CopyCat PSU with only NPN and op amps.

[novice_nerd]

So, I have the cash to just buy electronic test tools, however I'm trying to learn and build useful things.

To that end, I've used the PSU video 1 to copy/draw a functioning circuit in falstad/circuit.

I skipped around in the video, I've watched it all twice at least, I even added another op-amp to it for a current limit LED.

So my question is, if I were to try and build this thing, clearly my lack of knowing what op-amp to select, or even what transistors to select is my problem. In addition, I just learned that op-amps have bias control, etc etc. Things an "ideal" op-amp doesn't include.

I'm also curious where the heat would be dissipated? I assume it must be in the darling pair (I think that is what it is called.) Seems that there isn't much current moving anywhere else in this thing... I also selected some CRAZY 1meg resistors for my voltage dividers etc... Not really knowing why you wouldn't just select very high resistance, to keep the current flow to a minimum. (Unless of course the IC needs some minimum to function.)

Any who... Here is the link to the mess I made, and it all functions. I only put one cap in it for output stability, not really sure if I should get some Opamps, bread it up, and test it out. (Again, I have O scope, and all kinds of tools that I don't even know how to use well yet... Just figured I'd start, and why not just go big...) I also haven't sorted out what transformer to use, how I'm going to rectify it, how to do any type of back flow protection, how to visually read the current limit, and voltage setting. (I assume a short button, then read the current output, then just use a voltage meter across the output to set voltage... Uhh, not sure really, but makes sense.)

I realize I don't need the 2 bottom op amps, if I just use a voltage divider, and create some other reference voltage somehow... on and on and on... I'm just working my way thru this mess.

Here is the link to the sim I used... http://www.falstad.com/circuit/circuitjs.html

Here is the code you can import to view the seemingly functional mess I made... (File, import from text)

Any insight is appreciated, I've been also watching all the MIT 6.002 videos, playing with arduino, and I'm making a kiln control for my wife's 40amp electric kiln. (It kinda works, I just have to write a bunch of code, tested the K-type probe up to 1,000f yesterday.)

Code: [Select]

$ 1 0.000005 0.575460267600573 35 5 50
a 288 288 368 288 1 15 0 1000000
R 144 112 80 112 0 0 40 24 0 0 0.5
w 240 112 400 112 0
w 224 160 240 160 0
w 224 160 224 176 0
r 240 160 240 112 0 1000000
a 208 176 208 272 0 15 0 1000000
r 96 160 144 160 0 1000000
r 144 160 144 112 0 1000000
g 96 160 96 176 0
r 240 160 288 160 0 1000000
w 208 272 288 272 0
w 144 160 192 160 0
w 192 160 192 176 0
w 288 160 288 272 0
a 208 352 288 352 1 15 -15 1000000
w 288 304 288 352 0
w 288 352 288 400 0
w 288 400 208 400 0
w 208 400 208 368 0
r 368 288 400 288 0 5000
t 400 288 432 288 0 1 -5.300004852573978 -0.00005785304398211861 100
w 448 272 432 272 0
w 448 272 464 272 0
g 464 320 464 352 0
a 624 416 544 416 1 15 0 1000000
172 624 400 672 400 0 6 5.3 10 0 0 0.5 Voltage
w 624 432 624 464 0
w 624 464 544 464 0
w 544 464 544 416 0
w 544 416 544 352 0
172 208 336 160 336 0 6 1 1 0 0 0.5 Current
w 544 272 464 272 0
a 464 208 544 208 0 15 0 1000000
w 464 192 464 160 0
w 464 160 544 160 0
t 576 208 608 208 0 1 -16.940548975322614 0.6113634718535987 100
t 608 224 640 224 0 1 -17.55191244717621 0.726741484774637 100
w 448 112 608 112 0
w 608 112 608 192 0
w 608 192 640 192 0
w 640 192 640 208 0
w 544 160 720 160 0
w 720 160 720 240 0
w 400 112 448 112 0
w 640 240 720 240 2
w 432 304 464 320 0
r 464 224 464 272 0 10000
r 544 272 544 352 0 10000
r 192 112 224 112 0 1
w 144 112 192 112 0
w 224 112 240 112 0
x 29 305 272 308 0 24 Current Control 1V=1A
x 647 381 811 384 0 24 Voltage Control
x 335 58 777 61 3 24 Current and Voltage limiting power supply
g 752 256 752 304 0
x 744 242 797 245 0 24 Load
a 384 320 384 400 0 15 0 1000000
w 368 320 368 288 0
162 384 400 384 448 1 2.1024259 1 0 0 0.01
r 384 448 384 480 0 770
g 384 480 416 480 0
x 173 430 363 433 0 24 Current Limit LED
r 544 208 576 208 0 100
174 720 240 752 272 0 20 0.6287 Load
c 640 240 640 272 0 0.00001 5.299880578355635
g 640 272 640 288 0
w 432 320 400 320 0
w 432 320 464 320 0
o 55 64 0 2083 0.0000762939453125 1.6 0 -1 0


[T3sl4co1l]

Sims like this make me curious what solution method (and op-amp models) Falstad uses.  It seems to be DC-correct, but I expect the real build of this circuit will oscillate.

The usual remedy is a series resistor into the op-amp -in, with a R+C from out to -in.  The RC time constants are adjusted for optimal step response.

Tim

[novice_nerd]

Is there some other type of simulator? I assume the 'better' ones actually take the spec sheet of the parts, and simulate the parts you use. (I think partsim, has a list of parts, etc.) The problem I have with those types of sims, is while they are probably more accurate, I'm not good enough to build with them. I've seen other people use LTSpice I think it was called to do their sims. I assume it's from Linear Tech (I just looked it up) and includes their parts, so you can build more accurate virtual devices with less hassle.

http://www.linear.com/solutions/4449 <-- in regards to opamp osilation.

I also just told the omp-amps in my design they could go from 0 to 15 volts. Kinda cheated and just told them they are powered by a pure perfect DC supply.

More reading/watching time I suppose.

[T3sl4co1l]

Yes, lots of depth to cover... modeling is in itself a whole career!

The important thing to remember is what a simulator is: a numerical computation, solving a system of equations step by step.

LTSpice, and all other SPICE based engines, do this by drawing up the equations for the circuit (i.e., node 1 connects to node 2 with a resistor, so there's a current from 1 to 2 that comes from Ohm's law and is I_12 = (V(1) - V(2)) / R, and so on, for all pairs of nodes in the circuit, and types of connections...), and finding that solution (through various methods).

What they do not do, is give you anything real.  The model is a purely mathematical abstraction.  It is up to you, the modeler, to create a model that is representative of reality!

LTSpice is as prone to error as any other environment; but as a starting point, at least it comes pre-loaded with reasonably accurate models for LT's parts.  You certainly shouldn't find an op-amp circuit that doesn't oscillate when it should!

Tim

[novice_nerd]

Well I appreciate the responses.

I'm 3 weeks into watching/reading everything I can on electronics.

I bought an Arduino for giggles, and realized it was trivial to hookup a probe, and a 50amp solid state relay to control my wife's glass kiln. (They sell controllers for 700+ dollars.) Almost feels like cheating it is so easy. (Until I started to have to think about coding again with glass melting schedules. blah)

Then I wanted to build a few circuits to better understand DC to AC inverts, that started me down oscillators, then PWM, then I couldn't see frequency of the oscillators I built, down and down the rabbit hole. Few hundred dollars later, and I can at least build things. I'm still just buying parts like mad, to fill up a bunch of drawers, so I can test, experiment with, and learn more.

So I guess, I'll have to keep working on this PSU, tweak some things, and build it. I keep thinking of all these OP amps, and just simple logic devices in this setup... Sure I'm measuring analog signals, but I'm not moving any current to speak of with them, nor really anything else.

Seems actually dreadfully simple, but I'm sure the devil is in the details. Low voltages on my current sense, maybe feed back and oscillations would need dealt with R/Cs.

I guess I'll just keep pushing forward...

Again, thanks for the reply, I do appreciate it.