The first circuit seems to be more of a current limiter rather than current sink. Something that can be put on the output of linear power supplies to protect them from overload and whatnot.
The circuit is really from the TI datasheet where it is called Precision Constant-Current. With 5mA and a beta of 250 the collector current is 20µA smaller than the emitter current. This can be eliminated with the FET. What remains a sense current for the TL431 of 2-4 µA which is then below 0.1%. (It depends on what you call precision, maybe the term precision is just taken from the name of the TL431 which is called "precision programmable reference".)
'precise' is how many digits are known. 'accurate' is how close you arrive to the value you wanted. If you create a 5.00001V voltage source, it is very precise, but if you wanted a 3.3V source, it isn't very accurate.
One ancillary benefit of the older 2N7000 N-channel MOSFET is that its channel resistance won’t be that low with only 2.5V of VGS, so that will inherently limit the brief inrush when the DUT is initially connected.
Just to tease ... the 0.1% shunt causes the circuit to be _precise_, but if the wrong values are used it will not be _accurate_ 🙂 Just my favourite bug
Having gone through these types of circuits in the past (both for sourcing and for sinking current) one caveat of the TL431 is that it needs a minimum Ik in order to maintain the regulated 2.5V. That current from the datasheet is anywhere from 0.5 to 1mA. So for a current source, that is what you have to set with the extra resistor in order for the regulation to work, and will be added to the regulated current. There is another device though that can get you much closer to a "precision" current source, the TLV431 or LMV431. Those have much lower Ik requirements, on the order of 55 to 80uA. Thus you can use a much larger resistor in series with them, and the extra current added to the regulated one will be much smaller and less supply-dependent. There are 2 caveats though with using the TLV431: first, the regulated voltage is 1.24V not 2.5V; second, maximum cathode voltage it tolerates is 7V.
As I think many have already mentioned, it's a precision current limiter, not a precision current source. The TI datasheet uses the correct description. I use this part a lot but usually only for current sinks since it needs about 1-3mA to keep it in regulation. There is a better part ATL431, which is more 'precision' with Ika around 200uA or so.
Use a any 3 terminal voltage regulator as a current source or sink. Easy to trim if the current is not above dissipation of the trimmer. Use a LMl317 or any LMlxx voltage regulator. Can be used either way just put it in line. Have pretty good temp and voltage drift if you do not over load. If you want to measure zener voltage, use your power supply. Set the max current and voltage and connect across the zener. Read voltage on the power supply as supply goes into current limit. The max zener voltage to measure iis the max voltage of the supply.
The first time I saw TL431 was in an old bendix radar, they had replaced the power supply “zener tubes” with lots of them in series, never seen one there was defect. Brilliant little device.
I used one of those 431's in is a shunt regulator of a solar panel that powers my driveway gate. Your zener tester sounds like an nice little project. I often find my self testing zeners with a resistor and a power supply. You tester could also check LED's for forward voltage drop.
The limit and sink circuits are actually the same except the Io in the sink circuit is from Vcc through the collector avoiding the current through the TL431. Basically either connecting between Vcc and collector or between emitter and ground. One is good the other is bad.
The TL431 is usually advertised (and used) as a low current adjustable zener replacement [e.g. voltage regulator]. Try an LM334 for regulated current of 10 mA or less, or the venerable LM317 for higher I. Simpler circuitry to your above circuits and they both work well.
Yeah, trying to find a good way to set a specific current is one of the most challenging things for me to design. Mostly because I want to drop almost no voltage from the circuit while doing so.
Figure 10-14 shows nothing else but a current limiter. For an accurate precision (+) current source to ground potential IMO it might be worth looking at a current mirror circuit.
incredible, but I'm designing a crude volt to amp source with a pmos and an opamp, for measuring resistors. it's crazy that these circuits are so off, even on the datasheets of the components!
So, by replacing the BJT with a FET you're fixing the the problem that Rs is dropping both Ib and Ic but your device under test is only seeing Ic? Would putting the DUT in series between Ie and Rs get to a similar result?
Is the zener tester intended to be something the viewer would build and use? Otherwise, I'd think that you'd just pull out your curve tracer to evaluate a zener.
There are not too many differences between them, actually. Basically, tl431+resistor+power transistor=lm317-protection circuits. Only to note, except of lm317 protection, that, (1) minimum voltage difference is 2.5 and 1.25 volts respectively; (2) in lm317 npn Darlington transistor is a built-in and cannot be changed; (3) to build current sink you really need lm337 instead, as protection circuit of lm317 does not like sinking current.
"Precesion current source" from TL datasheet is not precesion, because current of TL ground pin is added to stable current from shunt. And this current depends on load voltage drop and V_batt. So... It is profanation * Current sink from datasheet - works
Why are you worrying about base current? It's irrelevant as it gets measured in Rs. The unknown is what the current through the TL431 is, that is what will throw you off, especially if it is voltage dependent.
0:59 I did go to school in Europe, so maybe the conventions were different there, but I was taught that a current source is two overlapping circles with an arrow in it...
you will see different ways to draw schematics between countries. US/Europe/Japan not just current sources. resistors, potentiometers grounds connectors...
The first circuit seems to be more of a current limiter rather than current sink. Something that can be put on the output of linear power supplies to protect them from overload and whatnot.
The circuit is really from the TI datasheet where it is called Precision Constant-Current. With 5mA and a beta of 250 the collector current is 20µA smaller than the emitter current. This can be eliminated with the FET. What remains a sense current for the TL431 of 2-4 µA which is then below 0.1%. (It depends on what you call precision, maybe the term precision is just taken from the name of the TL431 which is called "precision programmable reference".)
'precise' is how many digits are known. 'accurate' is how close you arrive to the value you wanted. If you create a 5.00001V voltage source, it is very precise, but if you wanted a 3.3V source, it isn't very accurate.
It's dependency on hFE makes it quite poor. I guess if you measure the actual hFE you can select a suitable R1
One ancillary benefit of the older 2N7000 N-channel MOSFET is that its channel resistance won’t be that low with only 2.5V of VGS, so that will inherently limit the brief inrush when the DUT is initially connected.
Just to tease ... the 0.1% shunt causes the circuit to be _precise_, but if the wrong values are used it will not be _accurate_ 🙂 Just my favourite bug
Having gone through these types of circuits in the past (both for sourcing and for sinking current) one caveat of the TL431 is that it needs a minimum Ik in order to maintain the regulated 2.5V. That current from the datasheet is anywhere from 0.5 to 1mA. So for a current source, that is what you have to set with the extra resistor in order for the regulation to work, and will be added to the regulated current. There is another device though that can get you much closer to a "precision" current source, the TLV431 or LMV431. Those have much lower Ik requirements, on the order of 55 to 80uA. Thus you can use a much larger resistor in series with them, and the extra current added to the regulated one will be much smaller and less supply-dependent. There are 2 caveats though with using the TLV431: first, the regulated voltage is 1.24V not 2.5V; second, maximum cathode voltage it tolerates is 7V.
As I think many have already mentioned, it's a precision current limiter, not a precision current source. The TI datasheet uses the correct description. I use this part a lot but usually only for current sinks since it needs about 1-3mA to keep it in regulation. There is a better part ATL431, which is more 'precision' with Ika around 200uA or so.
Datasheet named circuit correctly. Limiter not source.
Right.
Interesting analog proto board design - I never saw anything like this, and it looks fun for making old school discrete transistor fuzz pedals.
TLV431 is the same, just 1.25V instead of 2V5 and higher.
Use a any 3 terminal voltage regulator as a current source or sink. Easy to trim if the current is not above dissipation of the trimmer. Use a LMl317 or any LMlxx voltage regulator. Can be used either way just put it in line. Have pretty good temp and voltage drift if you do not over load. If you want to measure zener voltage, use your power supply. Set the max current and voltage and connect across the zener. Read voltage on the power supply as supply goes into current limit. The max zener voltage to measure iis the max voltage of the supply.
Power supply ripple will couple through the gate capacitance of the FET. So that tiny current he talks about is for a DC input.
The first time I saw TL431 was in an old bendix radar, they had replaced the power supply “zener tubes” with lots of them in series, never seen one there was defect. Brilliant little device.
I used one of those 431's in is a shunt regulator of a solar panel that powers my driveway gate. Your zener tester sounds like an nice little project. I often find my self testing zeners with a resistor and a power supply. You tester could also check LED's for forward voltage drop.
Very good excuse to buy an SMU btw :)
The limit and sink circuits are actually the same except the Io in the sink circuit is from Vcc through the collector avoiding the current through the TL431. Basically either connecting between Vcc and collector or between emitter and ground. One is good the other is bad.
The TL431 is usually advertised (and used) as a low current adjustable zener replacement [e.g. voltage regulator]. Try an LM334 for regulated current of 10 mA or less, or the venerable LM317 for higher I. Simpler circuitry to your above circuits and they both work well.
TL431 is ubiquitous in today’s consumer electronics. Used for reference voltage in small switching power supplies instead of zener diodes.
Yeah, trying to find a good way to set a specific current is one of the most challenging things for me to design. Mostly because I want to drop almost no voltage from the circuit while doing so.
Figure 10-14 shows nothing else but a current limiter.
For an accurate precision (+) current source to ground potential IMO it might be worth looking at a current mirror circuit.
the schematics in Figure 10-14 is actually wrong as he said, unless the load is placed between Vin and the collector of the transistor.
They did indicated figure 10-14 was a current limiter, not a precision source. Perhaps you should evaluate that circuit as such by varying the load?
It's very worst for change in temperature 😢
You could use an op-amp buffer with the input (of the buffer) tied to the output and the output (of the buffer) tied to the "anode" of the TL431.
Merci, very interesting!
incredible, but I'm designing a crude volt to amp source with a pmos and an opamp, for measuring resistors. it's crazy that these circuits are so off, even on the datasheets of the components!
Lm317 would work on the first circuit
They mistakenly left out a word. Figure 10-14 is a Low Precision Current Limiter...
Hmmm. Could this be used to provide bias to a tube circuit in the cathode line to ground?
So, by replacing the BJT with a FET you're fixing the the problem that Rs is dropping both Ib and Ic but your device under test is only seeing Ic? Would putting the DUT in series between Ie and Rs get to a similar result?
Is the zener tester intended to be something the viewer would build and use? Otherwise, I'd think that you'd just pull out your curve tracer to evaluate a zener.
Interesting Jenner tester device, how to implement it as an plugin for the tester?
lol nice haha😎
Any reason you’re not using an LM317 or any other LDO in source configuration? Asking because I’m probably the dumb one
There are not too many differences between them, actually. Basically, tl431+resistor+power transistor=lm317-protection circuits. Only to note, except of lm317 protection, that, (1) minimum voltage difference is 2.5 and 1.25 volts respectively; (2) in lm317 npn Darlington transistor is a built-in and cannot be changed; (3) to build current sink you really need lm337 instead, as protection circuit of lm317 does not like sinking current.
👌👌👌👍👍👍.....
"Precesion current source" from TL datasheet is not precesion, because current of TL ground pin is added to stable current from shunt. And this current depends on load voltage drop and V_batt. So... It is profanation
* Current sink from datasheet - works
5:15 Was yelling at the screen to use a FET... and you did! Yay.
Why are you worrying about base current? It's irrelevant as it gets measured in Rs. The unknown is what the current through the TL431 is, that is what will throw you off, especially if it is voltage dependent.
0:59 I did go to school in Europe, so maybe the conventions were different there, but I was taught that a current source is two overlapping circles with an arrow in it...
you will see different ways to draw schematics between countries. US/Europe/Japan not just current sources. resistors, potentiometers grounds connectors...
How about the "classic" LM317T current source circuit