Circuit building: stop using antique parts (2014)
sensitiveresearch.com
sensitiveresearch.com
I like FETs as much as the next engineer but if you're going to write about them, try to not make things worse.
In particular, this made me sad: the 10K resistor to ground isn't strictly necessary, but it ensures that the MOSFET remains OFF if the arduino is disconnected or it's pin is not an OUTPUT.
I was helping a high school student with a transistor project and he was complaining that "half the FETs he bought were 'bad'". And they weren't really bad, but they were destroyed. They were destroyed because, like our author, the student had no understanding of how the FET worked and so didn't realize that when you exceed the maximum Gate voltage on a FET it generally causes a current to "jump the gap" to the source and that permanently breaks it. What is even more, since the gate is essentially a capacitor, you can just touch the gate with your finger and pass enough charge (without feeling a shock or anything) to greatly exceed the gate voltage. Pick up the FET without being grounded and "boom!" dead FET.
Now there is almost no way to generate enough current by touching a Bipolar transistor to kill it, and so they continue to work for a long time while plugging them in and out of breadboards.
Do they dissipate more power? Absolutely. Are they difficult to run in parallel? Sure. But they are pretty robust parts. Sort of like the difference between alkaline batteries and LiON rechargeable batteries. Sure the latter are a "better" choice, you can recharge them after all, but if you use them wrong they catch on fire, if you short an alkaline battery it gets hot but doesn't go ballistic on you.
So those "antique" parts are generally very cost effective, very robust, and easy to get. So they make excellent tools to teach you the basics. Do you want to stay with them as you get more advanced? Probably not, but you're probably not skiing on the same skis you learned on either.
/endrant
Do you know of anything available like that?
Surface mount only (like most modern transistors), but SOT-23 isn't too hard to hand solder. There are lots of breakout boards for surface mount packages available for prototyping, or you could even solder some mod wire directly to each pin and put some hot melt glue on it to keep it in place.
Not sure why that made you sad (extraneous ' in its?).
But I agree about the static-sensitive disadvantage of MOSFETs.
Also, the author mentions that a TIP120 is not a transistor but a darlington pair (two transistors in one) but then misstates the Hfe.
The whole point of a darlington is to get a high Hfe (small base current controlling large collector current) and the Min. Hfe is 1000, at least according to:
Also if you throw away all your BJTs, what do you use to drive the gate capacitance of your fancy FETs?
Part of the reason, I think, is that a lot of "designers" of guitar pedals don't understand the math that determines component values for a particular transistor in a particular circuit so they don't know how to either select a modern transistor to replace an obsolete one based on datasheet parameters or to change the collector / emitter / source / drain / etc resistor values to work right with a new part.
A guitarist friend decided to "test" me once and put this in front of me and ask what it would sound like: http://el34world.com/charts/Schematics/files/randall/Randall...
My immediate response was: "That looks like a really great amp from the mid 1980's--looks like it was designed by someone who actually understands electrical engineering. It should have a pretty clean. But if you overdrive it, the clipping on those JFET's is gonna sound like fscking buzzsaw trash."
He, of course, laughed knowing that Dimebag Darrell used the amp for precisely that buzzsaw sound. You would have to add lots of "clipping" components to get the same sound out of depletion mode MOSFET's.
Now, whether that's a good sound is open to debate. :)
Digital is much more suited to time based effects. Analogue to harmonic distortion based effects.
You can keep your ADC/MCU/DAC distortion pedal. But if you are ever wondering why your guitar distortion sounds crap, this is it. But maybe you think it's fine. Anything sounds great with cotton-ears.
And tube amps only being for nostalgia? I wan't you to take your guitar, your digital distortion pedal and your digital amplifier and go to a rehearsal room with another guitarist who uses analogue distortion and a tube amp. Turn them both up loud. Jam together. Compare the tones. Compare the nature of how your amplifiers respond to your tone.
Sitting in your bedroom playing by yourself will never convince you of the short comings of your digital rig.
Shitty digital pedals from 20 years ago sound shitty, I agree!
Well modelled digital effects, however, can these days give you anything an analog pedal can. Most of the time, just run a high accuracy Spice model of the pedal in real time! Modern DSP is so cheap and fast that this is possible and works fine.
Lest you think I'm some kind of analog purist, I'm actually currently building a batch of pedals for sale that use an AVR MCU an its internal ADC to generate 3 different octaves of square wave output.
And yet smart engineers keep using Analog Devices SHARC DSP's for effects on audio interfaces.
So SOMETHING must not be quite as simple as your claim.
(The answer is that most of these guitar effects are VERY non-linear and simulating non-linear stuff digitally is very much NOT straightforward in real-time.)
I mean, nearly every tube head on the market has an essentially pointless standby/bypass switch because of the mistaken belief that you need to warm up your tubes for some arbitrary amount of time in a modern amp and keep them warm.
I forget which manufacturer it was that dropped the switch on one model and saw sales plummet and got freaked out support calls asking why the switch wasn't there. Next iteration had the switch back.
One standout anomaly was the Heathkit 250W amp, which just had an on/off switch and 2 meters.
*Using a cheap tube amp from CL as the basis for a new build and tossing everything except the chassis and transformers, and maybe the sockets if they are chassis mounted, is a great way to build a great sounding tube amp on the cheap.
The regular ordinary procedure for cold-starting high-power vacuum tube systems is to switch on the unit while the tubes are in standby.
This will allow the cathodes to heat to their characteristic glow, fully developing the ability to deliver the current you will soon be demanding.
But no high voltage is applied to the anodes (there will be no sound from audio systems) until removed from standby.
A minute or two later is good.
If the cathodes are too cold when applying the high anode voltage, especially at high current demands, it can strip the cathodes of their ability to provide the full performance when you need it most in the future.
Small tubes, most preamp types, run on lower voltages and currents, don't suffer noticeably, and were cheap to begin with.
It's the bigger more expensive tubes that standby is for.
Shut down by switching to standby while the main unit is still powered on.
After no more than a minute or two, fully power off the unit unless you would otherwise have to cold-start unreasonaby soon afterward. This is about 10 to 20 minutes for popular audio units, or much more if there is a true need for instant-on over a longer term.
24 hour standby is not very good for the tubes either, best fully powered off to save hours, or idling at a reasonable fraction of their ratings.
On standby tubes are still more vibration sensitive than fully cold, so it's seldom a good idea to leave them on standby while there is a source of strong vibration from other sources.
Never leave a tube amp in direct sunlight unless fully powered off. Standby will not save you. If it is allowed to heat up above ambient for any reason, let it cool back down to ambient before powering up. Gear that can run continuously at ambient 100degF can die at 70degF in the sun.
Without the sun, different gear (including differing power bias settings on the same gear) will have different maximum ambient temperatures beyond which only short-term operation is advisable, even for units which are otherwise capable of continuous operation under slightly milder conditions.
For musicians it's the class of thick-cone reinforced speakers that can take 20 or 30 minutes (near their power rating) to break in every time, that can throw you off. Speakers can not be "warmed up" silently, you have to make some noise. After this cranked-up break in, reach into an open-back cab and you feel how hot the magnets are, they were made for this but not in the sun either, especially closed-back cabinets where you can't touch the magnets but you know they're even hotter. But often it's not the warmth of the magnets (or the tubes) that makes a (beneficial) difference in the tone, it's the preflexing of the cones which makes them more articulate afterward. Most amps themselves warm up way faster than that, and can do it silently.
Yes, the community is really based in "older is always better" kind of mentality, but there is a reason some old components are so valued, and why respected pedal manufacturers commonly hand-pick parts that meet their designed specs best (especially fuzz and overdrive effects), and this includes everything from picking transistors with ideal leakage etc, to hand picking resistors (though the resistor bit I'm not sure on why they don't just buy high precision ones for a few cents more)
high precision resistors can get expensive, aren't always readily available (depending on your volume), or simply might not be available in the power rating you need. especially if you don't need absolute precision you might consider it wasteful. so it's good to know how to match resistors when you're designing/prototyping the circuit.
because of that, it's probably become hearsay and superstition, and is consequently overused.
As well there's the vintage issue...Sure, I buy NOS caps from the 50-70s for guitar tone circuits. I doubt side by side they sound noticeably better but if you're building vintage it only makes sense to keep the 'tronics true to the time period.
In vintage radio repair, they sometimes will replace old caps (usually electrolytics) with new ones, but put the new ones inside the old cans - to preserve the look of the device.
Obviously, though, this won't work with all parts...
You can also hollow out old carbon comp resistors and put metal film resistors inside them. As a bonus you could use resistor shells on the outside with color bands that don't at all represent the actual value of the resistor inside as a practical joke on anyone trying to clone the circuit.
There are even kits to replace the guts of old car radios with modern stereos while preserving the controls and dials and their function.
It turns out that the instruments needed the erratic 5 volts, else they'd "stick".
Any places I should get started looking?
There's a pretty classic book called Electronic Projects for Musicians that might be of interest: https://msu.edu/~dougl126/Electronic%20Projects%20for%20Musi...
I've got a decent tube combo amp, but it's just easier and more flexible to do it digitally. I mostly run S-Gear, which I find has by _far_ the best amp modeling, it's the only VST that really gets moderate overdrive right.
I'd buy some things, like germanium vs silicon transistors, but most of it I think is cargo cult behavior.
On top of that, BJTs excel in many analog applications. There is a good reason why many analog IC companies make parts using a BiCMOS process. If you are doing a discrete analog design (generally for performance reasons) you often need the performance of BJTs or JFETs.
I've a little aquarium and I've noticed how fish always get scared when lights turn suddenly on or off. So I took an Arduino and wrote a timer app for it, using PWM to slowly drive the output up over 60 minutes (sunrise), keep it at max for a few hours, then drive it down slowly over 60 minutes (sunset). I used a MOS-FET to drive the power LEDs directly from the Arduino. Works fantastic, and the FET is cold at all times. A 2N3055 would have been pretty hot most of the time.
I've been playing with electronics for a few decades now, so I can't be accused of change for the sake of change. The old stuff still works, sure, but there are new toys which are better in some cases.
I would certainly use MOSFETs for lighting (a power application). I probably wouldn't design an analog front-end for a specialized sensor with them though.
PT4115, pretty cheap on ebay and other usual places.
So I just hooked it up to 12V via the MOS-FET. The PWM signal from the Arduino (plugged into the MOS-FET) does the job of changing the apparent brightness.
No need for magic chips in this context.
You'll need to properly drive it though and not at too high a frequency (and for obvious reasons, not too low either because you'll notice the flicker).
The thing is, most of these larger BJT parts were designed to be used with heatsinks, not usually standalone. IMHO, there's nothing wrong with dropping some voltage and dumping some heat. Sure, it isn't efficient, but if you size everything properly, and your project can accommodate the heatsink - then where's the problem?
Heck, a long time ago you could get tiny heatsinks for some of the small round metal-can transistors that existed - I've got an old 16K minicomputer core-memory board from the 1970s that has drive electronics set up like this.
The author of this "rant" mentions not using L298 (and presumably the L293) either - but there's nothing wrong with these parts for hobbyists (outside of the L298's weird footprint, and the fact that heatsinks for them are almost unobtanium) - provided that the limitations are kept in mind (finding a low-cost mosfet 4A h-bridge in a thru-hole design is virtually impossible). These parts were originally meant for automotive applications, where having a 12-14V large amperage battery supply was the norm (people trying to use them for circuits with 6V or less always wonder why their small motors won't turn).
In a pinch: any piece of aluminum U profile will do, just drill a hole in it and use some heat conducting paste.
The frequency is either 500 Hz or 1000 Hz, I forget which (depends on Arduino type). I doubt there are any critters that can tell it's not continuous.
Looking at the code I wrote a while back to drive a four-digit common-cathode LED display (you know, a TV bomb timer!) as a clock, I find that after all the tuning I did to get rid of flicker, I ended up activating each digit for 1ms at a time. Granted, running under Raspbian this is going to flicker anyway from time to time because multitasking, but that aside, I found no PWM flicker perceptible, even at what is effectively 20% duty cycle over a 5ms period. (The decimals are implemented as a fifth "digit".) So I'm going to guess that .5KHz and up are fine for most humans. Fish I don't know about; (1) cites a flicker fusion rate between 2 and 40Hz for swordfish, but for goldfish I've no idea, other than to guess that .5KHz and up is probably fine for them too.
(1) https://www.google.com/amp/amp.livescience.com/3799-fish-eye... (Sorry for AMP, but taking it off the front nets me a timeout from livescience.com. Something broken there, I'm guessing.)
At default clock rates (clock/256) I can definitely see the flicker, especially through the lowest duty cycles.
The article would have been much better with a different title - "How to use FETs effectively and when to use them", not a binary oppositioning of "antique" versus "modern".
For analog applications, BJTs are still very much important, but for digital stuff (and especially power stuff), it is usually easier to grab a cheap FET than use a BJT.
MOSFETs do have some problems. They can draw large gate currents for the first few nanoseconds of turn-on, which can overload whatever is driving them. The gate input is very vulnerable to electrostatic discharge during handling. Device failure tends to be into the ON state.
I just went through a big struggle with MOSFET selection for a special purpose switching power supply. The big through-hole parts have too much gate capacitance and need too much drive at turn-on. Only in surface-mount could I get something that would work.
That's the real problem with antique parts. The new stuff is surface-mount only.
[1] https://www.fairchildsemi.com/datasheets/FQ/FQP4N20L.pdf
Let's say you wanted to do a special purpose switching power supply. Digikey doesn't have a column for gate capacitance (for the mosfet), or Q (for the inductor). All that manual work to read datasheets is terrible, especially if your design takes a little while to perfect and some of the components go end-of-life, so you have to replace them.
I personally like to graph the cost-efficiency curve so I actually have to tabulate lots of mosfets, inductors, and caps, then simulate them in a spreadsheet.
I know about the specialized design tools done by each manufacturer that will recommend parts and do a design for you based on your criteria. I want something that does a much broader search for parts though.
Maybe I should write one. :)
But, even the mosfet is old. IGBT (depending on loads, but especially for slower switching)
Which is why I pointed out. A lot of reasons people use antiquated components is probably because documentation and reference schematics are easy to find and have probably been refined to be very reliable.
Online electronics stores are no help because of the overload of possible components.
Is there some sort of Cookbook out there I could refer to for up-to-date examples of common circuits and electronic tasks? I mean, I otherwise totally would've used a 2N2222 transistor for some tasks, because I wouldn't know better.
The truth though is that it's not like the old parts got any worse, and are still often the cheapest choice. You have to add additional constraints (I want lower loss, better performance etc) for something else to be "better".
https://octopart.com/common-parts-library
I also often refer to Sparkfun's product list as a reasonable starting point.
I have very very little experience with electrical engineering (I'm in the Lego mindset of I plug things in according to things I find online until they work), but is there something I should be using instead with 120V ~3-5 A loads?
Relays are great though for motor controllers, especially when you need higher amperages (and don't want the huge expense a mosfet solution can sometimes bring); an h-bridge is very simple to build, and tying an n-channel mosfet (of appropriate amperage capability) on the low-side between the relay(s) and ground, and you get easy PWM control (just don't switch the relay while PWM'ing of course).
Paying a premium for a new part with predictable characteristics isn't always a mistake.
Another problem is if you use one to switch from say an antenna to a dummy load, you might find that you are leaking quite a bit of power onto the supposedly disconnected antenna link. RF is tricky, especially at higher power levels.
One trick you can use is to use only single contact relays and to bend open the contacts as far as they will go as long as they reliably close again when the relay is energized.
That way you minimize capacitive coupling between the contacts. DC relays are your best bet for a starting point because they already have larger gaps than most AC relays (to inhibit arcing).
As far as a replacement part? I don't think there are many alternatives there (at least with the same size, and easy to configure - there are a few stereo ICs I've seen used for class-D amps, but they weren't as simple to use or as small).
The 741 was introduced in the late 1960s. It's completely outclassed by modern general-purpose op-amps; there is absolutely no reason to still be using it today.
When in doubt, your best move is probably to see what Horowitz and Hill have to say. AoE3 has a lot of specific component recommendations.
This is a problem for me when building high end audio circuits. A lot of the modern components are only available in surface mount forms. I find this with matched transistor packages. Like a matched quad. There are modern components, like AD's MAT14 [1], but if you are building your own circuits and using through hole components your ONLY choice is discontinued components. They simply are not made in a through hole package anymore.
At some point I'll be forced to move to surface mount, but at the moment getting old components is a lot easier. And those old components are as good as the new ones. Just different packages.
If anyone knows of a modern precision matched quad like the MAT14, but in a traditional package, please reply with the component!
[1] http://www.analog.com/media/en/technical-documentation/data-...
I soldered one project with paste, a toothpick to apply it, and an unmodified hot plate controlled by hand. No stencil needed.
We have a proper oven but everyone hates using it.
You see, its not actually "easier". Its "easier" to you because you already have the equipment and are already doing it. Hey! I already know this! This is easy!
PCBs: You can make PCBs by toner transfer, phototransfer, or milling, just like you'd make PCBs for through hole designs, except you don't need the precision drilling you'd need for through hole parts. The transfer methods need an etch stage, which you can do at home with fairly safe chemicals these days, see for instance http://quinndunki.com/blondihacks/?p=835 for a very detailed tutorial. With a bit of care you can do double layer this way, but if you need lots of layer interconnections it's better to outsource PCB production. It's WAY easier to make PCBs for surface-mount designs at home because there is a lot less drilling involved.
Stencils: If you have a laser cutter available, cut them from construction paper. They'll only work once, but you can make a bunch. If you have one of those label cutting machines, those work too. If you're etching your PCBs, you can etch stencils from thin copper sheet, using the same method. Another excellent material that is already the correct thickness is the wall of an aluminium beverage can. Toner transfer works well for these and can give excellent detail. Remember most have a plastic coating on both sides, so file the outside with the image until you get naked metal (check with a multimeter), transfer on that and then etch. You can get presensitized sheets of various metals for phototransfer as well. If you only have a handful of pads you can also plop bits of solder paste on with a toothpick.
What about breadboards? The most common surface mount packages have breadboard adapter boards available. If there isn't one for the one you want, etch a board with a bunch of 2.54mm pitch pads along the edge (no holes, you want the board vertical so you have lots of breadboard space available). Stick any decoupling caps and pullup/pulldown resistors straight on the PCB so you don't clutter your breadboard with them, solder a bunch of header pins sideways, and there you go. You can put a label for the kind of part it is right into the copper.
Again, unless you're in a hurry or the board is very simple you probably want to outsource PCB production. If you do that, get a stencil made as well. If you can combine a bunch of stencils into one file, they probably won't charge you any extra for a stencil the size of 5 boards compared to one the size of one, so with a bit of planning you can get stencils EXTREMELY cheaply. And you can do tens to hundreds of boards with a single stencil if you take good care of it.
For alignment between stencil and board, if you're getting both manufactured for you, place two holes on opposite corners of the PCB that are exactly 3mm in size, and two identical holes in your stencil. Then you can use normal M3 bolts and nuts to fix them together, and it's super-easy to fine-tune alignment with the bolts in place but not yet tightened. There are pre-made stencils you can buy which have a selection of component footprints, so if you have say a little breakout board with just one difficult component you can apply paste to that with the stencil and do the others with the toothpick method or with a syringe dispenser (safety note: NEVER put solder paste in glass syringes, only use plastic ones).
My point in all this? If you can make through-hole stuff you can make surface-mount stuff, and probably with less effort.
It's easier now to order the ideal component. My original point is the day is fast approaching where there will be no more precision transistors in through hole. Even second hand or NOS. That's the day that getting an SMD setup will become easier for me. And then I'll get into it.
Do you really want to provide any more disincentive for budding hobbyists?
You can find a BJT about just as antiquated with a lower VCE(sat).
Also, maybe don't run a motor off 5V and then complain that the transistor is eating too much of that voltage.
(also, 2n2222's are like $3.49 for 50 on eBay US, via US sellers)
I use a PN2222 [1] transistor to switch a 5V supply through an infrared LED from a microcontroller. It works, but should I be using something better/smaller/more cost effective/more appropriate for the circuit [2]?
[1] https://blog.bschwind.com/2016/05/29/sending-infrared-comman...
[2] https://blog.bschwind.com/2016/05/29/sending-infrared-comman...
(the resistor in the photo is 680 ohms)
If you are trying to optimize for whatever parameters, then maybe there are better parts. But for something simple and basic like that, its fine.
also, you are likely making one circuit, not planning
on mass-producing 1,000,000,000 of whatever it is; who
cares if you spend a buck where a dime might do? you
want it to work first time every time, right? this is
art not technology (eg. industrial capitalism); we
have different design criteria here.
https://www.infineon.com/cms/en/product/power/igbt/igbt-modu...
If you want that special bridge chip, you start looking through online stores and datasheets (what a chore, why do they have to suck so hard?), trying to find the chip that has the availability, a reasonable price, the right specs and appropriate package for your use. Then you pay the shipping cost on top of the component, which might be more than the component itself. Then you wait a week or two for the thing to arrive. Then you pray it is intact and you can maybe start building. Hope you ordered the right part and didn't make a mistake in reading the specs.
Keep your TIPS, you might not necessarily use them for that brand new design of yours, but you'll probably use them for something.
[1] https://ocw.tudelft.nl/courses/structured-electronic-design/
Case in point: https://encrypted.google.com/search?q=lm317+hot
Can't be taken seriously if he's going to malign the utility of an entire class of components.