BNC output noisy, what to do?

Thread Starter

Adrienboub

Joined Jun 15, 2020
22
Hello everyone (forgive my very average english),

I did a PCB mount of a VCO that outputs a variable frequency of around 10MHz, so far so good.

However, when I put a simple wire for the output I get a pretty clean signal like this:
260358370_2849287262028285_4972505644931550642_n.jpg

Whereas when I put a BNC output with a BNC cable I rather get this:
259564217_590516125491764_3418065020418413044_n.jpg
What to do to remedy the problem, because I would like to have a BNC output? I have heard of impedance matching but where do I start?

thank you
 

MrChips

Joined Oct 2, 2009
24,610
When you say ”BNC” do you mean a coaxial cable with BNC connectors on both ends?

What is written on the black plastic of the cable?
How long is the cable?
 

crutschow

Joined Mar 14, 2008
28,165
A coaxial cable has a characteristic impedance (typically 50 or 75Ω) which it must see as a load (termination) at the end to prevent reflections in the cable.
Without that, you get the ratty waveform you see.
But to drive the cable, the oscillator output must be able to drive the low cable impedance, which may require a line driver circuit.
 
Please understand that a "50 ohm impedance " cable does not represent a 50ohm load to a signal source.
Also, the signal is 5 volts PP and at 50 nanoseconds per division, 10 MHz is a fairly high frequency. ALSO, there is a much higher frequency present at about 0.1 volt present on both traces.
How are you making the connection to the scope input when you do not use the BNC connection? Are you using a scope probe? Many scope probes include a 10;1 ATTENUATEOR, and if the first wave form is actually a higher voltage.
But the first problem is all of that high frequency oscillation on the signal. Something in your system is oscillating. And the first waveform in post #1 is NOT very good! It has a lot of much higher frequency signal present.
 

Thread Starter

Adrienboub

Joined Jun 15, 2020
22
Please understand that a "50 ohm impedance " cable does not represent a 50ohm load to a signal source.
Also, the signal is 5 volts PP and at 50 nanoseconds per division, 10 MHz is a fairly high frequency. ALSO, there is a much higher frequency present at about 0.1 volt present on both traces.
How are you making the connection to the scope input when you do not use the BNC connection? Are you using a scope probe? Many scope probes include a 10;1 ATTENUATEOR, and if the first wave form is actually a higher voltage.
But the first problem is all of that high frequency oscillation on the signal. Something in your system is oscillating. And the first waveform in post #1 is NOT very good! It has a lot of much higher frequency signal present.
" Please understand that a "50 ohm impedance " cable does not represent a 50ohm load to a signal source. "
Ok but crutschow seems to said the opposit.

" ALSO, there is a much higher frequency present at about 0.1 volt present on both traces. "
How can you see it ?

In the first photo: i use a scope probe, the RP2200.
There is a Compensation range with a capacitance trimmer 10 à 25 pF on this probe. I don't turn it because the 1st signal seemed correct to me before i read your message ^^

" And the first waveform in post #1 is NOT very good! It has a lot of much higher frequency signal present. "
I really don't know where this high frequency comes from... Because there is only the VCO.
 

DickCappels

Joined Aug 21, 2008
7,955
Regardless of other problems, If you run your signal through a cable you must either accept the ringing (artifacta) or terminate the receiving end of the cable to its characteristic impedance. If perfectly terminated, a mismatch at the sending end will not be apparent.
 
What I see on he signal is not ringing. It is a much higher frequency oscillation, and it is clearly visible as those little vertical lines all along the sloping part of the wave form, and as the thickness of the trace at the top. The oscillation may be above the frequency response of the scope, as I am not at all familiar with the model in the pictures. I suggest trying a much faster sweep and the oscillations will be more obvious if the rise of the waveform is spread over three or four horizontal divisions.
 

MrChips

Joined Oct 2, 2009
24,610
I see reflections in the second trace.

@Adrienboub still has not said what "BNC" means.

"Oscilloscope probe" is not enough information. Was it set to x1 or x10 attenuation?
You need to learn about transmission line impedance, reflections and termination.
You also need to learn how to adjust frequency compensation on your oscilloscope probe.

Ask if there is something you don't know.
 

Thread Starter

Adrienboub

Joined Jun 15, 2020
22
"What I see on he signal is not ringing. It is a much higher frequency oscillation, and it is clearly visible as those little vertical lines all along the sloping part of the wave form, and as the thickness of the trace at the top."
What I see on he signal is not ringing. It is a much higher frequency oscillation, and it is clearly visible as those little vertical lines all along the sloping part of the wave form, and as the thickness of the trace at the top. The oscillation may be above the frequency response of the scope, as I am not at all familiar with the model in the pictures. I suggest trying a much faster sweep and the oscillations will be more obvious if the rise of the waveform is spread over three or four horizontal divisions.
"What I see on the signal is not ringing. It is a much higher frequency oscillation, and it is clearly visible as those little vertical lines all along the sloping part of the wave form, and as the thickness of the trace at the top."
Of course my signal (around 8.6MHz) is made up of a sum of faster signals in order to achieve the almost rectangular 8.6MHZ signal.
I take photo of what you see(the sweep is maximum). The signal displayed is each time the signal from the RP2200 probe in X1 :
261093878_905863936729125_2942854252123109586_n.jpg

The signal above in a Spectrum analyser:260829243_631089814996377_7993165922743820903_n.jpg

So we can see the high frequency on the right side of the panel. I put a Low Pass filter and look at that:260745333_261766909162216_5631872699221107115_n.jpg

And i Finally obtain this sh*t :259964490_985722875712714_9926315957201263_n.jpg

I use the probe but my problem is noticeable when I use my BNC with my coax cable

And I see a reflection problem in the second trace too MrCHips. I present the 2 photos for comparing.

So how can i delete this reflection problem when I use my BNC and coax cable ? What to start with ?
 

Thread Starter

Adrienboub

Joined Jun 15, 2020
22
As previously stated, the coax needs to be terminated in its characteristic impedance (try 50 or 75Ω).
I don't understand. The coax have characteristic impedance of 50ohm, ok for that.
But when you say it needs to be terminated in its characteristic impedance, what do you mean ? Because the oscilloscope impedance of entry is 1 Mohm. How can i deal with ?
 

MrChips

Joined Oct 2, 2009
24,610
You are still not answering the questions.
We are not there to look over your shoulder. We cannot see your setup and what equipment you have.

Let's start with one question at a time.
What is the meaning of "BNC"?
 

Thread Starter

Adrienboub

Joined Jun 15, 2020
22
You are still not answering the questions.
We are not there to look over your shoulder. We cannot see your setup and what equipment you have.

Let's start with one question at a time.
What is the meaning of "BNC"?
This is on my pcb:

https://fr.rs-online.com/web/p/conn...6E6669673D3126&searchHistory={"enabled":true}

With coax cable of different longer. 20 cm, 50 cm, 1m as you want that goes to my oscilloscope or spectrum analyser.
The change of the longer of the cable, change the shape of the signal.
I want to delete this reflection
 
What kind of coaxial cable are you using? (Not the connector - the cable itself) Is there a type number on it?

If you can't tell, you can try connecting the wiper and one end of a 200 ohm ceramic trimpot and adjusting it for the best response. There! You have terminated it with the correct resistance. If you measure the resistance that will tell you the cable's characteristic impedance. (Note: The coax might be too short for this to work, but from the photo that does not appear to be the case.)
 

MrChips

Joined Oct 2, 2009
24,610
With coax cable of different longer. 20 cm, 50 cm, 1m as you want that goes to my oscilloscope or spectrum analyser.
The change of the longer of the cable, change the shape of the signal.
I want to delete this reflection
We understand your problem. We just need to get the correct information.

Ok. BNC is a type of connector.

Now you need to describe your coax cable.
There are identification marks on each cable.
Tell us what is printed on the cable, something like RG-58U, RG-59/U, RG-62U, RG-174/U, etc.
Every coax cable is different. Check all your cables to be certain.
 

MrChips

Joined Oct 2, 2009
24,610
While we are waiting for this information, learn how to set up your oscilloscope probe. This is a separate issue from your current problem.

Your oscilloscope input impedance is 1MΩ.
Your oscilloscope probe has two settings, x1 and x10 attenuation.
Do not use the x1 setting. Use the x10 setting. This will bring the input impedance to 10MΩ.

But before you use the probe you need to adjust it correctly. Learn how to do this by following instructions in your User Manual or from the internet. You only need to do this once when using the same oscilloscope. If you need help just ask.

btw, this is all very important stuff you are about to learn which they don't teach you in school. You are going to learn this by coming to the right place.
 
Top