscope and my circuit are garbage.

Hi,

That noise looks like it is somewhat systematic which means it could be coming from inside the scope itself and interfering with the signal. It's probably coming from the timing circuits inside the scope. What might help is to use some external amplification and set the vertical sensitivity a bit higher. For example, with no amplifier and the scope vertical set on 0.1v/div you might see some objectional noise riding on the signal, but if you use a 10x external amp and set the scope on 1v/div that noise should not be as pronounced as long as the external amp is designed right. Perhaps a high speed op amp will get you there, but i dont know the max frequency capability of your scope.

You could build a small op amp amplifier and power it with a battery or two and see how it helps. The input impedance for your new scope booster amp doesnt have to be super high for testing the transistor amplifier you are working on or for testing power supplies.
 
Last edited:
Hi,

That noise looks like it is somewhat systematic which means it could be coming from inside the scope itself and interfering with the signal. It's probably coming from the timing circuits inside the scope. What might help is to use some external amplification and set the vertical sensitivity a bit higher. For example, with no amplifier and the scope vertical set on 0.1v/div you might see some objectional noise riding on the signal, but if you use a 10x external amp and set the scope on 1v/div that noise should not be as pronounced as long as the external amp is designed right. Perhaps a high speed op amp will get you there, but i dont know the max frequency capability of your scope.

You could build a small op amp amplifier and power it with a battery or two and see how it helps. The input impedance for your new scope booster amp doesnt have to be super high for testing the transistor amplifier you are working on or for testing power supplies.

Thank you MrAl.
I would not have come to the idea to improve the scope in such way,
sounds like a good way to reduce the noise.
The scope model is DSO-5200 USB
Bandwidth 200MHz (-3dB)
200Mhz: analog bandwidth, 250MSa/S: real-time sampling.
Probably a ultra-fast op-amp would be needed.
I have no idea however, how exactly to connect the op-amp to the probe and how
to set the amplification up.
 
Hi again,

Well, it could work if the conditions inside the scope are such that the noise is either relative or the noise is injected into a stage before the variable gain of the scope vertical amplifier. You could probably get away with a 100MHz bandwidth but that's going to be a little bit of a challenge. If the noise is being injected after the variable gain however this wont help, so a simple test is in order to find out for sure. It could also be quantization noise which may or may not mean an improvement so you really need to do at least one simple test.

To test this, build up a simple op amp amplifier and see how the lower frequency rate signals look, like a 50Hz or even 1kHz square wave. The amplifier could have a gain of 10 which would be good for testing this principle on your scope. If it works right, you could look for a better op amp.

The circuit is quite simple really, an input resistor of 10k going from inverting terminal to ground, and a 90k resistor from output to inverting terminal, and plus and minus 9v batteries for the testing power supply. You then input your signal into the non inverting terminal and ground, making sure not to exceed an input voltage of about 8 volts. You could probably try this with the op amp you are using for your amplifier in this thread. If it works, look for a better op amp and maybe make some other improvements.

In theory if the noise affects the signal before the variable gain of the scope the noise reduction will be proportional to the gain of the external op amp provided of course the op amp isnt noisy. So for a gain of 10 the noise would be reduced 10 times, but even for a gain of 2 it would be reduced by 50 percent.

If you have a variable amplitude square wave, you could set it on 1v peak and see how much noise you get, then set it to 10v peak and see if the noise on the scope looks smaller relative to the signal itself. If the noise goes down, the op amp amplifier technique will help, but if the noise doesnt decrease, then the op amp might not help.
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top