Current-dropping current source! Please help!

[vicious5id]

Greets everyone,

I am not sure if this is the right place to post this in the forum, but I have a problem with a circuit that I designed at the implementation stage.

**broken link removed**

^ That's the circuit that I have designed. The purpose of it is to behave as a current source that supplies a range of current from 0 to 120 mA, to a 20 Ohm load (R11 in schematic). I have simulated the circuit within software known as SIMetrix and it works as I want it to under high temperatures, long periods of operation and if the 20 Ohm load alters in value. However once I built the circuit and tried it out I was surprised.

When I begin to rotate the pot, and stop it at a point where I need it say 52.2 mA (?!), the current will immediately begin to drop by 0.1 mA. Over time it will keep on dropping (over time = ~10 mins) but the 0.1 mA reduction gets very slow. I thought that it's becuase my resistors or something else was getting hot, however if I cooled the transistor more current began to flow!

I really don't want the current to drift by 0.1 mA at all! I want the current to stay exactly where I set it regardless of how much temperature changes. Also the current I set should stay constant for +/- 10 Ohms in the load. So if I were to set the current I want output to 80 mA it should stay constant over a 10, 20 and 30 Ohm load.

Well anyway I think this post is getting long enough and I don't want to start mixing my objectives and confusing everybody!

I really appreciate any help you guys have to offer.

v5 0ut.

 

[Russlk]

There is only 1.5 volts Vce at room temperature at the max current and Vce increases with temperature, so perhaps the transistor is going into saturation.

 

[hjames]

I'd check the reference voltage being fed into the potentiometer. If that moves by 0.1/52.2 = 0.2%, that'll result in the change that you're seeing. The circuit only tries to hold the transistor's emitter at the reference voltage, so if the 12V power supply varies by .2%, you'll see the that much variation.

To make it more stable, use a voltage reference thats isolated from the main power supply...

 

[vicious5id]

Russlk, if the transistor were going into saturation would it cause the current to decrease? I mean I always thought that the transistor would simply maintain that current.

hjames, I just want to make sure everyone, not only you is with me here; the current doesn't simply change or is unstable, it practically drops by about 1-2 mA over time. I'm also not really sure what you mean by voltage reference, are you talking about VIN+ on the op-amp or the combined resistance of R2, R9 and R4? Also how would I go about isolating this refernece from the main power supply?

 

[Nigel Goodwin]

I'm also not really sure what you mean by voltage reference, are you talking about VIN+ on the op-amp or the combined resistance of R2, R9 and R4? Also how would I go about isolating this refernece from the main power supply?

The current is dependent on the reference voltage applied to the opamp (via the pot), if that voltage changes then the current will change. You've made no effort to make that voltage stable, so the current isn't going to be stable at all. I would suggest using a precision voltage reference to feed the potentiometer?.

 

[vicious5id]

NG, I see where your coming from, and if I take that in mind, then I still don't see how the current in my 20 Ohm resistor begins to decrease and keep on decreasing, and only ever decrease and (almost) never increase. Surely the stability of a power supply is randomised, i.e. it fluctuates with increments and decrements.

I would suggest using a precision voltage reference to feed the potentiometer?

My friend found this ZREF12Z lying around at his Uni's lab so he swiped it ! This gives a 1.25 Vref. However I am not sure how to use it in my circuit as it has only two active pins, and a third floating one which I am supposed to tie to pin 1. Is this the right thing to use in my circuit and how do I use it, if it is. If it's the wrong component could anyone please suggest one? I'm sorry to be a pain like this, it's just that I have little experience in these specialist application components.

Many thanks, v5.

 

[Nigel Goodwin]

NG, I see where your coming from, and if I take that in mind, then I still don't see how the current in my 20 Ohm resistor begins to decrease and keep on decreasing, and only ever decrease and (almost) never increase. Surely the stability of a power supply is randomised, i.e. it fluctuates with increments and decrements.

The supply voltage is likely to gradually fall, and as it does the reference voltage will fall with it, and the current will obviously fall. It's essential to have a stable reference for anything like this - there may be other problems, but without a stable reference you've no real way of knowing.

My friend found this ZREF12Z lying around at his Uni's lab so he swiped it ! This gives a 1.25 Vref. However I am not sure how to use it in my circuit as it has only two active pins, and a third floating one which I am supposed to tie to pin 1. Is this the right thing to use in my circuit and how do I use it, if it is. If it's the wrong component could anyone please suggest one? I'm sorry to be a pain like this, it's just that I have little experience in these specialist application components.

Treat it like a zener diode.

 

[vicious5id]

Treat it like a zener diode.

So you mean I should stick it in under a shunt configuration? I mean should I simply place it in parallel across the pot I already have, and maybe take some resistance from the 100.5k resistor so I can get it to that 1.25 V value?

 

[Nigel Goodwin]

Place it in parallel with the pot, and reduce the value of the 100K to get the required current through the chip, the datasheet should tell you what value would be best.

 

[eblc1388]

That is a badly designed current source to begin with.

The sensing resistor not only measure load current, but also transistor base drive current too and this will cause big error if load resistance changes.

Looking at what Bob Pease did is to use a trilington(three NPN transistors combination) instead of a single 2N3053 to build an ultra high gain transistor switch and reduce the contribution of Ib to a very very low level. Using even a darlington setup or a MOSFET instead of 2N3053 would be much better in your case.

 

[vicious5id]

Woah!

That is a badly designed current source to begin with.

Man, I'm sorry LC, this is the first time I tried this and I'm only 17! I don't know if that still merits any sympathy, but its worth a try .

Looking at what Bob Pease did is to use a trilington(three NPN transistors combination) instead of a single 2N3053 to build an ultra high gain transistor switch and reduce the contribution of Ib to a very very low level. Using even a darlington setup or a MOSFET instead of 2N3053 would be much better in your case.

I couldn't find that article on Bob Pease by the way ("trilington" config.), eventhough I don't actually know who he is, until now (and after tapping his name into Google). However, your suggestion makes sense though, because the transistors that I am currently using offer a gain of about 50, but a TIP121 for example will churn out a gain of 1000.

EDIT (05/08/06 11:48 AM):
I tried using a darlington transistor in the circuit and I got similar results, sort of. The current seems to bounce around what I set it (set at 95.5 mA). I saw a drop of 1 mA in the first 10 mins of operation then the current immediately jumped back up to ~95 mA. After 53-60 mins of operation however, I recorded the current at 95.8 mA. This is good, but I really am not fond of the current varying by 1 mA at the very beginning. Would it be wise to go for a higher gain?

LC, I was hoping you could point me to a book or article that includes information regarding good, accurate variable constant current source design(s).

NG, I set up the voltage reference as shown **broken link removed** and it did work quite well, however the current dropped, but by only 0.5 mA in 40 mins, which is much better than what I had before (a drop of 2 mA in 10 mins). My apologies for the badly drawn half-schematic in Paint!

 

[eblc1388]

Man, I'm sorry LC, this is the first time I tried this and I'm only 17! I don't know if that still merits any sympathy, but its worth a try .

You don't have to say sorry. I have nothing against you and the comment is directed towards the circuit, not you. Does it comfort you that many Uni graduates can't put together a decent circuit? On the other end of the spectrum I knew one 15 year old guy designing circuits he sold to others as his bread and butter.

I couldn't find that article on Bob Pease by the way ("trilington" config.)

See image for what I meant by trilington. I can recommend you the book "The Art of Electronics" by Paul Horowitz and Winfield Hill, and perhaps also the following shows what is very interesting to watch.

**broken link removed**

I'll tell you you how to trouble shoot this problem. You cannot pin point the problem because you have not taken sufficient voltage readings to enable you to. You might need another voltmeter to fully debug your setup.

What you need to measure is the initial voltage on the two inputs of the opamp, after you have adjusted the pot for the desired load current. So you have three readings to take. The actual load current and the two opamp inputs volt reading. Draw up a table and write down the results.

Do this again when you have observed the current drifted and then it is very easy to point out where your problem is.

 

[vicious5id]

Does it comfort you that many Uni graduates can't put together a decent circuit? On the other end of the spectrum I knew one 15 year old guy designing circuits he sold to others as his bread and butter.

That's messed up, not comforting, alas you can probably assume as much, since this world is moving from analogue to "digital" (its all still analogue underneath). Even I'm considering a degree in digital electronics rather than analogue.

OK, I plan to measure the voltage of the op-amp inputs with ground, so Vin+ to GND and Vin- to GND. Then just take the current measurements as they appear in succession.

Also in your personal experience, is the basic idea of the circuit modified so far OK? Or is there a better alternative? I just want to know because I expect the load of 20 Ohms to change a bit, say +/- 5 Ohms.

**broken link removed**

 

[Nigel Goodwin]

What are the two 100 ohms doing?.

 

[eblc1388]

Also in your personal experience, is the basic idea of the circuit modified so far OK? Or is there a better alternative? I just want to know because I expect the load of 20 Ohms to change a bit, say +/- 5 Ohms.

**broken link removed**

The circuit is pretty standard and will suit your purpose just fine.

As long as you have a few volts across the C-E of the regulating element so it can do its job of regulating the current during use, then the circuit would not care a bit about the loading or load resistance changes.

One point I would like to make is why you have omitted the resistor from Opamp output to base of transistor. Doing so would prevent the output of the opamp to rise above I-sense voltage + VBe so it really is a bad idea. A series resistor of 2K2 would be much better.

 

[Roff]

<snip>One point I would like to make is why you have omitted the resistor from Opamp output to base of transistor. Doing so would prevent the output of the opamp to rise above I-sense voltage + VBe so it really is a bad idea. A series resistor of 2K2 would be much better.

Can you explain what you mean by this? I see no need for a base resistor, unless it is needed for frequency stability (usually in conjunction with another resistor and a capacitor).

 

[vicious5id]

What are the two 100 ohms doing?

I am using those to take some voltage from the transistor so that it doesn't get too hot (P = IV). Since the current's always the same from + 12 V to Collector, then the only thing I can use to limit power to a device are some extra resistors. For example, if the resistors were not there the transistor dissipates ~1 W, with the extra 50 Ohms, the transistor dissipates ~400 mW at optimum output.

Also just to make sure, I plan to use this circuit over extended periods of time like 2 hours non-stop, and I would really like for there to be no drift at all and if that's unavoidable then a drift of say +/- 0.1 - 0.3 mA I'll be happy with! That's also over the whole 2 hours.

 

[eblc1388]

Can you explain what you mean by this? I see no need for a base resistor, unless it is needed for frequency stability (usually in conjunction with another resistor and a capacitor).

I can offer no good explanation for the need of a resistor because the circuit would still work without one. The Opamp will make sure its output is just high enough to cause sufficient bias current to flow into the output transistor.

But from some of the circuits I have seen there is always a resistor used between Opamp output and current regulating transistor and I'm more comfortable using one so that the transistor regulator is not destroyed should the sense signal to the opamp input disappeared for whatever reason.

 

[vicious5id]

Results...

LC, I completed that test run this morning over a 20 min period. For the first minute I took readings every 10 secs, after 1 min, every minute then after 10 min one final reading at 20 mins. Collated data is attached as a .txt file, use a single-sized font type to view, for example Lucida Console or Courier New.

I punched the data in a spreadsheet and got a few graphs out, quantities, etc all indicated in titles.

Man I really hope this gets somewhere.

 

[eblc1388]

Good work, vicious5id.

From your results, I'm surprised that even the reference voltage changes. The reference voltage determine the load current and it has to be rock solid in order for you to have a constant load current. when reference voltage changes, everything changes.

Looking at the value of the circuit responsible for creating the reference voltage, I have calculated that the resistor value of 70K is definitely too large. The ZR423 can only function properly when current through it is larger than 50uA whereas now it only has 25uA passing through it and it does not function properly.

So you have to increase the current in the ZR423 and I would recommend about 100uA (datasheet min. 50uA or more) to get a stable voltage. I trusted that you can work out the correct resistance by taking the total current through the resistor and the voltage drop across it.

After changing the value of this resistor, hopefully your circuit will work to your satisfaction.