Transformers signal distorted

[i12know]

Hi guys,

I have built my own transformers which will increase my input voltage to about 10 times higher. I have also built a circuit according to the picture attached in the post. For my signal, I am generating a 10 pulse square wave signal in about 250us (40kHz at that period) and stay at 0 for say 500ms. And then I regenerate that 10 pulses again, and this goes on and on. And I would expect to see that 10 pulses at the output of the transformer as well, with higher voltage.

But my problem is I am losing the first 5 signal when I check the voltage at the output of the transformers. I see it as a constant voltage, and then 5 pulses at the end. Is that caused by the transformers itself? If it is, what is causing it? And can it be improved? NOTE: If I generate only 5 pulse, I will not see any pulse at the output of the transformers.

I am using a FET (as shown in Q40 in the diagram) before the transformer, and I have checked that without the transformer, the output of the FET seems to be fine, as I can see 10 pulses at the output, so I would expect it not to be the cause. Is that correct? Or I am missing something here where I can't check the output of the FET just like that?

Thanks a lot in advance.

 

[Grossel]

Given that the connection is correct: You must be aware that your transformer is nothing like an ideal transformer. There is capacitances all over that would make it resonate for certain frequenzis.

The same transformer will probably work better for other frequensis than your choice.

 

[crutschow]

I did a simulation and the problem appears to be related to using a transformer with a uni-directional pulse. This means there is some average DC current that will flow in the primary (which can saturate the core if it becomes too high). This seems to require a number of input pulses to establish this core current. During this time little appears on the output.

I did find that adding a resistive load seems to improve things. Do you have any load on the transformer output besides the scope probe?

 

[i12know]

Given that the connection is correct: You must be aware that your transformer is nothing like an ideal transformer. There is capacitances all over that would make it resonate for certain frequenzis.

The same transformer will probably work better for other frequensis than your choice.

I will try with different frequencies and see if it still does the same thing, but I guess there is no easy way other than trial and error to know what is the resonance frequency of the transformers, is there?

I did a simulation and the problem appears to be related to using a transformer with a uni-directional pulse. This means there is some average DC current that will flow in the primary (which can saturate the core if it becomes too high). This seems to require a number of input pulses to establish this core current. During this time little appears on the output.

I did find that adding a resistive load seems to improve things. Do you have any load on the transformer output besides the scope probe?

Yes, I do have a transducer that acts as a load on the transformer output, vibrating with the signal from the transformer.

And just want to clarify some confusion that I have regarding the transformer with a uni-directional pulse that you mentioned, did you mean that for that kind of transformer, it will always need some DC current in the primary before it can start transmitting its power to the secondary? If that is the case, anyway I can get around that? Or probably going to a bi-directional pulse transformer or something?

 

[ben7]

Well what exactly are the diodes on the output of the transformer supposed to do?

-Ben

 

[crutschow]

.......
Yes, I do have a transducer that acts as a load on the transformer output, vibrating with the signal from the transformer.

And just want to clarify some confusion that I have regarding the transformer with a uni-directional pulse that you mentioned, did you mean that for that kind of transformer, it will always need some DC current in the primary before it can start transmitting its power to the secondary? If that is the case, anyway I can get around that? Or probably going to a bi-directional pulse transformer or something?

You might try adding an additional small resistance load and see if that has any effect.

 

[i12know]

Well what exactly are the diodes on the output of the transformer supposed to do?

-Ben

From what I know, it is there to prevent the transducer's signal at the output going backwards and blow the FET. But please correct me if I am wrong, would be good to know that too.

You might try adding an additional small resistance load and see if that has any effect.

You mean adding a resistor or some other load in series with the transducers?

 

[crutschow]

From what I know, it is there to prevent the transducer's signal at the output going backwards and blow the FET. But please correct me if I am wrong, would be good to know that too.

You mean adding a resistor or some other load in series with the transducers?

All two back-to-back diodes will do is reduce the peak AC voltage by about 0.7V.

Add a resistor in parallel with the load. You'll have to experiment with the values.

 

[i12know]

All two back-to-back diodes will do is reduce the peak AC voltage by about 0.7V.

Add a resistor in parallel with the load. You'll have to experiment with the values.

I haven't had the chance to experiment with that yet, hopefully will be able to try that out today or tomorrow, but just want to know as well, what is the benefit of adding the resistor in parallel here?

 

[crutschow]

I haven't had the chance to experiment with that yet, hopefully will be able to try that out today or tomorrow, but just want to know as well, what is the benefit of adding the resistor in parallel here?

The benefit, based upon my simulation, is that it may minimize the problem you are noting.

 

[i12know]

The benefit, based upon my simulation, is that it may minimize the problem you are noting.

Yup I do figure that is the case, but what I am after is more about the theory of adding that resistor. Adding the resistor in parallel will reduce the overall impedance of the load, and draw more current from the primary. And what will happen next? I am stuck here.

 

[crutschow]

Well, I don't have a good understanding of that. It appears to allow the DC magnetizing current to built up more rapidly, which allows the pulses to appear sooner. But the theory behind that is not clear to me.

And even with that I'm still concerned about this DC magnetizing current saturating the core. A transformer should really be driven by an alternating plus and minus signal of equal value to minimize the DC current. One way to eliminate any DC is to drive the transformer through a series capacitor in the primary. The capacitor will not allow any net DC current independent of the duty-cycle of the waveform.

 

[i12know]

One thing that I noticed is that, when I changed the supplied voltage, the number of pulses that 'disappear' changes as well. When I lower the supplied voltage to 12V, the pulses appeared sooner (I can see 7 or 8 pulses at the output, with only 2 or 3 missing).

And I also noticed that the pulses are missing on the primary as well, not just at the output of the transformer.

 

[i12know]

Well, I don't have a good understanding of that. It appears to allow the DC magnetizing current to built up more rapidly, which allows the pulses to appear sooner. But the theory behind that is not clear to me.

And even with that I'm still concerned about this DC magnetizing current saturating the core. A transformer should really be driven by an alternating plus and minus signal of equal value to minimize the DC current. One way to eliminate any DC is to drive the transformer through a series capacitor in the primary. The capacitor will not allow any net DC current independent of the duty-cycle of the waveform.

I see. I am actually going to try magnetizing the core using DC current by adding a resistor parallel to the FET, so that when the FET is turned off, there is still some current going through the core. Not sure if it is a good idea or not, as it will be a waste of power to have current going through the core and the resistor all the time, but if that helps, then at least I would be able to narrow the problem down hopefully, and can then find other better solutions to work it out. Right now I am just waiting for some components to arrive to try out your suggestions.

 

[crutschow]

Adding a resistor in parallel to the FET is likely not a good idea since that will just increase the DC core current and the possibility of core saturation. DC core current is normally undesirable and you want to minimize or eliminate it.

 

[i12know]

Adding a resistor in parallel to the FET is likely not a good idea since that will just increase the DC core current and the possibility of core saturation. DC core current is normally undesirable and you want to minimize or eliminate it.

Well, I don't have a good understanding of that. It appears to allow the DC magnetizing current to built up more rapidly, which allows the pulses to appear sooner. But the theory behind that is not clear to me.

I am a bit confused here. I thought the idea to make the pulses appear sooner was to allow some DC current flowing through the core first before the pulses. Is the DC magnetizing current that you mentioned the same as the DC core current?