Building an RF Switching Unit

RE: Building an RF Switching Unit

neuromodulator — Sat, 11 Sep 2021 19:47:48 GMT

Great project Shabaz,

I was wondering why you chose to use GCPW instead of microstrip? And why enig instead of soldermask over the TL?

RE: Building an RF Switching Unit

DAB — Sat, 11 Sep 2021 17:47:07 GMT

Very good project.

DAB

RE: Building an RF Switching Unit

shabaz — Sat, 11 Sep 2021 16:51:30 GMT

I've attached rev 1.1 PCB files to the blog post, they correct the rev 1.0 errors, and also I took the opportunity to make the switch land pattern suit more popular normal tact switches, and also J6 power entry is now more convenient, the right-angle SIL connector will no longer be on top of the LDO (which would make it hard to desolder the LDO if there were ever a problem, without desoldering the right-angle SIL connector. Now that is corrected. [mention:333884685f3541cfa8a54caa82a8c2f5:e9ed411860ed4f2ba0265705b8793d05] hopefully you've received the EAGLE zip files in your e-mail.

This is what Rev 1.1 looks like (changes marked with arrows):

[View:/resized-image/__size/1328x926/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/5658.contentimage_5F00_211986.png:1328:926]

Also, I've attached rev 1.1 KiCAD files to the blog post, the EAGLE import was successful again, except the silkscreen text was moved, so I manually moved the text to where it should be. It looks like this in KiCAD, one difference I've seen is that the edges of the board have solder mask, whereas the EAGLE version didn't. It's not major, it would be important if it were going to be fitted into an enclosure. Maybe anyone who wishes to use the KiCAD version can remove the solder mask on the edges if they know how.

Since I've not used KiCAD before, error checks should be run (guessing KiCAD has error-checking built-in like EAGLE). But I'm optimistic, because it seems near-identical in all the board aspects where it matters.

[View:/resized-image/__size/1331x921/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/0044.contentimage_5F00_211987.png:1331:921]

RE: Building an RF Switching Unit

shabaz — Wed, 08 Sep 2021 16:10:11 GMT

Updated with the final board photos, and measurement of insertion loss. I think it's working reasonably well, at least to start making use of it! I have already used it a bit, and need to document that.

RE: Building an RF Switching Unit

shabaz — Wed, 25 Aug 2021 21:39:52 GMT

Here's the modification to get the power on/off function working with the current circuit:

[View:/resized-image/__size/827x461/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/0268.contentimage_5F00_208607.png:827:461]

Such an approach can be implemented on the current PCB (rev 1.0 board) by performing the following steps (as shown in the photo below):

(a) Rotate C16 to isolate one end of it, and solder that to +3V3 at ferrite FB1

(b) Lif pins 14 and 16 of U7, and wire the pins to +3V3 at ferrite FB1 too

(c) Lift pin 13 of U7, and solder a 470k resistor to the lifted pin, and the other end of the resistor is attached to U8 output pin.

This is still ugly, but at least allows for this revision of the board to now fully function.

[View:/resized-image/__size/1074x628/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/7115.contentimage_5F00_208608.jpg:1074:628]

RE: Building an RF Switching Unit

shabaz — Mon, 23 Aug 2021 01:40:20 GMT

I finally assembled this, photos to follow, and good news, it (mostly*) works!

it was stressful soldering the first RF connector, since if I messed them up then the board was scrap, since I would probably damage it during desoldering.

However, it went on fine, and I learned some tricks along the way how to solder them on, which I'll write up.

I've run a quick test, using the connections connected to open-short-load (OSL) it seems to be OK, and that will simplify life no end. The OSL turns the unit into a (semi-automatic) calibration tool.

* the one thing that definitely doesn't currently work, and not sure why, is the power on/off feature. (If anyone can spot any flaw in that portion of the circuit, it would be most appreciated). There is a single button to cycle between the four electronic switching positions, but a long press was supposed to toggle power on or power off the device. However, it seems to be permanently powered on and I can't switch it off. I've not probed it to see if the regulator enable line is toggling as it should.

It is minor I guess, I could wire in a slide switch, but I do want to make it work.

RE: Building an RF Switching Unit

shabaz — Tue, 11 May 2021 02:22:16 GMT

Getting close! It was the right idea to test the boards before soldering much else. So far, all I've soldered are the RF switches, and the 100nF supply decoupling capacitors for each of them.

Then I attached wires to the four control pins (labelled CTLA1, CTLA2, CTLB1, CTLB2 in the circuit diagram) and the 3.3V supply and 0V, as shown here.

[View:/resized-image/__size/799x712/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/6014.contentimage_5F00_208604.jpg:799:712]

Then I patched those wires into a breadboard, supplied the 3.3V power (current limited, the switches only draw tens of microamps), and then just wired the control lines to 0V or 3.3V to observe the switching behaviour. To observe it, I just used a multimeter set to ohms. From the J5 RF connector trace out to any of the other RF connectors traces, the resistance was about 7 ohm for the switched-in path (i.e. about 3.5 ohm resistance per switch IC), and about 45 kohms for the switched-off paths, so it's functioning at some level. There was slight variation between the ICs (e.g. say 4 ohm instead of 3.5 ohm), Incidentally I can imagine if there was lots of $$$ laying around then a whole reel of ICs could be purchased and a test jig could be made to sort them, so that they match per board, before soldering them on.

Anyway, that was for the good board, but the second board was definitely faulty, the resistances were low. To fix it, I reheated the board, lifted off one of the RF switches with tweezers, applied flux paste to the underside of the chip and re-placed it in position, and then gently tapped on top of the part to hope that all the pins of the IC were making contact. Then repeated for the next switch and so on. This whole exercise must have taken at least 15-20 minutes or so, but the chips survived the heat during that entire duration, because I'd used low-temp solder paste - the first time I've tried such solder with a stencil. I really like it, it seems great for people like me who don't do such parts regularly enough to be totally comfortable working quickly. It took me ages to nudge the parts just right, I kept over-nudging it. I didn't see the self-adjustment that is supposed to happen by capillary action, it must have been so slight it wasn't visible.

The particular solder I used was a 30cc syringe of Qualitek 670 (tin bismuth, Sn42/Bi58) which costs £25+tax and was purchased two years ago and not kept in a 'fridge. Surprised it all worked out so far.

My only slight concern is that maybe the flux on the second board wasn't totally used up under the ICs, it may affect performance (it was no-clean flux, but I didn't have good control or measurement of temperature, my pre-heater is very basic.

RE: Building an RF Switching Unit

colporteur — Mon, 10 May 2021 01:52:02 GMT

In 1997, when I left my RF career, there were no devices for switching RF. At least none that we were using. I worked in aviation engineering designing and implementing of VHF & UHF transmitter and receiver sites.

The sites didn't switch RF but rather bottled. Large suba size tune able RF filter tanks. The bottles had an input and output connector and two tuning rods that slide up and down inside the tanks (yes they looked like suba tanks). The bottles (filters) enabled multiple transmitters and receivers to use the same antennas.

Required a spectrum analyzer and RF generator to tune the suckers. We let the factory look after the tuning part. They were a *** to adjust if someone banged a tuning rod.

The power was a 100 times what you are working with 25 watts at the transmitter.

Your project intrigues me. It has been a long time since I have burned myself on RF.

RE: Building an RF Switching Unit

shabaz — Mon, 10 May 2021 01:16:59 GMT

I made an effort to start the PCBs today, by just attaching the RF switch ICs first. I attempted to make two boards, since I had six of the ICs (each board needs three).

The first board visually looks ok so far, although I still need to examine it in detail:

[View:/resized-image/__size/896x672/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/5556.contentimage_5F00_208600.jpg:896:672]

The second board doesn't look as nice, not sure why. I may reheat with a soldering iron tip.. Also considering beading some flux paste around them again and re-heating. I don't really know what I'm doing, since I don't work with QFN so much.

[View:/resized-image/__size/891x668/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/3312.contentimage_5F00_208601.jpg:891:668]

Another view of the second board:

[View:/resized-image/__size/890x667/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/6545.contentimage_5F00_208602.jpg:890:667]

The procedure was pretty identical although the heater was still semi-hot when I started the second board.

The pasting was good for both boards, I think I've perfected that part of the process, just not the rest! : ):

[View:/resized-image/__size/883x646/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/4401.contentimage_5F00_208603.jpg:883:646]

I'm going to try to test the boards before soldering on the RF connectors since they are the most expensive part by far, and I won't be able to remove them again if the board is beyond easy repair (they will get in the way of any subsequent reflow on the pre-heater I'm using - I didn't use an oven).

RE: Building an RF Switching Unit

shabaz — Thu, 29 Apr 2021 02:38:39 GMT

The boards arrived quicker than the components. All parts are in-stock/orderable (just about), but are arriving in several packages, very slowly : ( Definitely issues getting components these days.

Not looking forward to mounting the switch ICs but I will be using a stencil for that side.

[View:/resized-image/__size/863x647/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-4afa1b6f-82a2-4ee8-a9e0-d0ff4c4f855e/contentimage_5F00_208599.jpg:863:647]

RE: Building an RF Switching Unit

Jan Cumps — Thu, 15 Apr 2021 18:21:48 GMT

The JLCPCB front end is exactly the same as Seeed's.

Are they related?

RE: Building an RF Switching Unit

BigG — Thu, 15 Apr 2021 14:58:52 GMT

That's an excellent design explanation. I like the size of the IDT F2912 SPDT RF switch. Nice choice even if a little pricier than the others. Was puzzled by the logic behind having a Mode option with double throw switches.

Anyway, I await the testing once the boards are made up. Say did you try https://aisler.net/ for PCB fab and stencil. They have a 4-layer option. Delivery charge is included within EU & UK for standard shipping.