EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: ddavidebor on July 14, 2013, 11:39:53 pm
-
Hi guys.
I'm sniffing around for a good schematic for a little current supply.
I need 10mA fixed, with max 4.5v but should work a bit under 3v...
First requirement is precision and stability. I can trimm it, but it should remain stable over temperature and time. At least less than 0.1% over a few month.
I've successfull used this circuit https://upload.wikimedia.org/wikipedia/commons/2/23/Const_cur_src_112.svg
It need some little modification. First it need a lower voltage reference instead of the zener. Next i think i should use and identical transistor as diode (base-emitter only) for best thermal compensation. Same package should allow best thermal bonding if required.
But, i've no experience and i've a simple question?
Is this little circuit sufficiently precise or i need something different? If yes, what do you suggest.
I've no longer opamp and local shop is closed, but if really need it for best precision i will buy it
-
0.1%? Op amp. High precision can be achieved without them, but good luck.
-
You don't mention temperature range.
If the temperature is fixed, then the single transistor CCS show will hold 1000ppm - the tempco of the parts being the biggest source of error (I've used this design in many amplifiers over the years and it is pretty stable, I tend to regulate the voltage to the base reference circuit and add a cascode stage above it - when I need lower noise or flatter response - not really relevant to what your doing).
If you've got a LM334 in your parts bin, then great.
One other thing, the tempco on trimmers tends to be alot worse than for fixed resistors - so if you can achieve your 0.1% with a fixed resistor then great, else use the trimmer to make only a small adjustment (i.e. for 500ohm, use a 470ohm fixed and a 50ohm trimmer).
-
For your required level of precision and stability, you will need a good current source based on a high quality op amp (no Vos temperature drift to speak of), and a high stability voltage reference diode. A simple zener diode reference, with or without the extra diode 'temperature compensation', won't be anywhere near stable enough for your purpose.
The schematic for your simple CCS won't be able to reach the level of performance you require. Among other problems, then the base current contributes to the current through the sense resistor. The negative feedback from the single emitter resistor is also way too feeble for your purpose.
You can either use a precision opamp, which can sink/source the required current, or you can 'bootstrap' it with an extra NPN BJT, a JFET or even a small signal MOSFET. If for instance your load *must* have at least one terminal connected to the positive supply rail, then the easiest way to get around the error caused by the base current in a BJT, is by using either a N-channel JFET (J310) or ditto small signal MOSFET (BS170). Exactly how to build the CCS and what components to use would depend on whether you have dual supply rails, and how high they/it are.
-
Thanks a lot for the tips.
I've watched the lm334... 10mA is the maximun current so it will be to it's fisical limit. No good at all unfortunately, because is damn cheap.
I've found (based on the just learned keyword, i'm italian) the lt3092. But it's super expensive, 3.4$! And it require a 10uA 1.25 reference diode for best precision. Expensive as well...
So i need to search for an opamp solution. If anyone has suggestion, i'm open and grateful to hear them.
Tha supply is 2.7 to 4.5v, single or dual supply
-
Build one and measure it, or just simulate it.
Unless you define your temperature parameters it is going to be hard for you to analyse component datasheets and calculate drift.
-
Ok, thanks a lot!
-
You might look at some of the high precision voltage references. You might be able to use them like you can use an LM317 to make a constant current source.
The LM4140 (http://www.ti.com/lit/ds/symlink/lm4140.pdf) has pretty good overall specs. You might be able to use one of the low voltage parts (1.024V or 1.250V) and a circuit that is a combination of figure 29 (to boost the current) and figure 33 from the datasheet to get a precision current source. How precise is hard to say.
-
The circuit you linked is slightly load dependent, there is no correction for the early effect ... you can use a cascode, but with your low power supply voltage I doubt you want more voltage drop.
An opamp based circuit will be mostly load independent and the only temperature dependence you have to worry about is the voltage reference.
-
Ook, i suspect i will take the bus to the non local local electronic store.
Thanks a lot, you're righgt, every volt of drop more is a volt i cannot use.
-
I'd use one of the common 3-terminal voltage references like the TLV431 to replace your zener and diode. There's no need for an additional opamp, since the feedback pin on is actually the negative terminal of an internal opamp. Something like the attached screenshot will work well. Unfortunately, it won't compensate for the temperature dependence of beta, so you could use a fet instead, but that will have a higher compliance voltage
-
Ok, thanks a lot!
No, he is asking you for more information.
What is the working temperature range for this circuit? Be specific.
-
I'd use one of the common 3-terminal voltage references like the TLV431 to replace your zener and diode. There's no need for an additional opamp, since the feedback pin on is actually the negative terminal of an internal opamp.
Oops, didn't even realise these existed.
-
Ok, thanks a lot!
No, he is asking you for more information.
What is the working temperature range for this circuit? Be specific.
Dunno... 10-30 degree celsius is problably appropriated
-
I'd use one of the common 3-terminal voltage references like the TLV431 to replace your zener and diode. There's no need for an additional opamp, since the feedback pin on is actually the negative terminal of an internal opamp. Something like the attached screenshot will work well. Unfortunately, it won't compensate for the temperature dependence of beta, so you could use a fet instead, but that will have a higher compliance voltage
so, for what i've understandood, correct me if i'm wrong.
the tlv431 do it's black magic and keep the Vref-Vanode=1,24V by act on Vcathode-Vanode
sooo, in your circuit it keep the voltage drop on Rset (and the current) costant.
but, i don't understand "it won't compensate for the temperature dependence of beta"... you mean the transistor right?
how can it influence the current in this configuration?
-
You're right on the money with your description of it's operation. :-+ Yes, the current through Rset, and thus the emitter current of the transistor are held constant across temperature. However, the collector current of the transistor is a wee bit less than the emitter current, due to the finite current gain of the device, which manifests itself as base current.
So, here's the kicker: The beta of a BJT is significantly dependent on temperature. Look at Figure 1 on page 3 of www.fairchildsemi.com/ds/PN/PN2222A.pdf (http://www.fairchildsemi.com/ds/PN/PN2222A.pdf) There are three traces that show how the beta, or current gain, of the transistor varies versus collector current for three different temperatures. Note that at 10mA of collector current and across it's rated operating temperature rage of -40°C to +125°C the current gain varies from a low of ~100 all the way up to ~250. This means there will be a "error" or offset in the collector current of 1% at -40°C and 0.4% at +120°C. If we then assume that this variation is linear over temperature, then the resulting tempco of the current source will be (1% - 0.4%)/(-40°C - 125°C) ~= 0.0036 %/°C or 36ppm/°C.
This is similar in magnitude to the inherent tempco of the reference itself. This depends on how you define it, but according to Note 4 on page 6 of www.ti.com/lit/ds/symlink/tlv431b.pdf (http://www.ti.com/lit/ds/symlink/tlv431b.pdf) it is about (11mV/1.24V)/(-40°C - 125°C) ~= 0.0054 %/°C or 54ppm/°C.
So, really, I guess it actually comes out in a wash. And considering you're only worried about a 10-30°C temperature range, it truly results in a bee's dick amount of error, so I suppose it's not really worth worrying about after all. ^-^
-
Fantastic... This solution is really sex on a stick!
Thanks, i need to go to order some components.
-
I built a current source from a REF102 precision +10v voltage source a while ago. It's simple and it works really nicely. It is Figure 10 in the datasheet
http://pdf1.alldatasheet.com/datasheet-pdf/view/56859/BURR-BROWN/REF102.html (http://pdf1.alldatasheet.com/datasheet-pdf/view/56859/BURR-BROWN/REF102.html)
The voltage source is good for 10mA and 2.5ppm/degC so then it comes down to the temperature stability of the resistor in the circuit.
I built this up as a 1mA current source on a Friday afternoon a while ago. I built it all up on a scrap of breadboard, adjusted the trimmer to what I'd calculated would give me 1mA. Plugged it in, connected an ammeter and there, 1.000mA at ten to five on a Friday afternoon. Headed off to the pub feeling smug. Why doesn't prototyping circuits always work like that??? :-//
-
Interesting but a loooooooot out my need (price, precision, Vin, Vout)
-
I built a current source from a REF102 precision +10v voltage source a while ago. It's simple and it works really nicely. It is Figure 10 in the datasheet
http://pdf1.alldatasheet.com/datasheet-pdf/view/56859/BURR-BROWN/REF102.html (http://pdf1.alldatasheet.com/datasheet-pdf/view/56859/BURR-BROWN/REF102.html)
The voltage source is good for 10mA and 2.5ppm/degC so then it comes down to the temperature stability of the resistor in the circuit.
I built this up as a 1mA current source on a Friday afternoon a while ago. I built it all up on a scrap of breadboard, adjusted the trimmer to what I'd calculated would give me 1mA. Plugged it in, connected an ammeter and there, 1.000mA at ten to five on a Friday afternoon. Headed off to the pub feeling smug. Why doesn't prototyping circuits always work like that??? :-//
Nice part! Lot better circuit than the one I linked too as well.