My problem has been in getting the size of the pieces of PCB that I cut exactly the right size. I use a pair of tinsnips to cut it. (I have a paper cutter and have tried that but the paper cutter gets a lot of use and use for cutting PCBs doesnt seem good for it. My wife is an artist and she uses the paper cutter a lot so I dont want to ruin it for her.
Yeah get two shears, the one will get useless in short order...
I use tin snips most of the time, works fine. Scribe the line, measure twice cut once and all that. Watch and guide it carefully during the cut, it's easy to wander.
When it needs to fit tightly (like the above picture: it's an enclosure held with only two screws, with two clamshells that slide onto a chassis in the middle), cut slightly oversized, and use SiC sandpaper on a flat plate to grind it straight and to size. Tedious, but it works.
Another example, this with slots on both end pieces so the whole thing is held with just one screw:
https://www.seventransistorlabs.com/Images/Reverb2.jpgThe pushbuttons and MAX232 (tinned board, bottom right) were the hardest to position I think, followed closely by the four slots...
Sawed cuts are fine as well. Saws leave smoother cuts than shears, and if you cut straight, not much material needs sanding to get a flat, clean end.
Power tools are excellent of course; normal sawblades can be used, but better life is had with fine toothed carbide blades or even abrasive wheels. The dust and smell are obvious downsides. If the tool is compatible, water can be used to keep the dust down (consider a tile saw?).
Files work too of course, but don't use any sharp files that you use for metals, especially slippery metals like brass. A few touches to FR4 and it'll be too dull to cut brass again...
If I could cut PCB material to precise squares then I could use it much more properly. My main issue is that with RF construction, its the inside surface that needs to make contact with the sides of the PCB a fair amount of the time. For example, a device where I want the two sides of the PCB to also be RF shielded from one another, say ons side is RF, the other power and control.
And you want to solder both seams if possible, to get extra shielding? And it may not be feasible to solder either, in part or in full, for various reasons?
Edges can be lapped over with foil. If you don't have copper foil handy (which you do; copper tape works for this, obviously conductive adhesive being a huge bonus, if not a necessity), and don't feel like ordering any, you can peel it off PCBs with a bit of heat, and sand the (usually oxidized) face clean so it accepts solder.
You can always make multiple walls, just soldering one side of each. If they're like nested boxes, this makes things far easier to maintain, too!
Putting circuitry on both sides of a proto is already a bit of a construction nightmare. I prefer to build on single sides and stack things up as needed. Fly wires connect between boards, so they can be unfolded for maintenance. Or using connectors even, so they can be completely unplugged.
If I need something to fit tightly, I can try to optimize it, but I'd much rather just not, or order a proper PCB instead.
The most optimizing I think I've done is this,
https://www.seventransistorlabs.com/Images/CukHack5.jpg which has a board on the bottom with some power transistors and diodes, which connect to one side of the inductor; then on top of that, the smaller board with all the capacitors on it, and power in/out wires. The gate driver board floats in the middle. Control sits in the top enclosure half, with double stick tape. (The power board has a thermal pad to help dissipate power to the enclosure, and a bit of copper tape lining the enclosure to further help spread the heat. It's an ABS enclosure, not much dissipation at all, but it's something.)
Obviously, it's wide open from an RF standpoint. If it were a diecast enclosure, it would kick a fair amount of ass, and some additional filtering on the power leads would make it completely quiet. I wouldn't want to do a copper clad enclosure for this, partly because I already have the enclosure, and it wouldn't be as strong.
In order for the device to work well I now make a box out of stiff thin cardboard squares , apply copper tape to both sides of them, the inside and outside surfaces, using cardboard I can get exactly the right size and cutting the cardboard into squares (not bending it) means I can have nice sharp corners. Then I can use components like a feed through capacitor or an SMA connector fairly normally and get a decent electrical connection to both the inner and outer box. (it actually ends up being a double shielded box.) Cardboard coated with copper tape is also solderable to (little dabs, not more) . This works fairly well, for low power devices or passive things like filters. At least it keeps them out of harms way. But this is not suitable for anything that has even a remote possibility of needing to be waterproof or banged around, because its flimsy mechanically. Which is why I only make small boxes.
Yeah, that's not bad electrically; the substrate is just not very good mechanically. And if the adhesive doesn't make (or allow) very good electrical contact, you've just made an RF mummy full of seams, which will take so many dabs of solder to stitch up... (another good reason to keep it small).
I'd like to see a picture of this. I can see copper tape and polyamide tape being very useful in a filtering context because of the capacitance. It might act much like a feed through capacitor.
Doesn't look like much, because, well, it is what it is, plus there's stuff on top...
https://www.seventransistorlabs.com/Images/Inv1kW_Deck1.jpgThe TO-220s are mounted on copper tabs (heat spreaders), and brown electrolytics are sitting on top of some area. You can see a strip, and the blue wire, carrying power up to the inverter sections from bottom-left. The transformers (wrapped with yellow tape) are constructed in the same way, the primary in layers of copper foil tape, with the secondary (200V) wound on top. Topology is push-pull, so each pair of transistors connects to opposite sides of the primary, and the CT goes to the supply (with local ceramics SMT'd to the tape, and the electrolytics standing on top). There are up to five layers, four of which are relevant: the PCB is 2-sided copper clad, of course I'm not making use of the bottom side so we can ignore that. PCB top is GND, then (in what order I don't remember exactly) the three primary terminals, in foil tape.
Control board is to the left, on wires. Should've made them with connectors, honestly.
Side view:
https://www.seventransistorlabs.com/Images/Inv1kW_Deck2_1.jpgStill can't really see the right side transformer, it's in shadow here, but suffice it to say there are two down there. It's a duplicated circuit (ultimately to make +/-200V for a mains inverter on the top deck(s)). Can also see the control board mounted (facing down), with a DC-DC module on top of it. The module, and HV output, are all on plugs so they can be tested separately.
This whole thing is going to be enormous, about 8 x 5 x 6". It's made for serviceability. A commercial version would be a fraction of the size. I've fought myself plenty of times trying to squeeze more things into already congested layouts, I'm not doing it again.
I am thinking of getting a NanoVNA. Right now everything I do is just about seat of my pants, and I'm just fooling around. With a VNA suddenly I'll have tons more data about whats going on, so I think that will be a good investment, I'll suddenly be able to make devices which I can verify as working, which will be a big plus. Its insanely cool that suddenly I'll be able to make this huge leap ahead.
Haven't used one myself but the results others have posted here are impressive. Seems like it's comparable or perhaps even better than a classic boatanchor HP something or other, the VNAs with sticker prices
starting at $20k y'know? (Give or take range I suppose, the big iron going to some GHz usually?)
---- I have come questions for you about misc. stuff in your images directory. You made a bowtie antenna and balun. That looks very handy because of the easy rotativity. Do you use that for off the air TV or receiving ham radio horixontal polarization, also how does the balun work? I have a bunch of magnetic materials at this point. Are those sleeves #43 iron powder or ferrite - For VHF/UHF baluns I usually use #61 when I can - very small binocular cores ones work the best for receiving. What is the design frequency use for your bowtie?
Heh, it is actually quite rotatable at the moment, I've got it hanging from the ceiling on that wire and I can just reach over and twist the cable and spin it around... fun to watch different stations rise and fall as it rotates.
I don't do much radio at all so that's about the extent of it, watching signals on the spec and doing a few experiments here or there.
It's just long enough to pick up FM BCB, for which I've made a tube radio that receives quite nicely:
https://www.seventransistorlabs.com/Images/FMRadio2.jpghttps://www.seventransistorlabs.com/Images/FMRadio3.jpgSeems to be reasonably wideband, which is nice. I see much of the TV and cell UHF bands, I think some aviation (possibly ADS, I forget?) at higher frequencies. (My spec is only 1.8GHz so that's about all I care to look for, anyway.)
The balun is simply a 1:2 TLT. Twisted pair (~100 ohm Zo), two in parallel giving 50 ohm Zo at the feedpoint. They are equal lengths. One is ground to ground, signal to bal+; no ferrite bead needed but I used FBs symmetrically anyway. The other is ground to bal-, signal to GND; i.e., inverting, and a full, whatever core impedance, is required to give it useful bandwidth. I think, the beads are Laird #28 LF/wideband, and the binoc is #61 (maybe #43; it's NiZn in any case). Which is more than adequate for the >100MHz response it gets. So as a 1:2, the feedpoint is 200 ohms, which, eh, more or less right. I'm not transmitting so I don't care if the SWR is moderately ass.
I'd like to make something exactly what you have there to see if I can receive more hams on the VHF/UHF horizontal polarization /SSB/CW The RTLSDR is truly a crappy receiver for receiving really weak signals, especially on 6 meters where the birdies are really bad. 2 meters is pretty bad too. Although on the higher FM parts of 2M its semi acceptable.
Sure, easy enough to make; I'd recommend against making it from wire like I did, I was hoping it was as simple as tacking the crossings with solder, but wire this loose is rather finicky, and rather floppy too for that matter, hence the crossing wires (which probably do something for or against reception, but without a reference antenna or background to measure it against, I have no idea). About half the joints are twisted in place.
This is actually a good place to use your favorite cardboard and foil tape method -- super easy to make something big and usefully shaped. Could even use tinfoil and adhesive spray (or double stick tape) for most of the area, saving cost; then make a good crimped joint to copper foil tape, and use that to carefully shape the vertex, and solder the feedline to it. (The root is where all the high frequency response happens; if you want it flat to some GHz, you need clean geometry on the same length scales.)
Obviously, you'll need to go a lot bigger to reach 6m, or you might go with a range of dipoles (like the multi-element kind so it's effectively overlapping bands, or maybe a short log periodic) which might fold up better than a huge foiled-cardboard book will. Or a tuner.
It's almost hopeless. It may also be my QTH, its very noisy and it seems as if the noise is coming from someplace out of my control nearby. Maybe its light ballasts in the big industrial building near me or something. the worst noise seems to come and go.
Tell me about it; I've got some low-VHF conducted noise in my lab here, and I don't know from what, if it's something of mine here, or a neighbor, conducted or induced; it's conducted locally anyway, and is pretty annoying any time I'm probing something with the scope, when that something is also grounded to my PC. It's literally conducted, if I lift a ground anywhere it goes away. Argh...
Looks and sounds like some shit SMPS or charger with no filtering.
There's not usually much switching noise above 100MHz, so I don't know what your situation is about exactly, but who knows; computers without cases, SMPSs pushing higher and higher frequencies; radio (ab)users, licensed and unlicensed; even just old fashioned distribution line corona.
Anyway, I am desperate to improve this. Baluns decople the antenna from the feedline and are really essential. I like the way that is set up. So I am going to try to see how that works.
Also your baluns. I am interested in things like that. Thank you for making it available.
Any ideas on making better baluns or antennas, I am very interested in.
I'm quite fond of TLT theory, and I wind transformers in that way whenever possible. Like the DC-DC in the picture above, the small ferrite transformer and bobbin that you only see the top of,
https://www.seventransistorlabs.com/Images/Inv1kW_DCDC.jpgit's wound with (salvaged) triple-insulated wire in twisted pairs. The primary pairs are wired in parallel. It's also 24:12+12+12 but the secondary pairs are all 1:1 right, so the primary has its own pair which is wired in series, making it a combined 2:1 and 1:1:1:1 TLT. Circuit is just an ordinary flyback converter. Barely needs any snubbing at all, thanks to the very low leakage of this design!
Another flyback with a similar aim:
https://www.seventransistorlabs.com/Images/DCDC_800V.jpg12V in, 100-800V out (adjustable). Primary is foil, secondary is two layers of however many turns, wired in series, with the CT taken out to a pin. Secondary is rectified as +/-400V, and the "-400" is simply grounded so it's a unipolar output.
The transmission line structure can be considered as a very low impedance TL for the primary (the foil over itself, and over the secondary), which is probably dominated by the stray wiring really; and the secondary is two TLs with respect to the primary (which looks like ground plane), of whatever length the turns are. Because the secondary is a complementary pair, their displacement currents cancel out, giving no common mode noise from the secondary! The primary still gives full CM noise of course, which is "merely" 80Vpp.
Not that CM noise matters much on this common-ground application, but it's one less mode to have to worry about, and which will inevitably get into the output, demanding more filtering than otherwise.
The windup looks like so:
https://www.seventransistorlabs.com/Images/DCDC_800V_FoilWindup.jpgThe layers alternate P:S:P:S:P, so the S's are trapped between planes of P foil. This gives a fairly low impedance, probably 60-100 ohms, for a given turn of the secondary. Pretty low for a 400V output (it dominates as stray capacitance), but the winding length is also kept short so I don't have to worry too much about it.
I forget if I measured leakage inductance; it's not in the note below the figure, obviously. Those notes are ungapped and gapped primary magnetizing inductance.
It should be around, let's see, mean turn length is about 68mm so 15 turns is about 1m of wire, about 0.6uH per half of the secondary. OTOH, primary strays are probably around 20nH, or 4.5uH secondary referred, so, like I said, guess which one is dominant...
I think TLTs may be most interesting in switching applications, as the impedances and ratios can be all over the place -- in radio you're mainly matching small signals over modest ratios, in the range of say 20 to 1000 ohms. OTOH, SMPS don't have to worry too too much about wide bandwidths, but that is perhaps changing these days. But also you don't get many applications where it can be done at all; there are just too many ratios and values of transformers to possibly supply them COTS, so there aren't many opportunities to apply theory.
At best you can practice it with custom windups (but probably not with twisted pairs like I did above, as the labor will be costly), or planar transformers. Which are fantastic, but are nigh impossible to wind with high impedances due to the broadside-facing nature of planar stackups, and the dielectric constant of FR4. That makes them attractive for LV applications at various power levels, and mains applications at high power levels.
Tim