Manual Calculation of filter

[JimB]

Interesting, thank you.

That is more or less what I was expecting, the two pi sections make a lowpass filter.
I am surprised at the amount of ripple in the passband, it looks bad, but as this is a linear scale a quick calculation shows that it is about 3dB, which is a bit more than I would expect, maybe 100 Ohm is not a good value for the characteristic impedance of this filter.

When I first saw this circuit, I was instantly reminded of a so called "half wave filter" where a single pi section has components which have a reactance equal to the desired characteristic impedance at the cut-off frequency.

I would still love to know where this whole circuit came from.

JimB

 

[MrAl]

Interesting, thank you.

That is more or less what I was expecting, the two pi sections make a lowpass filter.
I am surprised at the amount of ripple in the passband, it looks bad, but as this is a linear scale a quick calculation shows that it is about 3dB, which is a bit more than I would expect, maybe 100 Ohm is not a good value for the characteristic impedance of this filter.

When I first saw this circuit, I was instantly reminded of a so called "half wave filter" where a single pi section has components which have a reactance equal to the desired characteristic impedance at the cut-off frequency.

I would still love to know where this whole circuit came from.

JimB

Hi Jim,

Now that you mention it that's a good point about the ripple. Why design a filter that has that much loss in the central area unless maybe that was part of the goal.
We know that a 6th order filter should have a sharp cutoff, but it seems like this is not a very well designed filter at all. However, due to the high frequency maybe the designer took into account parasitics we dont happen to have on hand that change the response greatly. As you mentioned, we cant know the details without knowing the application and that's too bad for us

 

[sallehzxc]

Hi guys, sorry was away for the week for a school project. Unfortunately the application details known to me by my mentor is that it is meant to filter out a high frequency output from a syntehsizer chip used. Thus, the ADF4350 chip used, generates 4 signals and the output(in this case showcasing one of the four) is passed through the filter to only allow the signal at the specific frequecny to pass.
Unfortunately, the mentor attached to me also had a very bad understanding of the schematic. Upon inspection of the schematic, unfortunately i am unable to disclose, R11 was found to be a 50ohm resistor.

if it makes sense, i will show you the throught process of the flow of the functional diagram.

again, i am not sure how to determine the functionality of each component within the circuit. for example the capacitors. upon first glance, i assumed that it was a generic T section bandpass. however, the presence of the capacitors at the start are something i never saw before thus i am unsure.

When I first saw this circuit, I was instantly reminded of a so called "half wave filter" where a single pi section has components which have a reactance equal to the desired characteristic impedance at the cut-off frequency.
JimB

Thanks you guys fo ryour help. I am still trying to do simulations on Matlab or something to make sense of this circuit. Thank you!!

 

[sallehzxc]

additionally, now i am curious about the function of C22, C23 and C5. what would be their purpose if the capacitors are arranged in that order?

 

[JimB]

OK, I now understand where your ideas are coming from.

I am now sure that your circuit is bizarre and you are overcomplicating something which is very simple.

I don't have time to sit at the keyboard discussing this at the moment, but I will be back later.

A quick question for you:
Is this just a simulator exercise, or do you also have some real hardware?
ie an ADF4350 development board and a spectrum analyser?

JimB

 

[JimB]

the ADF4350 chip used, generates 4 signals and the output(in this case showcasing one of the four) is passed through the filter to only allow the signal at the specific frequecny to pass.

The output of an ADF4350 is a distorted square wave.
Looking at the datasheet (I have Rev B) on page 3 are the RF Output Characteristics.
The harmonic content is stated as -19 or -20dBc for the second harmonic, and -13 or -10dBc for the third harmonic.
(dBc is decibels relative to the carrier, ie the fundamental).
Looking at the output of my ADF4351 (a close relation of the ADF4350) using a spectrum analyser, the harmonics spread out a long way.

To set up the output of the ADF4350, look at page 27 of the datasheet Output Matching.
Use a 50 Ohm resistor to create a 50 Ohm output impedance.
50 Ohm is the generally accepted "standard" for RF systems.
You will now have an output which is rich in harmonics, so add a lowpass filter.

To get a single frequency from the ADF4350, you simply put a lowpass filter on the output to remove all of the harmonics.

As I mentioned in an earlier post, a simple "pi" section LPF can be designed as follows.
Choose the 3dB cutoff frequency.
Choose the filter impedance. In this case 50 Ohms.
Calculate the inductor value using Xl = 50
Calculate the capacitors using Xc = 50
So if we made the cutoff frequency 60MHz:

Xl = 50 = 2 x 3.14 x 60 x 10^6 x L
Which gives L = 0.133 uH

Xc = 50 = 1/(2 x 3.14 x 60 x 10^6 x C)
Which gives C = 53 pF

The example I have shown here is something which I used recently with an ADF4351 which I was using for a 50MHz source which had to be free of harmonics.

If you are intending to create frequencies of 100s of megahertz or into the GHz range, the form of the filter would have to be selected accordingly.

Try building this simple filter in your MATLAB simulator and see what happens.


JimB

 

[sallehzxc]

Wow, thanks sir for the insight. Sadly i don't have any of the hardware and my task has been purely simulation. Thus, I feel like it has been hard for me to visualize things without seeing it happen but i guess that's no excuse for my weak theoretical understanding....

The purpose was from my supervisor to create a filter that only allows the specified frequency to run and also prevent harmonics from intererfering with the signal, thus the use of a bandpass- sorts of filter. My supervisor, like myself, is also unclear and hopes with the simulations, things will make more sense haha.

Thank you so much sir for your wisdom. Additionally, i was wondering from your experience, given a filter-like schematic, what would be the first text to be done? I did LTspice simulations and saw the peaks resonated fairly close to the carrier frequencies to be used. However, can i do a simulation in LTSPICE in such a way to mimic the outputs being interfered by harmonics from the four outputs without a presence of a filter?

Sorry for my bad phrasing...

Thank you as always sir.

 

[JimB]

Time for a little bit of "Show and Tell"...

My ADF4351 test setup.
It has many things which are not very good in terms of RF performance, but it does what I wanted it to do.

Close up on the ADF4351 Evaluation Board.
The coax cable coming in on the left hand side is from the 32MHz oscillator mounted on the PIC board.
For what I wanted to do, using as 32MHz clock rather than the 25MHz clock built in to the eval board, made programming the frequencies that I wanted much easier.

The Low Pass Filter.
Some of the coils are a bit battered due to rough handling.
This is basically the same as the three pi sections shown in post #26 above.

Using an oscilloscope to look at the 50MHz output of the ADF4351 we see this:

A rather distorted square wave.

On the spectrum analyser, it looks like this:

If we now pass the output of the ADF4351 through the LPF
On the oscilloscope:

On the spectrum analyser

A much cleaner signal with the harmonics suppressed better than 60dB.

Using the spectrum analyser and tracking generator we can plot the response of the simple LFP

A reasonably flat response up to 60 or 70MHz, a steep fall off into the stop band with 60dB attenuation up to about 250-300MHz when the response starts to rise up to about -45 dB at 1000MHz.
Probably to be expected from a simple filter with no screening between stages.

Sometimes we need a better filter.
Like this one:

Looking inside:

Uses larger components than the simple filter, but this one was intended for use with a 50Watt transmitter.

On the spectrum analyser/tracking generator, its response looks like this:

A nice flat stop band with better than 60dB attenuation up to 1000MHz.


Does any of this help?

JimB

 

[atferrari]

Using the spectrum analyser and tracking generator we can plot the response of the simple LFP.

JimB

Hola Jim
If it is not deemed derailing the thread, could you outline briefly how you did it.
Thanks.

 

[JimB]

Hola Jim
If it is not deemed derailing the thread, could you outline briefly how you did it.

Hola ATF
Si, de nada.

The spectrum analyser (SA), the tracking generator (TG) sits on top of it.

The cables on the right hand side connect various oscillator signals from the SA to the TG so that the TG outputs a signal on the frequency to which the SA is tuned.

Connect a short cable from TG output to SA input, and adjust the various controls to give a reference level trace on the SA screen.
My preference is to use the -10dB line (second from the top) rather than the very top line of the screen.

Now replace the short cable with the filter to be tested.

Looking good, but we can improve the display and get rid of all that "grass".

Switch in the video filter to remove the low level noise, and set the display to storage mode to give a nice stable image for the photograph.

And that, briefly, is how I did it.

JimB

 

[atferrari]

And that, briefly, is how I did it.

JimB

Well explained as usual with you, Jim.

Had to look what a tracking generator is, BTW.

Gracias.