I am trying to change the voltage level of a 10MHz clock signal. For this I am using LM7171 as a non-inverting summing amplifier, to DC shift the input signal, with some positive gain to reach the required voltage levels. The circuit has been soldered on a general purpose PCB and is operating at +/- 5V.

Though the simulations on LTSpice showed no problem, my output on the oscilloscope is being rippled and distorted. The distortion is not very prominent at low frequency of 100KHz, but still present.

Can someone tell me why is this happening? Do I have to make changes in my circuit design or take a new different IC? Please suggest the way out.

 

Hi mva,
Welcome to AAC.
What are the source impedances of the clock source and DC offset source.?

E

 

You have a peak in the gain at 50MHz, probably due to the input capacitance.

What is the input voltage? What output voltage do you need? What power supply voltage do you have?
I think a simple Resistive attenuator would do a better job, or perhaps a comparator.

 

Hi mva,
Welcome to AAC.
What are the source impedances of the clock source and DC offset source.?

E

I am producing the ac signal from a function generator in High Impedance mode and giving the DC signal from a lab bench power supply whose source impedance I do not know.

FYI I am probing using 1x SMA to BNC cables.

 

You have a peak in the gain at 50MHz, probably due to the input capacitance.

What is the input voltage? What output voltage do you need? What power supply voltage do you have?
I think a simple Resistive attenuator would do a better job, or perhaps a comparator.

Please see the attached schematic for the input and output voltages. As mentioned, I am operating at +/- 5V.
Will the resistive network work at such high frequencies? What about loading?

 

Hi mva,
I can reproduce the waveform shown in your post by adding a low-value capacitance load on the OPA output.
What is the OPA output connected too?

E

 

Please see the attached schematic for the input and output voltages. As mentioned, I am operating at +/- 5V.

I see 0V to 2.45V for the output, but the input isn‘t too clear.

 

Hi mva,
I can reproduce the waveform shown in your post by adding a low-value capacitance load on the OPA output.
What is the OPA output connected too?

E

The output is directly connected to the DSO via SMA to BNC cable. Is this problem somehow related to soldering of the board on general purpose PCB?

 

I see 0V to 2.45V for the output, but the input isn‘t too clear.

sorry for that!

The input clock signal is a 10MHz square wave varying between -0.84 to 0.91V voltage level. To compensate for the DC shift, I'm adding the positive DC of 0.84V at the other input of non-inverting summing amp.

 

From what I have been reading here, the problem is probably that your amplifier is not set up to drive the capacitive load of the cable +the SMA input.

You might need to use a technique like the one below to isolate the feedback from the capacitive loading and possibly add a little capacitance across RF.

Edit: Gracious @atferrari , the third time is the charm (?)

 

Have you considered a 74LVC04 running on 2.5V, and using a pull-up resistor to bias the input?

 

From what I have been reading here, the problem is probably that your amplifier is not set up to drive the capacitive load of the cable +the SMA input.

You might need to use a technique like the one below to isolate the feedback from the capacitive loading and possibly add a little capacitance across RF.
View attachment 251163

I don’t know where you found that diagram, but the op-amp inputs are back-to-front!

 

Back-to-front? I am anxious to know what that means.

Thank you for calling my attention to the circuit, I see they have the inverting and non-inverting inputs swapped. Now corrected.

 

hi mva.
A simple example of what the effect of a load capacitance will have.
The OPA is the nearest type i have.
E


Update:
A simple resistive fix is often used, 2nd image

 

From what I have been reading here, the problem is probably that your amplifier is not set up to drive the capacitive load of the cable +the SMA input.

You might need to use a technique like the one below to isolate the feedback from the capacitive loading and possibly add a little capacitance across RF.
View attachment 251163

Hola Dick. I get error 404.

 

Though the simulations on LTSpice showed no problem, my output on the oscilloscope is being rippled and distorted.

Try inserting a resistor in the ground lead to dampen ringing caused by ground lead inductance.

You can do a quick test by enabling the 20MHz bandwidth limit option.

 

The output is directly connected to the DSO via SMA to BNC cable.

The high capacitance of that cable is likely the problem.
You need a cable driver matched to the BNC cable characteristic impedance (likely 50 or 75Ω) to minimize overshoot and ringing.

 

hi mva.
A simple example of what the effect of a load capacitance will have.
The OPA is the nearest type i have.
E


Update:
A simple resistive fix is often used, 2nd image

It worked like a charm. I soldered an 81Ohm resistor at the output of my OpAmp.
Thanks for going above and beyond by simulating it and showing

I am bit confused on why this is working though, is this resistor matching my circuit impedance to the SMA cable? I didn't understand where many were pointing out that my OA is unable to drive capacitive load.
And how should one select this resistance value (I just tried with all resistors less than 100Ohms)?

 

hi mva,
Pleased to hear you have it working.
As you may know, most OPA have an internal capacitor that helps set the response.
The 47R helps to isolate the cap from the loop.

There is an online video, which explains, see if I can find, will post.

E

Update:
No success with video... yet
But this links explains
https://www.analog.com/en/analog-dialogue/articles/ask-the-applications-engineer-25.html

 

is this resistor matching my circuit impedance to the SMA cable?

Yes, the 81Ω resistor is likely near the characteristic impedance of the cable.
If a cable is not driven by its characteristic impedance, there will be ringing and overshoot in the signal.