Choosing Transistors is Bewildering

Discussion in 'Homebrew and Kit Projects' started by N5HXR, Jul 30, 2020.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: Left-3
ad: abrind-2
ad: L-MFJ
ad: Subscribe
  1. N5HXR

    N5HXR Ham Member QRZ Page

    Yeah, sorry if I confused things with my various questions -- I'm only interested in HF right now. My mid-term goal is ultimately to make a QRP digital mode transceiver from scratch.

    For me, at least, it's not enough to just go copy the PA from some other project and move forward. For example, I've assembled a QCX, which has a ~4W output with 3 BS170s, but I can't say I really understand what design process went into making it. I just assembled it, according to instructions. I'd like to build a thing, and be able to pop the lid open and tell my kids what every part does and why it's there :).

    I've been most inspired by EMRFD and a couple of individuals' project sites because they describe their design process and at least enough theory that I feel like I'm learning things. I have a working receiver chain that I've studied, built, tested, and measured (not that I won't learn some more and improve it, but hey, one thing at a time!).

    So now I need to make baby steps on the transmit side. I've narrowed down which kind of super-basic class AB amplifier I'd like to build, but the part listed in the article seems unobtainable. Hence, this post :). If that BFU590G would work in a braindead-simple amplifier circuit for HF, then I'm good to go, and can build one up and play with it.
  2. N1OOQ

    N1OOQ Ham Member QRZ Page

    The real secret is... Most designers have a handful of favorite transistors that they use for everything. I try to build most HF stuff with 2N3904s, for instance. Now the problem with ham magazine articles is that a lot of authors have a stash of old-timey components. Just check Digikey/Mouser when you read an article and see if they're still available. If so, buy a stash, especially of anything that you can get 100 for a few buck$. Volume discount!

    Otherwise, check around some of the surplus vendors, like Jameco. Sometimes you can find the older stuff, but much of the time it's just whatever they got from the big distributors' surplus.
    N0TZU likes this.
  3. VK4HAT

    VK4HAT Ham Member QRZ Page

    the value of books, Josh, like emrfd, is not the ability to replicate the circuits part for part, but rather the information about the design process. I treat schematics now, not as things to copy, but as inspiration. Often you can stuff any old part in there and it will work, mostly.

    The key with BJT's is ALL MATH and a little reading the datasheet. The math being calculating bias points. If you can get to grips with that, you will then understand all there is to know about BJTs in a practical sense. Armed with that knowledge, you can look at a circuit and go, OK, OM Billy Bob used a 2n90001 unobtanium. Datasheet, tells you FT of the device, bias current etc. Reverse math the circuit and see where OM Billy Bob biased the transistor. Then you can put any BJT in that circuit, bias it correctly and get the same gain that OM Billy Bob did, assuming he was right the first time. BJT's are all about the current though the emitter and emitter bypassing to set the gain.

    Now there are the book ways of calculating bias points, and I am not a fan they tend to be way to nerdy for someone like me who is more of a practical person than theoretical. If you fall into the practical way of doing things, I recommend watching Charlie Morris ZL2CTM on youtube, he runs though the math for biasing, inductor values, dc blocking cap values etc etc over and over in his videos. He uses a simplified method of the math and its what I use. Actually once i understood the math, i just made it all into a spreadsheet and I drop in some values and it spits out the resistor values for biasing, its not perfect but good enough for the kinds of girls I go out with.


    You still need to work out the transformer windings, blocking caps and emitter degeneration R16 and C30 below. But that is rather simple.


    OH one last thing Josh, if you do not have LTspice, download and install it. Its where I do all my tinkering with values and checking my math is correct. It saves on parts and solder. Its also free and hugely powerful. I have some PDF's on my website that I wrote to explain how to use it, if you need them. If you can follow Charlie Morris's math and grok LTspice, you will be well on your way to victory.
    N1OOQ likes this.
  4. KB0MNM

    KB0MNM Ham Member QRZ Page

    By the way ( BTW )- "Lumped circuit values" and "Parasitics" are both ways of describing how a given 'passive' component ( or for that matter any conductor or semi-conductor ) has properties other than you may have expected where higher frequencies are involved. The term 'parasitics' refers to how the overall circuit suffers as if it were a mammal with a bunch of mosquitos on it- hampered by a parasite. "Lumped circuit values" is more about 'dissecting' or 'separating' how much variance there is at different frequencies- resistance, phase change ( whether more inductive or capacitive ), etcetera. This is where an oscilloscope and knowledge of resonant behavior would contribute to your understanding with real-world experiments. If not, you should be looking at filter design soon. The phrase "ELI the ICEman" comes in handy when describing the effects of inductors (L) and capacitors (C) on the phase of a circuit. In this memory aid, E represents voltage ( Electro-magnetic force or 'pressure' analogy to water ) and I represents the 'actual' flow known as 'electron current', measured in amperes. In theory, at the atomic level, 'currents' of electrons and lack thereof seem to move in opposite directions, much like the empty places on a checkerboard when every other one is populated. That is the basis for what we call 'conventional' current compared to other... In terms of particles- the polarity + that Mr. Benjamin Franklin is said to have assigned- well... That is another story. Outside the battery...
  5. SM0AOM

    SM0AOM Ham Member QRZ Page

    In a previous occupation, soon 40 years ago, I did such designs for a living. But I had enough self-preservation as a newly minted EE to listen to "my Elders" who knew better.

    Out of this a few rules emerged;
    • Standardise on as few types of active components as possible
    • Do not put too much gain in a small box
    • Use an Ft about 10 times the operating frequency
    • Never "skimp" on cooling of transistors, always design of an adequate margin of power dissipation
    • Watch out for unwanted resonances in DC feed circuits, ferrite beads may "do wonders"
    For BJTs at lower frequencies and power levels to about one watt, the BC140 or 2N3053 families were used, with the BC337 used for general purpose switches and small-signal amplifiers.

    The 2N3866 and BF173 were used for frequencies up to 100-200 MHz and power levels to 2 W and 100 mW respectively. Higher frequencies and power levels were decided case-by-case.

    I want to "chime in" with G0HZU and VK4HAT that some care should be exercised so devices with very high bandwidths are not used in lower frequency circuits, as it becomes very easy to get an unwanted oscillator if the Ft should exceed the operating frequency with a factor of 100 or so.

    Further, the design process should not be overcomplicated.
    Use "rule-of-thumb" for determination of DC bias values and input/output impedances, and s-parameters if available for getting a "preview" of gain and stability conditions.

    It is much better to use one additional transistor stage to attain a desired gain value if trying to "squeeze" the last dB out of one stage means that the stability margins are compromised.

    KB0MNM, N1OOQ, N0TZU and 1 other person like this.
  6. VK4HAT

    VK4HAT Ham Member QRZ Page

    Always good to share in your wisdom Karl-Arne. And I agree with your points about high gain and high Ft and care needed to make sure i does not become an oscillator, something I overlooked to mention in my posts, but is important to remember. I had lots of fun "Taming" MPSH10's the first time I used them, crazy amounts of gain with a fairly high Ft and not enough knowledge LOL
  7. N7EKU

    N7EKU Ham Member QRZ Page

    Hi N5HXR,

    To continue on your quest. For your last question, since it seems you are aiming for HF, for a 0.5-1W amp there is no reason not to try the 2N2222A in a TO-18 case. I expect these could do 1W in class AB in push-pull for most HF frequencies. Their fT is 300MHz and they are fine to operate at 12V to get that power level.

    The latter could be a "show stopper" when looking at data sheets for RF transistors. Some are made to operate at higher voltages so you should observe carefully the voltage specified in the datasheet for what they used for power out and for the fT spec. If you want to stay at 12V this could limit you a bit.

    Once you move up to the 5-10W level, you start to become limited in what you can buy in a modern part for HF in a BJT transistor. Many manufacturers have switched to MOSFET's so you may have to also. That should not be a problem as they are simpler to bias and not prone to thermal runaway. They just require some different strategy for managing their different input impedance compared to a BJT. The problem would be though to find something if you want to stick with an older design like from Harry who used CB transistors.

    You can parallel lower watt BJT's as I said before. If you want also, I have a good stock of CB transistors so I can send you a couple. They have a different footprint than the TO-220 so you'd have to adapt the PCB pattern if you have one for that.


  8. N5HXR

    N5HXR Ham Member QRZ Page

    Wow, that's interesting. I just filtered for all the different packages for the 2222, and there are a lot of options that are beefier than the little TO-92s that I have scattered about.

    I guess the deprecation of all of these classic BJTs shows that in the market place. Once I finish this project, I'll investigate the FETs. I know I'm not taking the fastest path to making leading edge transceivers. But it's probably the physicist in me, instilled by my father when I was a kid, that wants to move one step at a time and be sure I understand things.

    Thanks for your help!
    VK4HAT likes this.
  9. N5HXR

    N5HXR Ham Member QRZ Page

    OK, I've stumbled my way through simulating filters and things with ngspice, but had avoided ltspice because I don't really have any Windows machines around the house. On your recommendation, though, I tried ltspice under Wine, and it seems to work great. I've now been happily tossing things in and working through the simulation, and it's far easier than ngspice has been. Thanks for suggesting it -- this is really helping me figure out what might make sense. I know that simulation is not the same as reality, but this helps me.
    N0TZU, K6CLS and VK4HAT like this.
  10. VK4HAT

    VK4HAT Ham Member QRZ Page

    yeah simulation just helps me to visualize what effect changes have on things and it allows me to make calculations like current draw, impedance, bandwidth etc to be able to compare ideal with reality. I learned a lot about biasing, negative feedback and gain control playing with amp values in spice and find it good for evaluation schematics you find on the net. Lots of copy this copy that out there and often they are all wrong.
    N0TZU likes this.

Share This Page