Hi, I have the following question. I’m reworking a project for an antenna tuner. This is a modification of an existing project. Currently, I have redrawn the schematic from a PDF into KiCad and imported all PCB layers from the Gerber files, except for F.Cu and B.Cu, which I recreated manually.
Now, I need to do the following in KiCad:
I’d like to modify the schematic and PCB so that a large part remains as is, but the connections leading to the microcontroller need to be routed to a connector or pin header (about 20 connections). I plan to create a test PCB for this part, as it’s an HF circuit. These 20 connections would then go to a breadboard where I can test different solutions, program, etc.
Is there a way in KiCad to have everything in a single schematic, but divide the project into one part for the real PCB and another part for the breadboard, and avoid errors in **
Interesting circuit. So sampling forward and reverse power and switching elements in/out for best match or something like that?
I’d be tempted to add freewheel diodes across the relay coils, gate-source resistors on the fets, and filter the control lines – even a ferrite-bead per line located at the connector. For 20 lines, I might opt for a 2x20 100mil header and 40-conductor ribbon cable (easy to make to any size). Put signals on odd pins and ground all the even pins, and the ribbon has ground wire interleaved every other conductor to keep things cleaner. FFC is a smaller-footprint way to do the same thing with some fine-pitch soldering needed.
Yes, it’s an automatic antenna tuner known as the ATU-100. It works excellently, as verified by thousands of users, but I’m trying to make some modifications to enable remote control via Bluetooth or WiFi. That’s why I reverse-engineered the schematic from a PDF and the PCB from Gerber files without copper layers, which I had to recreate manually since they couldn’t be imported. My goal was to create a complete original project, which I finished on Friday.
Yesterday, I worked on separating the design into two parts: the tuning section, which I plan to keep as it is, and the control section, which I’ll modify. Whether I can do it or not, I don’t know yet. By the way, in the picture, you can see the original tuner in the foreground and my test setup behind it, where I partially simulated the FWD and RVS signals and experimented with programming the relay control.
I’ve reached a point where DC testing is no longer sufficient, so now I’m working on creating a test PCB specifically for the HF section. For example, in the new design, I’ve already added gate resistors for the MOSFETs. What you see here is still the original circuit. Once I resolve all the errors, I’ll start transforming it into my own design. However, since I’ve never worked with KiCad before, I consider this a learning process before finalizing the design.
Thank you all for your help and support. I managed to successfully divide the schematic using the hierarchical sheet. I do still have a few errors related to the power supply, but I’ve addressed them in a separate topic:
Wow, that is a big breadboard! Glad you got hierarchical sheets working – that is critical to any schematic that has more than a dozen parts. You also have buses working which are invaluable. Remember that power/ground symbols are global across sheets, and avoid global labels for signals (local labels within a sheet is fine). I would still add freewheel diodes on the relay coils, else you will get di/dt spikes on the fet drain at turn-off (which can be a hundred volts or more). Most any diode will do:
However, some relays integrate them internally, so look at the datasheet. If they are internal, the coil now has a polarity, as the cathode needs to go to positive supply.
ps: I never bother with kicad erc – more hassle to keep it happy than it is worth imho.
Have fun! Gil (not active, but still af7ez)
I do multiple boards together all the time, sometimes with a main board I need in quantity plus little proto boards tagging along, sometimes as two boards of a product.
I generally have the fab house score a V-groove to snap the main board from the other(s). Can also cut it yourself (even with a hacksaw) if you leave a little space for the saw cut.
This one just had some protos added to the main board:
This next one was a production board fabbed and smt-assembled at jlc, where both boards were part of the same product. V-score situated on edges that can be rough after snapping (ie: not an edge that needs to accurately snuggle up to the face of a housing or whatnot).
These are all done in a single schematic – one or more of the schematic pages belong to the “extra” board(s), and I use different power and ground symbols on the different boards (you can use the same power connections and just ignore the not-routed ratsnest lines).
Sometimes the fab house adds a fee for “extra boards.” This is fine when it is a modest fee for the v-score or routing they are going to do, but some places want to charge for extra boards even when I ask for no cuts, which really pisses me off.
I had a small-quantity project where I built an array of six little boards together, and was going to just cut them apart on my bandsaw and tidy the edges on the belt sander. The quote came back with an extra 50 bucks or so for each of the six little boards. They were doing absolutely nothing to earn that. So I changed the board such that the bottom ground plane covered all of the boards – it looked visually like one board to those weasels, and then I got a lower quote. So get creative.
Looks good. In my case I cando it without hierarchy now. I draw the schematic for the tuning part now with connection to a connector. Now I work on the PCB just for this part. When I am done I will send it tho the production and in the meantime I will add MCU schematic part and when all is done and tested I will add to the existing PCB the MCU part.
Me too. But it looks that relays switch high frequency signals. May be capacitors are intentionally used to avoid fast voltage change at relay coil because of some reason we don’t know…
The tuner has been extensively tested in real-world conditions—just try googling ATU-100 N7DDC. It has been proven reliable through years of operation. I personally own three units, one of which is mounted on a balcony, exposed to temperature fluctuations from -10°C to +60°C and high humidity, yet it functions flawlessly.
The reason I want to modify it is that I believe it can provide more data and support remote control via Bluetooth or WiFi. The current version lacks proper remote display functionality.
I’ve already experimented with basic remote control using Arduino and ESP32 via Bluetooth, which worked well, but the issue was with displaying data remotely. That’s why I decided to redesign the control section to improve remote management. You can see some of my earlier attempts on my page: www.ok1tk.com.
If I understand your issue correctly, you want to keep part of the circuit only as visuals to not get confused during routing or for better visualization in the schematic? This is something I often see beginners want; they want part of the schematic not to be part of the PCB.
I personally don’t like this strategy, it has the opposite effect on me. Because now the schematic does not represent the finished product, and it’s confusing at least for an outsider. And as I understand it, this is only an advantage to the designer, and only during a specific phase of the design.
This is why every large electronic project should have architectural drawings. You can do them directly in KiCad (without creating a layout from it) or using visio, draw.io, google drawings etc. If it’s multiple PCB design, there’s an architectural drawing showing how the PCBs are connected (connector to connector). If it’s a large PCB with multiple sheets, an architectural drawing showing the building blocks (usually sheets) and the interfaces between them, per PCB. If I were to do the same project as you’re doing here, I would clearly label the connector with pin names same as the arduino headers. If you still need more context about external connections, add graphics.
Yes I know this doesn’t answer your specific question because you have already decided on a solution. Just like you don’t seem to want input on the electronic design itself, because you lean on how it’s been used reliably for so long. To this, all I have to say is that you can simply ignore the input from the talented engineers advising you, or you can choose to at least evaluate the proposals, and maybe you find you can improve the product or your process.
Honestly I’m not sure what these words mean.
Do you think it is pro EMF or against EMF?
And EMF means EM Field or EM Filtering?
I have Ethernet driver at PCB. According to my idea (not fully checked) the simplest way (can be done with 0603 ferrite beads and not big ferrite clamps) to limit emissions is by adding impedance into all (semi DC) wires going out of PCB (supply, inputs, outputs, RS485 communication). As wires going out of relay can curry 1A (and sometimes more) I decided to cut the high frequency GND loop between my PCB GND and relay coil.
When I was with my device in some company their internal EMC lab they said: We envy you that you can use such good connectors with integrated transformers, because our bosses tell us to use the cheapest ones (as they are our competition I didn’t talk about my solution).
This doesn’t prove anything (may be really the connectors were better), but I think it confirms my idea to some extent.
