Simple series LC BPF, the results are promising!

Hi I use rfsim99 and I have found very good and symmetrical response when simulating a single series LC from input to output. By adjusting the values, I can also get very sharp responses at various frequencies.
I could just not get these sharp responses using a parallel shunt LC.

I know the question is a bit general but, are these characteristics generally realistic for a series LC or am I doing something wrong?

Please consider how the quality factor of a LC circuit is determined, and how it's related to source and load impedance in different circuits. That's very basic AC circuit theory.

If you show the respective test circuits with component values, we can discuss it less generally.

FvM said:

Please consider how the quality factor of a LC circuit is determined, and how it's related to source and load impedance in different circuits. That's very basic AC circuit theory.

If you show the respective test circuits with component values, we can discuss it less generally.

Click to expand...

For example see this image. graph is 1-30MHz and filter peak is at 13.7MHz.
In the second example more details are shown (another frequency peak).
All scales are linear.

It seems that these can be mage very selective if large LC ratios are present.
Setting the Q-factor setting for the L in that program from default (100@100MHz) to 30@30MHz only affects the peak response (loss) not the bandwidth.

Your graphs are linear in voltage, switch to dB scale for something more meaningful (Half way down those graphs is only -6dB).

Also, what of the shunt capacitance of the inductor, a 1mH inductor likely has more then 10pF of inter-winding capacitance.

Sims with unlikely component values tend to look interesting until you model ALL the parasitics, at which point reality disappoints.

Build it, and stick it on a VNA to get a feel for what really happens.

Regards, Dan.

As young students, all of us have been excited about filter design with ideal elements. Simple designs can do amazing things - but unfortunately it all breaks with real world component data. That is also true for your resonator, where Q factor does limit the bandwith.

Dan Mills said:

Your graphs are linear in voltage, switch to dB scale for something more meaningful (Half way down those graphs is only -6dB).

Also, what of the shunt capacitance of the inductor, a 1mH inductor likely has more then 10pF of inter-winding capacitance.

Sims with unlikely component values tend to look interesting until you model ALL the parasitics, at which point reality disappoints.

Build it, and stick it on a VNA to get a feel for what really happens.

Regards, Dan.

Click to expand...

I have not thought of inter-winding capacitance. Thank you, you are right!

- - - Updated - - -

volker@muehlhaus said:

As young students, all of us have been excited about filter design with ideal elements. Simple designs can do amazing things - but unfortunately it all breaks with real world component data. That is also true for your resonator, where Q factor does limit the bandwith.

Click to expand...

Thanks a lot!

neazoi said:

I could just not get these sharp responses using a parallel shunt LC.

Click to expand...

When you think about it, your series LC is not too different from the LC tank. Both resonate at a single frequency, which means the loop conducts the greatest current at that frequency.

With the series arrangement you are observing this effect directly. (This assumes you do not place a lot of components in the loop.)

----------------------------------

It is common to make the L:C ratio large. Voltage swings are greater with large L, and small C.

It also operates on smaller current, which is handy if you don't wish to consume a lot of power.

It also tends to give you higher Q.

BradtheRad said:

It is common to make the L:C ratio large. Voltage swings are greater with large L, and small C.

Click to expand...

True, but there are sane limits, and 5 orders of magnitude difference is pushing it for reasonable effective Q...

You often find filter networks that start by doing the L match thing to raise circuit impedance, then do the work, then step back down at the far end of the chain.

Seriously, a handful of parts and a VNA will give you a better feel for this stuff then any simulator.

Regards, Dan.

Q factor of the LC circuit (series or parallel) varies a lot with LC ratio. And implicit will vary the 3dB Bandwidth. Always should be a reasonable compromise choosing the LC ratio.

For example at 1.5MHz, L=5mH and C=2pF will give a Q factor about 50000 and BW(3dB)=300Hz, when at the same frequency, with L=5uH and C=2nF gives a Q factor about 50 and BW(3dB)=30kHz

In real life, using inductors with very high inductance (mH's) is not a reasonable choice. Only the equivalent LC resonator of the Quartz crystals have an equivalent inductor in mH's range, getting Q about 100000.