EEVacademy #8 - Howland Current Pump

[@dough9950]

Thanks, the family is very proud of this circuit...

[@power-max]

Username checks out. :P

[@michsol]

damn, why was the username changed though xD

[@deusimperator7692]

I used one of these in a design at work recently. I needed a very tightly regulated low current output that was tolerant of wide variations in load, both resistive and inductive. This circuit worked great.

[@steverobbins4872]

Nice video. I gave it thumbs-up. But I'd like to offer a few constructive criticisms: 1. You never showed the original (basic) Howland circuit. In the basic Howland, the load is connected to the non-inverting input of the opamp, NOT the output of the opamp. When you insert an extra resistor between the load and the non-inverting input (like you did) that is the "improved" Howland. 2. You didn't explain the restriction associated with the improved Howland. The resistors have to conform to certain ratios in order to keep it balanced. This is important for explaining why a second opamp overcomes those restrictions. 3. You didn't explain the biggest advantage of a Howland: it's output is bipolar (two-quadrant operation).

[@RLDacademyGATEeceAndAdvanced]

Excellent

[@istvan.design]

A little bit of feedback: 1. You should start with interesting applications of a circuit, for instance we need an improved Howland pump for either a dc constant current source (boring) or if we need to measure skin impedance (impedance imaging/tomography), we can also control muscles or stimulate nerves very precisely even if the impedance changes. 2. An improved Howland pump is difficult to implement in real life, you usually need to isolate it with a linear optocoupler or an opamp if you want to drive it with an audio signal. 3. You should actually warn people NOT TO breadboard a howland pump because it will most likely oscillate and show this on the scope, the resistors must be 0.1% matched in real life, this is very important, 1% doesn't cut it. The 10nF caps are actually mandatory because they are cheap to add and they eliminate a source of oscillation. 4. You could show how to buffer a howland pump for higher current output or how to bridge higher voltage (many opamps won't cut it in the 25-30V limit for most applications). 4. With infinite resistance or higher resistance than you can afford with the opamp's capability the circuit will oscillate and draw maximum current (even kill the opamp). 5. High voltage opamps are expensive, low voltage opamps are not that useful.

[@DrFrank-xj9bc]

Nice video with good, lucid explanations. 4 very versatile aspects have to be added: 1. the current source differential output resistance will become infinite, only if these four 100k have pairwise the same ratio. So add a trim pot between 2 of the 100k, wiper into the OpAmps input, to trim the circuit for a precision constant current source, i.e no change of current at varying load resistor . 2. the circuit operates symmetrically to GND. Trivial, but explicitly to be mentioned, therefore it's a GND related current source. It also can act as AC constant current source, if a constant AC voltage is applied at the input. 3. The circuit acts as a precise voltage to current transformer, so it's also possible to create a DC + AC current by providing an additional summing input for separate VDC and VAC inputs. 4. There's a more precise circuit, called modified Howland current source, implementing a 2nd OpAmp, so the common mode voltages at the 1st OpAmp are zero, and the input voltages are related to GND also, that's more convenient for AC / DC voltage sources. Additionally, the small error from the current through the 100k vanishes, AND you can realize the current shunt resistor as a decade switching ladder, so to switch the output current decade wise.

[@electronicsNmore]

He did a great job making this video.

[@6alecapristrudel]

This is probably the best explanation of this circuit I've seen so far! Can you do more analog stuff like this? Maybe OTAs or impedance converters?

[@Audio_Simon]

More analog please!!

[@uwezimmermann5427]

good work there! Just a small little comment from teacher to teacher - you should perhaps start with an introduction or a goal, like saying more clearly that this circuit _is_ a constant current source for the current through the lower of the two resistors at the output.

[@andruloni]

+1 It's easier to see the effect if we know what it is

[@ricardonunes6724]

I agree. I also think that you should explain the advantages and disadvantages of the circuit compared to similar ones. I can think of many other constant current circuits.

[@joinedupjon]

cheers. Liked the format but I think you should look at ways of zooming up on the circuit you're talking about.

[@davidledger5941]

Yeah whooops lol

[@JBPerformanceTV]

So true xD

[@VoidHalo]

I love CircuitMod. Playing around with it is a HUGE reason why I got into electronics.Of course there are other things that played a fundamental role in me getting into electronics. EEVBlog being another big one. Only just found this series and I love it. I was a bit weary with watching a video not hosted by Dave, but you did a great job. Cheers.

[@deanolium]

Good video, and quite a clever system. Showing how it's derived is really useful, and a good way to learn.

[@philiprogers5772]

What software are you using?

[@davidledger5941]

Falstad its an online tool www.falstad.com/circuit/circuitjs.html

[@WacKEDmaN]

Thanks David..good little simple tool...that said, ive seen some people make some strange circuits with it, that wouldnt really work in the real world ("free energy")

[@h0lx]

I self taught electronics to myself with it. Took a piece of a working schematic and figured how it worked by replicating it there

[@philiprogers5772]

What a great bit of software. If only I had this when I was at school, how much further on in electronics would I be by now?

[@amirs5512]

DaveCad 3.0

[@pinolec]

Well done Dave. I like it.

[@pocoapoco2]

I found I was able to further simplify the circuit by replacing the resistor between the voltage source and the non-inverting input and the resistor immediately following the op amp output and replacing them both with wires. An then moving the constant current output from the op amp output to the non-inverting input. changing the value of the resisrtor between the ouput and non-inverting input then determines the current.

[@ThyAnon]

We need more EEV academy

[@siddacious]

Great video for someone like me who is still learning the fundimentals. More like this!

[@Mustakari]

Excellent! I like this style. Clearly explained and the tempo is just right.

[@1973jdmc]

Thank you sooo much for this video. I have an exam next week and couldn’t see the woods from the trees- THANK YOU

[@groveh1]

Nice! Thanks for introducing me to the Howland current pump!

[@NetworkXIII]

Good video David, enjoyed it a lot. Op-amps are amazing devices.

[@JamesPotts]

a) I did learn something, and b) that end picture alone is worth a thumbs-up.

[@NorbertHarrer]

Good job. Explained very well. Looking forward to more stuff from you, David.

[@lezbriddon]

great vid, i have seen something like this in cheap ebike brushless motor controllers to get a stable 12v for the gate drives, because (apparently) normal voltage regulators dont respond quick enough to input voltage drops, which leads to the 12v dropping, not fully turning on the mosfets, and then them exploding due to internal heating....

[@final182]

I like this format, I would love to see more of these videos where circuit building blocks are explained

[@gregfeneis609]

BTW, David, I went to the website for falstad, and just tried rolling the mouse scroll wheel while the mouse cursor is in the black circuit space. It appears that you can easily zoom the circuit this way. I did this on my Win 7-64 laptop with Opera and Chrome browsers. YMMV.

[@gregfeneis609]

SYDFD

[@Edson_Staack]

I laughed with the photo in the end! Unbelievable! :)

[@toniturnwald9890]

Thank you, for uploading, from your films I can learn pretty good the electronic stuff. Cheers and bye bye Toni

[@miketel01]

Awesome VID. Yay David2 is back

[@shomiahmed5144]

Hi there, at the last you talk about some sources of error in terms of a very small current passing through the positive feedback. What if we retained the previous circuit i.e. with a buffer instead of using high impedance? Will that eliminate the small current issue through the positive feedback?

[@joopterwijn]

It would be Nice to show it on a board next to the theorie! For instance load a super cap for a flash?

[@dentakuweb]

It would have been nice to know what the final outcome of the circuit is before even starting. SO... I did an experiment and connected two LEDs to the output of this circuit (using 5V instead of 1V) AND a separate circuit with just a voltage source, 1k resistor and two LEDs. The one with the simple resistor is dimmer and if I go from 1 to 2 LEDs the current reading goes down and 3 LEDs obviously doesn't even light. With the Howland Current Pump I can string a bunch of LEDs together and a COSTANT 5mA keeps flowing through it even with 5 LEDs in series. It starts going down if I add a sixth LED. You see, it would have been nice to connect this circuit and see it actually do something useful like lighting LEDs.

[@fabimre]

Interesting. Btw, which software did you use?

[@fabiStgt]

Great explanation and also very nice software setup you're using to visualize the circuit!

[@teslakovalaborator]

If I get it it we can made something like integrator if we charge cap with current proportional to the V difference and measure the time constant

[@davidledger5941]

Yes! you can. It is one way to calculate the value of a capacitor.

[@simonscherz2869]

So what's the advantage of the Howland Current Pump to a current regulator with Zener diode and transistor?

[@RobertShaverOfAustin]

What simulation/circuit tool are you using for this tutorial. I love how it looks and how easily you made changes. *CIRCUIT QUESTION:* in the latter part of this tutorial you show a "load" resistor on the right side of the diagram connected to ground symbol. What I don't understand is where the current returns to the source. How is that circuit completed? The ground symbol is just a shorthand way to show a connection. Can you replace this symbol with a wire to show the return path? *TUTORIAL QUESTION:* What is the objective of this tutorial? You explained many things but I did not always follow why they were important to know. One example is the _virtual ground_. I think you expected me to understand why this was important and why it would be useful.

[@RiyadhElalami]

Good job David, that was a good reminder for a very helpful circuit. You should do more!

[@gilbertjuly]

nice video, wonder what is the circuit simulator you are using? looks awesome!

[@irawarnaca8133]

Thanks David, great building block. It would be helpful to see how to leverage this in a circuit. Show us an example please. Thanks again, I really do appreciate your videos, I would like to see more of them.

[@henrywilloughby6965]

Love it! I’ve been hoping David would start getting screen time!

[@PilotPlater]

I wonder with the 100k resistors if it's susceptible to noise/interference? I've boj'd together timer circuits with BJTs and capacitors/resistors and to get long enough timer delays you need very high resistances, the BJTs were very sensitive to noise. In my case I stacked a couple stages of BJTs because I wanted a binary on/off, so that's probably where the over-sensitivity to noise came from.

[@k1ngjulien_]

What is the Advantage of this over a normal V to I CC OPAmp circuit?

[@newsogn5148]

Thank you for this!!! I wanted to know more about the homeland current source for a 4-20 ma output in industrial controllers, what software is that???

[@jd_flick]

really interesting explanation

[@ryebis]

Haha that end shot

[@MedSou]

thank you 👍👍👍👍👍

[@dshshajskj6143]

Can i use RC parallel as load and will it effect the current in Howland Current Converter

[@EDGARDOUX1701]

A very good explanation David! Thanks for sharing and hope to see more on OpAmps. I would loke to see opamps with strain gages, I do have an issue with those. Thanks again

[@yced101]

More of these vids pls :) Just gr8

[@jmannUSMC]

Really like how you teach!

[@tomwimmenhove4652]

Very well explained. Good work.

[@athousandprojects]

Great video - thanks

[@thiagoennes]

the picture of dave2 at the end.... always cracks me up. I really like this format. How about some power audio amplifiers?

[@ACTlVISION]

That was interesting but can someone ELI5 what applications is this used for? When would this be used for an LED over just a current limiting resistor?

[@ПетяТабуреткин-в7т]

Thumbs up for the Falstad simulator.

[@Sixta16]

Well, what I completely miss here is the analysis of what effect the accuracy of the resistors and the input offset voltage has on the behavior. Because in most cases, the Howland current pump is a useless circuit that does not give enough accurate current output in many applications. Also the completely floating input reference voltage is not how you would implement the circuit in reality. Sorry, if I have missed something.

[@GadgetUK164]

Great job Dave2 =D

[@janzkrs8006]

I like the short, simple format :)Question though. Isn't this more of a constant voltage source, I mean as said, if the ressistor is not constant the current wont be eather!?

[@PeregrineBF]

Upvoted just for the use of Falstad's circuit sim. Rodrigo Hausen made some improvements (github /hausen/circuit-simulator) which can be handy.

[@amciaapple1654]

@David: Fine video but it begs the comparison with other current sources, such as based on a single JFET and even BJT. Also, application examples would be nice...

[@6alecapristrudel]

Another nasty source of error is mismatch in the diff-amp's resistors. This causes the amp to have some common mode gain as well so it's not only amplifying the difference between the inputs. If the inputs are shorted together and then we connect them to some input voltage the output will still change with the input even though the difference between the inputs is zero. Even worse is that this common mode gain can be inverting or non-inverting depending on the exact ratio of the resistors in the two input sides.

[@davidledger5941]

OH YEAH, thats a pretty good point which I completely forgot about mentioning.

[@Sixta16]

Pretty much fail, as due to the mismatch the Howland current pump is one of the circuits that mostly works only on paper and in reality gives poor performance.

[@gregfeneis609]

Good work, Sir. As I understand it, a HCP can be used to emulate inductors to other circuits. Would you care to make a video about that?

[@EdwinNoorlander]

Why is everything so small. Why don’t use the whole screen?

[@vant4888]

What is a practical usage of such current ?

[@p_mouse8676]

Nice little circuit, although it would be interesting to compare this to some othere options in sense of price, stability and noise. In this case the resistors need to match very well. You can make a decent current source with tge TL431, which are cheap as chips

[@davidledger5941]

Yeah it does end up being quite comparable, resistors being so low cost, I prefer this topology when you need the current source voltage controlled, from a DAC or such.

[@p_mouse8676]

Why wouldn't that be possible with a TL431 or similar?

[@davidledger5941]

Well with this circuit you don't need any extra parts, wouldn't you need some extra's for a TL431 current source? You might not need much if you did a TL431 current sink. How would you do it?

[@p_mouse8676]

Maybe, don't know just right out of the top of my head. But in general, just a basic voltage controlled current source only needs an opamp, some kind of MOSFET or transistor and just 1 sense resistor (maybe two passives to get better stability) Like I said, this is a cute little circuit, but not very precise because of the resistor mismatching.

[@davidliddelow5704]

How accurate do the resistors need to be?

[@urugulu1656]

well explained although i remember the howland source a bit different in topology...

[@davidledger5941]

This is the improved howland current pump, very slightly different.

[@Audio_Simon]

P.s. Thanks for the easier topic this time around!

[@abc9999ful]

Wich spice do you use??

[@andreschneider3147]

Hi, Dave and Dave! Can you give us some update about EEVdumpsterHack channel?

[@topherteardowns4679]

Awesome video. Even awesomer outtro pic

[@johnpaulsalazar5382]

What op amp did you use?

[@TheDefpom]

What software was that Simulator? EDIT never mind you already answered it.

[@stonail665]

Big fan of Falstad ,also there are 2 mobile apps called icircuit and Everycircuit which is the same software. nice tuto great explanation.

[@Biela2008]

More of this please! :D

[@gorillaau]

Nice Video, David.

[@station240]

6:38 Couldn't you replace the unity gain buffer with a simple diode ? That would stop the current flowing the wrong way, and as you stated earlier the entire circuit counteracts voltage offsets.

[@SiloxMedia]

Yeah, I think that would work.

[@domenbull]

It wouldn't work, since you need to have gnd; There must be current (no matter how small of a current), because you need the drop of voltage on the resistors. Useing a diode you would basicly make one resistor useless; I hope you understand

[@domenbull]

tinyurl.com/y9ubrjzy here is simulation of it

[@yavuzturan6830]

damn. whole circuit is in the url guess.

[@domenbull]

Resistance of the load should be many times smaller than the resistance of the of the circuit, basicly circuit must have as high of impeadeance as posible (6:40)

[@abcvideoyoutuization]

Could you show a working pump?

[@Apocalypse033]

wow,it's a amazing software online

[@bm830810]

excellent explanation, please increase the size though

[@gorillaau]

What's the software you are using.

[@kaizen9451]

www.falstad.com/circuit/circuitjs.html

[@proluxelectronics7419]

Why so small??

[@davidledger5941]

The application doesn't zoom well. Wanted the video to be in native resolution, for maximum crispness.

[@christophertstone]

What doesn't zoom well, Falstad? It's vector based... And there's horrific letterboxing. This looks like something my grandma forwarded from the Facebook app on a Nokia 3310. Excellent content, but the production quality is seriously distracting.

[@steverobbins4872]

I too could not see the circuit very well at first. Then I changed the KZbin settings to HD and I could see the circuit very clearly. (David, I would suggest for your next video you start by telling your viewers to switch to HD.)

[@stephenkemp6870]

Yes, switch to HD. Worked for me. Thanks David.

[@salerio61]

Christopher Stone Everyone's a critic

[@emcgon]

More like this, please: I miss tutorial-style videos on EEVBlog since Fundamentals Friday stopped happening. One suggestion: it would be nice to mix the theory with a practical demonstration of an implementation on a breadboard or something like that. Loved it, though.

[@liviupeles8666]

What's the name of the simulation software ?

[@marianoaldogaston]

Great work, I Think it will be great next time after explanation, do the circuit on breadboard and make some measurements

[@illustriouschin]

didn't learn what it does and how it is used but pretty good video.

[@gianlucalocri]

Hi Dave! What about connect the load directly on the non-inverting terminal of the opamp? Like this fourier.eng.hmc.edu/e84/lectures/HowlandCurrentSource/node1.html

[@AttilaAsztalos]

But but but... what happened to the Old Pump of Howland...?

[@牛贺龙]

what is the application?

[@justinnamilee]

Yeah, great content, just zoom up on the subject matter a little more~! Keep it up Dave2.

[@viktordunaew5119]

How the software is called ?

[@Audio_Simon]

3:30 Looks very similar to the 'Super Balanced' audio input in a book by Doug Self. Designed to load each half of the diff input equally. www.douglas-self.com/ampins/balanced/balfig12.gif

[@pirateman1966]

That software looks so much like Circuit Wizard, but free and online!

[@DeRaafMedia]

Needs more breadboards!!

[@casperhito]

Dave 2 working on the goatee. nice

[@gorillaau]

Casper Myers I reckon he'd be able to pull off wearing a beret.

[@RLDacademyGATEeceAndAdvanced]

Good

[@MajorAbuse]

Please redo with a larger schematic.

[@chrisbalfour466]

Bah, that's what an LM317 wired as a current regulator is for. Discrete opamps are just a pain.

[@6alecapristrudel]

This can be used as an AC current source and can do small currents, it even has a differential input on top of that. The LM317 current source is DC only and has a minimum current requirement of 10mA-ish to work, though granted it can do much higher currents.

[@chrisbalfour466]

Except a (non-polarized) capacitor in series with a load given an AC source creates a nearly 100% efficient alternating current source (for small-ish loads like a pair of anti-parallel LEDs). The average current is determined by the capacitance. Just three components to build an LED light that runs straight off of 120V 50/60Hz ;)

[@6alecapristrudel]

Good for you. The Howland is pretty much useless for driving LEDs that really need the constant current since they're power LEDs that need much more current than the 20-ish mA an opamp can give you. You can get away with just a resistor for low power ones. What you can really use the Howland for though is oscillators, waveform generation and such. Have it drive a cap, feed in a square wave and out pops a triangle wave.

[@chrisbalfour466]

Sorry, opamps make me cringe. I have an electronics degree, but I'm an unemployed hobbyist and I've only needed an opamp once in the past ten years, for measuring brief voltage spikes too fast for an ADC to handle. For a waveform generator I'd actually ad-hoc something with a PWM through a capacitor or an H-Bridge and PWM. Only if either of those wasn't adequate would I consider an opamp. If I needed a 1MHz or higher frequency, I'd might have to break down and give the Howland a shot... But I'd exhaust all other possibilities first.

[@6alecapristrudel]

Sadly the cheap opamps are kinda shit at MHz frequencies, they can work but accuracy is terrible. Hell some can't even make it to 1MHz at all. The ones that can are expensive, though usually they're also much less hassle, I also do just hobby stuff and I want them all because they're magic.

[@johnsenchakinternetnetwork2025]

Couldn't you just connect a transistor to have gain and then use it for a current source from emitter to collector ?

[@davidledger5941]

There are many ways to do it, this is one way. One of its advantages is that its voltage controlled. And the control voltage could be ground referenced.

[@conansilva426]

WR KITS.. BRAZILIAN CHANEL ELETRONICS