Ldanielrosa

There is a safe way- in a grounded steel box that also contains the device being powered with no wires leaving without being isolated. If there is no denying that you need a small, light power supply for more than 30V and less than 30mA then build one. If you do then be sure that you use insulated knobbed banana jacks (or something similar) to reduce the shock hazard as much as possible.

Be aware that connecting neutral to earth at the device, as advised in Microchip's appnote, is likely to cause a GFCI breaker to trip.

MP, I reccommend both wires regardless of the configuration (hot-neutral-ground or hot-neutral/ground-hot) because this is for a single phase bridge rectifier. This means that on the negative half cycle the "hot" input will be clamped to ground while it pushes the "neutral" positive. This requires both inputs to be isolated. If it were for a voltage doubler, then the neutral could be left alone- but I wouldn't advise it. There is no guarantee good enough that the socket was wired correctly. Only on a device with no chance of human contact and no need of ground would one capacitor (and one fuse) do. Audioguru, you're right- it is an imperfect solution. One (possibly arrogant) assumption I make is that size and weight are issues because the device being powered is small, light, and possibly portable. If this is the case then hopefully it will not require a lot of power. Kevin Weddle, I'd set the limit at 10uF. Anything beyond this and the capacitors will be too expensive and hard to find. Not to mention that they may be bigger than the transformer they replace. Siddarth, the chief concern is what you are powering. A zener regulator is best designed for one device only. The calculations can't begin until the requirements are known. Also, with a limit of 10uF on the capacitors, you'll be limited to about 40mA.

Have you considered using a microcontroller for this?

If you can manage it, you should consider putting a capacitor and a fuse inline with both mains leads between the cord and the rectifier. This will offer some safety margin (not enough to bet your life on) and allow one end of the output to be tied to ground.

It depends on what you're using it for. If it's for PWM control of a motor, then 1kHz may serve you. It can be divided from most oscillator crystals that are likely to be used with uCs (sorry, lazy- didn't look for the code for 'mu').

Wow, what a mouthful! But seriously, I've looked for articles on this by the name above and by binary chain decoding. I haven't found anything written for my level (lower than it may seem). What drives me is that an encoder wheel of 2^n bits could be fashioned such that reading n bits would give the position and would only need one ring instead of n rings. I know I'm getting way ahead of myself here, but the idea seems so elegant. Comments?

Good morning MP, and thanks for showing interest in my (possible) future project. I guess I'm having trouble communicating clearly, and made the classical (arrogant) mistake of trying to meter the information to narrow the response. Last question first. I'll be using a 16c73a. I obtained about eighty that have been used, but can be overwritten and are still servicable in a limited capacity. I figure I'll have enough resources on board to take care of four steppers, two on port b and one each on ports a and c. I'll be able to keep and update a profile on each motor (maximum speed, acceleration, run duration in steps or cycles, slave to another motor, etc). It'll definitely be slow by anybody's standards, at 1k steps per second, but should suffice for my (yet to be) needs. What I'd really like to see is how others structure the data they send to their controllers. I'll write my own protocol if I have to, but I'd rather use one that's established.

By protocol I mean that I'd like to know how the request/command is structured for a serial bus, and what all information it includes. For servos I found a three byte protocol, the first of which only synchronizes the bus. I'd like to see more. For steppers I hope to find one or more protocols that include (at the very least) speed, direction, duration, motor ID, etc. Compact is good. I'm sure that whatever I come up with will cover most of my needs, but I'd like to use an established protocol for portability.

Hey folks! I'm thinking about making a dedicated controller for steppers and another for servos. I'm having trouble wading through the articles and advertisements to find the protocols being used. I did see one on servos though. Links are best, thanks.

You might look into using an smps. Check out the "black regulator" at http://www.romanblack.com/

Second question: Yes. First question: Three external components help with the task of step-up voltage conversion. One inductor, one capacitor, and one schottky diode.

It doesn't get much simpler. The chip can be a 40106 with minor changes. There is something that uses all discretes at http://www.romanblack.com/ One downside to these designs is that the frequency is not controlled, and as such may have some annoying side effects. For a motor control this is not likely to be a problem. Another point lacking is that interfacing these to some automatic control will be difficult.

Microchip and Maxim give away free samples. Just answer their questionnaires honestly. Unfortunately the packages are all SMT. As for the discretes, if you can get some used disposable cameras with flash units (some places will give them away since they're garbage) the charging transistor is ideal. The toroid core is not critical. One place I neglected to mention is http://209.67.62.186/ubbthreads/ubbthreads.php Here you can find people whose passion for LED flashlights borders on psychosis. There are articles on many design topologies that surface frequently. If your locale is really starved for parts connections, I may be able to help if the postage isn't too high. PM your address and I'll look into it.

For driving white LEDs from a lower voltage than Vf there are chips from Maxim, Microchip, and (probably) National. You'll have to do a parametric search for them. http://www.maxim-ic.com/index.cfm http://www.microchip.com/1010/index.htm http://www.national.com/ There are also discrete component circuits for them. These have several topologies, each suited to specific needs. http://elm-chan.org/works/led2/report.html http://members.shaw.ca/novotill/LedSupplyRegAA/index.htm http://www.satcure-focus.com/hobby/page2.htm http://www.elektor-electronics.co.uk/miniproj/download/010130uk.pdf The first three are variations on a blocking oscillator. The last one is different- it does not require a tapped inductor, but does need two transistors.

Good morning Mixos. Either my groggy eyes deceive me or the colors aren't rendering correctly on my computer. Your entry for 100K looks like1G0. Also you showed 010 for 100R. You may want to put a "G" for the third band on the sub-10 ohm values.

I'm surprised there aren't alternatives for tilt switches in your neighborhood. I have trouble finding a catalog that doesn't have something in this category. Granted, a steel ball rattling about inside a cage won't make as good a connection as a puddle of mercury, but it's something to start with. It does bring up a puzzle though. It'll have nasty bounce, so you'll have to write some pretty aggressive software to figure out what's being done to the device.

Yes to the last one. The circuit is more complicated. The term "negative resistance" is used to describe the behavior (oops, most of the folks here are in Europe, so that's behaviour) of the device as the voltage (or current) varies within the range so described. Instead of voltage and current increasing or decreasing together (linearly or otherwise), and increase in one is the cause or result of a decrease in the other.

A higher rated bridge will more easily tolerate voltage transients and startup surges. Also it may have a slightly lower voltage drop.

Greetings asciifish! Your profile doesn't say where you hail from. I burned a few iron-on transfers for a JDM type programmer (serial port). They're not tested yet, but when they are I'd be glad to send one to you if the postage isn't unreasonable.

Probably best to use an instrumentation amplifier. These can be a bit spendy, but one can be cobbled from three opamps (and they'd best be from a quad for better device matching). From that a current loop driver would be good, and an optoisolator too. I'll try to find an illustration I have laying about this afternoon, but I'm off to work in a few minutes. Okay, I'm a day later than I promised. The image is from page 1-19 of Designer's Handbook of Integrated Circuits by Arthur B. Williams. The relevant text indicates that the gain will be [2(R2/R1) + 1](R3/R4). "Resistor values should be kept low to minimize DC offset." Another illustration shows 10k in positions R2, R3, and R4. The gain can be varied with R1 alone. Pretty neat, I think I'll use it on something. I hope this helps.

All of the capacitors in a voltage multiplier contribute to the impedance. Until I read otherwise I'll go with my belief that they work best when all of the capacitors are of the same value.

Disclaimer: This is somewhere between conjecture and fantasy, so a schematic of my ideas is a bit difficult to produce. That much taken care of... russlk: I'm constrained by a few realities. The wire size is about 22ga, and I'd like to locate some gadgets 50m or more from the power supply. Some of them may have devices that want a higher voltage than 5V, or a split supply. If these are _ever_ subject to inspection, I want to keep _all_voltages low enough to be exempt from permits/regulations. I'd also like to load the power supply evenly regardless of the device in question. all: The last point gets a bit sticky. To keep true with this objective, I may have to run uCs on a

I'll confess that I haven't seen more than a dozen articles so far, but from what I have seen it seems that 38kHz is the most popular. Why? Did I miss something? Left to my own devices, I'd've picked 40kHz because a uC running at x.000 MHz can get the right frequency dead on. Also this allows a preexisting tv remote to be used (I'm under the impression that the appliances here use 40kHz, I may be mistaken).

MP: I understand that filtering can remove a lot of interference. I'm hoping to make that unnecessary by aligning the bits (timewise) with the center of the flattened peak of the power (when the ringing is the least). Staigen: Yes to the first question. At the moment this is conjecture. Since the devices would be remotely located, trying to supply 5VDC to all of them from a central supply is unrealistic. Since I may want to use some analog components, a split supply may be necessary. Raising the frequency allows for smaller capacitors.