#374

[@arashghasemi]

I wish I could make a video like this. I teach at a 4your US university and I tried to make a video similar to your technique. But I could never do that! The video begins with a crystal clear explanation and very nice voice tone completely balanced, and the it gently proceeds to testing and validations and finally a relaxing moments of summary and conclusion. Sir, I want to tell you that you have a very unique talent that can't be copied or mimicked by others ❤

[@w2aew]

As you likely discovered - videos like this take a lot of time to prepare and produce. For example, this video consumed approximately 4-5 hours of work (concept/idea generation, investigation, experiments/testing on bench, preparation of notes/talking material, filming (about 25 clips), editing & rendering of video.)

[@HansVanIngelgom]

​@@w2aewthanks for your time, I really enjoyed watching it!

[@WestCoastMole]

​@@w2aewAnd I imagine when you first started out videos of similar length and complexity took quite a bit longer. Thanks so much for this video and your efforts over the years. AA6XE

[@w2aew]

@@WestCoastMole Actually, my earlier videos took a little less time because I did much less editing, and the "shoots" were essentially one long unbroken recording. I take more time now because I shoot multiple clips and try to prepare good, complete notes.

[@warplanner8852]

Alan, these are a tremendous resource and I find myself accessing your videos as "refreshers" the more senility over takes me. Making your notes available is such a bonus as well! I even downloading the notes and print them on three-holed graph paper for that Forrest Mom's effect. Very gratefully, William, k6whp

[@danielkohwalter5481]

I LOVE this kind of video! We need to come back to basics eventually to bring back what's almost lost for this new generation.

[@warplanner8852]

Alan, these are a tremendous resource and I find myself accessing your videos as "refreshers" the more senility over takes me. Making your notes available is such a bonus as well! I even downloading the notes and print them on three-holed graph paper for that Forrest Mom's effect. Very gratefully, William, k6whp

[@gregorymccoy6797]

This is very much the sort of tutorial style I appreciate. Discrete component fundamentals get glossed over in many modern day projects that focus on ICs or Controllers. I can pop open my textbooks but this is so much more palatable.

[@jimbesselman7939]

Thanks for another great explanation video. You re-awaken some things I learned years ago. I often get more from your videos than from some of my engineering courses 50 years ago! Thanks

[@mahoneytechnologies657]

Yor series is a Great Learning Resource for Everyone! Thanks!

[@sasines]

Alan your videos are the absolute best. The continuing education you provide through these videos is priceless. As I often do follow along at home this was an inexpensive exercise. The ones that require more expensive equipment i.e. mini circuit devices are great too but a bit more expensive. The good thing about those videos is they justify my purchase of some cool test equipment! Keep up the great work i look forward to the next one. 73 de W3AL.

[@technobubba4]

Great presentation from the "Master Jedi". You are the bees knees Sir !

[@jp040759]

Great to see B2B videos again. I just love you printing and your ALAN-CAD sketches.

[@andymouse]

Fascinating, never realy used a JFET so I learned something today...cheers !

[@chrisabad6666]

Each time I see a new Back to Basics video from you just makes my day! Each one is really a pleasure to watch and you make this so easy and straightforward to set up on the Bench and get to work ourselves. Thanks so much Alan!!

[@ats89117]

Excellent video, as always. After I saw this, I compared the cascode arrangement to an arrangement with J111, J112, J113 in series with grounded gate and a degenerative resistance. The latter provided a higher impedance source with the same burden voltage in LTSPICE. Of course, as you mentioned, the manufacturers never figured out how to make JFETs with consistent performance, so a lot of testing would be required to see if this really works in practice. This was always a problem when manufacturing low noise transducer preamps. Of course you could pick out ones with similar characteristics, but then you would have to hand solder them to the boards. Then, the medical imaging people came in with big bucks and insisted on low noise JFETs that had consistent characteristics. So now they are available for about $100 apiece for a JFET with 0.8 nV per rt Hz performance that has been binned by the manufacturer prior to inclusion on a reel. I guess this is acceptable on an MRI machine, but not in any of my applications...

[@anders4u222]

Thank you! Have followed you for years and learned a lot.

[@dl7majstefan753]

Thanks for this basic video; it´s good to repeat "old" knowledge not to forget it in our digital days!

[@Batmule]

Good info. Nice presentation. Thanks!

[@terrynicklin417]

Another wonderful BtB, thank you. As always, great theory without excessive maths, and practical application.

[@DasIllu]

FETs and BJTs where topics in trade school, but i always felt like either the teachers didn't understand them beyond the Wikipedia article. Stuff like this is what you'd want to know. But they would come up with more complex solutions, that, yes would do a more precise job and so on. But this really helps to understand the behavior of parts better. And (and this is where my job comes into play) it helps you understand circuits in devices that you are tasked to repair. The world is full of devices where they pull off that one trick that stops you in your tracks when troubleshooting. Thanks

[@TheStuffMade]

I've had a few projects where I wanted to use JFETs as current sources, but I found JFETs to be very hard to get hold of these days, so I've been using LM334 instead.

[@w2aew]

Yes, there are a lot of parts that can be difficult to find these days.

[@keithfisk5641]

Great video. Siliconix used to produce a range of JFET current sources. I don't know if they are still available. They were a JFET with gate to source resistor integrated into the package and just 2 leads brought out. Worked great and helped reduce the component count for designs.

[@chitlitlah]

I think those are called current limiting diodes.

[@stevejagger8602]

Some years back I made a Triax cable tester which had a 9v battery supply and LED indication. It needed a current limiting circuit so I used the cascode constant current source you have described. It was easy to set up and the current remained constant over the typical range of battery voltage variation.

[@DavePKW]

Wonderful presentation as always. Thank you.

[@garygrove6643]

Thank you for posting this video on FET's ! I just found a bad one in my IFR -1000S. Gary Grove / 73's

[@dslee365]

Another great video! Your back to basics vids are always helpful and clearly explained!

[@JoseSilveira-newhandleforYT]

Thanks for another very interesting and informative video, Alan!

[@malcolmbeale4970]

Always an excellent educator... many thanks.

[@SuburbanDon]

Great job Alan. I may try this at home. Thankyou.

[@teamtacoslick]

I love these back to basics videos. Thank you.

[@dimitrioskalfakis]

useful trick with the cascode if you have large power supplies otherwise the compliance of the output is a serious limit for small VDD.

[@ke4est]

Another video from Alan, well I know what I am doing for the next 15 minutes!!

[@zxborg9681]

This was a really clear and useful explanation. When I did analog board design, I'd almost always use a bipolar current mirror as the core of my current sources, but I can see how this this might have saved me a few components. Nice.

[@w2aew]

They can save a few components, but they also are not as predictable (lots of part-to-part variation).

[@alaricsnellpym]

Nicely explained, thank you! And I'm oddly happy that the quiet rattly scrapey sound that occurs at least once in the background of all your videos is still there after all these years. In a world of change and uncertainty, some things can still be relied upon 😅

[@w2aew]

The sounds of the hot water boiler and electric flue damper.

[@SuburbanDon]

@@w2aew Great. Don't mess with it.

[@spagamoto]

I swear, next time I'm stumped on an electrical design problem I'll play that noise as a way to trigger contextual memories. Plus it will calm me down :P

[@frankywatte5646]

As always, interesting stuff. Thanks Alan📡

[@anorthernsoul153]

Wish my uni profs could teach like you. Excellent video! 👏

[@saarike]

Simply excellent!!! Thank you.

[@thomasw6169]

Love the basics series. Excellent content.

[@nickxia3209]

I enjoyed your video a lot, watched most of them, learnt a lot, thank you very much, a fan from china.

[@jgalindes]

Nice and helpful video. Best wishes Sir.

I've used the BF245C for classic 20mA LEDs as current limiter for years...

[@Pentode3000]

Great video again! The link to the Application note needs to be 70596 instead of 70595. 70595 is about Biasing. The information about the cascode could have been a bit more detailed in the relationship between the top and bottom Fet as the top Fet should have equal or higher Idss so the bottom Fet doesn't starve.

[@w2aew]

Thank you - I fixed the link...

[@Stefan_trekkie]

I am at work, watching this in between the calibrations and what I see.. I use the same exact bench multimeter model by Keitley 😃

[@ruisacramento9181]

Great content!!

[@VorpalForceField]

Fantastic content as always ... Thank You for sharing ... Cheers :)

[@W1RMD]

Thanks Alan! So my question is if there is so much variation in components you could theoretically substitute an mpf 102 in place of a j 310 within reason and assuming that you stay within max ratings. Am I correct with this thinking? Take care and I always enjoy your videos. 73's!

[@w2aew]

Sure, an MPF102, and just about any JFET would work the same way, but all will have different I(DSS) and V(GS) Cutoff differences.

[@RexxSchneider]

You didn't explicitly explain that having a source resistor provides negative feedback; to be precise any tendency to increase Id will result in an increase in Vgs, thus opposing that increase (and obviously the same applies for any tendency to decrease Id). One of the results is to sharpen the knee of the curve relating Id to Vd as well as flattening its slope, which represents the output impedance. The flatter knee allows Vd to be less while remaining a useful constant current source. The cascode circuit has the other disadvantage that the lower JFET only sees the Vgs of the upper transistor as its Vds, and that puts its operating point on the knee, giving a lower output impedance. That's why the cascode doesn't offer as much improvement as you might hope.

[@TomSny35]

Very nice video. Thanks so much.

[@bobdoritique7347]

Merci for this video, very interesting.

[@coffeecuppepsi]

Thank you for this

[@_droid]

Thanks! I have been meaning to play with JFET's since I was a kid but because I mostly use junk parts I don't think I've even seen one in real life. I need to buy some. Was that your gas heat turning on? Already that cold?

[@w2aew]

Yes, my gas boiler which is about 8 feet from my workbench.

[@VoidElectronics]

Nice video as always! I think it would have been nice to use the curve tracer to characterize the current sources, not just the plain JFET. This way, we could have seen a graphical representation of the output impedances of the various topologies that you've covered. :D

[@w2aew]

Yeah, I thought about that, but didn't want to make the video too long.

[@w2aew]

I went back and tried the 50uA Current Source, like the one to calibrate Simpson 260 VOM, built using a single JFET and the cascode design, traced out on the curve tracer, including a zoom in to 1uA/div: www.qsl.net/w2aew/youtube/50uA_curves.pdf

[@VoidElectronics]

@@w2aew Nice! The zoomed in curves look exactly as I expected. Thank you so much for going through the effort of taking these pictures. 😁

[@FXGreggan.]

Fracking youtube just stop showing videos from certain people, you've made A LOT of videos since I last saw one from you - I knew something was missing in my life :)

[@jimomertz]

I have used JFETs a couple of times and am aware of the large variability between devices. Why are these so hard to make to narrower tolerances?

[@erikdenhouter]

BJT's have also very large differences, but in 'amplification factor' (beta or hfe). It can vary per device from 10x to 600x. After manufacturing these are measured and divided in groups (like A, B, and C types), and the worst are probably thrown away. I guess they don't that with JFET's. The art seems to be to design stuff where the variation is not important or ruled out, like a self biassing circuit.

[@piconano]

I made many CC circuits using your past video on mirroring using NPN (2N2222A or 2N2904) tranys, with each circuit driving a slow flashing colour LED. Maybe I missed it in the video, but why use JFET instead of bipolar, since both are operating in their linear region?

[@w2aew]

The reasons can vary. Sometimes the JFET is just a simpler solution, or it is what is available, etc.

[@ydonl]

Then there's that low "base" current! Bipolar transistors have pretty low base current, for sure, but the gate current of a JFET can get closer to zero. That's really handy if you want to look at a voltage without disturbing it -- transducers (including musical devices), test instruments and that sort of thing... many kinds of things.

[@ghlscitel6714]

There were current "diodes" available 50 years ago. They were probably made by Motorola. Are there similar devices sold today by some companies?

[@w2aew]

There still are. Just take a look at this listing from a quick search on Mouser for "current regulator diodes": www.mouser.com/c/semiconductors/discrete-semiconductors/diodes-rectifiers/current-regulator-diodes/

[@ivolol]

If you put your audio into a free program like Audacity, you should be able to use the noise removal effect to get rid of things like a small fan constantly going in the background

[@McTroyd]

This is really cool, Alan! I could see this being useful driving LEDs, particularly as the season of holiday lights approaches. Not being an EE, I wonder how much heat this would dissipate? I assume it would be a function of the FET's "on" resistance?

[@w2aew]

Power dissipation is simply the drain current multiplied by the drain-source voltage. Given that it is a current source, the current is a fixed value, and the voltage across the drain-source will simply be the supply voltage minus the sum of the voltages across the devices connected between the supply and the drain. For example, if you were running a 12V supply and the current source was set to 5mA, and you had 3 LEDs in series, with a voltage drop of 2.5V each, then you'll have 12-(3x2.5) = 4.5V across the drain-source. So the JFET would be dissipating 4.5V * 0.005A = 22.5mW.

[@McTroyd]

@@w2aew Oh. Easy enough! Thanks for the explanation.

[@grimreboot]

Thank you for the upload! Out of curiosity, what application would something like this be used for?

[@w2aew]

There can be a lot of applications for these circuits. A constant current feeding a capacitor results in a nice linear voltage ramp which can be used in a lot of applications in timing, etc. A current limiter can be used to control how quickly a device charges/discharges, etc.

[@grimreboot]

Thank you! :)@@w2aew

[@M0NTVHomebrewing]

Thanks Alan. Really helpful content as always. 73, Nick M0NTV

[@aduedc]

Another great instructional video. I learned how to cascode JFETs from this program. Therefore, it seems that JFET cascode device can easily be used on top of other devices such as BJTs, and MOSFETs without much biasing network. As you know we go to great length to bias the BJT and MOSFET cascode device to make sure the casecode device stays in saturated region; whereas, here you do not need much biasing network, and the cascode device always stays in saturated region naturally. I wonder if you have two matched JFETs, how can you make current mirror out of them? Anyone knows how? Oh by the way I used JFET current limiter for safely discharging high voltage capacitors, instead of resistor, I use LED, when the LED goes off I know the capacitor is discharged to few volts, and another JFET current limiter with resistor in the source, which is connected in parallel with the LED current limiter will complete the discharging the capacitor.

[@w2aew]

Yes, matched devices can be used to make a mirror - similar to how it's done with BJTs.

[@aduedc]

@@w2aew Thanks for the reply. However, I have some concern that it would forward bias the JFET. i.e . If we connect Gate to Darin of JFET like MOSFET to generate reference voltage for the mirror JFET. Plus there is no feedback mechanism to set the correct Vgs voltage value. The only way that I can think of is to put a battery ( i.e. voltage source) between drain and the source. Say put a 3V battery with positive connected to drain and negative connected to gate. Which means 3V zener diode connected from Drain to Gate and the gate connected with high impedance say 1MOHm to the V- supply, and the source connected to GND with a +V GND -V type of supply. The the gate of this JFET can generate reference voltage for mirroring JFET, with its source connected to GND.

[@RexxSchneider]

Matched pairs of JFETs are near impossible with discrete devices -- there's just too much variability in the fabrication, and then you have the effects of temperature being difficult to match. I have some JFETs from the 1970s where two devices are fabricated on the same die and they match quite well when used aa a differential pair (this was before the TL072 series were commonly available and I needed an FET-input opamp). You could use enhancement mode FETs (generally mosfets) to make a current mirror, but as you say, the biasing needed to do the job with JFETs is too complex when a pair of BJTS (preferably matched in a single package) will do the job simply.

[@aduedc]

@@RexxSchneider Thanks for your reply. I am an Analog IC designer, so I usually use transistor for current mirrors ( such as MOSFET, Finfet, BJTs, ...) each of them has its advantages and disadvantages. We match the transistors easily by using several transistors close to each other in common centroid configuration to average out their characteristics. So, instead of using one to one transistor we use say 10 to 10 transistors to get a better match. The problem with MOSFET is flicker noise (1/F noise). JFET is very similar to MESFET (Metal-Semiconductor -- without the oxide layer -- Field Effect Transistor), So we can use JFET circuit topology for MESFET circuit. MESFET, similar to JFET, has much lower flicker noise. That is why I am interested in the JFET circuits. Note that JFET uses PN junction for gate whereas MESFET uses Schottky diode for the gate.

[@RexxSchneider]

@@aduedc Thank you, that's interesting. Of course, you can take advantage of the inherent matching of devices fabricated on the same chip to obtain results that are near impossible to achieve with discrete devices, but the idea of paralleling 10 devices to do the job of one in the same chip had never occurred to me. When I designed analog circuits many years ago, the toughest part was always to make sure that thermal mismatches didn't push the circuit out of spec. I never knew whether a production pcb would ensure that both semiconductors in a pair would experience the same temperature variations unless I specified they were bound together on a heatsink, or the whole module was potted, and both of those cost money. I know that FETs have an advantage over BJTs in flicker noise, but I would still consider that any depletion mode FET would be tough to bias effectively in a current mirror configuration.

[@JaenEngineering]

It's channels like this that remind me that the internet maybe wasn't such a bad idea afterall.

[@romainchristophe1796]

Perfect

[@Jim202030]

Love the vid.

[@KX36]

Would have been nice to see the circuits back on the curve tracer as 2 pin devices

[@w2aew]

I went back and tried the 50uA Current Source, like the one to calibrate Simpson 260 VOM, built using a single JFET and the cascode design, traced out on the curve tracer, including a zoom in to 1uA/div: www.qsl.net/w2aew/youtube/50uA_curves.pdf

[@KX36]

@@w2aew Thanks, that's very kind of you! What's the source of the wiggle on the zoomed in plots? Is it artefact? oscillation?

[@w2aew]

@@KX36 I think it is an artifact - likely some switching logic cross-talk or a power supply filtering issue inside my old 576.

[@garygrove6643]

I had heard that there is such a circuit as SCR's that are operated in the oscillation mode. Although there is no need to have them work at RF frequencies, they have been known to oscillate in the audio spectrum, but I haven't seen any circuits that accomplish this. If you have anything on this, please do a video. If a circuit of such does exist, I just gotta see it! 73's, Gary Grove

[@w2aew]

I've only ever seen relaxation oscillators made with SCRs.

[@garygrove6643]

@@w2aew The one circuit that I did see was using the ac line as its power supply, but that was not an "honest" form of oscillation.

[@accountant1653]

👍👍👍👍

[@TheSiliconchip]

Super (AP2SM)

[@waynegram8907]

When the FETs output voltage is NOT Flat, this is the FETs output impedance? I'm confused when the FETs output voltage is NOT flat what types of problems can it cause with other circuits?

[@w2aew]

When the FET drain current is not flat vs. drain-source voltage, that is an indication of a lower-than-ideal output impedance. This would cause issues only in circuits that rely on a constant current regardless of the voltage at the load. One example might be precise timing circuits that use the constant current to charge or discharge a capacitor to create a linear voltage ramp.

[@waynegram8907]

@@w2aew When the FETS current is not flat the FETS output impedance is lower? I'm confused how the FETS drain voltage is FLAT while the FETS drain current is "Not" flat at the same time. I would think the drain voltage and drain current would both be either flat or not flat but they are also be opposite at the same time which doesn't make sense to me.

[@w2aew]

@@waynegram8907 It is all shown in the curves. The drain current represented on the vertical axis is relatively constant (flat) as the drain-source voltage, represented on the horizontal axis, varies. The "flat" portion exists for drain-source voltage above a few volts. BJT transistors are similar.

[@waynegram8907]

@@w2aew The Horizontal axis is the drain voltage which is not flat it has a SLANT. I don't understand why the drain voltage is not flat. I'm not sure why the FET output impedance would vary or contribute to the drain voltage to be slanted not flat. I guessing the doping resistance or junction resistance or the FET channel resistance is causing or contributing to the drain voltage being slanted not flat?

[@w2aew]

@@waynegram8907 You're not interpreting the graph correctly. The graph is showing a change in DRAIN CURRENT in response to the DRAIN VOLTAGE change. The SLANT is on drain current in response to drain voltage value.

[@ornithopterindia]

👍

[@kippie80]

But what is uniquely different about a JFET? I had bought some JFETs from China, they were fraud devices, they were really BTJs. They did not work in my device (a DIY capsule microphone that needed a balanced circuit). The knock-off was un-balanced. I bought a genuine JFET and it worked.

[@kippie80]

The circuit I'd made was RF.AMX10 .. search for that. That circuit WILL NOT WORK without a JFET. Not sure why exactly. Somehow the JFET 'floats' whereas a BJT is biased to ground one one side.

[@w2aew]

@@kippie80 This only works with a JFET because it will self bias with a negative V(GS) due to the voltage drop across the source resistor. A BJT wont work because it needs a positive V(BE).

[@tze-ven]

I like your American spelling of "Exercise".

[@R2AUK]

👍❤