Small-scale BGA soldering

If you have a PC. This is for a very low cost industrial gauging system. I agree in general about the appearance. That's mostly because it's a Franken board (TM)--I did the fast stuff and ChesterW did the rest, then he came up here and we bolted the head to the body.

It didn't do "Putting on the Ritz" with Gene Wilder, but it works like the bomb.

Cheers

Phil Hobbs

Reply to
pcdhobbs
Loading thread data ...

I could send you a Dremel for Christmas.

formatting link

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

I already have one with the swoopy gooseneck handpiece and all, but thanks. ;)

I've posted photos of my usual proto style recently. I use protos the way you do, to try out new ways of misusing parts or things I'm not sure will work. Other times, of which this was one, I build whole proof-of-concept instruments like that.

Normal dead-bug rocks for that--it's many times faster than perf board (as well as less flaky) and has similar performance to a ground-plane PCB.

Cheers

Phil Hobbs

Reply to
pcdhobbs

On a sunny day (Sat, 14 Oct 2017 16:58:06 -0500) it happened ChesterW wrote in :

Do not the round isles have far less capacitance between them than the square enes?

Reply to
Jan Panteltje

On a sunny day (Sat, 14 Oct 2017 16:34:15 -0700) it happened John Larkin wrote in :

120 ps is only about 8 GHz

I did not design this, but, lessons te be learned:

formatting link

Top left: 8 to 12 GHz preamp Bottom right: 2 oscillators (for different bands) Top middle: mixer, IF filter Top right: IF amp

The issue now is to modify those oscillators by filing one of those resonators. to get it to work in the HAM ATV band...

Price was good....

Wonder what soft they used to design those band filters.

BTW there is a 10 GHz heterodyne scope hidden there I would think...

Reply to
Jan Panteltje

On a sunny day (Sat, 14 Oct 2017 17:52:07 -0700) it happened John Larkin wrote in :

Is that silver paint?

Reply to
Jan Panteltje

No, it is a self-etched PCB, tined in the sampler region. If I had to do it again, I'd just Dremel it on gold-plated FR4.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

More like 3.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

On a sunny day (Sun, 15 Oct 2017 08:25:17 -0700) it happened John Larkin wrote in :

1 / 120E-12 = 8333333333.333333 ;-)

OK, up, stay up, down, sort of thing you mean?

I have done those SMD flying in air at 1.5 GHz, but it is very fragile. There is no reason not to prototype on some veroboard, or other low loss material, works just as well.

Reply to
Jan Panteltje

On a sunny day (Sun, 15 Oct 2017 08:23:56 -0700) it happened John Larkin wrote in :

Yes that gold dremeled PCB looks nice and solid.

Gotta try it some day, I have the dremel... Now the gold... Maybe later, although gold will probably keep going up, war threat.

Reply to
Jan Panteltje

Old scope formula:

BW = 0.35/risetime = .35/120e-12 = 2,916,666,666.67 = 3 GHz

Reply to
Steve Wilson

There's not much gold on an ENIG PCB, something like 10 microinches over electroless nickel.

Every year or two I have one of our PCB houses make me a couple of square feet of gold plated FR4. They charge us $100 a square foot, which sounds pretty arbitrary to me.

Gold leaf is cheap, around $12 a square foot.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

A sinewave (or any other periodic waveform) needs a rise and a fall. And more time to settle; risetime is measured 10-90 per cent, or sometimes 20-80.

The 0.35 number works about right for critically damped, Gaussian, and

1st order RC responses.
--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Besides the cost, I wonder about skin effect. I don't know the thickness of the nickel coat, and the skin depth at low GHz is likely to be deeper than the gold coating. So you are in two films that have less conductivity than copper.

But you need something to protect the bare copper against fingerprints and corrosion.

I'd think of coating the bare copper with a thin coat of Krylon Clear Spray. Apparently you can solder right through it. This would be a lot cheaper than gold, and may give better grounding.

Of course, you chop up the copper with a dremel, so grounding is not so much of an issue. But you can use Manhattan style using small circular plugs from a pcb hole punch. A 1/4" dia 0.032 copperclad will have about

1/2 pf capacitance, which will have little effect below 10GHz. This will keep the conductivity of a clean sheet of copper, and you don't have to worry about trying to map the ground currents in an arbitrary shape of copper.
Reply to
Steve Wilson

On a sunny day (Sun, 15 Oct 2017 17:16:44 GMT) it happened Steve Wilson wrote in :

Interesting. If something happens in 120 ps, then it seems to me there must be a 8.3 GHz component generated? Say something keeps happening every 120 ps, then the frequency is 8.3 GHz, a spectral line appears. How do you arrive at .35 ??

Reply to
Jan Panteltje

On a sunny day (Sun, 15 Oct 2017 10:31:49 -0700) it happened John Larkin wrote in :

OK, THAT I can follow.

Reply to
Jan Panteltje

On a sunny day (Sun, 15 Oct 2017 10:26:39 -0700) it happened John Larkin wrote in :

I could melt some Maple Leafs..

Reply to
Jan Panteltje

That still doesn't tell you where the 0.35 comes from. Why is it 0.35, and not 0.55 or some other number?

Also, the environment is assumed to be an analog circuit with LCR components that give critically damped, Gaussian, or 1st order RC responses. You can plot the frequency response of the circuit and look at the rolloff. You can even do it in LTspice.

There was an old HP application note that gave the derivation for the 0.35 constant. That was many years ago, and I don't know if it is still available or how to search for it.

It is not clear how this applies to a digital circuit with the risetime determined by semiconductor switching time. How do you plot the frequency response?

So we are taking a rule-of-thumb equation far from its roots, and calling it good.

Reply to
Steve Wilson

n
d

formatting link

Reply to
Lasse Langwadt Christensen

On a sunny day (Sun, 15 Oct 2017 18:49:53 GMT) it happened Steve Wilson wrote in :

Yes, if you look closer, and say 'rise time' is from 10 % to 90 %, then the actually 'event' is much much longer than 120 ps. Nevertheless, in the fast rising part, say for example, from 40 % to 60 %, there surely IS a 8.3 GHz component. He basically says: it is a low-pass, and the 8.3 GHz is attenuated enough to be ignored.

That is a specific case, and not always true, you can have this:

Not to scale: ...... ..... ... .. . ,

Reply to
Jan Panteltje

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.