The 2N3904 says to the 2N3906, "nice package". The 2N3906 says to the @2N904, "nice legs". That's an example of a "complimentary pair". But when the 2N3904 says to the 2N3906, "you complete me". Now that's a complementary pair!
If you use two diodes on the input to bias the transistors, the crossover distortion will be very small. If used with the op amp, it will be phenomenally small. With the op amp but without the diodes, you will still get noticeable distortion.
the Roland System 700 VCA has that, i have build that, using the same values of resistors (50hm) and at idle, i noticed they get hot, but i used a cloned by me LM13700. Now, i have to unsolder the 90º headers to be able to debug the circuit.
The diodes forward voltage drop and temperature coefficients will have to be almost exactly matched to the transistors for that to work or you will risk thermal runaway or distortion. It's better to use the B-E junction of identical transistors to act as the "diodes" do the biasing. Might cost a few pennies more in the price of the transistors over the diodes, but it will save much more in time and energy trying to match diodes.
Yes, an engineer can put together a design to nominal/typical parameters without a simulator. However, SPICE and other tools are useful, if for nothing else, to check the tolerances of the devices in the design. While any engineer can get a prototype working in the lab, the usual requirement is to produce good yields through manufacturing and reliable products in the field. Part variation is expensive and time consuming to verify in the lab, but tools like SPICE and other simulators, as well as mathematical tools, can calculate extremes and determine margin relatively inexpensively. This can also check for violation of maximum characteristics, such as your overcurrent situation. Since in my career with many companies and being responsible for many products, I was always tasked, as part of the requirements, to create reliable and manufacturable systems, and I would have had to fire any engineer who would not simulate or calculate the variations. Luckily, I never had an engineer working on my team that did not know how to prove his work and seek to do just that. My first six years of my career was mostly spent fixing and cleaning up the empirical designs that other people pretending to be engineers did before me. I have a low tolerance for low quality work, because it costs my company a lot more than any savings because engineering did not take that extra hour to verify it. Tools should never be a crutch, that is true, but the right tool should be used to do a job right. If you are doing analog design, SPICE is easy to learn and often the CAD tool chain can create the deck for you. There are free versions of SPICE that work for many uses, but if your project supports it, licensed tools often have some worthwhile advantages. In digital design, simulation is also useful, but I also suggest trying formal methods as well, wherever you can fit it in schedule. Most designers and managers underestimate the value of verification, but it helps preserve schedule and budget and produces better results as long as you have people on the team that know how to get the value from it.
The most difficult part of designing for production is to work out what the worst-case scenario is. You need to have done lots of calculations over years to intuitively know how particular parameters affect final outcomes, and that gives you the ability to see what combination of parameters produce the worst-case result. It's beyond reprehensible to produce a design that works for most samples of the parts used in it, but can fail when one sample has a particular extreme combination of parameters. Sod's Law says that the circuit with those extreme parameters is the one that gets installed on a panel mounted 50 foot above the floor in a steelworks that's 300 miles away from your base. And you get the job of fixing it.
Once upon a time, I took an Electronics 101 course at university, mostly because I already knew my way around the topic, so a very easy 'A+'. The Professor, an EE at the PhD level of course, gave us the complete lesson and process to calculate the required resistor values for a simple Class A amplifier circuit. Did all that, Load Lines, the whole thing. Built the prototype, and it still needed tweaking for both bias and gain. I complained that I could have just started with a simple example circuit and made similar tweaks. Even if the 'solution' is yet-more analysis to account for another dozen pages of possible variables, there comes a point where it's arguably better, faster, cheaper just to start with a typical circuit and tweak. Modern analysis tools will certainly change the balance from what we had in 1979-1980. This is more of an anecdote than a lesson for others in the present day. Carry on !
Very true. I learned small signal analysis at university. It is possible to design an opamp or multi-transistor circuit on paper, to an accuracy of 5%. When I learned how to do that, during my sophomore year, I wrote a Basic program that iterated through standard 10% resistor values and came up with the design of a common emitter amplifier with the closest gain or the collector current to what was specified. I later re-wrote the program in C and added common emitter, common base, cascode, and emitter coupled pair circuits, along with an output to an OrCad schematic. The low frequency and high frequency roll-offs were calculated as well. A few times I built the designs and tested them. The actual circuits agreed within 5% of what my program calculated. When doing incremental analysis of a 1 or 2 transistor circuit, there are several open circuit time constants that appear in the node equations. The high frequency response of a circuit can be closely estimated by taking the sum of the inverse time constants. Spice, which has much more accurate device models, uses a matrix calculation, which produces a more accurate result. I have worked with technicians who used Spice all the time. I have noticed some had had a very superficial knowledge of how transistor work. Their designs, which were done entirely with Spice, had unecessary components and were no where near optimal. The more a technician or EE learns about electronics, solid state physics, and how to design a circuit on paper, the better use they can make of Spice. The best analogy I can think of is someone relying on the self-driving capability of their Tesla, because they are not a good driver.
Be careful that the popular BC547/8/9 have the opposite pinout (swap C and E). Always check datasheets from a semiconductor manufacturer not random webpages.
Yep... Random pages are dangerous !!! If you check them alweys check many sources (!!!) - NOT Just 1. I did that .. source said: " That amp. Works on 17V... input ... and Because the original power supply didn't work ... I just hooked up 17 V power. Put it ...ON ..... . and 💥💥💥 few sec .later big El. CAPACITOR exploded, and hit me in the knee . Lucky ... no big consequences.... just lessons. But I learned something out of it.
It will eventually fail. At high temperature, the dopants migrate faster and it will fail sooner. At high temperature beta increases to the point of instability and it will be on until cooled even without base current (thermal runaway).
Looks like you'll need to add a bias circuit to get past the dead zone when switching between high side and low side. Sure the op-amp can mask some of it but for audio, you need it.
You can heat sink those to92 packages to get a bit more power out of them. I have used a binder clip for that. You can also bend them over and glue them to the ground plane to remove quite a bit of heat. Also I think the die is thermally more bonded to the collector lead, so keeping leads short and having a large collector copper pour can also act as a heat sink.
It's a plastic package, you're not going to be able to move a significant amount of heat through that, or a lead... and, I'm not sure, what's limiting the current to be speced at 200 mA, for 2N3904. In the short-run failure, at over-current, it's probably thermal run-away... but, who knows what the long-run effects of limited over--current are? (Is their research on that?) You want more power, us a 2N2222A (it's in a metal package). 2N2222A is 10 times the price of 2N3904; but, it's still less than half a buck.
@@willthecat3861 its not for current, its for power dissipation. The 625 milliwatts is at ambient 25C with a 5 mw derate for each C over. With a small heat sink you can add at least 20 percent or more to the margins even though it is a plastic package. You can get commercial to92 heat sinks as well, the binder clip is a quick and dirty substitute.
@@argcargv In this context, no current... no heat. Potential is the potential to do work. That there can be a difference in potential, and yet no work done, indicates the fundamental cause of heat is current. Heat and power are not the same thing.
IMSAI Guy, I want you to know that you are my FAVORITE electronics channel and how much I appreciate your dedication to producing educational, entertaining and high quality videos on really interesting and relevant topics. You actually do things like I or some other experimenter or engineer might do them at their electronics workbench. Thank you for this treasure trove of goodness you have and continue to provide. That said, this episode in particular spurred me to post this comment because transistors are such a fundamental part of the world of electronics, and related circuits are so easily accessible to us all, that I wanted to ask if there is a possibility that you could do a series covering the various implementations of BJT, FET, JFET, MOSFET etc characteristics and circuits, going into as much detail as possible. I personally have been fascinated by transistors since I was a kid, and 50 years later, that fascination has not waned. It’s nice to see how you go think about things and solve problems in general, would be great to see even more specifically on transistors. I know you covered some of this in your past videos, but a refresher with maybe a new twist would be AWESOME. Once again, THANK YOU for your dedication and GREAT WORK!
I just did a Jfet: ruclips.net/video/rJb1eVR30dY/видео.html here is some BJT stuff: ruclips.net/video/Z8JDsvfZjL8/видео.html a mosfet will be a video soon
Since last week, I'm playing with a portion of a circuit that Roland used in a very famous synthesizer. Transistor, Fets and mosfets are quite more complex than just switching. Also the transistor totem pole arrangement is pretty special, there pretty much no current usage on the base. Both transistor cancel each other out on the base, pretty amazing, like you would be using a mosfet / fet.
At 12:00 There's no "sometime" about it. You either need to read the datasheet and do the maths and make sure at worst-case you're within limits for voltage, current and dissipation, or you've read the datasheet and done the maths so many times that you have a feel for what's reasonable. There are not "two schools of thought". There's a sensible middle way between creating a spice model for every simple circuit that you've built a thousand times, and just sticking any old voltages and resistances into a circuit to "suck it and see" with no prior expectation of the outcome. And if I was the R&D manager and I saw you just sticking any old circuitry together that was bound to fail, I'd fire you.
The way I remember the pinout for these is that the shape is a C, not a D. So from looking down on the C, the top of the C is the collector. The base is the middle, and that leaves the emitter as the remaining pin.
I have worked with them for a long time. Seen single PCBs with over a hundred. Even for repair purposes, even for my personal stock I always purchased them in hundreds. As far as I am concerned, they are THE jelly bean, small signal transistor. I just look at the flat and say E B C. And that is by far the most popular pin-out for bipolar transistors in the TO-92 package. That's why it is so impressed on my memory cells.
As soon as you drew in the 50ohm load at 4:15, I jumped on the calculator and thought this is going to get interesting. Time to sit back and wait for the smoke to escape :)
I'm fiddling with a design for a big-output-capacitor-less audio amplifier (eliminating the highpass effect w/ the speaker as well as an extra 6dB) that push-pulls the speaker with inverse-phased class AB stages like this (TL084 is my op amp) and I'm using just 2N2222s (basically a little beefier 2N3904) and 2N2907s (lil' beefier 2N3906) and even though they're just little TO92 signal transistors they're actually surprisingly good at filling a room with sound.
0:45 For 3-leg components (transistors in this example), I'd recommend just skipping the 1 2 3 pin numbers and go straight to simply memorizing EBC (for common examples like these). To help clarify thoughts, does anyone use pin numbers for diodes, or straight to Anode and Cathode ? Polarized capacitors, easier straight to + and -. I'd suggest that pin numbers only add value for a larger number of pins. In this case, they make memorization arguably slightly more difficult.
Dude, your dissipation calculations are (slightly) off. With he load you measured about 3.8V across the load. That would give you a current of about 75mA. The voltage drop across the transistor was 12-3.8=8.2V. Therefore the power being dissipated by the transistor was 8.2x.075=615mW. Jus' sayin'. But it's close enough to AMR to cause real trouble.
I always first design using rules of thumb using good old pen and paper, but before I built it I also do a spice model. Why? Especially important for stability analysis. E.G, the circuit you drew of the fb for an amp? How do you know that the parasitic capacitance of the transistors will not affect the fb, even make ringing with the parasitic inductance of the wires? Right, spice model. Before you build. Always. Rules of thumb are nice, but nothing compares to a good spice model 🙂.
As with the other comments adding base diodes would help a little by biasing the transistors on and negating base emitter drop, could also get it to oscillate. I was curious was to how much the base voltage was changing due to current in the through the potentiometer. You didn't mention if it was a 470K pot or a 100 ohm pot, obviously not 100 ohm.
Please suggest another complimentary pair with a higher power capacity. I've considered these two to replace proprietary power design transistor because they share high hfe's. I can't locate other transistors with similar abilities. How do I search without reading every data sheet?
I always wonderd about this: There are two types of to92 packages: The one with straight leads and one with bend outwards leads. Which do you recommend for breadboarding? And is there a reason why there are two types?
hi, I have trouble understanding how both PN junctions can be shorted in the bases, in some other circuits I see they use 2 external dropping diodes in the bases. can you elaborate?
Something to keep in mind is that the maximum power dissipation rating on the datasheet, is at 100% duty cycle. If this arrangement would have been used as intended, each transistor would have around 50% duty cycle and could handle that current easily without overheating. There are actual charts that show this on some transistors, but I was unable to find one for the 3904/06.
@@davidknightaudio934 Please explain. To my knowledge, the maximum power dissipation rating on a transistor is limited by the maximum collector-base junction temperature. Are you saying that a transistor that only has current passing through it 50% of the time has an operating temperature the same as one operating at 100% of the time as in a class A amplifier?
5:27 11.4V÷51Ohm= 0,235294118 Amp. that's quite a current consume for these small boys not using a resistor on the base of them. What would you do? since the HFE is 100 average, if you need 100mA, you need to limit the base current to 1mA, correct? Did you try the Falstad simulator, it quite correct and good, buy a Moog Filter clone diy kit, build and simulate it, if will respond the same.
Sorry, I don't get it. Why wouldn't you play around with anything but the simplest circuit in Spice before trying to build it? It's not perfect, and it's no 100% substitute for real world component experience, but it'll still teach you a lot. It'll allow you to mess with amplifier biasing and gain before even warming up the soldering iron. Maybe I missed your point.
You are limiting your decision to function of a device to last as long as possible ... that's not something to consider in a missile guidance ... so it is telling what industry you worked in ... I'm just saying sometimes macgyverneering is good enough and fits the bill ... planned obsolescence is a thing that surpasses the binary thought process.
Spice shy? Why? Back in the 1980's (and early 90's) ... more than 30 years ago (a hyperbolic infinity in semiconductor design and modeling) ... back then, Pease, and his cronies, had their good reasons. But in the last decade, at least, I can't think of an undergrad (and especially a grad) EE school where the labs don't start with SPICE simulations, and even more specialized simulation packages, before going to the lab to build. Maybe in the EE School of Old Farts,, they're still just stuffing breadboards. Dunno. Seems like small businesses, run by old farts, might be doing the same. But if you want to work in analog design, and even with discretes, and you're SPICE shy, then get a job as a plumber's apprentice. (You'll probably get a job faster, and keep it longer.) Seems to me, someone's been out of school, and a job, for a long time.
Yes, I am an old fart, guilty. I do believe (and have seen) many grads that understand the theory and are great at cad models, but are a disaster in the lab, trouble shooting or inventing new ideas. I know of good ones that can do both. I agree with you that spice is very valuable (and I do use it too).
Наука
The 2N3904 says to the 2N3906, "nice package". The 2N3906 says to the @2N904, "nice legs". That's an example of a "complimentary pair".
But when the 2N3904 says to the 2N3906, "you complete me". Now that's a complementary pair!
you must be a dad.
😊
@@copernicofelinis Did you catch the difference in spelling? (that was the point of my post) :)
I like the "drive bigger trucks over the bridge until the bridge falls down" method of circuit design.
If you use two diodes on the input to bias the transistors, the crossover distortion will be very small. If used with the op amp, it will be phenomenally small. With the op amp but without the diodes, you will still get noticeable distortion.
the Roland System 700 VCA has that, i have build that, using the same values of resistors (50hm) and at idle, i noticed they get hot, but i used a cloned by me LM13700.
Now, i have to unsolder the 90º headers to be able to debug the circuit.
Or a Vbe multiplier
The diodes forward voltage drop and temperature coefficients will have to be almost exactly matched to the transistors for that to work or you will risk thermal runaway or distortion. It's better to use the B-E junction of identical transistors to act as the "diodes" do the biasing. Might cost a few pennies more in the price of the transistors over the diodes, but it will save much more in time and energy trying to match diodes.
Yes, an engineer can put together a design to nominal/typical parameters without a simulator. However, SPICE and other tools are useful, if for nothing else, to check the tolerances of the devices in the design. While any engineer can get a prototype working in the lab, the usual requirement is to produce good yields through manufacturing and reliable products in the field.
Part variation is expensive and time consuming to verify in the lab, but tools like SPICE and other simulators, as well as mathematical tools, can calculate extremes and determine margin relatively inexpensively. This can also check for violation of maximum characteristics, such as your overcurrent situation. Since in my career with many companies and being responsible for many products, I was always tasked, as part of the requirements, to create reliable and manufacturable systems, and I would have had to fire any engineer who would not simulate or calculate the variations. Luckily, I never had an engineer working on my team that did not know how to prove his work and seek to do just that.
My first six years of my career was mostly spent fixing and cleaning up the empirical designs that other people pretending to be engineers did before me. I have a low tolerance for low quality work, because it costs my company a lot more than any savings because engineering did not take that extra hour to verify it. Tools should never be a crutch, that is true, but the right tool should be used to do a job right.
If you are doing analog design, SPICE is easy to learn and often the CAD tool chain can create the deck for you. There are free versions of SPICE that work for many uses, but if your project supports it, licensed tools often have some worthwhile advantages. In digital design, simulation is also useful, but I also suggest trying formal methods as well, wherever you can fit it in schedule. Most designers and managers underestimate the value of verification, but it helps preserve schedule and budget and produces better results as long as you have people on the team that know how to get the value from it.
The most difficult part of designing for production is to work out what the worst-case scenario is. You need to have done lots of calculations over years to intuitively know how particular parameters affect final outcomes, and that gives you the ability to see what combination of parameters produce the worst-case result. It's beyond reprehensible to produce a design that works for most samples of the parts used in it, but can fail when one sample has a particular extreme combination of parameters. Sod's Law says that the circuit with those extreme parameters is the one that gets installed on a panel mounted 50 foot above the floor in a steelworks that's 300 miles away from your base. And you get the job of fixing it.
Once upon a time, I took an Electronics 101 course at university, mostly because I already knew my way around the topic, so a very easy 'A+'. The Professor, an EE at the PhD level of course, gave us the complete lesson and process to calculate the required resistor values for a simple Class A amplifier circuit. Did all that, Load Lines, the whole thing. Built the prototype, and it still needed tweaking for both bias and gain. I complained that I could have just started with a simple example circuit and made similar tweaks. Even if the 'solution' is yet-more analysis to account for another dozen pages of possible variables, there comes a point where it's arguably better, faster, cheaper just to start with a typical circuit and tweak. Modern analysis tools will certainly change the balance from what we had in 1979-1980. This is more of an anecdote than a lesson for others in the present day. Carry on !
Very true. I learned small signal analysis at university. It is possible to design an opamp or multi-transistor circuit on paper, to an accuracy of 5%. When I learned how to do that, during my sophomore year, I wrote a Basic program that iterated through standard 10% resistor values and came up with the design of a common emitter amplifier with the closest gain or the collector current to what was specified. I later re-wrote the program in C and added common emitter, common base, cascode, and emitter coupled pair circuits, along with an output to an OrCad schematic. The low frequency and high frequency roll-offs were calculated as well. A few times I built the designs and tested them. The actual circuits agreed within 5% of what my program calculated. When doing incremental analysis of a 1 or 2 transistor circuit, there are several open circuit time constants that appear in the node equations. The high frequency response of a circuit can be closely estimated by taking the sum of the inverse time constants. Spice, which has much more accurate device models, uses a matrix calculation, which produces a more accurate result. I have worked with technicians who used Spice all the time. I have noticed some had had a very superficial knowledge of how transistor work. Their designs, which were done entirely with Spice, had unecessary components and were no where near optimal. The more a technician or EE learns about electronics, solid state physics, and how to design a circuit on paper, the better use they can make of Spice. The best analogy I can think of is someone relying on the self-driving capability of their Tesla, because they are not a good driver.
Be careful that the popular BC547/8/9 have the opposite pinout (swap C and E). Always check datasheets from a semiconductor manufacturer not random webpages.
I fried a chip like this. The bc548 was fine, the chip got cooked that it was connected to.
Yep...
Random pages are dangerous !!!
If you check them alweys check many sources (!!!) - NOT Just 1.
I did that .. source said: " That amp. Works on 17V... input ...
and
Because the original power supply didn't work ...
I just hooked up 17 V power. Put it ...ON ..... . and 💥💥💥
few sec .later big El. CAPACITOR exploded,
and
hit me in the knee .
Lucky ... no big consequences.... just lessons.
But I learned something out of it.
Always good to review the basics. Thanks.
It will eventually fail. At high temperature, the dopants migrate faster and it will fail sooner. At high temperature beta increases to the point of instability and it will be on until cooled even without base current (thermal runaway).
Looks like you'll need to add a bias circuit to get past the dead zone when switching between high side and low side. Sure the op-amp can mask some of it but for audio, you need it.
Your tutorials are great !...more please....cheers.
You can heat sink those to92 packages to get a bit more power out of them. I have used a binder clip for that. You can also bend them over and glue them to the ground plane to remove quite a bit of heat. Also I think the die is thermally more bonded to the collector lead, so keeping leads short and having a large collector copper pour can also act as a heat sink.
It's a plastic package, you're not going to be able to move a significant amount of heat through that, or a lead... and, I'm not sure, what's limiting the current to be speced at 200 mA, for 2N3904. In the short-run failure, at over-current, it's probably thermal run-away... but, who knows what the long-run effects of limited over--current are? (Is their research on that?) You want more power, us a 2N2222A (it's in a metal package). 2N2222A is 10 times the price of 2N3904; but, it's still less than half a buck.
@@willthecat3861 its not for current, its for power dissipation. The 625 milliwatts is at ambient 25C with a 5 mw derate for each C over. With a small heat sink you can add at least 20 percent or more to the margins even though it is a plastic package. You can get commercial to92 heat sinks as well, the binder clip is a quick and dirty substitute.
@@argcargv What else, but current, causes heat to be dissipated, to the surroundings, in a 2N3904 transistor?
@@willthecat3861 voltage. If you need to run at a higher voltage it doesn't take much current to get to 625milliwatts
@@argcargv In this context, no current... no heat. Potential is the potential to do work. That there can be a difference in potential, and yet no work done, indicates the fundamental cause of heat is current. Heat and power are not the same thing.
Nice quick vid i enjoyed it! Thanks
IMSAI Guy, I want you to know that you are my FAVORITE electronics channel and how much I appreciate your dedication to producing educational, entertaining and high quality videos on really interesting and relevant topics. You actually do things like I or some other experimenter or engineer might do them at their electronics workbench. Thank you for this treasure trove of goodness you have and continue to provide.
That said, this episode in particular spurred me to post this comment because transistors are such a fundamental part of the world of electronics, and related circuits are so easily accessible to us all, that I wanted to ask if there is a possibility that you could do a series covering the various implementations of BJT, FET, JFET, MOSFET etc characteristics and circuits, going into as much detail as possible. I personally have been fascinated by transistors since I was a kid, and 50 years later, that fascination has not waned. It’s nice to see how you go think about things and solve problems in general, would be great to see even more specifically on transistors. I know you covered some of this in your past videos, but a refresher with maybe a new twist would be AWESOME.
Once again, THANK YOU for your dedication and GREAT WORK!
I just did a Jfet: ruclips.net/video/rJb1eVR30dY/видео.html
here is some BJT stuff: ruclips.net/video/Z8JDsvfZjL8/видео.html
a mosfet will be a video soon
@@IMSAIGuy Let me guess: 2n7000?
@@robballantyne3 nope
APPRECIATED.
YOU REMINDED ME OF MY ENGINEERING DAYS IN UNIVERSITY OF ENGINEERING AND TECHNOLOGY, LAHORE, PAKISTAN OF 1974-1978.
Since last week, I'm playing with a portion of a circuit that Roland used in a very famous synthesizer.
Transistor, Fets and mosfets are quite more complex than just switching.
Also the transistor totem pole arrangement is pretty special, there pretty much no current usage on the base.
Both transistor cancel each other out on the base, pretty amazing, like you would be using a mosfet / fet.
Amazing channel.
Good explanation :) Pretty amazing what op amps can do.
Excellent! Thank you.
Like what you said about rules of thumb when quickly prototyping.
At 12:00 There's no "sometime" about it. You either need to read the datasheet and do the maths and make sure at worst-case you're within limits for voltage, current and dissipation, or you've read the datasheet and done the maths so many times that you have a feel for what's reasonable.
There are not "two schools of thought". There's a sensible middle way between creating a spice model for every simple circuit that you've built a thousand times, and just sticking any old voltages and resistances into a circuit to "suck it and see" with no prior expectation of the outcome.
And if I was the R&D manager and I saw you just sticking any old circuitry together that was bound to fail, I'd fire you.
The way I remember the pinout for these is that the shape is a C, not a D. So from looking down on the C, the top of the C is the collector. The base is the middle, and that leaves the emitter as the remaining pin.
I have worked with them for a long time. Seen single PCBs with over a hundred. Even for repair purposes, even for my personal stock I always purchased them in hundreds. As far as I am concerned, they are THE jelly bean, small signal transistor.
I just look at the flat and say E B C. And that is by far the most popular pin-out for bipolar transistors in the TO-92 package. That's why it is so impressed on my memory cells.
As soon as you drew in the 50ohm load at 4:15, I jumped on the calculator and thought this is going to get interesting. Time to sit back and wait for the smoke to escape :)
Interesting video. I have 1 thing to point out. You are the first person that I've seen using an HP. I really love using RPN.
Nice idea for AGC
I'm fiddling with a design for a big-output-capacitor-less audio amplifier (eliminating the highpass effect w/ the speaker as well as an extra 6dB) that push-pulls the speaker with inverse-phased class AB stages like this (TL084 is my op amp) and I'm using just 2N2222s (basically a little beefier 2N3904) and 2N2907s (lil' beefier 2N3906) and even though they're just little TO92 signal transistors they're actually surprisingly good at filling a room with sound.
These 2 are plentiful in my stock. Also 2N7000
me too
me 3
I can remember using but can't remember what for. A transistor lockin amplifier perhaps
Cool calculator.
0:45 For 3-leg components (transistors in this example), I'd recommend just skipping the 1 2 3 pin numbers and go straight to simply memorizing EBC (for common examples like these).
To help clarify thoughts, does anyone use pin numbers for diodes, or straight to Anode and Cathode ? Polarized capacitors, easier straight to + and -.
I'd suggest that pin numbers only add value for a larger number of pins. In this case, they make memorization arguably slightly more difficult.
BTW, emitter arrow direction: NPN = Not Pointin' iN, PNP = Pointin' iN Please. You'll never forget.
That's a great idea for trying to remember the pin information.
Dude, your dissipation calculations are (slightly) off. With he load you measured about 3.8V across the load. That would give you a current of about 75mA. The voltage drop across the transistor was 12-3.8=8.2V. Therefore the power being dissipated by the transistor was 8.2x.075=615mW. Jus' sayin'. But it's close enough to AMR to cause real trouble.
Who just happens to have a 51Ω resistor lying around? ;)
I always first design using rules of thumb using good old pen and paper, but before I built it I also do a spice model. Why? Especially important for stability analysis. E.G, the circuit you drew of the fb for an amp? How do you know that the parasitic capacitance of the transistors will not affect the fb, even make ringing with the parasitic inductance of the wires? Right, spice model. Before you build. Always. Rules of thumb are nice, but nothing compares to a good spice model 🙂.
As with the other comments adding base diodes would help a little by biasing the transistors on and negating base emitter drop, could also get it to oscillate.
I was curious was to how much the base voltage was changing due to current in the through the potentiometer. You didn't mention if it was a 470K pot or a 100 ohm pot, obviously not 100 ohm.
10k
Classic
How do you figure the voltage and current of the 3906?
Please suggest another complimentary pair with a higher power capacity. I've considered these two to replace proprietary power design transistor because they share high hfe's. I can't locate other transistors with similar abilities. How do I search without reading every data sheet?
This video is going to get the 2N2222(A) / 2N2907 fanboys riled up. 🙂
I always wonderd about this: There are two types of to92 packages: The one with straight leads and one with bend outwards leads. Which do you recommend for breadboarding? And is there a reason why there are two types?
I find the bent legs easier to space on BBs
@@the2d Thank you!
I like the straight legs. many of the boards I have use a triangular pad layout.
hi, I have trouble understanding how both PN junctions can be shorted in the bases, in some other circuits I see they use 2 external dropping diodes in the bases. can you elaborate?
ruclips.net/video/slIOOJqZ3nw/видео.html
Something to keep in mind is that the maximum power dissipation rating on the datasheet, is at 100% duty cycle. If this arrangement would have been used as intended, each transistor would have around 50% duty cycle and could handle that current easily without overheating. There are actual charts that show this on some transistors, but I was unable to find one for the 3904/06.
you have much to learn
@@davidknightaudio934 Please explain. To my knowledge, the maximum power dissipation rating on a transistor is limited by the maximum collector-base junction temperature. Are you saying that a transistor that only has current passing through it 50% of the time has an operating temperature the same as one operating at 100% of the time as in a class A amplifier?
U would want a focused product 4 this though, right! so what will this generic circuit go in? good luck! mad scientist engaged!
5:27 11.4V÷51Ohm= 0,235294118 Amp.
that's quite a current consume for these small boys not using a resistor on the base of them.
What would you do? since the HFE is 100 average, if you need 100mA, you need to limit the base current to 1mA, correct?
Did you try the Falstad simulator, it quite correct and good, buy a Moog Filter clone diy kit, build and simulate it, if will respond the same.
1:58 Why are no base resistors required? Is it because there is a load? What if there was no load, would the transistors be damaged?
best to have one
The pot provides variable base resistance to some extent, until it approaches the rail voltages.
You're going to need a bigger transistor.
Sorry, I don't get it. Why wouldn't you play around with anything but the simplest circuit in Spice before trying to build it? It's not perfect, and it's no 100% substitute for real world component experience, but it'll still teach you a lot. It'll allow you to mess with amplifier biasing and gain before even warming up the soldering iron. Maybe I missed your point.
You are limiting your decision to function of a device to last as long as possible ... that's not something to consider in a missile guidance ... so it is telling what industry you worked in ... I'm just saying sometimes macgyverneering is good enough and fits the bill ... planned obsolescence is a thing that surpasses the binary thought process.
Spice shy? Why? Back in the 1980's (and early 90's) ... more than 30 years ago (a hyperbolic infinity in semiconductor design and modeling) ... back then, Pease, and his cronies, had their good reasons. But in the last decade, at least, I can't think of an undergrad (and especially a grad) EE school where the labs don't start with SPICE simulations, and even more specialized simulation packages, before going to the lab to build. Maybe in the EE School of Old Farts,, they're still just stuffing breadboards. Dunno. Seems like small businesses, run by old farts, might be doing the same. But if you want to work in analog design, and even with discretes, and you're SPICE shy, then get a job as a plumber's apprentice. (You'll probably get a job faster, and keep it longer.) Seems to me, someone's been out of school, and a job, for a long time.
Yes, I am an old fart, guilty. I do believe (and have seen) many grads that understand the theory and are great at cad models, but are a disaster in the lab, trouble shooting or inventing new ideas. I know of good ones that can do both. I agree with you that spice is very valuable (and I do use it too).