BTW, this should be one of my best quality videos ever, no extra compression using Handbrake. The file size was maybe 3 times bigger than my usual uploads.
***** It is not wasted. Many people like to watch in full HD on a big screen, and often they see many little details. This is my new workflow, so all videos will be like this.
No problems with video here. Another great video Dave! What I like is not only that you discussed it's design and built it up for us, but you went to the trouble to explore the little traps as well as examining the error and noise. Education at its finest!
When I first clicked on the video, I complained "WTH another precision current source video? This is boring..." But after 5 minutes, I am completely absorbed in the video. Awesome videos as always!
Hey wait a sec... 19:24 the non precision resistor is 9.91 Mohm, so the resulting current should be 10V/9.91 = 1.009uA, not 0.991uA that Dave says. Which we finally see measured at 23:20.
I get the impression that you can do all of this kind of stuff in your head and don't actually need to breadboard or measure anything, you just do it for demonstration purposes, thanks for taking the time and trouble !
Here YT let me select 1080p, thanx dave ! Great vid by the way learned a lot on the constant current thing. This is not on my focus, so i really do not care much about constant current sources. But you have done a very good job on telling us how they work.
I know this is old, but Dave you just solved an issue I’ve been having for weeks building a precision programmable constant current source for testing LED Luminosity. Used a rail to rail op-amp configured as a voltage follower and never have been able to get the damn thing to drop its current below 500 micro amps.
My guess : Either Dave needs a way to check the accuracy of the µCurrent, or we'll soon see a µCurrent Platinium edition with 1A, 1mA, 100µA and 1µA current sources built in... :)
I know this is an old one, but thanks, it is exactly what I need right now. If I can only find a source for the REF102 at 0.025%. Everybody seems to be out of stock. Paul A.
Yeah, took a while for the video to finish processing, we're a bunch of geeks who just can't wait to see the latest EEVblog :) Just youtube processing taking its time to finish, as this is another fairly long vid.
Hi Dave Are you planning a "micro Ohm" after the microcurrent? The precise low current sources could be used to measure resistors with low values using the voltage range of the multimeter just as you did with the microcurrent. It would also fill a nice gap since multimeters are not very good for measuring resistances well below 1 Ohm.
Enjoying your well done videos and projects, thank you for the excellent quality, theory, testing, and techniques. There are a handful of providers doing this and you are one of those in the "Top Ten" in the world from my view. As a revisit, these would be great for the "legacy old school analog VOM/AVO meters list the Simpson 260/270 or the Triplett 310 series handheld VOM. One using 50ua/250mv and the other using a 60ua. Great Videos, theory review, etc. Oh by the way, love the eevblog multimeter ( Xmas Gift from my son. )
I'm currently in Engineering school studying energy engineering, love watching your videos and hope to achive the same intuitive understading of circuitry as do you have :D
Yep, add a diode in series, then you've got quite a current source and a temperature sensor (with some specific noise distribution) in series with the load. Would it not have been better to use bipolar supplies or lower the ground of the load?
Thank you for your video! I use this precision current source to generate 10 uA current. The current changes by 0.5 % when the load resistance changing from 50 kOhm to 150 kOhm, but I use unscreened wires. How can I additionaly improve the accuracy of current?
Probably the opamp input bias current I assume. Use a good opamp with low offset voltage and input bias current like an LT1012 (is not single supply sadly)" I know, it's a bit late.
the absolute accuracy doesn't matter so much in these things, more important is tempo and long term stability; the sense resistor needs to have low tempco, but also excellent load and shelf life stability
Do you know of any silicon part that can do precision current out of the box like there is for voltage? The reason I ask is because it looks to me like the biggest weakness in this circuit is the fact that you need to rely on 2 precision devices to make this work.
I'm not shure, but there isn't so precise current sources. You can get only something about 0.5% (not better) current source, with a small choice of nominal current values.
Hi I have made a constant current source using LM317 on bread board. I am generating a constant current of 100mA. The input to the regulator is 3.3vdc. 100mA constant current is fed to a resistor of 100 ohms. I should measure a voltage across resistor of =10mA x 100= 1000mV=> 1V, but the voltage across the resistor I am measuring is 3.3vdc. why so? I want to use the constant current source to find the unknown resistor using current and voltage How can I do this?
Could you not reduce the input bias current error by a couple of orders of magnitude by adding a (normal 0.1% precision) 10k resistor into the negative feedback path (ie match the resistors, feeding the input bias currents)? That would push the output of the opamp up by a tiny amount (via opamp action), and therefore the 10V output from the reference up by the same. And that would provide (almost) exactly the additional amount of input bias current for the +ve input of the voltage follower? Standard trick worth mentioning? Doesn't deal with input offset voltage, or input offset current obviously, but eliminates a major source of error?
Input bias current doesn't necessarily goes into the opamp inputs. It depends on the input stage topology, right? I mean, a PNP long tailed pair has a bias current of opposite direction of a NPN one.
@EEVblog Hahaha! I love the phrasing of the opamp input bias current explanation at about 17:30 !!! From now on when i'm trying to visualise electrons flowing a lattice my mind will be replacing the boring 'e' with a little bee! How much did the components for this build end up costing? I'm looking to make something very similar, but i want to make it switchable in 0.5mA intervals from 0.5mA up to 2mA. I'm guessing i can just interchange values for R to do that without disrupting the rest of the cct? By the way i love your vids, they are a great resource for me in my studies.
do some arduino + lcd builds ..i recently built a pc fan control with 4 temperature readings for cpu/sys/gup and so on. i have even designed a pcb for it all :)
maeanderdev I guess that's good enough, and you don't have to make a dual polarity power supply. Although after I've thought about it, you could also put a diode in series with the op-amp negative supply and use the diode drop to raise it 0.6V above the power negative and make a pseudo negative rail.
You lost your accuracy as soon as you added that first diode to the circuit and it only got worse from there when you added more diodes. I have run into this exact problem before but in my case I could not 'lift' the ground voltage at the load end, and I was stuck with using a single supply op-amp.. I also solved my problem using a diode but I used it where small changes in voltage across it due to temperature and current did not affect circuit accuracy. I put a schottky diode in series with the ground to negative power supply loop and connected the neg side of the op amp to the negative power supply terminal, this gave me a few hundred millivolts of negative rail to compensate for the limitations of the rail to rail op-amp. The best part was that it didn't matter what the actual voltage across the diode was, as long as it was more than the output voltage of the negative rail of the op-amp.... . You might want to consider doing something like that in your next design.....
BTW, this should be one of my best quality videos ever, no extra compression using Handbrake. The file size was maybe 3 times bigger than my usual uploads.
***** It is not wasted. Many people like to watch in full HD on a big screen, and often they see many little details. This is my new workflow, so all videos will be like this.
Was a bit confused with place the load on a input pin. What are these current source used for? EEVblog
Keep it up Dave! As I am one of the people that like to watch on a bigger screen
It's got better resolution than the real world!
I'm watching in 1080p and can still see what looks like compression artifacts in some places...
No problems with video here.
Another great video Dave! What I like is not only that you discussed it's design and built it up for us, but you went to the trouble to explore the little traps as well as examining the error and noise. Education at its finest!
When I first clicked on the video, I complained "WTH another precision current source video? This is boring..."
But after 5 minutes, I am completely absorbed in the video. Awesome videos as always!
I told my science teacher to check you out, because you are so awesome. Keep up the good work, and happy engineering!
Hey wait a sec... 19:24 the non precision resistor is 9.91 Mohm, so the resulting current should be 10V/9.91 = 1.009uA, not 0.991uA that Dave says. Which we finally see measured at 23:20.
I get the impression that you can do all of this kind of stuff in your head and don't actually need to breadboard or measure anything, you just do it for demonstration purposes, thanks for taking the time and trouble !
Here YT let me select 1080p, thanx dave !
Great vid by the way learned a lot on the constant current thing. This is not on my focus, so i really do not care much about constant current sources. But you have done a very good job on telling us how they work.
You are awsome. These videos must end up as classic mandatory educational material - thanks thanks
I know this is old, but Dave you just solved an issue I’ve been having for weeks building a precision programmable constant current source for testing LED Luminosity. Used a rail to rail op-amp configured as a voltage follower and never have been able to get the damn thing to drop its current below 500 micro amps.
My guess : Either Dave needs a way to check the accuracy of the µCurrent, or we'll soon see a µCurrent Platinium edition with 1A, 1mA, 100µA and 1µA current sources built in... :)
where do I get one?? I think I need it!!
I know this is an old one, but thanks, it is exactly what I need right now. If I can only find a source for the REF102 at 0.025%. Everybody seems to be out of stock.
Paul A.
thanks again for showing how to read through test-config induced noise!
Yes, we enjoyed them!
Awesome work as always!
I always wanted to learn something new in this way!
Yeah, took a while for the video to finish processing, we're a bunch of geeks who just can't wait to see the latest EEVblog :) Just youtube processing taking its time to finish, as this is another fairly long vid.
I really enjoyed this one Dave. Thank you.
good video. need more videos explaining why datasheet circuits often dont work
Hi Dave
Are you planning a "micro Ohm" after the microcurrent? The precise low current sources could be used to measure resistors with low values using the voltage range of the multimeter just as you did with the microcurrent. It would also fill a nice gap since multimeters are not very good for measuring resistances well below 1 Ohm.
I think that's a great idea and also a circuit for low amps to measure parasitic draw .
Enjoying your well done videos and projects, thank you for the excellent
quality, theory, testing, and techniques. There are a handful of
providers doing this and you are one of those in the "Top Ten" in the
world from my view. As a revisit, these would be great for the "legacy old school analog VOM/AVO meters list the Simpson 260/270 or the Triplett 310 series handheld VOM. One using 50ua/250mv and the other using a 60ua. Great Videos, theory review, etc. Oh by the way, love the eevblog multimeter ( Xmas Gift from my son. )
just a great lesson learned, thanks for sharing.
Good audio and video here. Thumbs up.
Awesome, I'm going to build one of these to play with
I'm currently in Engineering school studying energy engineering, love watching your videos and hope to achive the same intuitive understading of circuitry as do you have :D
Yep, add a diode in series, then you've got quite a current source and a temperature sensor (with some specific noise distribution) in series with the load. Would it not have been better to use bipolar supplies or lower the ground of the load?
22:20 - That app note suggests there's a -Vs available. Couldn't you just tie pin 4 of tie op amp to -Vs rather than adding several diodes?
Thank you for your video!
I use this precision current source to generate 10 uA current. The current changes by 0.5 % when the load resistance changing from 50 kOhm to 150 kOhm, but I use unscreened wires. How can I additionaly improve the accuracy of current?
Probably the opamp input bias current I assume. Use a good opamp with low offset voltage and input bias current like an LT1012 (is not single supply sadly)"
I know, it's a bit late.
It depends of current shunt resistanse value too.
the absolute accuracy doesn't matter so much in these things, more important is tempo and long term stability; the sense resistor needs to have low tempco, but also excellent load and shelf life stability
You rock Dave!
Remember when we had to use a mercury battery for precision voltage reference?
FYI, this circuit also has bootstrap issues and as a result is picky about what op-amp you use.
At some point during the video my brain just said "nope, f you, I'm not keeping up with those decimals anymore".
Do you know of any silicon part that can do precision current out of the box like there is for voltage? The reason I ask is because it looks to me like the biggest weakness in this circuit is the fact that you need to rely on 2 precision devices to make this work.
I'm not shure, but there isn't so precise current sources. You can get only something about 0.5% (not better) current source, with a small choice of nominal current values.
At ~23:00, try lifting your circuit up off that static mat by a few inches (assuming it is properly grounded).
Excellent video! tell me for this circuit what value would the load resistor that you used be? Thanks.
Thanks Dave very interesting
Why not use a negative voltage rail for the OPA instead of diodes/resistors
I know its a off the wall thought .. could you wrap your leads to a Ref Ground to reduce your off set noise ?,, kinda quick fix :) ?
Thanks for the video) Is it possible to source up to 30ma constant current with this schematic by adding a pnp bjt like in fig7 from ref102 datasheet?
Hi I have made a constant current source using LM317 on bread board. I am generating a constant current of 100mA. The input to the regulator is 3.3vdc. 100mA constant current is fed to a resistor of 100 ohms. I should measure a voltage across resistor of =10mA x 100= 1000mV=> 1V, but the voltage across the resistor I am measuring is 3.3vdc. why so? I want to use the constant current source to find the unknown resistor using current and voltage How can I do this?
Very nice video thank you. Would it be possible to modify these to pulsed recession low current sources?
Could you not reduce the input bias current error by a couple of orders of magnitude by adding a (normal 0.1% precision) 10k resistor into the negative feedback path (ie match the resistors, feeding the input bias currents)?
That would push the output of the opamp up by a tiny amount (via opamp action), and therefore the 10V output from the reference up by the same. And that would provide (almost) exactly the additional amount of input bias current for the +ve input of the voltage follower? Standard trick worth mentioning?
Doesn't deal with input offset voltage, or input offset current obviously, but eliminates a major source of error?
Please, something for 10^4 times less current? 100uA constant current source, with thermal compensation (as LM334)
hi, I have 24v and 3.3 v supply but I want to 1 ma constant current source. How can I do it.
Input bias current doesn't necessarily goes into the opamp inputs. It depends on the input stage topology, right? I mean, a PNP long tailed pair has a bias current of opposite direction of a NPN one.
@EEVblog Hahaha! I love the phrasing of the opamp input bias current explanation at about 17:30 !!! From now on when i'm trying to visualise electrons flowing a lattice my mind will be replacing the boring 'e' with a little bee!
How much did the components for this build end up costing? I'm looking to make something very similar, but i want to make it switchable in 0.5mA intervals from 0.5mA up to 2mA. I'm guessing i can just interchange values for R to do that without disrupting the rest of the cct?
By the way i love your vids, they are a great resource for me in my studies.
7 minutes up and i´m *not* the first - that´s can only happen to/for dave´s videos ;)
do some arduino + lcd builds ..i recently built a pc fan control with 4 temperature readings for cpu/sys/gup and so on. i have even designed a pcb for it all :)
Or you could use a split +/- 10v supply if you've got one... :)
Lol, once someone in school started kinda argouing with me that diode drop is 0.7V, not 0.6V lol ...
Hi sir ... if this circuit work with the human hand load instaed of load in circuit shematic
UA-cam problem people- change your video settings and it will appear.
Can I use this to drive 405nm laser diodes at 50mW?
Do you have any videos on noise?
no video over here. just audio.
i see video but only 360p...i will wait :)
11:34 - 8 millipercent!
Is it possible to make a prescision 1nA current source?
Yes. It just won't be so easy as 1 uA. 10 MOhm resistor, but you need an other opamp.
Am I the only one saying run the op-amp from a dual supply?
good option. But a series resistor/diode does equally good.You can get the opamp into it's range either way
maeanderdev I guess that's good enough, and you don't have to make a dual polarity power supply. Although after I've thought about it, you could also put a diode in series with the op-amp negative supply and use the diode drop to raise it 0.6V above the power negative and make a pseudo negative rail.
He explains why about 9 minutes in :-)
@@therealjammit That simply moves the diode around on the schematic, not in the actual circuit. It's the same thing.
I don't think Digikey will send me sample resistors when they cost $100 a pop ? Stick with normal Vishay 0.01% types for under $10 each.
The diode stuff is a big disappointment! It turns out not better than a LM134 based regulator, and the parts price is a bit higher.
The video is dark, engineers like to work in the dark. :)
You lost your accuracy as soon as you added that first diode to the circuit and it only got worse from there when you added more diodes.
I have run into this exact problem before but in my case I could not 'lift' the ground voltage at the load end, and I was stuck with using a single supply op-amp..
I also solved my problem using a diode but I used it where small changes in voltage across it due to temperature and current did not affect circuit accuracy.
I put a schottky diode in series with the ground to negative power supply loop and connected the neg side of the op amp to the negative power supply terminal, this gave me a few hundred millivolts of negative rail to compensate for the limitations of the rail to rail op-amp.
The best part was that it didn't matter what the actual voltage across the diode was, as long as it was more than the output voltage of the negative rail of the op-amp....
.
You might want to consider doing something like that in your next design.....
I'm having trouble visualizing what you mean? Could you draw this?
Not seeing any video, just audio.....
Tried again and it works now, must have been Utube server slowdown...
no audio over here just video
I made the mistake of not setting this video private when I was uploading, so it automatically got released before it had finished processing.
EEVblog
mistake...? No it´s a *feature* ;)