Building the pulse generator

September 18, 2012 8:48:01 PM CDT

In yesterday's post, I talked about building and testing the pulse generator to make sure the design actually worked before publishing it. While I was working, I took photos of the process as a sort of mini-tutorial for anyone who wants to build one.


Here's my standard work setup for small SMT projects:

up to work

The grey base plate is a baking sheet. As I mentioned previously, the bases for my wire clips have magnets in them. Working on a piece of sheet steel keeps the work steady.

The labels on the component tapes aren't just for your convenience. I've spent enough of my life finding a random tape on the bench, peering at the parts through a magnifying glass, then flipping through datasheets trying to remember which SOT-353 package is marked '7Y', thank you very much. Besides, red 0805 LEDs look exactly the same as green 0805 LEDs until you run power through them.

Putting in the power LED:

Speaking of LEDs, that's where we'll start.

Blobs of 
solder for R2 and LED1

The rules for hand-soldering SMT are sort of backwards from the rules for through-hole parts. With through-hole, you place the components then solder. With SMT, you start with a blob of solder then set the component in that. The photo above shows the pads for R2 and LED1 with one pad for each device holding a blob of solder.

LED2 and 
R1 tacked down

To place a component, you pick it up with your tweezers, melt the blob of solder with your iron, set the part in place, then align it before pulling the iron away.

You're not going to get any action photos of the soldering because I don't have a hands-free camera to take them. I could make one, but I have enough projects in the pipeline as it is.

The photo above shows R2 and LED1 tacked down on the right side, but still unconnected on the left.

LED2 and 
R1 in place

Once the parts are tacked down, things get easier. You can flip the board around and solder the remaining pads. You don't have to worry about alignment because the solder joint you've already made holds the part securely in place.

You can fiddle around with the parts long enough for the solder on the far pad to melt, but you shouldn't.

Another "do the opposite of what you were taught for through-hole" rule applies to the way you bring in the solder: when you were taught though-hole soldering, someone probably told you, "touch the solder to the component, not to the iron." The idea is that you need to get the pad and component lead hot enough for the solder to adhere, and you can build a big-ol' blob of solder on the end of your iron without ever getting adequate heat to the component.

For SMT, heating time isn't an issue. As soon as the component touches molten solder, it's hot. The big risks of SMT are overheating and using too much solder, so for hand-soldering SMT, the rule is, "use the iron to carry solder to the part."

Dip your solder in flux, tap it to the tip of your iron, melt off a small blob, then touch that blob to the joint you want to make. Surface tension will pull enough solder from the iron to make a good joint. After you do it a few times, you'll get a feel for the size of the blob you need. Once you're good at it, you can fly through making connections.

The photo above shows R2 and LED1 with their remaining connections soldered.

Finally (at least for this part), I'm a big fan of incremental testing:

the LED

It's a lot easier to find a bad connection or part if you test the circuit after putting in each piece than it is to play, "how many faults do we have and where are they?" with a completely built circuit that doesn't work.

the photo above shows that yes, I did put the LED in the right way around.

Putting in the chips:

The process for putting in chips is mostly the same as for resistors and LEDs. You start with a blob of solder on a single pad:

Blobs of solder for the chips

(The photo above shows blobs for both the 74LVC1G126 buffer and the 74LVC1G14 inverter)

And here's the buffer tacked into place:

126 buffer tacked down

The major difference between chips and simpler components is the need to spend time fiddling with the alignment. Not only do you need to get the part centered across the pads, you need to make sure all the pins line up with their respective pads.

Start by tacking the first pin down, then look at the board to see where the chip needs to go. Then grip the part with your tweezers, tap the iron to the joint to melt the solder, move the part, remove the iron to let the joint harden, and check the alignment again.

This is actually one of the places where the reflow people have an advantage over us hand-soldering types.. if the solder for all the pads melts at the same time, surface tension will pull the part into the correct position. Doing hand adjustments isn't that hard though, the main trick is learning how to move parts a few tenths of a millimeter at a time.

The key is proper bracing. Don't try to do fine adjustments from your shoulder. Put a block as close to the part as possible, rest your hand on that, and get the shortest mechanical connection you can from that block to the tip of your tweezers. When you're so thoroughly braced that you could only move a couple of millimeters if you tried, you're good.

Once the part is aligned to the pads, you can solder the remaining pins:

126 buffer in place

It's a good idea to start with the pin diametrically opposite the one you tacked. That's the one that will have to move the most if you need to make any final adjustments, and having opposing pins tacked down will give you the strongest possible connection.

The picture above shows residual solder on the traces from where I attached wires to test the buffer (it worked).

Then came the challenge:

Schmitt inverter in place

The buffer (left) is a SOT-753 package. The inverter (right) should be too, but all I had on hand were SOT-353s. It's possible to fit a '353 package onto a '753 footprint, but it isn't easy (or, as you can see from the photo above, pretty).

It's in, it works, and that's all I really cared about for a smoke test.

The timing network:

With the chips in, we're back to passives:

Blobs of 
solder for R1, C1, and D1

The photo above shows blobs for R1, C1, and D1.

R1, C1, 
and D1 tacked down

This one shows the parts tacked down on one side. At this point, an error in the board layout came to my attention: the diode is a SOD-123 package, while the pads were laid out for SOD-323.

R1, C1, 
and D1 in place

You can see the offset better in the photo above, with the parts tacked down on both sides. The connections all work, but the layout had to be redone before the design was ready to go.

The finished board:

The final step was to add a wire for the output:



Of course, the really important part is to see whether the thing even works.

Turns out it does. With a 3v supply, the time between pulses is 400 milliseconds:


and the pulse width is 100 microseconds:

Pulse width 

With the circuit itself tested and known to work, all I had to do was modify the PCB layout a little to make the design ready for publication.

This really isn't that exciting a circuit, but as you can see, even small projects can hit snags.

Random brain cookies:

One Page Principle: A specification that will not fit on one page of 8.5x11 inch paper cannot be understood. -- Mark Ardis