[SOLVED] Understanding Output waveform ?

[jayce3390]

Hi,

I am using an RF 5-500MHz 2W amplifier. A perfect 50 MHz pulse signal is driving this amplifier.

Could you explain me why I got this output waveform? (The green trace.)

See attached picture.

Thank you.

 

[biff44]

Looks like it is oscillating. Do you have 50 ohm broadband loads at input and output? Since the signal is ringing so much, I would guess not.

ALso, could be DC bias problem, as you are passing a DC based signal thru an AC coupled amplifier.

 

[jayce3390]

Yes output & input are 50 ohms teminated.
I guess the answer might be the second point you mentionned but I didn't get it actually. The mean voltage of the input pulse signal is zero volt (centered). Could help?

 

[biff44]

You have 3 different traces. Could you explain what each trace is?

 

[jayce3390]

Yellow is related to input signal just before the PA and green is output. The other one is a mathematical treatment of the first one...

 

[BigBoss]

I think it's a power supply problem.The amplifier is not supplied sufficiently and therefore supply voltage is dropping and ringing because of protection circuit..
It's my guess..

 

[Azulykit]

You might also want to see if the disruption is due to some nearby circuit. Often strange signals originate from "elsewhere". Move into a screen room ..... or just move elsewhere.

Are you saturating the amplifier? What if you increase the input amplitude? What if you decrease it?

Is your input really pure? Are you sure?

 

[biff44]

I think it's a power supply problem.The amplifier is not supplied sufficiently and therefore supply voltage is dropping and ringing because of protection circuit..
It's my guess..

Yeah, I might go along with that line of thinking. Try putting 470 uF of capacitors on the DC supply, and see if the ringing goes away.

 

[jayce3390]

You might also want to see if the disruption is due to some nearby circuit. Often strange signals originate from "elsewhere". Move into a screen room ..... or just move elsewhere.

Are you saturating the amplifier? What if you increase the input amplitude? What if you decrease it?

Is your input really pure? Are you sure?

I m not saturating the amplifier, a variable attenuator takes place just behind the amplifier. I tried to reduce the amplitude but the problem remains.

@ bigboss & bif44, I am adding a decoupling capacitor on the power supply however I noticed the voltage drop came when the input signal duty cycle was small.

 

[jayce3390]

Come back. 470 uF capacitor on the dc power has no effect, the output voltage waveform is still ringing. :-(

Is it easy to amplify a pulse square waveform?

 

[tony_lth]

I remember that I had read a EMC book, that describe how to eliminate the ringing, but the book is not at hand, pls wait several hours...

 

[tony_lth]

Maybe the output is not standard 50 ohms, you can test it with VNA and match the output.

Maybe you can read book: <signal integrity: simplified> by Eric Bogatin. The book recommend many ways to diminish ringing.

 

[kayaoo]

The picture is not clear enough to check the ringing frequency. How about adding an attenuator before the input?

 

[jayce3390]

The picture is not clear enough to check the ringing frequency. How about adding an attenuator before the input?

The ringing frequency is around 2 MHz.
I put an attenuator (50 ohms) just before the amplifier (to prevent saturation), waveforms have been printed with this attenuator.

 

[G0HZU]

What equipment are you using to produce this readout? Is it a high sample rate digital scope?

Could the 2MHz signal be an alias term in the scope?

Maybe you should look on a spectrum analyser to see if the amp is producing a very high frequency term signal that is producing an alias that looks like 2MHz on the scope.

Otherwise, if it is 2MHz then it looks suspiciously large to me.

Is it a commercial amplifier or one that is being developed?

It may be that the bias network is not decoupled adequately and this allows a form of relaxation oscillation although this usually happens at much lower frequencies. You could try placing a 1uF capacitor on the cold part of the amplifier bias line.

There does look to be some very high frequency ringing and this is a clue that the amp is going unstable somewhere.

What happens if you gradually reduce the drive level? eg reduce the pulse amplitude going in.
Is there a point where is suddenly cleans up and the ringing stops?

Also, why is the 50MHz input pulse train showing an inconsistent pulse width?

How are you generating the pulse? Are you deliberately changing the pulse width as per the yellow waveform?

Can the pulse driver actually drive a 50 Ohm attenuator/amplifier? Could it be generating the ringing problem as some form of interaction with the amplifier?

 

[FvM]

Having an amplifier with 5 - 500 MHz bandwidth doesn't necessarily mean, that it's able to reproduce pulse waveforms. Apparently, the amplifier is near to it's stability margin at high frequencies. If it's actually terminated correctly, I suspect a bad designed amplifier. In any case, a frequency domain gain characteristic would be interesting.

You say, the yellow trace shows the signal before the PA. Does this mean, the original input signal has already a varying pulse width? If so, the low frequency part of the ouput waveform won't be surprizing.

 

[jayce3390]

What equipment are you using to produce this readout? Is it a high sample rate digital scope?

Could the 2MHz signal be an alias term in the scope?

Maybe you should look on a spectrum analyser to see if the amp is producing a very high frequency term signal that is producing an alias that looks like 2MHz on the scope.

Otherwise, if it is 2MHz then it looks suspiciously large to me.

Is it a commercial amplifier or one that is being developed?

It may be that the bias network is not decoupled adequately and this allows a form of relaxation oscillation although this usually happens at much lower frequencies. You could try placing a 1uF capacitor on the cold part of the amplifier bias line.

There does look to be some very high frequency ringing and this is a clue that the amp is going unstable somewhere.

What happens if you gradually reduce the drive level? eg reduce the pulse amplitude going in.
Is there a point where is suddenly cleans up and the ringing stops?

Also, why is the 50MHz input pulse train showing an inconsistent pulse width?

How are you generating the pulse? Are you deliberately changing the pulse width as per the yellow waveform?

Can the pulse driver actually drive a 50 Ohm attenuator/amplifier? Could it be generating the ringing problem as some form of interaction with the amplifier?

I m using a wide bandwidth scope (6 GHz) with appropriated probes. The amplifier is a commercial product "Mini Circuit 5-500MHz ZHL-1 2W amplifier".

The input pulse signal (yellow trace) is basically a PWM signal, that's why pulse width is moving, I m generating this signal with an AWG, output is 50 ohms.

---------- Post added at 07:47 ---------- Previous post was at 07:45 ----------

Having an amplifier with 5 - 500 MHz bandwidth doesn't necessarily mean, that it's able to reproduce pulse waveforms. Apparently, the amplifier is near to it's stability margin at high frequencies. If it's actually terminated correctly, I suspect a bad designed amplifier. In any case, a frequency domain gain characteristic would be interesting.

You say, the yellow trace shows the signal before the PA. Does this mean, the original input signal has already a varying pulse width? If so, the low frequency part of the ouput waveform won't be surprizing.

FvM, Do you mean harmonics spreading of the input pulse signal makes the amplifier unstable?

I found the 5-500 MHz amplifier datasheet, is it allowed to post the link?

 

[FvM]

I guess, the amplifier type is ?

I think, the envelope waveform at the output can be explained by the PWM, because it drives the amplifier with a modulation of about 2 MHz, which apparently can't be reproduced by the amplifier. The high frequency ringing however isn't understandable from the given gain characteristic. But as already mentioned, I would like to see an actual frequency domain gain measurement for reference.