Using a discrete current mirror is useful for more than a single mirrored current. Say you want to drive 10 LEDs, and dim them all together. You can have 10 of them, all mirrored on the same transistor, all with the same current. Doing that with other technology is fairly hard, but simple with a current mirror. See "The Art of Electronics" for more information and examples. It has a great section on current mirrors.
It is still January, and I found an application, the Jensen 990 op amp - an old design but a classic in its field. A mirrors help balance the differential pair in later versions. I like them myself, just so elegant when you're only passing a few mA.
I made an array of 12 current mirrors for 4-20mA current loop testing. I used PNP's with a heatsink on Q2 for thermal runaway and pots in series with a resistor on Q1 for trimming the I1/I2. Works like a charm for 5 months now - and the PLC's seems to love it too :D Only limit is that it has to be fed with 12v only. If needed I think I'll redesign it with an opamp/current source in mind. -Nice to see you are back with the kitchen LEDs projects.
Thanks for the video, I have the same question. think you missed the point of the question. Even with a constant current power supply, parallel strings of COB LEDs can have slight differences in the current on each string. My system that I'm building has 400 3W LEDs that run at 500-3000 mA. I want to use BGT transitors to keep the current constant in each of the 16 strings. Its all built on a sheet of aluminum. do you know how to size them? shopping is hard if you don't know what you need. If I just wired all the gates together, would they all mirror each other?
If I'm not mistaken, voltage controlled oscillator (VCO) circuits use them. Since oscillators often have non-linear response to control voltage (for instance the pot setting the frequency), the current mirror seems to act as an antolig amp (or log amp?). Will try to find an example.
Toxi Com this guy is doing a great job explaining this stuff. I'm an electronics engineering student and there is a lot to this stuff but he is doing a great job in my very humble opinion. Some background info might help though like watching some videos about diodes and semiconductor basics
I think that a current mirror would work to power 1 led, using a current of around 20mA. You would need a higher current transistor device though like a Darlington transistor, but couldn't that work to light a single LED? We always use current limiting resistors with LEDs so it would be cool to see an application where it didnt need the resistor :) Also the voltage drop of an LED is around 2.5 V so maybe just providing the current might not work correctly to power it? No idea though... I kind of want to try it out
The problems of stabilizing a current mirror can be handled by incorporating negative feedback into the circuit. Furthermore, current mirrors can produce multiples or fractions of the current that's passing through the reference transistor, so it's not necessary to waste large amounts of current in the reference transistor. Uses? Driving inductive loads -- speakers, motors, switched-mode power supplies, etc. -- whose behavior is determined by the currents through them.
As far as i know, current mirrors are used in audio amp preamplifier stages. Btw, could you do video about biasing a constant current transistor or mosfet?
This is best explanation on TH-cam, my confusing is over,Thanks. By the way, I have a question ,would it be OK if we use two mosfets instead of two transistors in a view of thermal runaway because more heat on mosfet make it less conduct in oppose to two transistors.
Ooooh one of my favourite subjects! First packet of questions, hopefully someone will answer 00:44 How do i know what current is that? 1:07 How do i know the Vt and Von? 1:18 how do i calculate that? What are the all other things being equal? 1:33 How did you reach that conclusion? Also, SAME or EQUAL? Small difference in words, big difference in actual reality. And another thing, under WHAT conditions will those currents be equal? M2 can't possibly create a current out of nothing, only thing it CAN do is adjust the "resistance" across D-S. What happens? 1:55 How do you know it's that and not anything else? Do we set it? Is it always that? 2:45 finally some hard stuff! That's scary? Anyone who finds that scary should stay away from metallurgy books, some empirical equations in there span multiple rows. 5:10 what happens when the rate at which the transistor generates heat ultimately matches the rate at which that heat is dissippated to the outside world? Won't that be an equilibrium? A bastard, evil, hacky equilibrium?
Hello. If I may I would like to try to answer some of your questions: 1) You set the current by adjusting the resistor, thus adjusting the bias voltage at the gates of the MOSFETs. 2) I think Vt and Von are provided on datasheets. (I'm not sure though) 3) Assuming you know what current you want, you need to calculate what bias voltage you need to provide that current. So if you know the voltage drop across the M1 and you know the source voltage and you assume the source is grounded, you just use Ohm's Law to calculate the resistance needed. With these calculations we are assuming we are using CMOS as opposed to individual MOSFETs so we assume the parameters for M1 and M2 are equal. 4) We can reach the conclusion that the current through M2 will equal the current in M1 because both MOSFETs will have the same bias voltage at the their gates. An interesting aspect about MOSFETs is they can force current based on the bias voltage at the gate no matter what the voltage drop across the MOSFET because of the way MOSFETs work. 5) For BJTs, Vbe is almost always assumed to be about 0.7 V (I was taught 0.7 V, but I guess others were taught 0.6 V) 6) Maybe scary to hobbyists because they do not usually think about the detailed math behind the components. But looking at these components as an engineer there are some fairly long equations associated with MOSFETs. 7) Maybe, but will the BJT last until that equilibrium is met? Also, how likely is it one can set the circuit to create that equilibrium?
4) Will it? How can the M2 force a current? Suppose i do not stick a power supply at M2... what will happen then? What *property* of M2 changes which ALLOWS the current to be that value? Please do not use the word 'force' because unless we are talking about an inductor, it can *NOT* *FORCE* a current. 5) what if i stick a voltage across those terminals? like 9 V? will it still be 0.7 V - 0.6 V ?
George Tsiros 4) You mean: will the currents be equal? In theory yes, in real life the currents will be very close. How close? Depends on the materials of the CMOS or the individual MOSFETs and imperfections in the manufacturing. For nMOS, when you apply a positive voltage at the gate, a certain amount of current must go from the drain to the source (technically source to drain) because the positive charge attracts the electrons from one n-type substrate, through the p-type substrate then the other n-type substrate attracts the electrons. Thus the amount of current flowing through an nMOS depends on the gate voltage. In theory, with no voltage source there would be no voltage drop and, according to Ohm's Law, no current flow. In real life, doing so would likely damage the CMOS chip because there is almost no voltage drop (maybe pV) and the device is trying to push a current. As far as I know, no property changes. M2 can push that same amount of current as M1 because they have the same gate voltage. Sorry, "force" is a term one of my profs used and I did not realize it is incorrect to use it when referring to MOSFETs. 5) 9V at the base and gnd at the emitter? You would likely fry the BJT. The base and emitter are like a diode, which is why we assume a 0.7 V drop (like we do when assuming 0.7 V drop diodes), so doing what you ask would push too much current through the BJT and cause a problem because there is a 9 V drop when there should be a 0.7 V drop, approx. Also, looking back at my old notes: 0.3 V
I'm gonna push you more and i hope you've got some juice left, because we've not even started yet. "In theory, with no voltage source there would be no voltage drop and, according to Ohm's Law, no current flow. In real life, doing so would likely damage the CMOS chip because there is almost no voltage drop (maybe pV" So the transistor would be damaged... because... ? Too much power? But there is no current, so there is no power dissipated. No current flows anywhere. Not G-S, not G-D, not D-S. There is no power dissipated anywhere. There is no high voltage applied anywhere. Think it a bit more, i will be waiting. "the device is trying to push a current" What is the mechanism behind that pushing? As far as i know, the mechanism via which the transistor ALLOWS current to pass through is gathering freely-moving charges between the D-S to create a channel. "Sorry, "force" is a term one of my profs used and I did not realize it is incorrect to use it when referring to MOSFETs." My professor is doing that too and when i asked him what does it even mean he couldn't respond. When i asked him "but the transistor doesn't actually FORCE a current, because it CAN NOT force a current?" he replied via verbal abuse towards me. Great teacher. 5) ok then, 9 V, AND a 10 μA current limit. what would happen then? What is the Vbe voltage then?
I have an example: there is a big 555 timer kit. This kit, with discrete components, emulates the die of the 555. Since the actual die of the 555 ic houses multiple current sources, they made current sources with discrete components. Et voila: an example ;). Cheers
it sounds like people are trying to use a current mirror were they should be using a constant current cct. current mirrors are used in power amp for diff reasons
Perhaps this is a dumb question. But I am looking for a way to control an ac motor, namely an o gauge train layout with an arduino. Would this work? Control the voltage of the ac current by having it mirror the dc voltage? Or do you have something else you could recommend.
+Kevin Rutledge It won't even work that way. And anyways current mirrors are used in 99% in ICs instead of resistors to set the bias currents, because it's easier to do transistors than resistors on silicon die. End of story:)
Kevin Rutledge there is always a way to do it but I dont know if theres a way to get from DC to AC, as opposed to the other way around. There is certainly a way to do what you want to but it might not be practical or even safe :)
You would use an AC-AC converter (aka a single phase AC controller) such as a triac, or two SCR/thyristors. The digital pin from the Arduino will allow current to conduct in specified direction at specified interval. There's some example code of AC controllers. This has absolutely nothing to do with current mirrors though
Pretty late to the party, I hope to God you're still active here cause I need some assistance on a research project. If your load was a DC motor, could you vary the current through it by varying the resistor on Q1? Assuming voltage is constant
You dont HAVE to use a constant current on one side, many circuits can call from use of having current on one side be the same as the other such as current monitors and compactors
Not sure his 350mA argument is valid for normal leds. All the leds in the string will carry the same current, and the same current will pass through the transistor, only 20mA
I be interested in your thoughts are on why a discrete configuration would not be suitable for the following application: Regulating the current through both triodes in a dual triode vacuum tube (such as an ECC88) when being used as a balanced phase splitter [see: digilander.libero.it/essentialaudio/hybrid_circuit.htm]
Dave Jones EEVblog did a video on a 555 timer kit. That had some current mirrors in them. watch?v=U6qZPx4uD0g#t=2296 Not sure if the real 555 has current mirrors also...
Can somebody give a list of applications current mirrors are regularly used in? What types of instrumentation or devices? Kinda curious to where they are useful not where they are not lol
They are used all analog ICs such as comparators and opamps. They are also used to generate the voltages in voltage regulators, using the so called bandgap reference technique. As far as analog mirrors, I have used them to allow dimming of lots of parallel LEDs all driven from a single low voltage source.
***PRACTICAL APPLICATION wiki.analog.com/university/courses/electronics/text/chapter-11 Figure 11.21 A combination of the Widlar mirror and peaking current source provides further improvement in the regulation of a variable input current can set up a charging cycle for lipo charger
I do not understand why you chose this terrible camera angle !!!. Didn't occur to you that things to the right of the board will be a bit harder to read ? ..
11:41 to say this is a current mirror. They have serious flaws. They provide no value as each application that might need current limiting has better alternatives. Mmm I think that was a waste of 11 minutes of our lives. Sorry.
Using a discrete current mirror is useful for more than a single mirrored current. Say you want to drive 10 LEDs, and dim them all together. You can have 10 of them, all mirrored on the same transistor, all with the same current. Doing that with other technology is fairly hard, but simple with a current mirror. See "The Art of Electronics" for more information and examples. It has a great section on current mirrors.
second part!
9:37 sounds good in practice, or theory?
10:06 active load for differential pair at an audio amplifier's input stage?
It is still January, and I found an application, the Jensen 990 op amp - an old design but a classic in its field. A mirrors help balance the differential pair in later versions. I like them myself, just so elegant when you're only passing a few mA.
I made an array of 12 current mirrors for 4-20mA current loop testing.
I used PNP's with a heatsink on Q2 for thermal runaway and pots in series with a resistor on Q1 for trimming the I1/I2.
Works like a charm for 5 months now - and the PLC's seems to love it too :D
Only limit is that it has to be fed with 12v only.
If needed I think I'll redesign it with an opamp/current source in mind.
-Nice to see you are back with the kitchen LEDs projects.
Like your user name haha!! lol
Thanks for the video, I have the same question. think you missed the point of the question. Even with a constant current power supply, parallel strings of COB LEDs can have slight differences in the current on each string. My system that I'm building has 400 3W LEDs that run at 500-3000 mA. I want to use BGT transitors to keep the current constant in each of the 16 strings. Its all built on a sheet of aluminum. do you know how to size them? shopping is hard if you don't know what you need. If I just wired all the gates together, would they all mirror each other?
How about a emitter resistor...
If I'm not mistaken, voltage controlled oscillator (VCO) circuits use them. Since oscillators often have non-linear response to control voltage (for instance the pot setting the frequency), the current mirror seems to act as an antolig amp (or log amp?). Will try to find an example.
Why hasn't there been any new episodes? I miss these monthly videos.
Thank you for posting such an instructive video on current mirrors.
watching this actually causes more confusion
Toxi Com this guy is doing a great job explaining this stuff. I'm an electronics engineering student and there is a lot to this stuff but he is doing a great job in my very humble opinion. Some background info might help though like watching some videos about diodes and semiconductor basics
I think that a current mirror would work to power 1 led, using a current of around 20mA. You would need a higher current transistor device though like a Darlington transistor, but couldn't that work to light a single LED? We always use current limiting resistors with LEDs so it would be cool to see an application where it didnt need the resistor :)
Also the voltage drop of an LED is around 2.5 V so maybe just providing the current might not work correctly to power it? No idea though... I kind of want to try it out
Watch some more videos about this topic and it makes so much sense. Current mirrors belong in an ic, not with discreet components - he says.
The problems of stabilizing a current mirror can be handled by incorporating negative feedback into the circuit. Furthermore, current mirrors can produce multiples or fractions of the current that's passing through the reference transistor, so it's not necessary to waste large amounts of current in the reference transistor. Uses? Driving inductive loads -- speakers, motors, switched-mode power supplies, etc. -- whose behavior is determined by the currents through them.
As far as i know, current mirrors are used in audio amp preamplifier stages.
Btw, could you do video about biasing a constant current transistor or mosfet?
Thanks, love your style. Really straight talk.
yeah you can make a tempeture sensor using a simple diode or you can make external opam maybe
When will the new video for February be available?
Just use a constant current souce with single transistor for your LED driver
This is best explanation on TH-cam, my confusing is over,Thanks. By the way, I have a question ,would it be OK if we use two mosfets instead of two transistors in a view of thermal runaway because more heat on mosfet make it less conduct in oppose to two transistors.
Current mirrors are especially useful in various signal amplification designs, as well as some variants of ADCs (such as sigma delta, etc.).
mopar3502001
I have what I think is a practical use for a CM. Are you still interested ?
Ooooh one of my favourite subjects! First packet of questions, hopefully someone will answer
00:44 How do i know what current is that?
1:07 How do i know the Vt and Von?
1:18 how do i calculate that? What are the all other things being equal?
1:33 How did you reach that conclusion? Also, SAME or EQUAL? Small difference in words, big difference in actual reality. And another thing, under WHAT conditions will those currents be equal? M2 can't possibly create a current out of nothing, only thing it CAN do is adjust the "resistance" across D-S. What happens?
1:55 How do you know it's that and not anything else? Do we set it? Is it always that?
2:45 finally some hard stuff! That's scary? Anyone who finds that scary should stay away from metallurgy books, some empirical equations in there span multiple rows.
5:10 what happens when the rate at which the transistor generates heat ultimately matches the rate at which that heat is dissippated to the outside world? Won't that be an equilibrium? A bastard, evil, hacky equilibrium?
Hello. If I may I would like to try to answer some of your questions:
1) You set the current by adjusting the resistor, thus adjusting the bias voltage at the gates of the MOSFETs.
2) I think Vt and Von are provided on datasheets. (I'm not sure though)
3) Assuming you know what current you want, you need to calculate what bias voltage you need to provide that current. So if you know the voltage drop across the M1 and you know the source voltage and you assume the source is grounded, you just use Ohm's Law to calculate the resistance needed.
With these calculations we are assuming we are using CMOS as opposed to individual MOSFETs so we assume the parameters for M1 and M2 are equal.
4) We can reach the conclusion that the current through M2 will equal the current in M1 because both MOSFETs will have the same bias voltage at the their gates. An interesting aspect about MOSFETs is they can force current based on the bias voltage at the gate no matter what the voltage drop across the MOSFET because of the way MOSFETs work.
5) For BJTs, Vbe is almost always assumed to be about 0.7 V (I was taught
0.7 V, but I guess others were taught 0.6 V)
6) Maybe scary to hobbyists because they do not usually think about the detailed math behind the components. But looking at these components as an engineer there are some fairly long equations associated with MOSFETs.
7) Maybe, but will the BJT last until that equilibrium is met? Also, how likely is it one can set the circuit to create that equilibrium?
4) Will it? How can the M2 force a current? Suppose i do not stick a power supply at M2... what will happen then? What *property* of M2 changes which ALLOWS the current to be that value? Please do not use the word 'force' because unless we are talking about an inductor, it can *NOT* *FORCE* a current.
5) what if i stick a voltage across those terminals? like 9 V? will it still be 0.7 V - 0.6 V ?
Oh and thanks for taking the time!
George Tsiros
4) You mean: will the currents be equal? In theory yes, in real life the currents will be very close. How close? Depends on the materials of the CMOS or the individual MOSFETs and imperfections in the manufacturing.
For nMOS, when you apply a positive voltage at the gate, a certain amount of current must go from the drain to the source (technically source to drain) because the positive charge attracts the electrons from one n-type substrate, through the p-type substrate then the other n-type substrate attracts the electrons. Thus the amount of current flowing through an nMOS depends on the gate voltage.
In theory, with no voltage source there would be no voltage drop and, according to Ohm's Law, no current flow. In real life, doing so would likely damage the CMOS chip because there is almost no voltage drop (maybe pV) and the device is trying to push a current.
As far as I know, no property changes. M2 can push that same amount of current as M1 because they have the same gate voltage.
Sorry, "force" is a term one of my profs used and I did not realize it is incorrect
to use it when referring to MOSFETs.
5) 9V at the base and gnd at the emitter? You would likely fry the BJT. The base and emitter are like a diode, which is why we assume a 0.7 V drop (like we do when assuming 0.7 V drop diodes), so doing what you ask would push too much current through the BJT and cause a problem because there is a 9 V drop when there should be a 0.7 V drop, approx.
Also, looking back at my old notes: 0.3 V
I'm gonna push you more and i hope you've got some juice left, because we've not even started yet.
"In theory, with no voltage source there would be no voltage drop and, according to Ohm's Law, no current flow. In real life, doing so would likely damage the CMOS chip because there is almost no voltage drop (maybe pV" So the transistor would be damaged... because... ? Too much power? But there is no current, so there is no power dissipated. No current flows anywhere. Not G-S, not G-D, not D-S. There is no power dissipated anywhere. There is no high voltage applied anywhere. Think it a bit more, i will be waiting.
"the device is trying to push a current" What is the mechanism behind that pushing? As far as i know, the mechanism via which the transistor ALLOWS current to pass through is gathering freely-moving charges between the D-S to create a channel.
"Sorry, "force" is a term one of my profs used and I did not realize it is incorrect to use it when referring to MOSFETs." My professor is doing that too and when i asked him what does it even mean he couldn't respond. When i asked him "but the transistor doesn't actually FORCE a current, because it CAN NOT force a current?" he replied via verbal abuse towards me. Great teacher.
5) ok then, 9 V, AND a 10 μA current limit. what would happen then? What is the Vbe voltage then?
I have an example: there is a big 555 timer kit. This kit, with discrete components, emulates the die of the 555. Since the actual die of the 555 ic houses multiple current sources, they made current sources with discrete components. Et voila: an example ;). Cheers
it sounds like people are trying to use a current mirror were they should be using a constant current cct. current mirrors are used in power amp for diff reasons
Perhaps this is a dumb question. But I am looking for a way to control an ac motor, namely an o gauge train layout with an arduino. Would this work? Control the voltage of the ac current by having it mirror the dc voltage? Or do you have something else you could recommend.
+Kevin Rutledge It won't even work that way. And anyways current mirrors are used in 99% in ICs instead of resistors to set the bias currents, because it's easier to do transistors than resistors on silicon die. End of story:)
Kevin Rutledge there is always a way to do it but I dont know if theres a way to get from DC to AC, as opposed to the other way around. There is certainly a way to do what you want to but it might not be practical or even safe :)
You would use an AC-AC converter (aka a single phase AC controller) such as a triac, or two SCR/thyristors. The digital pin from the Arduino will allow current to conduct in specified direction at specified interval. There's some example code of AC controllers.
This has absolutely nothing to do with current mirrors though
Pretty late to the party, I hope to God you're still active here cause I need some assistance on a research project.
If your load was a DC motor, could you vary the current through it by varying the resistor on Q1? Assuming voltage is constant
What happen to the "Pete" videos?
According to Pete has been going this whole time?! Why wasn't it in my subscriptions page?! I have some catching up to do...
Looking at this 10 years after it was put out as a total beginner to electronics, everything else PALES in comparison!
Baby photo runaway is a problem with current mirrors.
You dont HAVE to use a constant current on one side, many circuits can call from use of having current on one side be the same as the other such as current monitors and compactors
nice lecture sir
sir can u please suggest me a book to refer for basics on mosfet technology
Not sure his 350mA argument is valid for normal leds. All the leds in the string will carry the same current, and the same current will pass through the transistor, only 20mA
COB LEDs use multiple amps and high voltage.
More videos pete they are awsome
I be interested in your thoughts are on why a discrete configuration would not be suitable for the following application: Regulating the current through both triodes in a dual triode vacuum tube (such as an ECC88) when being used as a balanced phase splitter [see: digilander.libero.it/essentialaudio/hybrid_circuit.htm]
Dave Jones EEVblog did a video on a 555 timer kit. That had some current mirrors in them. watch?v=U6qZPx4uD0g#t=2296 Not sure if the real 555 has current mirrors also...
Arduino switching PWM mode... I am thinking about that right now.
people are confusing current mirrors with constant current cct which cct are very similar but different
The user asked about parallel LED strings, and this guy began to explain the problem of thermal runaway with a single string. Yeah no.
Can somebody give a list of applications current mirrors are regularly used in? What types of instrumentation or devices? Kinda curious to where they are useful not where they are not lol
They are used all analog ICs such as comparators and opamps. They are also used to generate the voltages in voltage regulators, using the so called bandgap reference technique. As far as analog mirrors, I have used them to allow dimming of lots of parallel LEDs all driven from a single low voltage source.
***PRACTICAL APPLICATION
wiki.analog.com/university/courses/electronics/text/chapter-11
Figure 11.21 A combination of the Widlar mirror and peaking current source provides further improvement in the regulation of a variable input current can set up a charging cycle for lipo charger
LM3909 has a current mirror.
i need ur help
man this guy looks like johnny depp
I do not understand why you chose this terrible camera angle !!!. Didn't occur to you that things to the right of the board will be a bit harder to read ? ..
why are you laughing?
11:41 to say this is a current mirror. They have serious flaws. They provide no value as each application that might need current limiting has better alternatives. Mmm I think that was a waste of 11 minutes of our lives. Sorry.
🌟👎🌟
It's good.....
but you are a bit distracting.. you know....
You know, this guy could almost be interesting if he didn't always sound so condescending.
He doesn't sound 'condescending'. I understand though what you confuse AS condescending.