Soldering to Ni/Au pads

The zinc doesn't evaporate? It boils about 1200K, iron melts at 3100K or so.

On 1 Aug 2015 05:40:57 GMT, Jasen Betts Gave us:

Yeah, ya dope, it "boils", but it isn't water. It doesn't phase into a gas the same way. You melt the iron first, then add solid form zinc to it, and more than you actually end up with. Then you must cast the "alloy" before it "boils away". This is how alloys get made.

Otherwise, the Iron would "boil away" when it gets alloyed with Tungsten. Etc., Etc.

I found my soldering book. What they said is that IBM's standard process for soldering gold-plated leads was to flux and tin them first to dissolve and remove most of the gold, and then a second time in a second (cleaner) solderpot to ensure complete wetting and complete removal of gold.

Joe Gwinn

While breathing zinc vapour, brr.

Cheers

Phil Hobbs

I just used SN63/PB37 (Kester No-Clean) on 50 u-inches gold over 150 u-inches nickel. Worked surprisingly better than anticipated.

Assuming that in your line of work these diodes must be expensive, can't you re-bond them or is the carrier now all toast? If you will have to do this a lot it might be worthwhile to invest in a used manual wire bonder. It doesn't have to be a high-end Kulicke & Soffa.

Yes, the first one I tried survived. Of course you had twice as much nickel and half as much gold as these ones nominally do.

These diodes are about $65 each (GPD Optoelectronics GAP60CS). I bought eight of them for $520 plus tax.

The gizmo is actually pretty interesting--it's a lidar for detecting and mapping individual submicron tin particles inside the chamber of an EUV light source. Those suckers can be moving at 3 km/s (Mach 9 at room temperature), so the front end has to be quick. (This one is only good to about 4 GHz, which is equivalent to Mach 4.5 radial velocity.)

Cheers

Phil Hobbs

Sometimes the phase diagram indicates gas (assuming some pressure, I guess), but the evaporation isn't unmanageable, at least for the copper alloys, which usually melt near, and are poured well above, the boiling point. A pot of brass or bronze has an eerie blue-green glow over it. :) And yes, stay out of the fumes...

Tim

Luckily 150/50 u-inches was the thickest standard process at the PCB fab, else it would have been expensive. Got 250 bond carrier boards for a little over $1k.

Ouch! If the #1 bond wire is still there you might be able to connect to it with silver-epoxy. Unless the #2 pad has been eaten up but in the datasheet it looks like it spills over in the radius a bit.

Sometimes I use RadioShack "Bare Paint" to make such iffy contacts or connect directly to bond wire. Bare Paint presents a few ohms once dry and doesn't have much mechanical strength but for prototyping and testing it has been ok so far. And the (remaining) local RadioShack lets me park my trusty old road bike in the store if I need new stuff, like yesterday.

I love those kinds of challenges when they present themselves. The GAP60 should be faster. Why only 4GHz?

Seems extreme. That wouldn't be feasible for gold-plated circuit boards, or for Phil's gold-plated ceramic.

This doc

has a graph that says that nickel doesn't diffuse into tin below about

But if one prowls the web for a few minutes it's easy to find a lot of contradictory opinions about soldering to nickel/gold platings.

The diodes aren't much faster than that in real life, I don't think, and besides, the back end for the demo is my TDS694C scope, which is 10 Gs/s.

The optics are more challenging than the electronics--it has a working distance of 600 mm, which means that both the Tx and Rx spots are only

Cheers

Phil

Is this for you-know-who? Daniel and his friends?

Yup. I got super busy, so it's gone very late, but I'll have it done this month. My other main project (the NA 6.4 microscope) is giving good data, so I just have to take some resolution target images and demonstrate the deconvolution, which doubles the resolution. That should take only a couple of weeks' work. The microscope is basically a

The 2-Dness causes a lot of extra complexity, both in the signal processing (a table covered with connectorized Mini Circuits stuff and several Bud boxes full of hand-wired PLLs and such) and in the optics. When I have a chance I'll put some photos on my web site.

Assuming the customer's customer ponies up for the second phase, this will all get crunched down considerably, with DDSes and such, but I needed flexibility for the demo because I didn't know what the spurs were going to be like.

Cheers

Phil Hobbs

They measured them at 5V bias. I wonder how much more one would get at the abs max of 25V. But since in your case the scope limits the BW there wouldn't be a point.

Probably nobody is allowed to sneeze within 100ft so nothing can vibrate.

Some stuff never gets old, and some does. I am still amazed how much in precision one can milk out of a TDR that contains less than $5 in parts. Yet ultrasound has become a bit boring for me after doing those designs for almost 30 years.

Many photodiodes do speed up a lot, suddenly, when you get to full depletion. The speed limit is usually diffusion rather than capacitance.

It can be a bit ambiguous when you read datasheets, because both go like the area, but for different physical reasons.

It's actually pretty robust once you get it lined up. (Epoxy is wonderful stuff.) The system is either perfectly aligned or obviously broken. It's the intermediate stage, where you get reasonable-looking wrong answers, that you want to avoid, and this one does so.

I'll build a TDR myself one of these days, but I have four SD-24 TDR heads, which is enough for everyday use. ;)

Cheers

Phil Hobbs

If the 50 photons represent one event maybe you could use a comparator with a uC adjusted threshold. Similar to this ...

... except the time of flight could be measured by a ramp from a reference point to when the comparator switches (current source stops after it switched). You could even logically stagger several comparators. Of course, if more than half a dozen would be needed that method quickly gets old.

Because the particles don't slow down much in the time it takes them to cross the beam, you can use a filter bank to find them. My original version (1992) was intended for slow-moving particles (~1 m/s or less) orbiting the plasma sheath in semiconductor etch tools. It used four LC filters and window comparators, and an interesting adaptive thresholding technique that relied on the Gaussian statistics of shot noise.

The threshold crossing rate of Gaussian noise can be calculated accurately, so the idea was to use two thresholds: a lower one, where the false count rate was ~5 kHz, far beyond any true count rate; and a higher one, where the false count rate was on the order of one per day. The trick was to servo the lower threshold to make the false count rate _exactly_ 5 kHz, and derive the upper threshold as a fixed multiple of the lower one. (I used a voltage divider.) It would never have worked except that the laser noise canceller gets rid of excess laser noise so very well that all that's left is signal and true shot noise. The servo idea is explained in

The 2015 system uses windowed FFTs to make a much larger filter bank, and software to do the detection. Fun stuff.

Cheers

Phil Hobbs

Nice. False alarm rate tracker and all that reminds me of Radar. It took a lot of convincing just to try some Radar techniques in medical ultrasound. I guess with your projects you can mostly start with a clean slate and bring in all kinds of ideas.

Didn't know you had two middle names. AFAIR only the Dutch have more, usually three.

IBM was dealing with gold-plated component leads, not hybrids. But what they felt they had to do is instructive.

For the gold on ceramic, I bet that spot welding is the usual approach in manufacturing, and indium solder is used for repair.

Yes.

I prefer the books on soldering. Their authors do have their opinions, but they also have the space to explain the various rationales, so one can decide which best fits the problem at hand.

Joe Gwinn

On Sun, 02 Aug 2015 11:51:25 -0400, Joe Gwinn Gave us:

The absolute best soldering experts are NASA scientists and engineers.