LM317 and op-amp Constant Current Source

[Ron H]

1206DX, if I understand your post, you think you need an LM317 current source as a preregulator. You can't preregulate a current source with another current source. You WILL need some sort of voltage regulator (LM317 or other) for your op amp. And if you don't want to wait for warm up and then have to calibrate every time you turn this thing on, use a MOSFET instead of a BJT. Today's MOSFETs are pretty robust, or you wouldn't find them in practically every switching regulator on the market.
I can't see how the positive TC of the MOSFET would have any relevance to this circuit.

[Dimbulb]

Chris;
I would like to keep the lower voltage compliance, the amperage range could be less.
The LM6142 has lower Vos and Vsupply down to a battery friendly 1.8 V but it is not cheap.<p>The level shifting would still be needed why not have the LM358 operate at higher current output applying a big wire wound resistor on the output reducing current 3 to 1 and filtering at the same time.<p>
RonH;
Electronic enthusiast just won't use a slow warm-up semiconductors any more and this is a good point, what I have seen are like you mentioned 8 to 10 watt power TO-220 mosfets. IRF540 511 ect.
What about a dual mostfet using the other opamp as a pre-regulator ? <p>When you mentioned warm-up time, 20 - 30 seconds should be alright for most applications I would think. Sure takes alot of fussing to get a circuit to behave accurately.<p>[ May 14, 2003: Message edited by: 1206DX ]</p>

[analogee]

RonH: (caution! long, irrelevant post coming) your post provokes an interesting (well, to me anyway) theoretical question. I've always enjoyed thinking about 'duality' in circuits. You know, like how a current source is the dual of a voltage source; everything is kind of opposite, but similar -- a current source has infinite output resistance, while a voltage source has zero output resistance; you can't short the output of a voltage source, while you can't open the output of a current source; etc. Or, like how an infinite capacitor looks like a short circuit, while an infinite inductor looks like an open circuit. The dualities appear all throughout circuit theory.<p>So, your assertion that you can't pre-regulate a current source kind of struck me. Why not? If you can regulate voltage, why not current? I mean, I know the practical end of it; I've never seen a current pre-regulator. And if you tried to put 2 current regulators in series, unless they had precisely the same value, they would fight each other in a theoretically impossible fashion.<p>But, thinking in duals, could a current pre-regulator be possible? I think so. Think of a voltage regulator. It generally puts 2 elements in series, wherein one of them "wastes" excess voltage, leaving the regulated remainder across the other element. Straight duality thinking would say that a current regulator would have 2 elements in PARALLEL, wherein one of them "wastes" excess CURRENT, leaving the regulated remainder to go THROUGH the remaining element.<p>And that lead me to think of a differential pair. The tail current splits into 2 paths, essentially in parallel. 1 path could be the regulated current, and the other path could be the excess bleed-off. It's not important for this discussion, but you control how much goes where by the differential input voltage. Then, at least conceptually, you could cascade as many of these things as you wanted, with the current decreasing at the output of each successive stage, just like if you have enough input voltage, you can cascade a virtually unlimited number of voltage regulators, each one putting out less voltage than the previous one.<p>Well, that's that. How is it useful? Other than as a mental exercise, that's a tough one. Cascading 2 voltage regulators definitely, and pretty much obviously, improves the line regulation. Would cascading 2 current regulators have any performance advantage? Not clear to me that it does. Probably why we don't see too many of them. But I'll be sketching these things on my engineering pad next time I get bored. Maybe something will come of it.<p>Regards,
Todd

[Dimbulb]

concider this short html called
"Precision current sink costs less than $20 "
http://www.e-insite.net/ednmag/archives ... 8/06di.htm<p>Looking at the schematic I am wondering why a run of the mill LM358 costing .10 and several inches of heavy nichrome wire for R4 could do 1/10 or 100mA as a low voltage version typically run from two AA batteries and illiminating the reverse protection diode.<p>The loop would be extended from the emitter of the TIP112 to include the circuit whose current is being controlled.<p>Further a big mosfat could replace Q1.
looking at figure two a for 3V possibly even LM358 and a 10K vernier would source hopefully lowering that part of the slope that resembles a dog leg below 10mA with a standard offset adjustment and leaving a small led running in loop so it has something to chew on.<p>[ May 15, 2003: Message edited by: 1206DX ]</p>

[Ron H]

Todd, I have also always been fascinated by the voltage-current duality. I briefly considered parallel current sources before my last post, but couldn't see the advantage. The "excess current" concept implies a source and a sink in parallel. Both would have to be high precision if you wanted the output to be high precision, which is not the case for series voltage regulators. Does the duality break down here, or is there something I'm missing?<p>1206DX, I still don't understand what you really are trying to do.

[Dimbulb]

A nice article by Clayton B Grantham called
"Current source has high output impedance"
http://www.e-insite.net/ednmag/index.as ... 69&stt=001<p>In this configuration an LM117 and LM4130 are very simply connected. The point I would like to highlight was a the term [current error over temperature] as this is often used since temperature degrades impedance.<p>Not that this circuit is super spectacular but is identifying the source of error.<p>The temperature constraint could be dispersed by using parallel circuits in this senario impedance would approach 300 Mega Ohm with a zero temperature coefficient.<p>Some may say is'nt this just a voltage controlled resistor, whats all the fuss just get a big wire wound and call it good. Looking at the voltage compliance the input can vary while maintaining a very even output current this is evident since a voltage that controls a voltage controlled resistor in affect adjusts the resistance so in concidering this as approaching an ideal resistor it needs a very stable voltage.<p>On page 6 of 6 figure 15 shows a simple cascaded
Jfet and table 2 shows typical ohm values.
Looking at for example the 2N5486 or its SMT counterpart. If you buy in quantity it is easy to sprinkle these about like CLDs. I am adding this cascaded example simply to show how two jfets are used to handle current. It is called an attenuator and we are discussing DC. Comments welcome.<p>http://www.qsl.net/n4xy/PDFs/Semiconduc ... _AN105.pdf<p>Constant current for most hobbiest acts as a cushion or damper so it is possible to use parts that require smaller current. However not getting too small is practical the plus side is having parts to use and taking advantage of the precision they can offer. This is why I am concerned with CC and low voltage battery powered circuits.<p>[ May 15, 2003: Message edited by: dimbulb ]</p>