PCB based Filters

Рет қаралды 34,974

Күн бұрын

Пікірлер: 29

@donepearce Ай бұрын

I've designed PCB hairpin filters at 12GHz that only have 1dB of loss. Your problem here was entirely an over-ambitious bandwidth choice. You really need to be over 10% bandwidth for these filters to work. And filter design software never gets the centre frequency right. You need a proper electromagnetic analysis programme like HFSS to understand what a filter will do before you build it. But the basics here are dead right, and thank you for covering it.

@Klaus-b1u 17 күн бұрын

Thx for sharing!

@debrainwasher Ай бұрын

This video remembers me to things, we have done in university: The Kuroda-transform. In simple words: How the create matched impedance traces on PCB, transmission lines and filters from RLC-networks on a PCB. Excellent video that covers most common filters.

@ItsTristan1st Ай бұрын

Adding the can does indeed increase the capacitance but to a tiny degree. However, it also reduced the inductance which is where the higher frequency response comes from. Band pass filters are almost always a two way game between capacitance and inductance. Generally increasing the gap between the signal and ground will reduce capacitance but increase inductance and the opposite will also hold true.

@FesZElectronics Ай бұрын

I guess this is also why the overall insertion loss increased - the filter was designed to be 50ohm without the shield; with the added shield the trace impedance also changed, going to a lower value

@m1geo Ай бұрын

Nice video, Fesz. I was always taught never to go through the core on anything needing tight tolerances. I'd be very interested to see you redo this using a 4-layer board, with SIGNAL on Layer 1 and GND on Layer 2. I think that'd help, too. Especially with repeatability.

@FesZElectronics Ай бұрын

I'll have to try that at some point. With this design I went with the idea that a large distance trough the dielectric should help with lower losses since the traces are wider. However, as you mentioned the tolerances should be higher, so its a compromise...

@m1geo Ай бұрын

@@FesZElectronics Oh, I totally get it. It's just my understanding that the core material is the lowest tolerance and the core thickness the least well controlled.

@Mike-H_UK Ай бұрын

Excellent video - many thanks.

@stevenbliss989 28 күн бұрын

RF magic that blows both my and any cave man's mind (that even know what a radio is)! :)

@MuhammadAbdullah-zb8ci 23 күн бұрын

❤

@Soupie62 Ай бұрын

There's microstrip, and then there's stripline. You started with microstrip, then (when adding a shield case) made a form of stripline. For "true" stripline, select 4-layer board from JLPCB, put the traces on layer 2, and make layers 1 and ground planes. Tighter shielding makes a better design, but you can't trim track lengths. Alternatively, stay with a 2 layer design - but replace your shield with a blank (1 sided) PCB, and screw / solder that over the top.

@renelefebvre53 Ай бұрын

Filters at 1GHz are like magic for me ! Thank for this vidéo .

@pavelgromovikov5846 Ай бұрын

Congrats, this video was featured on Hackaday!

@michaelalex5235 Ай бұрын

This is really cool stuff - thanks very much.

@trophosphere Ай бұрын

I would see how much insertion loss is being contributed by the connectors and whatever microstrip that is not part of the filter by making essentially a test coupon. That way, you can get a clearer idea of where the majority of insertion loss is coming from.

@erikbertram6019 Ай бұрын

It seems curious that the fr4 is so much more effected by the shield compared to the teflon.

@rogeronslow1498 Ай бұрын

The teflon board is probably much thinner than the FR4 board resulting in a lower E-field flux in the air above the board. You can see this by the width of the tracks being narrower on the teflon board. Presumably they are the same impedance.

@akosbuzogany2752 Ай бұрын

Did you cut back the pins of the SMA connector? Did it make a difference when shielded?

@FesZElectronics Ай бұрын

I did not cut the SMA connectors - they where left as is, just soldered in; I guess they present a bit of a discontinuity, but it shouldn't be too bad

@pravardhanus Ай бұрын

Nice video. Good concept as we can reduce the price of a product without using a SAW filter or any other components. I think better calibration can be done using laser trimming.

@FesZElectronics Ай бұрын

Once you include laser trimming and a low loss pcb, i'm not sure if you can get cheaper than a SAW filter; however, with a PCB based filter you can get any center frequency or response type, things that are usually unavailable as standard components.

@johnaweiss Ай бұрын

Seems fantastic. Why don't more systems use PCB filters?

@FesZElectronics Ай бұрын

I would say that the size taken up on the board, the special proprieties of the board needed to minimize loss and finally the number of development cycles needed to get the design right - with a commercial SAW filter or similar, it will work the first time, with this sort of PCB design, it will take a while to get it right; however if the design is produced in large enough quantities it should be cheaper.

@johnaweiss Ай бұрын

@@FesZElectronics Would this make more sense for an RF signal between transceiver and antenna? Or better for filtering after a signal is buffered or amplified? How's efficiency and other performance factors compared to a SAW or RLC filter?

@andrewmcfarland57 Ай бұрын

Excellent video!

@ranaharsh365 Ай бұрын

Hi, Please make a video on SMPS feedback loop compensation design in LTSpice simulation and in real life also based on actual plant response and tuning parameters, specifically for FLYBACK DCM & CCM mode with a voltage mode and current mode PWM controller Ic. & how to design transformer for a wide range of output voltages. & also explain what to look for in bode plot for the best stability criteria.