Maximum Current for Discrete 741 & 555
  • This may be more of a physics question than a support one, but I appreciate whatever info is offered.

    Since the discrete transistors are hundreds (thousands?) of times larger than the tiny transistors that are etched (printed?) in silicon on an IC, shouldn't the kit versions of these circuits be able to handle vastly more current than the IC ones?  After all, a through-hole resistor can handle more current than a surface mount one.  Power dissipation in resistors seems to correlate closely with physical size, and I believe that's true for capacitors, inductors, and diodes?  But as far as I can tell, these kits have similar current ratings to the classic ICs they're modeled after.  I'm sure that's intentional, but I'm curious if its relatively easy to modify these kits so they can handle much more current.  For example, if I replaced the 3904 & 3906s with 2222 and 2907 transistors would the 555 and 741 now handle 800ma?  For that matter, would using ZTX968s & 1049s let it handle 4 amps (assuming the resistors and PCB traces were large enough)?

    Those three transistor families all come in TO-92 packages, so I guess I'm also don't understand why they have such different capabilities either.  Is the size of the PN and NP junctions in the ZTX968 16 times larger than in the 3904?  I know that most higher power transistors come in physically larger packages (TO-220 etc), but is that just to allow a larger heat sink surface area, or are the silicon boundaries physically larger?  And for the ones that do come in the same package size, is there an advantage to the lower ones other than cost?  Is it cheaper to make a 3904 than a 2222, or can the 3904 switch faster because it's smaller?

    I know that's a lot of questions for one post and that this isn't an electronics classroom.  But again, I appreciate any insight.
  • While the set of discrete transistors may occupy a volume thousands of times that of the 555 or 741 die, the difference in size is largely one of packaging. 

    Here is an Xray through a TO-92 2N3904: https://www.flickr.com/photos/59935244@N03/5475873168

    As you can see, the actual die is tiny-- only about 0.5 mm (.020 inch) across, including the areas used for wire bonds. A modern 555 variant, the TLC555, lists its die size as 1.65 x 1.27 mm
    (source: http://www.ti.com/lit/ds/symlink/tlc555-die.pdf ).  So, a naive estimate would be that one transistor has about 1/8 the area of a whole 555.  

    Now, take a look at the die for that TLC555. There's a photo here: http://www.bunniestudios.com/blog/?p=3542 . The big black circles are the wire bonds. And, there are several very large areas that look like intersecting combs. THOSE areas are the power transistors that handle the outputs.  Yes, some of them are 1/8 the size of the die, or close to it.   And, that's on a modern chip -- a "shrunk" die, based around super-efficient mosfets, not big and clunky bipolar transistors like the 2N3904.

    So, a first inspection suggests that the actual area of the power transistors on the original chips is of the same order of magnitude as the area of our discrete transistors. It's certainly not a difference measured in the hundreds or thousands of times. On the contrary: the sizes are quite comparable.

    Getting back to your original question: Could you replace some of the transistors with more powerful types, and make a 555 or 741 with much higher current carrying capability? Absolutely!  Just watch the rest of the properties of the transistors, to make sure that you're keeping with the same kind of device.

    It is certainly possible to make more efficient devices than the 2N3904/6, through modern fabrication techniques and using larger areas.  

    I don't have any specific data about its die, but given that the ZTX968 is about 12 times the price of the 2N3906 (including packaging costs, which are fairly constant), it would make perfect sense economically if the die were 16X as large, 2 mm x 2mm, which would still fit nicely within the TO-92 package.

    And yes, often smaller devices are preferred, because they have lower capacitance, and can be driven more quickly. 

    As to our particular choices: the 2N3904 and 2N3096 are simple, inexpensive, and common "general purpose" transistors that work well for these kits in most respects.  There are a few places here or there that using a specific transistor with higher current or voltage characteristics would have been helpful, but we eventually decided (for both kits) that the simpler BOM was the better option.
  • Wow!  Thanks so much for the quick and detailed reply.  I've never seen an x-ray of an electronic component and find them strangely pretty, especially the TLC555.  I had no idea the actual "goods" inside a discrete transistor were such a tiny portion, although I take some comfort realizing that size matters here just as elsewhere, and that there's even a logical reason that a 3904 costs less than a 2222 which costs less than a 968.

    Looking forward to putting together the discrete 741 when it arrives and then doing the 741 experiments in Charles Platt's Make More Electronics with it and my oscilloscope.