 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| HomeDiscussion GroupsElectronicsBasicsRepairDesignCADComponentsEquipmentElectrical Engineering ElectronicsKB.com Contact UsLink To UsSearch & Site Map |
Electronics Forum / Basics / August 2007Simple Linear Supply (for teaching my son)
I picked up a few 16VACrms, 1Amp transformers from a junk table at a local supermarket store (I think they are used for doorbells and the like) for $2/each, new. Cheap. Couldn't resist. Anyway, I just do this stuff for a hobby but I'm thinking of teaching my son some of the design issues for a completely discrete, bjt-only, linear power supply that supports variable voltages from something just under 1VDC to perhaps maybe 12V, at no more than say .6-.7Amps. Here is the schematic I came up with to teach from: http://www.infinitefactors.org/misc/images/ps1.gif I intend to start by using a 3.3k ohm resistor for the "BASE DRIVE" circuit, instead of the current source shown. Just by way of showing that it can work either way and some of the trade-offs. I also will start without including the current limit circuit, but adding that in later on. Then once we get the current limiter put on the high side, I'll also do the same thing for the circuit but instead linking the current limiter into the ground return instead (and change where the voltage control circuit 'grounds') so that I can illustrate its operation in either configuration. Might even make this into small, separate component boards that can be linked together in pieces. That way, these steps will be easier to add and pull apart. I'd like any thoughts, though. Questions asked here would help me make the presentation better, later on. Or perhaps I'm not thinking about something I need to deal with. Thanks for any thoughts, Jon > I picked up a few 16VACrms, 1Amp transformers from a junk table at a > local supermarket store (I think they are used for doorbells and the [quoted text clipped - 26 lines] > make the presentation better, later on. Or perhaps I'm not thinking > about something I need to deal with. Sounds like a plan, but substitute 1 amp rectifier diodes like 1N4001 (or 1N4002, 1N4003, 1N4004 or a 1 amp packaged bridge rectifier) in place of the 200 mA 1N4148 shown on the schematic. Those tiny diodes might blow just charging the storage capacitor the first time. >> I picked up a few 16VACrms, 1Amp transformers from a junk table at a >> local supermarket store (I think they are used for doorbells and the [quoted text clipped - 32 lines] >schematic. Those tiny diodes might blow just charging the >storage capacitor the first time. I was just using those as placemarkers for the schematic and forgot to change them out before posting it up. Thanks for the reminder, though. I'm thinking of two different "blocks" for the drive supply -- one is just a 3.3k resistor and the other is the current source shown. Each of these can be swapped in. Also, two or three blocks for the voltage limit -- one a fixed voltage one using a pair of resistors and a bjt, then one adding the ripple compensation cap only, and then the one as shown on the schematic. Those can be likewise swapped. Finally the current limit block, just to remind, can be attached either on the high side or the ground side (with slight other wiring differences.) What do you think of the schematic layout? Are the purposes and blocks clear? Thanks, Jon >>> I picked up a few 16VACrms, 1Amp transformers from a junk table at a >>> local supermarket store (I think they are used for doorbells and the [quoted text clipped - 49 lines] > Thanks, > Jon Jon, there are lots of good 'bad ideas' to be improved on there :-) so this indeed is a good circuit... to start with. * Temperature sensitivity. How to improve his? * Then, how to improve this while still satisfying the 1V min requirement? * What will happen to Q3 if you suddenly dead short your supply while it's set at 12V output? * What will happen to Q4 if you suddenly dead short your supply while it's set at 12V output? * Is Q2 sized adequately? * What is the transient resposne of this supply? * Q1 will benefit from a E-B 1K or so resistor. >>>> I picked up a few 16VACrms, 1Amp transformers from a junk table at a >>>> local supermarket store (I think they are used for doorbells and the [quoted text clipped - 53 lines] >there are lots of good 'bad ideas' to be improved on there :-) >so this indeed is a good circuit... to start with. My son basically understands resistors, Ohms' law, less-than-EM1 hybrid-pi model of BJTs, a little about inductors (the differential equation) and a little about capacitors (also, the differential equation.) He knows about Thevenin but hasn't really had to apply it much, and knows about the two basic forms of Kirchoff's somewhat-obvious-today laws, though again he hasn't had to apply them all that much. He also has a passing exposure to ideal opamps. One of the more important things I'm trying is to avoid hitting him with too many details and to allow him to see just how much reasonable performance can be achieved with only a little bit of quantitative theory and knowledge. Although I want him to see this in the broader behavioral blocks (excluding having to look into exact details of each), I also want him to be able to quickly approximate values within a block almost on sight and without the need for much use of a calculator. I also want him to see what can be done with just a few simple components, which is part of why I'm completely avoiding anything that looks like an IC. (No MOSFETS, for now, too.) Plus, I like being able to swap in supposedly improved bits and pieces and see how they help (or hurt) things. He and I have already covered, in some detail, the issue of a bridge rectifier and following capacitor -- including what happens to the peak current when you just hang on a huge capacitor, because of the much shorter moment available then for recharging it (load current remaining the same.) Once we get through some variations on this theme, and I feel he has mastered the details well enough, that will be the time to start asking some new questions to complicate the result a bit. >* Temperature sensitivity. How to improve his? A large source of heat is with the 2N3055 and keeping it away from other elements will be helpful. The other BJTs won't heat much on their own -- I think they are probably in the few tens of milliwatts, or so. I expect some dissipation in the bridge. All BJTs have a lower Vbe with higher temperature, I think. Perhaps dropping 0.23V in 100C change, just going from some vague experience I recall. The main problem will be with the voltage comparison, as it entirely depends on this Vbe as a reference value. As this Vbe goes lower with higher temp, the output voltage will be driven lower as a result. The output is about Vbe(Q3)*((R3+R4)/R5+1) so as Vbe(Q3) drops with rising temp, it directly affects the output voltage, dropping it as suggested in the formula. With R3=10k, this would be some 2.8V change. At R3=0, the change would be only .2V. The current source is a BJT, too, so the current there also goes down with lower Vbe -- higher temperatures would tend to reduce the source magnitude. But I set the base drive current source far above the needed drive current, more than double what I expect to use. It would be trivial to increase that even further to provide more margin. Most of the improvement here would come from improving the voltage control section. (Assuming one can live with the current limit section's drift over temp.) >* Then, how to improve this while still satisfying the 1V min requirement? Actually, I'd like to hear from you about this. One quick possibility comes to mind, by adding a couple of diodes and capacitors to the existing bridge to add a negative rail for some headroom. >* What will happen to Q3 if you suddenly dead short your supply while >it's set at 12V output? >* What will happen to Q4 if you suddenly dead short your supply while >it's set at 12V output? C2 (now that I look, I see I didn't renumber the capacitors... oh, well) will deliver a bit of a short current pulse. But are you referring here to the equivalent C from emitter to base and base to collector and the stored charge there at the time? Or? >* Is Q2 sized adequately? Estimate beta=30 at .5A-.7A for Q1. Call it 600mA/30 or 20mA. But probably less, really. Q2 will be at a beta of about 200, or so. So we are talking about 20mA/200 or about 100uA. Again, probably less. I think the 2N3904 can handle the 20mA collector current just fine. >* What is the transient resposne of this supply? That would be well beyond where my son is at and probably something for which I wouldn't mind an education and would likely miss some important considerations as I went through what I know. Remember, I've had no training whatsoever. It's just something I enjoy thinking about, at times. >* Q1 will benefit from a E-B 1K or so resistor. Yes! Jon >I picked up a few 16VACrms, 1Amp transformers from a junk table at a > local supermarket store (I think they are used for doorbells and the [quoted text clipped - 29 lines] > Thanks for any thoughts, > Jon I have read the other comments as well. But some suggestions I have are: 1. The current limit as shown requires a power pot. You could use a single sense resistor and a small pot to adjust the limit, but the resistor would dissipate more power at higher current levels. For a variable limit, it's probably best to use a small sense resistor and low level comparators or op-amps, 2. High-side series pass elements are more easily implemented with PNP devices, although they are usually more expensive and not as rugged. The base drive is then just a current path to the common return. 3. You can get very good regulation by using a zener reference and a simple op-amp type circuit comprised of BJTs. I have attached an LTspice model of a simplified version, and I used an AC source for the reference, but you can substitute a zener and a pot. I have left off capacitors and other components that might be necessary for stability, but this circuit should work well, and can be modified with higher gain to produce almost any level of regulation desired. It does not work too well at very low output voltages, but that can be fixed if needed. (Hint: add a power diode). Paul ASCII file follows: Version 4 SHEET 1 880 680 WIRE 240 -128 -144 -128 WIRE -144 -64 -144 -128 WIRE -144 -64 -320 -64 WIRE 16 -64 -144 -64 WIRE 368 -64 112 -64 WIRE 432 -64 368 -64 WIRE 464 -64 432 -64 WIRE 368 -16 368 -64 WIRE 64 96 64 0 WIRE 240 96 240 -128 WIRE -320 112 -320 -64 WIRE 0 144 -160 144 WIRE 368 144 368 64 WIRE 368 144 304 144 WIRE 432 176 432 -64 WIRE -48 224 -160 224 WIRE -16 224 -48 224 WIRE 64 224 64 192 WIRE 160 224 64 224 WIRE 240 224 240 192 WIRE 240 224 160 224 WIRE 160 272 160 224 WIRE -160 288 -160 224 WIRE -320 400 -320 192 WIRE -160 400 -160 368 WIRE -160 400 -320 400 WIRE 160 400 160 352 WIRE 160 400 -160 400 WIRE 432 400 432 256 WIRE 432 400 160 400 WIRE -320 448 -320 400 FLAG 464 -64 Vout FLAG -48 224 Vin FLAG -320 448 0 SYMBOL npn 0 96 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL npn 304 96 M0 SYMATTR InstName Q2 SYMATTR Value 2N3904 SYMBOL pnp 112 0 M270 SYMATTR InstName Q3 SYMATTR Value BCW68G SYMBOL res 144 256 R0 SYMATTR InstName R3 SYMATTR Value 100 SYMBOL res -176 128 R0 WINDOW 3 -26 67 Left 0 SYMATTR Value 1k SYMATTR InstName R1 SYMBOL res 352 -32 R0 SYMATTR InstName R2 SYMATTR Value 1k SYMBOL res 416 160 R0 SYMATTR InstName R4 SYMATTR Value 5 SYMBOL voltage -320 96 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value 15 SYMBOL Misc\\signal -160 272 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V2 SYMATTR Value SINE(2.5 2.5 60 .02) TEXT -288 504 Left 0 !.tran .1 >>I picked up a few 16VACrms, 1Amp transformers from a junk table at a >> local supermarket store (I think they are used for doorbells and the [quoted text clipped - 35 lines] >sense resistor and a small pot to adjust the limit, but the resistor would >dissipate more power at higher current levels. I'd be using a wire-wound 10 ohm. They are available for about $3, but I might also just wind one myself and make a wiper for it. (I used to make my own diodes for my crystal radios using galena crystals I'd sink into molten lead. And I tried my hand at winding wire for earphones using cylinder magnets and small bits of steel as diaphrams. Worked, though not as good of course as those I could buy.) >For a variable limit, it's >probably best to use a small sense resistor and low level comparators or >op-amps, I am intentionally avoiding opamps. I have already spent some time with him so that he understands their idealized operation (trivial to explain), but before he begins to use them in earnest I would rather that we spend some time with a differential pair of BJTs and gradually grow into more complex arrangements (emitter current sink for the npn variety and the idea of using a current source on a collector leg to increase the useful gain, etc.) >2. High-side series pass elements are more easily implemented with PNP >devices, although they are usually more expensive and not as rugged. The >base drive is then just a current path to the common return. Yes. But this was very easy to explain as a start and, also, the arrangements for controlling the voltage from the ground reference work as simply designed negative feedbacks against the current source at the top. If I were to use a PNP (aside from the 'more expensive' and 'less rugged' issue), I'd probably need to add more parts and the details needed to fully apprehend the arrangement would increase, as well. >3. You can get very good regulation by using a zener reference Indeed. But I eliminated that for several reasons. One of them was nothing less than it was fewer parts to start. Simply less to deal with, at the outset. I don't want to teach by including at the beginning everything required for a "good design." Just one that is workable, from a beginners' point of view. Understand the basic structure, first. Then worry about adding detail later on. I also like to learn by seeing what doesn't work so well, too. If you have something simple to start and then you discover a few "problems" with the result, you have some motivation for finding something a little better to fix that without breaking other things in the process. Then as you get more sophistication, it all makes good sense in the end. Too much at once reduces motivation to understand the quantitative details because they are so daunting. >and a simple op-amp type circuit comprised of BJTs. Yes. We will eventually get to making up an opamp out of discrete parts. Once he has a basic idea of that, I won't have much problem with just using commercial opamps for later things. >I have attached an LTspice model of >a simplified version, and I used an AC source for the reference, but you [quoted text clipped - 3 lines] >of regulation desired. It does not work too well at very low output >voltages, but that can be fixed if needed. (Hint: add a power diode). Thanks. I have done such things and have often added another diff amp section to get good performance. There is a temperature issue that needs compensation also (the diff amp Vbe), when I use a zener reference to one of the bases of the diff amp. I usually just add a diode elsewhere to get a matching effect to compensate for that. I want to avoid a design like this, though, at first. Listening to me about the diff amp would make him "go to sleep," just now. Got to start a little easier. Jon >>2. High-side series pass elements are more easily implemented with PNP >>devices, although they are usually more expensive and not as rugged. The [quoted text clipped - 24 lines] > in the end. Too much at once reduces motivation to understand the > quantitative details because they are so daunting. When I was a teen, my father had me build a simple power supply with a couple of 2N3055s with the base controlled by a pot across a zener, in an emitter follower configuration. I still have and use the supply today, although it could benefit from more modern design. It is about as simple as can be. I don't quite follow the logic of omitting a zener to reduce parts count, yet using a forward B-E junction as a reference. The whole circuit would need to be redone to do it right. I would just use a pot with a series resistor across the raw supply at first, to demonstrate adjustability and show the poor regulation. Then add a capacitor across the pot to show capacitance multiplication by using the gain of the BJT, and then add a zener for voltage stabilization. The emitter follower still suffers from poor load regulation without direct feedback (you have indirect feedback via the relatively constant B-E drop). So you add a diff-amp with output voltage sense, and then you have a stable and useful supply. Add current limiting, then make it fold-back, and you have a good all purpose supply. After that you can go further and design for higher power and greater efficiency by adding an SCR phase-fired front end to optimize the voltage drop across the series pass transistors, and then you will need to add inductors to minimize switching spikes. Also, as the design progresses in complexity, things like transient response and feedback can be learned. Good luck. A power supply is a solid basis upon which to build. Paul >>>2. High-side series pass elements are more easily implemented with PNP >>>devices, although they are usually more expensive and not as rugged. The [quoted text clipped - 30 lines] >although it could benefit from more modern design. It is about as simple as >can be. I've done that one to death with him, already. It was my first shot at this and he "gets it" very well. >I don't quite follow the logic of omitting a zener to reduce parts count, >yet using a forward B-E junction as a reference. The whole circuit would >need to be redone to do it right. Because it works sufficiently to show the operation of a summing junction. Which is the next step in what I'm trying to show. I could be wrong about that approach, though. So as always, I love to have informed criticism about anything I'm trying to do. Thanks for your kind time and that of others here, as well. Much appreciated. >I would just use a pot with a series >resistor across the raw supply at first, to demonstrate adjustability and >show the poor regulation. Assuming you are talking about "with a bjt," here... done that. Well, actually I didn't need to. It was easy enough to show by equation form and he is gaining proficiency at differential equations so he had more than enough math skill to have no problem seeing the details just from this relatively simple algebra (ignoring details like temperature variations.) >Then add a capacitor across the pot to show >capacitance multiplication by using the gain of the BJT, and then add a [quoted text clipped - 3 lines] >voltage sense, and then you have a stable and useful supply. Add current >limiting, then make it fold-back, and you have a good all purpose supply. I'll take this advice. Worthwhile doing. >After that you can go further and design for higher power and greater >efficiency by adding an SCR phase-fired front end to optimize the voltage >drop across the series pass transistors, and then you will need to add >inductors to minimize switching spikes. Also, as the design progresses in >complexity, things like transient response and feedback can be learned. By that time, I'll be learning as I go, as well. These are things I yet need to apprehend well and haven't had time in my own hobby life to explore well. But that's part of what makes this a good thing for both of us to share. At some point, I'm pushing myself to keep up some pace and we both learn together. >Good luck. A power supply is a solid basis upon which to build. I think so, too. There is plenty detail to master in something as "simple" as a linear power supply. You can dig deep and continue to find nth-order effects to try and cope with until you lose interest, I suppose. Thanks for the thoughts and additions. Appreciated. Jon >[...](I used to make my own diodes for my crystal radios >using galena crystals I'd sink into molten lead[...] Aaaaaaa! That's even worse than using solder. http://www.google.com/search?q=cache:2VojQZ0mo_gJ:en.wikipedia.org/wiki/Cat's_wh isker_diode+*-*-high-temperature+501+Wood's.metal+can.damage.many.crystals http://www.google.com/search?q=define:Wood's+metal >>[...](I used to make my own diodes for my crystal radios >>using galena crystals I'd sink into molten lead[...] > >Aaaaaaa! That's even worse than using solder. I used what I had that was cheap. At the time you could buy 5lb semi-round blocks of lead for about half a dollar. Used to use them as door stops, in fact. Jon > I picked up a few 16VACrms, 1Amp transformers from a junk table at a > local supermarket store (I think they are used for doorbells and the [quoted text clipped - 29 lines] > Thanks for any thoughts, > Jon This may just be me, but it seems a little more intuitive if you show the voltage control block to the right of Vout with its output feeding back to the left. Voltage control happens after Vout - it seems to me that putting it on the right of Vout reinforces the idea that it responds to Vout before Vout responds to it. Ed ><snip> >This may just be me, but it seems a little more intuitive [quoted text clipped - 3 lines] >right of Vout reinforces the idea that it responds to Vout >before Vout responds to it. I thought that, too. And in fact, my first incarnation of it had it that way because that is how it is often shown. However, I think this way is a little better because the "wire" gets in the way of understanding. One thinks too much upon what is connected to what and doesn't realize that it is all separable as components. So breaking the wire and using a symbol instead allowed me to focus the attention on each block better and what it examines and delivers. In any case, my preference is NOT to show a wiring diagram but a functional one. Avoiding showing bused wires where they don't add meaning is only one pet peeve of mine. Another is that I like to show electron flow from bottom of sheet to top (no side-directions unless I can defend it well) and to always show signal flow from left (input) to right (output.) In this case, there is feedback so there must be a connection from the output to the input (backwards direction.) The way I arranged this keeps the left-to-right signal flow arrangement, but just doesn't show the actual wire going backwards. But I removed it on the idea that the actual wire is more distraction to the eye from the block function than it is a contribution to understanding. But perhaps that is just me. Jon > This may just be me, but it seems a little more intuitive > if you show the voltage control block to the right of Vout > with its output feeding back to the left. Voltage control > happens after Vout - it seems to me that putting it on the > right of Vout reinforces the idea that it responds to Vout > before Vout responds to it. Both voltage and current control are feedback loops, so it cannot be done that all signals flow from left to right. There has to be a jump back from right to left, somewhere. [story snip] > Or perhaps I'm not thinking > about something I need to deal with. > > Thanks for any thoughts, > Jon Just a basic querry: Is it _you_ or the child who wants to engage in this. "You can bring the horse to water............" and so on. Let him do it his own way and wait for questions. Have fun Stanislaw >[story snip] >> Or perhaps I'm not thinking [quoted text clipped - 4 lines] > >Just a basic querry: Is it _you_ or the child who wants to engage in this. Actually, it's the style I've been using with him all his life. I work at home, so whenever he has any question or interest I drop what I'm doing and give him whatever time he wants. As soon as his interest wanes, I drop it and go away. He's old enough now that he drives this process more than before and can stay on a subject much longer (an hour, sometimes two.) Sometimes, he would come home from school with a question his teacher couldn't answer on physics and we'd go over the details. His interest didn't wane a bit, then. Yes, this is _my_ hobby, too. So I'm not going to deny an interest, myself. But this is his first year out of high school and he asked me if he could take electronics at college. So actually he is the one driving me to produce something to discuss. However, what that subject actually is comes from me. >"You can bring the horse to water............" and so on. >Let him do it his own way and wait for questions. I think that's what I've been doing. But I am the one who knows what he knows and what he can handle learning in spurts and this particular project is at the right size for all the circumstances and to keep his interest, without overwhelming him. I am glad you are pressing that button, because there are a lot of people who need that kind of kick in the pants. But in this case I'd just say that I'm aware of what you are talking about and keeping it ever in mind and I think I'm being sensitive to him and not letting my own inclinations or interests blind me to what is going on with him. Thanks, Jon >> [story snip] >>> Or perhaps I'm not thinking [quoted text clipped - 35 lines] > Thanks, > Jon Good luck! Stanislaw |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.