Watson's eBlog

Toroids for Joule Thief, List of Commonly Available
...
...
...
...
...
It seems that a common complaint from neophyte JTers is how to make the coil. I decided that I would make a list of the ones I've found to be easily obtained and what I used to wind them.

I bought the Fair-Rite 2673oo2402 and 2643002402 cores from Mouser and I have used them many times. They small diameter, only 3/8" or 9mm. The '73' is a high mu (permeability) material and takes very few turns. The '43' is a medium mu material and works well with about 20 to 30 turns.

I wound a '73 core with 10 turns of 24 AWG solid cconductor insulated telephone wire, and 9 turns of 26 AWG stranded (I just couldn't get any more wire on the core). The primary measures 155 uH. In a conventional JT circuit with a white LED, a 1k resistor and a BC338 transistor ran at 33 kHz and drew about 90 mA. The same core in a supercharged JT ran at 156 kHz and drew 55 mA.

I wound the '43' core with four 16 inch lengths of 30AWG enameled magnet wire, quadrifilar wound. One winding was used for the feedback, the other three were connected in parallel for the primary. Each winding as about 150 uH. This has become a standard coil commonly used in many of my JT experiments because it's easily repeatable ( I can wind a new one in a few monutes), it's cheap (cores are a dozen cents in small quantities) and they're small cores that're convenient to work with.

Next is the T231212T cores that are from CMI Magnetics. I bought these from Surplus Sales for a quarter each, and they seem to have a lot of them. They're small, less than a quarter inch outside diameter. Everything about them can be found in my previous blog. These cores are high mu, and the datasheet says they are used for LAN coupling transformers. This leads me to believe that some of these cores could be scrounged from Network Interface Cards or motherboards that have a LAN port. I've scrounged cores from similar potted transformers, but I had to hacksaw the case and epoxy resin, and crunch up the case with a pair of Vise Grips. The hacksaw won't cut through the glasslike ferrite, but I broke one when I crunched it with the pliers. Since then I've been more careful and haven't broken any. I suppose I could heat it up with a torch and melt most of the plastic off it. But I haven't tried that yet.

TOR-23 available from All Electronics. This seems to be a popular core among JTers. Ten for a buck is probably the main reason - they're cheap. They seem to be made of iron, not ferrite. The bag they cam in has turned brown inside from the material rubbing off. I'll have to dig up the info on these. I think I blogged quite awhile back, but I'll have to check.

There are other toroids in All Electronics website, but they often sell out and then those are never available again. I grabbed a bunch of TOR-54s that seem to work okay, but they are a bit low as far as permeability goes. Seill, they do okay when wound with enough turns of solid enameled wire.

I ordered some medium size toroids from Electronic Goldmine. I'll be receiving them next week, and I'll have to wind a few and put them in a JT and SJT to find out how well they work. I often find surplus toroid cores that are low mu. probably because higher mu cores are good for RFI and EMI suppression, so small companies scoop them up for use on cables and leads inside equipment.

Tiny Toroid in Supercharged JT Gives High Inductance
...
...
(this picture is of another conventional JT, not the SJT discussed here, but it uses the same tiny toroid.)
...
I bought some tiny toroids from Surplus Sales of Nebraska. They're 25 cents apiece or 100 for $20, not as cheap as the Fair-Rite 2643002402 but still reasonable. Their stock number was ICH-T231212T. Along with these, I bought some 7/8 inch toroids for a buck and a quarter apiece, which I'll experiment with in a later blog. Their online ad says:

(ICH) T231212T
Ceramic Magnetics ferrite toroid has a co-efficient of thermal expansion = 11.5 x 10-6 /ºC. MN60 material, Mn-Zn ferrite. Max flux density is 4500 gauss. µ=5000 / 6500. 24,000 available!!

• Dimensions: 0.12" I.D. x 0.12" H x 0.24" O.D

These tiny little toroids measure only 5.7mm or 0.228 inch O.D., 2.9mm or 0.116 inch I.D. and 3.15mm or 0.126 inch tall, but they pack a lot of inductance: it takes only a dozen turns to give over 300 microhenrys. I was quite surprised. A straight piece of wire put through the core gives 2.5 uH on the inductance meter. That's a lot! It means these cores are high permeability, which is what the "µ" means, above.

I wound one of the cores with about a dozen turns of enameled wire, 28 AWG for the primary and 30 AWG for the feedback windings. Each winding was only about 6 or 7 inches long. These fit comfortably on this core. The inductance was 400 mH for the primary winding. The high permeability allows us to use fewer turns of heavier wire, which reduces losses in the resistance of the wire.

I connected it into the standard supercharged Joule Thief circuit seen in Fig. 2 here. The transistor was a BC337-25, the capacitor was 680 pF, the resistor was 1k. I got 13.8 mA for the LED current, 45 mA for the supply current, and a frequency of 203 kHz.

But the white LED wasn't very bright for 13.8 mA, so I changed it to another white LED. The new LED was much brighter, LED current increased to 21.5 mA, the supply current went up to 60 mA, and the frequency was 236 kHz. Apparently the first LED had an abnormally high internal resistance, and I'm glad I swapped it out. It's now in the trash.

This LED current is about what it should be, about 20 mA. If I divide the output power (3.2V * .0215A) by the input power (1.5V * .06A), I get an efficiency of 76.4 percent. That's a nice, respectable number, too, for a circuit that I tack soldered together in a few minutes. I would recommend these cores for making a very compact JT, or the Fair-Rite cores for slightly larger toroids.

I tried experimenting with the value of the capacitor. I tried values below and above 680 pF, and found that they didn't make that much difference in the performance, thus 680 pF seems to be a good choice. The 1k resistor also seems to be a good choice. Varying it will vary the current through the LED, and at 21.5 mA, it's already at the right value.

The picture is of a conventional JT, not the supercharged JT I've discussed. It can be seen in the picture that this conventional JT is small enough to be taped to the side of the battery. The size can be reduced by cutting the leads sorter and repositioning the parts (the 2.2 ohm is not used). The parts can yhen be taped to the side of the battery with electrical tape. The only thing left is an on/off switch. I've made this out of two wires taped close together, but I have to hold my finger on the wire contacts or the LED will go out.

Size matters
My size measurements of the cores were different than what Surplus Sales gave in their ad, so I googled for the T231212T part number and found the toroid info at the CMI website. My dimension measurements were consistent with the values given in the spec sheet. The sheet also says that these cores are used for LAN coupling transformers. That's interesting. Someday I'll have to remove the coupling transformer from an old NIC card, and open it up and see if it has one of these cores in it.

I found other interesting and useful info on their website, including a summary of the properties of various ferrite materials. Another thing they have that I've never heard of before is a squaroid. It's a square toroid core, which seems reasonable when the wire is only a single turn straight wire going through it. They also have gapped toroid cores, which have advantages according to those who know about magnetics.

I should point out that if more inductance and power are needed from a small toroid such as this one (or from any toroid core), two toroid cores can be stacked, and the wire wound around both cores. This gives double the inductance and twice as much core, so more power can be handled. This is essentially what has been done when you use an RFI/EMI sleeve; it's the same as putting two or three toroid cores over the single wire. Thus there is more core around the wire, more loss, and more of the RF interference is absorbed.

I've heard of several toroid makers, but never heard of CMI. They're in Fairfield, New Jersey - that's halfway between Passaic and Parsippany. Where's that at? Well it's roughly 60 miles from where I lived for almost a year, Ft. Monmouth, N.J., which is the home of the Army Electronics Command, which is halfway between Long Branch and Red Bank, not far from Asbury Park, which is Bruce Springsteen's home town. Not only do New Jerseyans have these funny named towns, they talk kind of funny, too. Like they say "New Joisey". And "cah", yeah, like "I gotta go fill up the cah with gas." Speaking of cah, one thing that sucks in New Joisey is the toll roads. We've got a few in So. California, but just a few and they can be avoided. But not in N.J. Anyways, I couldn't find any prices for the cores on their website, so I may have to call up and talk to one of those New Jerseyans and ask them how much they cost.

Back to experimenting...

Penny & Vinegar Battery Powers Flasher Circuit
...
...
...
...
I built some penny and vinegar batteries to power my Flasher Circuit. I got the idea from a YouTube video of a Joule Thief that ran off a lemon with several pieces of sheet copper and zinc stuck in it. I built a similar battery using pennies and vinegar.

Before 1981, all U.S. pennies were made of solid copper. Then to save money, the U.S. mint started making pennies (and other coins also) out of zinc with a copper coating to give it the look of a penny . You can hear the difference when you drop the penny on a hard surface, the zinc has a much duller sound.

As all chemistry students should know, two dissimilar metals separated by an electrolyte can make a battery. I made these penny batteries as follows.

I took a modern copper coated penny outside on the cement sidewalk and sanded the back side down with a piece of sandpaper laid on the concrete. You can also sand it directly on the concrete, but I had some extra sandpaper. I sanded it (and my fingernails) down until I could no longer see any copper color, only the steel gray of the zinc.

I picked out a solid copper penny, 1980 or earlier. I was tarnished so I dropped it into a dish and poured some vinegar on it, and left it for a few minutes. When I removed it the tarnish was mostly gone. Instead of the vinegar, I could have sanded the solid copper penny until the tarnish was removed.

I tore off a corner of a paper napkin, a bit larger than the pennies. I sandwiched it between the two pennies, and trimmed the paper to the size of the penny.

I then arranged the pennies so that the zinc penny was on the bottom, zinc side up, the paper napkin, and the solid copper penny. I wetted the paper with a drop of vinegar and put the sandwich together, and held the two leads of the flasher to the pennies, the negative lead on the bottom zinc penny and the positive on the top copper penny. The flasher I used is Figure 2 in this blog.

The flasher started to glow, so I got a DMM and measured the voltage across the pennies. I got 1.3 something volts, but it depended on how much I pressed on the pennies. I put another drop of vinegar on the paper, and changed back to the flasher. It was brighter, and when I connected it up again, it started to flash.

I think part of the problem is the penny battery has a higher internal resistance. I think a bypass capacitor could help a lot in making the flash more stable. The penny puts out enough current to flash the flasher, but the vinegar dries up after a few tens of minutes. Maybe another layer of paper would help keep it wet for longer.

I didn't think that a single penny battery would make enough power, so I had done two penny batteries. It turned out that one was enough, but I put two in parallel, and the light got a bit brighter. After a minute or so, the LED tends to dim, probably because the vinegar electrolyte is getting used up. I'm not sure what can be done to reduce this, but if I take the flasher off for a minute or two, it seems to recover somewhat.

One other thing that can be done to save aggravation is to solder some leads onto the pennies. Holding them with the fingers is not very practical. I guess some rubber bands might work okay, but a little solder makes them semi-permanent.

This project yielded results that were better than expected. I really didn't think that I was going to get that much out of such a small battery. But then the flasher draws very low current, so it's not that great a load. The commenter asked for a picture, so I attached a pic of the pennies and paper. I used vinegar, but lemon juice and other acids should work, too. Try it and see. This is a neat little science fair experiment that kids might like to do.

I googled for penny battery and found that Wikipedia has an article on it. There are instructables on it also. Apparently this isn't new, it's been around for awhile.

Back to experimenting...

10mm 1W White LED Lifetime Test
...
...
...
...
...
I have affixed four of the 10mm 1Watt white LEDs to a piece of wood with washers and screws, in series so the total voltage drop is about 12VDC. I powered them with a 9V wall wart, which when measured with no load was about 13 and a fraction volts -- being unregulated, the fraction of a volt varied with the incoming 120VAC line voltage, which tended to vary with the load on the power grid. I put a 1 ohm resistor in series with the LEDs, and found that it measured about 0.130 volt, which meant the current was 130 milliamps. This is considerably under the maximum current for a 1W LED, but I think it is about right for testing these LEDs. The washers and screws get warm, but not hot. I want to get an idea of how long these LEDs will last while running at a current that is realistic for real world applications.

I took initial measurements of each LED with my Weston light meter and got readings about 80 to 90 on a scale that has values only at 50 and 100, with 100 being at about 1/3 of maximum. Thus the readings were not very accurate, but a lot better than judging brightness by the naked eye.

I expect from past experience, that after 1500 or so hours, sometime after New Years, the LEDs will have dimmed considerably, and will have lost more than half their brightness after 2000 or so hours. Updates will be made in a month or two.

Another Subject
Meanwhile... I've been looking at some Tesla Coils on various websites and Youtube, and found that a small tabletop sized one can be made fairly easily. But it seems it will take a few hundred dollars to make a decent one.

The secondary coil is a piece of 2 inch PVC tube about a foot and a half long wound with wire. I did some calculations and found that it will take less than a half pound roll of 26 AWG enameled wire to wind about a thousand turns on the tube. The half pound of wire is under $20, and the PVC tube and other parts should not bring it to more than $30. It will be time consuming to wind the wire on the PVC tube.

The capacitive hat on top can be made from two small metal bowls, I would guess maybe $10 or $15 for both. If I'm lucky I might find something at the thrift store.

The base, primary coil, spark gap, and a few other things can be made from stuff I have on hand. I figure it may take another $20 for stuff I might need.

Many TC designs use a high voltage transformer, usually a neon sign transformer or NST. A typical NST is 10kV at 30 mA. These are available on Ebay, for example, for under $100.

The last thing seems to be the capacitor. There is more than one way to accomplish this. One is to build your own. Some use salt water capacitors. A glass bottle is covered on the outside with aluminum foil, and filled with salt water. A heavy wire is immersed in the salt water, and forms the top electrode. Several of the bottles are set in an aluminum pan, which forms the bottom electrode.

Another method is to make a MMC or multiple mini capacitor. Dozens of plastic capacitors, each rated for 400 volts for instance, are connected in series to give a capacitor that is capable of withstanding twice the NST voltage, or 20 kV. This can cost a lot if you have to buy several dozen capacitors, each at up to a dollar apiece. But I have more than a hundred 1 uF, 250V plastic capacitors, so I could make one for fairly cheap. But it will take a lot of manual labor to assemble properly.

The easiest and cleanest way to go is to buy a high voltage capacitor. So far I've seen prices of from $40 to over $130 for a single capacitor, depending on the capacitance and voltage rating. I found that Surplus Sales of Nebraska has some high voltage capacitors that I can actually afford. Two of the .01 uF 10kV in series might do the job. That should come to somewhat under $100 when I include shipping.

I haven't decided on which way to go. But then there are other things that I have to do, too. The total cost looks like it will come to over $200. And a thorny problem often crops up during the initial phase of getting a TC running. Things don't always go according to plans. Since we're dealing with high voltages and high power, the result is often damage to one or more components. So the cost of getting this running may creep above the initial cost. How much creep, I don't (want to) know.

Back to experimenting...

Salvaging Components - Identification
...
...
...
...
...
Identification
The problem of identifying the part may now be a stumbling block. The neophyte may have to take some time to gain experience with what parts look like. Visiting an electronics store (such as Fry's in California) that carries the NTE line of parts can be helpful. One can see each part on the shelf, and usually the store has a crossreference guide to help. It's also available online at nteinc.com. I recently found another online resource, a brief list of common semiconductors. Some other online resources are radio amateurs AKA "hams", and the Usenet newsgroup sci.electronics.components among the sci.electronics newsgroups.

In addition to all the online resources, I have an extensive collection of old manuals and data books, many of them for semiconductors. Many of the big names, such as Motorola and Texas Instruments, had heavy tomes chocked full of information. The Semiconductor Data Book from Motorola had information on most transistors up to the 2N5000's in the early 1970s. The old adage "An hour in the library is worth a day in the lab" still holds true. It's better to find out about something that someone else has already done rather than reinventing the wheel yourself.


Passive components - resistors, capacitors and inductors. etc.
Some simple components may be measured to find their value. Almost all DMMs have resistance ranges. Some DMMs have capacitance ranges, especially better quality DMMs. One of my favorite meters is the LC Meter IIb, from AADE. It's about a hundred dollars in kit form, and a bit more assembled. It's very accurate, and runs off a 9V battery (see note at bottom).

Most of the time resistors have their value marked on them with the color code bands or in print. If not, then just measure the unknown part with a DMM on a resistance range.

Capacitors also are often marked with the actual value, or the common but cryptic value in characters that conform to industry standards. You may see "103M Y5P" on a ceramic disc capacitor. The 103, similar to resistor color code, means 1, 0, and 3 zeroes, or 10,000 and the units are in picofarads. This is the same as .01 uF or 10 nF. The M is the tolerance, and J, K, M are often seen. J is 5%, K is 10%, and M is 20%. Sometimes the maximum voltage is given, such as 1KV. There are exceptions to this, especially when the cap is below 100 pF, and the numbers are the actual value. You can find a guide online that explains what the markings are, just search for the marking with Google, and you should also find what the other markings mean. I have also found that looking at a parts catalog (Mouser is one example) is very helpful. It's available online.

Inductors are sometimes marked with three digits, just like capacitors. Or sometimes the actual value is printed. And some small chokes are color coded like resistors. This often causes confusion because they look like a resistor, but they are usually shorter and bigger around and the body color is usually light blue-green. The inductor's DC resistance can be measured with a DMM, and has a resistance value that is usually low, and is not the same as what is marked. At least you know that the part is not a resistor or capacitor.

Transformers are one type of component that is worth salvaging. They are expensive, hard to find, and are heavy and can cost a lot to ship. More often they are found mounted on a PC board for convenience. If a transformer is mounted to a board with screws and has some wire that can be soldered to, then it may be possible to clip the wires at the circuit board and remove the transformer and then later splice on some longer wires with some heat shrink tubing. If the transformer is mounted to the board, it may be necessary to get a reall y heavy duty solder ing iron or soldering gun to heat up the metal tabs if they're soldered. In my case, I was lucky that the only things holding the transformer to the board were the pins of the bobbin. I could heat up a few pins and ease the transformer out a few pins at a time.

Some transformers have very heavy wire, such as the ferrite transformer found in most switching power supplies. It's especially difficul because the copper wires conduct the hat away quickly, requiring a lot of heat. But they are wrth salvaging because they make great ferrite cores for Joule Thiefs and other fun projects. Sometimes the wire can be unwound and used elsewhere, maybe even tie your tomato vines up?? ;-)

Active Components
Semiconductors or active components can be difficult to ID. If you're lucky, it's marked with an industry standard part number. If not, then things may be more difficult. Either way, a good place to start is the online search engine. I Google the part number followed by the word datasheet. This brings up a lot of hits, but my favorite is datasheetarchive.com. You can get an idea as to whether or not the part matches the specifications by looking over the (usually .PDF) datasheet, especially the picture, which gives the dimensions and pinout of the component.

Industry standard part numbers for different parts of the world are as follows. For the U.S., the standard JEDEC part numbers start with 1N or 2N, or rarely 3N or 4N. The 1N are 2 terminal devices, usually diodes and Zeners, but sometimes DIACs. The 2N parts are 3 terminal devices, commonly transistors, including BJTs (bipolar junction transistors), JFETs (junction field effect transistors), and MOSFETs (metal oxide silicon FETs). 2N numbers also include thyristors, SCRs, TRIACs. The 3N parts with four leads are most commonly dual gate MOSFETs. The 4N parts are commonly optoisolators.

The Japanese have their own industry standard part numbers. The three terminal devices all start with 2S, but these two letters are usually left off the part because they are assumed. The third letter (first on the part) is A, B, C, D, or J or K. A and B are PNP, C and D are NPN, and the first of the pairs is high frequency, the second is low frequency. But nowadays the high or low frequency is blurred because most silicon devices are capable of several tens of megahertz, so most devices are labeled with A or C. The J and K are for FETs. The Japanese use 1S for two terminal devices, but I've seldom seen these parts on boards. Most small signal diodes are 1N914 or 1N4148 - both JEDEC, or their surface mount equivalents.

The rest of the world often uses the "ProElectron" part number system. All germanium parts start with an A, but germanium parts are seldom seen nowadays, most semi makers stopped making them when the industry converted to silicon by the early 1970s. Thus most ProElectron part numbers begin with B which means it is silicon. And since almost all of the devices are silicon, this letter is usually not put on the part. You might see C547C on what looks like a transistor, and it is actually a BC547, a common transistor, with the C at the end to indicate its current gain rank. You will find out what this letter means when you read the datasheet.

Back in the first few decades of the semiconductor industry, most transistors were in metal packages because germanium chips had to be sealed in a 'hermetic' air-tight package. Then silicon chips predominated and the makers could coat the chjps with an oxide passivation layer and seal the chip in a plastic package, which was much cheaper than metal. The makers then marketed their own line of plastic epoxy cased devices, which eventually replaced the metal ones in most equipment. For instance, Motorola marketed their line of semis in plastic epoxy packages, and the parts began with "MP", often seen as MPS for transistors or MPF for FETs. Other makers then marketed parts with the same specs and part numbers. Texas Instruments had a line of transistors beginning with TIP. National Semi and Fairchild Semi marketed parts in plastic that were the same as the 2N parts, but named them PN because they were in a different package. Other makers started making them too.

So far I've just touched on the simple semi's with 2 or 3 leads. it gets more complicated with SSI, MSI and LSI devices (small, medium and large scale integration).

Note: The LC Meter IIb model I have has some quirks with how it acts when it's first turned on, requiring all the buttons to be out to calibrate. Since it runs off a battery, every time I used it I had to go through the same calibration process. I thought this was tedious and extra wear and tear on the buttons, so the first thing I did was to connect it to a 9V wall wart so it could be left on all the time. But for occasional use, the battery is okay.

Salvaging Used Components - Tutorial
...
...
...
I've salvaged used components from many pieces of equipment, and saved a lot of money. I have also been frustrated by components that have been out of tolerance or just plain defective. Is it worth It? I think so, especially in these hard times.

First, some advantages and disadvantages of removing used components.

ADVANTAGES:
Inexpensive - sometimes even free.

Sometimes components are unique and can't be easily found unless you special order them. These might be components that have precision and/or special values.

Quick - nothing like having a widget right at your fingertips, just for the price of a little unsoldering.

DISADVANTAGES:
Bad components - you may be unsoldering a defective or out of tolerance component that can cause you all sorts of headaches.

Disables the board or equipment from which it's being removed. But if it was already on its way to the trash, then who cares?

Newer equipment has surface mount components, which are difficult to work with.

Heat from unsoldering can damage the component. Also, mechanical stresses during removal can break pins or damage the component internally.

Static electricity can damage the component. For static sensitive components, anti-static precautions should be used.

Older equipment was not ROHS compliant, nor were its components. But that's about the lowest item on my list of concerns

Identification difficulty - often parts are house numbered, and can't be IDed easily, or even unnumbered (some paranoid makers don't want anyone to know about their circuits - or maybe they want to monopolize all repairs for themselves).

There are probably a few that I left out. Let me know and I'll add them.

First off, there is one thing that should be decided before salvaging any components. Is the board expendable? In other words, removing the components will most likely damage the PC board. If you must save the PC board, then it is best to destroy the component in the process of removing it and not damage the board. Components with many pins will be very difficult to remove unless you have specialized removal equipment. The usual way to remove a multi-pinned component without damaging the board is to cut off all of the pins and remove the component, then unsolder the pins one pin at a time.

If you decided to sacrifice the board, then you're ready to remove the components. Heat up the soldering iron and get the tip really clean so it can make good contact when heating the pads up on the board. I have to use flux paste to get all the tip well tinned with solder.

The way that components are removed in a factory is to use a solder sucker, and in stubborn cases, a length of solder wick to remove as much solder as possible. Then if the leads are free and can be moved, the component should come right out. But for Q&D removal, I just heat up the leads, and gently pull on the component. it usually comes free in a short time if the leads are heated. The key here is to heat it quickly, and get it out without damaging it. The more the leads, the more difficult it is to remove. I removed 40-pin CPU chips with a solder sucker that ran off compressed air, and I could lift them out of the board without pulling out any of the plating in the holes. But I had to be careful, since it was easy to damage a PC board.

Almost no one has a super duper solder sucker at home, so they would probably use one of the syringe type solder suckers. These can be cocked by pushing down on the plunger to compress the spring. Then when the solder joint is hot, the tip is applied and the plunger is released, and PLOP! The quick suction pulls out most of the solder. The pin may have to be teased with a sharp tool such as a dental pick or X-acto knife to loosen it from the hole. When all of the pins have been loosened, the part should come out easily.

Some components such as large capacitors, inductors and transformers may have pins that are strong enough to allow one to pull on the part while several pins are being heated. When some of the pins are loosened, they may come out partially. Then some other pins may be pulled out partially. Finally when the pins are far enough out, they can be loosened and the whole part removed. I have been removing over a hundred transformers from their boards, and I have not had one where I have damaged the part other than a bent pin. All of them have come out without problems, even though they have seven pins.

Nowadays, it seems that parts are being soldered with lead-free solder,which is more difficult to melt than 60/40 solder. I often add some 60/40 solder when I'm unsoldering it to make it easier to remove.

After removal, the solder may need to be removed from the pins. I heat it and use a stiff bristle brush to knock most of the solder off. I also shake the part to cause the solder to fly off. But I've had some of the solder fly off and hit another project, causing a short. So one needs to be very careful where the solder will land.

One other thing that bears mentioning is the problem with sensitive components being damaged by static electricity. The soldering iron should have a tip that is grounded. Some solder suckers have teflon tips and it seems that some parts can be damaged by this tip. So the tip should be antistatic. Your body can generate a lot of static electricity, so you should use a grounding strap on your wrist when handling sensitive components. And last but not least, your bench should not generate static electricity. Some techs put a piece of carpet on the bench to prevent scratching, and this carpet can generate static. It's a good idea to spray some antistatic on it. I've used Downy diluted in a spray bottle to reduce static. But here, it really doesn't get dry like it does in the the winter in some places, so static isn't so much of a problem.

Identification
This section got so long that I moved it to its own blog following this one.

Back to experimenting...

10mm 5 chip white LED Lifetime Test
...
...
...
...
These are the 290,000 mCd (or more appropriately, 290 Cd) LEDs that I mentoned in my previous blogs. I drilled some holes in a piece of scrap wood and mounted four of these with screws and washers. The idea was to let the screws and washers act as small heatsinks. I probably should have mounted them with big long screws and stacked up a bunch of nuts and washers on each screw to make a better heatsink, but what you see in the picture is all you get for now. My fingers sense only a slight warming after a few minutes.

I soldered wires to the leads to wire them all in series. For the first test I just connected them directly to a HP power supply that doesn't have adjustable current limiting. I adjusted the voltage to about 12V and the current meter read 80 mA, but it's very sensitive to the voltage. I put a current limiting resistor in series, a 50 ohm 3/4 watt which will drop about 4 volts (actually three 150 ohm, 1/4 watts in parallel). At 80 mA total, each chip in the LED is handling about 16 mA, which I thought would be a typical value for this LED to be handling.

I checked them with the Weston light meter and all four read between the 50 and 75 marks on the low part of the meter. That's not as high as the other LEDs I've tested, but the hole for the LED is about 5mm, so these 10mm LEDs won't fit into the hole, and more than half of the light falls outside of the hole, making the reading lower. But I'm not interested in the actual reading, I'm only interested in comparing readings at the beginning of the test with the readings a thousand hours or more later. To satisfy your curiosity, I taped a piece of cardboard over the light meter. First I cut it out and taped it together in the shape of a cone, with a 5 mm hole at the tip. This holds the LED at a fixed distance from the meter, and keeps out ambient light.

Now comes the Long Wait. Future updates in a month or two.

On another subject, the FTC says it's going to change the rules regarding light bulb labeling to emphasize the brightness in lumens, instead of watts. About time!
http://www.ftc.gov/opa/2009/10/lightbulbs.shtm

Back to experimenting...