Slashdot Mirror

Just like the 2N3055.. Yes, old. but in some cases, absolutely perfect for the job. Usually as linear pass transistors for power supplies :D

There is plenty of reason to avoid the 2N3055, well actually to avoid the TO-3 packaging. The package flexes when mounted, which can cause poor thermal transfer to its heat sink. TO-3 are manufactured to be curved so as to help ensure good thermal contact, but the curvature isn't preserved if the TO-3 is removed.

References: OnSemi's Application Note 1040, and Burr-Brown (TI) Mounting Considerations for TO-3 packages.

The device at U16 on Raspberry Pi 2 v1.1 appears to be an ON Semiconductor NCP6343 DC converter provided in a WLCSP-15 package.

Like all CSP packages, the bare die is photosensitive and needs to be protected from incident light if fault-free operation is expected. Usually such devices are embedded in closed cases like cellphones which prevent light ingress.

However, if the normal operating environment includes uncased bare boards or transparent cases (which are both common and normal for Raspberry Pi), then it is imperative that CSP-packaged dies be protected from light by other means such as opaque epoxies or caps, otherwise such devices cannot be expected to operate within specification.

It is a normal part of the engineer's job to understand their product's operating environment and the components they use, and to design accordingly.

Probably an electronic fuse, something similar to this

Yaright. Let's work that one out, shall we?

Design for a 100 W (max) light bulb and 120 V line input. First off, you'll need to put the diodes in series so their voltage drops knock off 12 V (10%). Then you'll have to put an identical string of diodes, reversed, in parallel with the first stack, because you want current to flow both ways. If you don't, you'll knock the brightness down by over 50%, since it will conduct only half the time.

An incandescent bulb is pretty much a resistor (more like a thermistor, but let's say it hit steady state already). Thus the equation P = VI holds, so I = P/V = 100/120 W/V = 5/6 A, thus we need diodes rated for at least 1.0 A. As a first approximation, diode forward voltage drops in this current range are in the ballpark of 1 V. So 1 V * 1 A = 1 W for the DC case. To figure AC power, you have to integrate over a cycle, but to simplify it, let's say the AC waveform is a square wave, so each diode conducts full power half the time, thus dissipating 1/2 W.

Now that we have our specs, we'll head over to Diodes Inc. and get the data sheet for the 1N4003. It has Vf = 1.0 V, Io = 1.0 A, and Vr = 200 V. Worst case, assume that the ambient temperature in the area is 40 deg. C. Thermal resistance, junction to ambient, is 100 Kelvins/W. So our 1/2 W, with a 40 deg. C ambient, gives us a 90 deg. C junction temperature, which is below the 150 deg. C maximum temperature, so it won't burn up, but it will get quite hot. Touch it and you could burn yourself, so you'll need to enclose it, and you'll need air vent slots or the diode's ambient temperature will likely rise well over 40 deg. C.

So far, so good. But since each diode only drops 1 V, we'll need a chain of 12 diodes each way to drop the voltage 10%. Each way. 24 1N4003 diodes can be had from Digi-Key for a total of $3.26.

You could try it instead with a pair of 11V Zener diodes connected back-to-back (11V reverse voltage on one diode + 1V forward voltage on the other = 12 V total drop). Unfortunately, the best I can find is the 1N5348, but it's only rated to 5 W, so you'll need a heat sink. You're dissipating almost 7 W per diode, so you'll need a thermal resistance less than (50 K / 7 W = ) 7 Kelvins/W. Looking at Figure 1 (on page 2) on the data sheet, we see that we'll need to attach the heat sink (a rather big one, since a small one will add more thermal resistance) right where the lead enters the package. Maybe we'll have to put a fan on it? At this point, I'd say this was unworkable. And don't say you can put more diodes in parallel to spread out the power; because of production tolerances in Vr and Vf, it is very unlikely that the current will be shared evenly; a 90%/10% current split wouldn't surprise me.

Conclusion: It would be far easier, and cheaper, to buy a 90 W light bulb.

And, while it may seem obvious, I have to say for safety's sake: Don't try this with fluorescents or other non-incandescent bulbs.

Yeah, no kidding. Smells like another Phoenix/Firebird/Firefox to me. For instance:

Yes and no, depends how your operating the transistors. For example, ECL (Emitter-Coupled Logic) runs quite fast and doesn't saturate the transistors, contrasted to what TTL does. By not saturating they're able to switch states quite quickly, but they dissipate power like crazy. As of 7 years ago you could easily find ECL lines (For example this AND/NAND chip can work at least to 3 GHz. This is a discrete component, so you can do logic this fast onto the pins.

But the trick is to exploit Shannon's theorem, and possibly work in base 4, base 16, or similar. You obviously need a higher SNR, but you won't need to clock as fast. Of course designing for base 2 is hard enough, base-4 components would be really difficult, and you'd have to come up with quite clever designs.

More interestingly it might be possible to have each 'bit' ride on a Microwave or higher carrier frequency, with the digital information modulating it. This way you could employ dense wave-division multiplexing, like in communications, to have multiple bits riding on each carrier line. Of course you'd need to design microscopic receivers/transmitters/processors to work on these signals, but it might be possible. The trick would be keeping the CPU size small, such that the registers/ALU/cache can all communicate with each other at a decently fast clocking rate (obviously limited by speed of light).

If your laptop has a vehicle power cord for charging from a car, you can use it. What you need to do is three-fold:

1. Rectify the AC from the bike. Your AC alternator output is not a problem for a full-wave bridge rectifier. You'll probably want to add the filter capacitor in the last graphic. Yes, this means that your positive and negative will not be referenced to the bike frame. So what? The computer's case is plastic.
2. Limit the current so that the alternator is not overtaxed. This is easily done with a resistor, though you will have to get something with a high enough power rating that it won't burn out. More on this below.
3. A voltage regulator to prevent over-powering the electronics. This is probably not an issue for most circumstances, but you want to be careful.

Next, the resistor. You want to limit the maximum current so that the bike will keep running even while the laptop is pulling as much current as you'll let it. Suppose you want to limit the power to 20 watts peak, with the alternator cranking out 17 volts and the laptop pulling enough current to bring its end down to 10 volts. Your current supply from the alternator is 20 watts/17 volts = 1.18 amps. Two diode drops from the alternator gives you 15.6 volts peak; assume another 0.7 volts for a low-dropout linear regulator and you get 14.9 volts, and the difference between 14.9 volts and 10 volts is 4.9 volts. 4.9 volts/1.18 amps = 4.2 ohms. Peak dissipation will be about 6 watts, though the average will be lower. If you can get a 4-ohm, 5-watt resistor and put it somewhere that it gets cooling air, you should be fine.

Last, the regulator. Wiki wimped out and failed to put a schematic of a linear regulator on-line, but I found this data sheet for a 3-terminal version. Whether you go with a 3-terminal regulator or roll your own with a power transistor (the simplest usable circuit has all of five components), you'll want to get about 12 volts out. Feed this into your laptop's vehicle power cord, and you're cooking with gas. Oh, and don't forget the fuse! An indicator LED would be a nice touch, but isn't essential - if you do this, run it from the regulated output so you can verify that power is getting to the end. For bonus points, put a 9-volt zener diode in series with the LED so that it doesn't light up until you see close to 11 volts at the output. If you do the LED trick, you'll need to adjust the value of its limiting resistor appropriately; if the regulator puts out 12 VDC and the LED's on-voltage is 10.7, you'll want a ~65 ohm resistor to limit the LED current to 20 milliamperes. This value is not sensitive, and a 1/4 watt resistor should be plenty.

Caveat 1: The series resistor and the regulator will dissipate significant power, and will need to be heat-sinked. You may want to build this affair into a small electronics box and mount those components onto a finned aluminum heat sink forming one side. Mount this where it will get some cool air.

Caveat 2: You'll want to check charging performance before you depend on this thing. Run your laptop down, plug it in and go for a ride.

A few years ago (in the wake of the Iridium fiasco) they already spun off part of their Semiconductor Products Sector (specifically the division that made discrete components, SSI glue logic, power electronics, and similar stuff) into the company that eventually became ON Semiconductor. Now the rest of SPS is following! At this rate, what will be left of their company?

In the last few years logic speeds have approached this, and you can now by GHz-level gates as part of the ECLinPS family.

Ah, crap. I pulled that link out of my bookmarks, and it's out of date now. For the datasheet, go here: http://www.onsemi.com/pub/Collateral/MC33441-D.PDF

Let me prefix this with I fully agree that the closer you can get to a True Sine Wave UPS, the better. However, I fully feel that my original comment is correct and true.

A couple of Data Points:

Almost all UPS manufacturers include a Equipment Protection Warranty. If the UPS's bad waveform fries your equipment they will replace it. See http://www.apcc.com/support/service/equipment_prot ection_policy.cfm for an example. If the pseudo-sinewave that they put out really caused that much grief do you think they would warrant your equipment against damage up to $25,000?

Also, the first thing all modern switching power supplies do is to rectify the incoming AC and then filter it with a capacitor, ending up with about 370 or so volts DC, which it then chops up to produce the lower +- 3.3-12 Volts used in PC's. The resulting output of the rectifier and Filter Cap (and power factor correcttion, EMI/RFI filtering, etc. etc. etc.) is virtually identical regardless of the input waveform - whether sine or pseudo-sine. I would have to do some nasty math to determine which waveform would be better - my guess is that it would be a toss up.

For a more technical information than you probably ever wanted to know about switching power supplies, I wholehartedly recommend the ON Semiconductor SWITCHMODE Power Supply Reference Manual at http://www.onsemi.com/pub/Collateral/SMPSRM-D.PDF

I'll also stand by my statement about motors. I'll restate to make myself more clear: Switchmode Power Supplies and UPS's get along just fine. Anything else you might have problems with.

I seem to remember that Noise Killer used to be a fairly popular product back in the 80's. (At least in Europe) This temperature controlled fan voltage regulator could turn most PC "tractors" quiet.

A) Extremely high temperatures and

B) Power supply maximum load.

Regulating the speed according to temperature makes a lot of sense, since these extremes are rarely encountered. In most systems this would also prolong the life of the fan.

Looking at their webpage , I find them quite expensive, and wonder if anybody has found less pricy alternatives?

Other than that, using a resistor to control fan speed can be tempting, but because of the relatively high start current of the fan, the fan might never be able to get started. A voltage regulator is therefore a much better choice.

However, I did at one point successfully use a 100ohm resistor to slow down a small, particularly noisy fan at the front of an old CD-R. To ensure that the fan would start, I simply mounted a 1000uF capacitor in parallel with the fan, which gave it the kick it needed to start.