another hack

On Wed, 09 Dec 2015 20:12:22 -0500, DJ Delorie Gave us:

If the solder melts on both sides of a non-plated thru hole, then your iron is WAY too hot. Especially at lass than 13 mils for the wire size.

On Thu, 10 Dec 2015 02:21:25 GMT, Bruce S Gave us:

CRC handbook of chemistry and physics.

Silver oxide isn't as good a conductor as silver, but it's one of a minority of metals whose oxides are also metals (i.e. have partly filled conduction bands). Other examples are vanadium and tin.

Nickel oxide is an insulator, but just barely.

Cheers

Phil Hobbs

Hmm... they're 600F Metcal tips. I suppose I could use 500F for those but I don't consider 600F to be "too hot", esp since I have an 800F tip for those big jobs.

And melting on the far side is less likely if the wire stays put so you can get the job done quicker :-)

Well, it seems like it takes about 5 minutes to do, which wouldn't be awful if it needed it every other year, say. Now that the kids are (mostly) gone, we don't use the big platters very often.

Might try that too. Getting rid of the plating solution afterwards is always the parlour trick, and with silver you waste a lot due to the metal wanting to precipitate.

Cheers

Phil Hobbs

Brown Scotchbrite has finer grit, if you can find it.

Cheers

Phil Hobbs

Whew. I highly value what's left of my reputation and thought I had offended you in some way.

With PIM, the major culprit is conduction through the connector shields. At high currents, anything magnetic goes non-linear causing it to mix whatever signals are present, thus creating intermod. Since most of the RF conduction is on the surface due to skin effect, the outer plating is the prime culprit. I can see where a oxidized layer on a nickel plated connector mating sandwich might be a problem. However, PIM is only a problem at high RF currents and you're talking about microvolts. At the power levels involved, I suspect that any surface diodes formed would be instantly vaporized by the high power levels.

Also, I did some crude tinkering with my home made PIM tester to see if nickel plated connectors are a problem on the receive side of the puzzle. I built up duplexer-circulator-cavity contrivance, that had silver plated connectors on the transmit side, and the worst nickel plated 7/16 DIN cables on the receive side. I then replaced the jumpers with equivalent albaloy plated jumpers. As far as I could tell, there was no difference. However, when I inserted a nickel plated jumper in the transmit chain, there was plenty of intermod. According to the connector specs, "ordinary" 7/16 DIN connectors, which I guess means stainless steel, produce about -155 dBc IMD, while the low PIM variety claim -175 dBc. Since I could only see down to about -160 dBc (with a cavity notch filters to remove the carriers), I could not verify the claims[1].

My conclusion was that it takes high RF power levels to create PIM. However, because the duplexer removes the transmit signal from the receive side of the system, I can't be sure that any diodes formed on the nickel surface are not a problem.

Yep. My weapon of choice was two crescent wrenches used to tighten the connector pairs. Brute force, probably with some metal migration, made a big difference in reducing IMD. I can add my own anecdotal evidence that obtaining a stable PIM rating was impossible as the reading was effected by literally everything and was only repeatable if I used exactly the same hardware. Beating on the setup with a stick changed the readings. Even cleaning the connectors produced changes. If I could obtain a stable reading within +/-5 dB, I was happy although +/-10 dB changes were observed.

Incidentally, my two transmitters would drop in output as they became hot. The heat would also cause the notch filters in the same box to drift. It wasn't just the connectors that were drifting around.

Now I see why you were working with nickel oxide diodes. By low impedance, does that mean that they can handle high RF currents? If yes, methinks you may have identified yet another IMD causing mechanism and possibly provided me with yet another headache. Thanks much.

Acronym and buzzword decoder: PIM = Passive Intermodulation Distortion IMD = Intermodulation distortion Albaloy = White Bronze (tin-zinc-copper) that looks very much like silver plating when new.

Get the white stuff, which is finest, IIRC. McMaster has it, among many other sources. IIRC there's two browns and one is coarser than green. Hmm, they might call that one "tan." The finer one might be called "maroon." And since I last shopped they either changed colors or descriptions to "light gray" for the finest.

Or quite possibly I was a cheap bahstid and bought a non-3M product, since they seem to have the colors-to-grit I recall in unbranded "Nylon Mesh Sanding Pads" - but enco lists Scotch-Brite 7445 "aluminum silicate super-fine" as a white pad (they also sell by onsies) so perhaps I shopped there (they don't even list the green pads) and they do have a

...And that's why I try to remember to shop at least Mcaster, Enco and MSC (even though MSC and Enco are "he same company" at some level) when looking for this type of stuff... For a more mechanized approach (hardly needed for most of us) there are also "Nylon Mesh Quick-Change Sanding Discs" for a rotary-power approach, or "Nylon Mesh Hook and Loop Sanding Discs" if you prefer that approach.

Use SoftScrub or toothpaste.

We had a pcb milling machine. For the sorts of boards that I posted, my Dremel can beat it by about 20:1 on elapsed time. For more complex stuff, you really need proper vias and usually a multilayer board with a ground plane. We got rid of the machine.

We have a beautiful Tormax, and we've done a couple of PCBs with it, but it's seldom worth the setup and programming hassle. It's great for hogging big hunks of aluminum, which the Dremel can't do.

We have a great classic fully restored Bridgeport, which you can fire up and use instantly, no programming required.

Sherline makes a nice tabletop nc mill.

What are the numbers?

Well, I was using junctions around 0.01 square microns, and the good ones were around 100 ohms. Not as low as metal, but pretty darn low.

Cheers

Phil Hobbs

Krylon makes the same acrylic with a clear green dye in it. If you use this, it looks as if your board was solder-masked :-)

The green stuff seems to solder-through just as well as the clear.

I've heard all sorts of horror stories about via rivets. Reportedly, after enough thermal cycles, the solder fillet is prone to suffer a ring fracture and the via goes open-circuit (often intermittently).

Yup. At high temperature the FR4 gets squashed out from under the swaged ends of the eyelet/rivet, and it doesn't flow back in as the board cools. Eventually you get an intermittent open.

Cheers

Phil Hobbs

I keep carving circuits that don't work, 5 so far, so my Dremel technique is getting really good.

I should have known better than to waste my time.

On Fri, 11 Dec 2015 08:34:19 GMT, Bruce S Gave us:

TLT, TLX, & TLY are far better substrates for frequencies at modern levels.

TLY looks real good...

Do you design picosecond stuff? Or are you speculating?