Transistors: The Transistor as a Switch

RE: Transistors: The Transistor as a Switch

D_Hersey — Mon, 16 Jan 2017 16:20:02 GMT

What happens if you put a little more effort into the CCS? How about a Wilson mirror?

http://www.diodes.com/_files/datasheets/ds30437.pdf

Or one made of a transistor, sense resistor and op-amp?

RE: Transistors: The Transistor as a Switch

balearicdynamics — Sat, 14 Jan 2017 20:24:41 GMT

John, I had only now time to read your post and I loved it! As many of us I suppose, we use transistors to switch signals in a lot of situations but this is a great reference point. I think that newbies can be very grateful to your easy to read and complete presentation of this use.

Enrico

RE: Transistors: The Transistor as a Switch

jc2048 — Sat, 14 Jan 2017 13:59:51 GMT

This is a follow-up the the blog above. Here I'm going to implement a constant-current source to replace the 1k collector load resistor to see if it will do a better job. This was at the suggestion of [mention:d8c3045f03b446e4be30edf8ec9c4677:e9ed411860ed4f2ba0265705b8793d05] in his comment above.

Here's the circuit diagram. The circuit replaces the 1k collector resistor I had in the original circuit, so the collectors of the two transistors join together.

[View:/resized-image/__size/600x334/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183314.jpg:600:334]

And here's the circuit as I built it. Beauty is in the eye of the beholder as they say and I say it's beautiful (feel free to differ if you want to - either way it will do the job).

[View:/resized-image/__size/600x450/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183315.jpg:600:450]

I better quickly say what it does, since it may not be obvious. The potential divider roughly splits the supply, giving a base voltage of something like 4.5V (in practice it will be a little higher because the transistor base will source some current). The emitter follower 2N3906 follows that voltage giving a voltage at the bottom of the 470 of somewhere just above 5V. That then means the 470 has a constant voltage across it and a current of approximately 9mA will flow through it. Most of that current will find its way to the collector, giving us our constant current source.

The 47n capacitor is there because I discovered that, when the switch turned on, the base-emitter capacitance in the 2N3906 was such that the base was pulled all the way down to ground before recovering again. The capacitor counters that. With a 47n it just twitches slightly - in a practical circuit you'd probably choose a value a bit larger (I just used what was quickly to hand in the clutter on the bench).

Now for the results.

[View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183316.png:480:234][View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183317.png:480:234]

And here are the original traces with the 1k load resistor as a comparison (watch out for the different horizontal speed in the one on the right above).

[View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183318.png:480:234][View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183319.png:480:234]

Now that we have a constant current going, it's evident that all of the transistor fall time, when the switch turns off, comes from capacitance - the previous exponential is now a nice linear ramp (though it starts to deviate a bit above 5V when the current source stops regulating the current properly).

Unfortunately, the constant current source doesn't really help at all. The capacitance of the collector of the PNP transistor combines with that of the NPN switching transistor to give an longer fall time even though it's now being charged more efficiently. Admittedly, we could increase the current, but we could do that with the resistor too without all the extra circuit complexity.

Final note. The constant current generator isn't too stable and is getting thrown around - here's an accumulated trace:

[View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_183320.png:480:234]

RE: Transistors: The Transistor as a Switch

mcb1 — Sat, 14 Jan 2017 12:10:27 GMT

Nice intro.

I've used the cap across the series resistor before when driving a 2N3055 using a 555, but never bothered measuring it.

I knew it worked because the transistor ran a lot cooler ....

mark

RE: Transistors: The Transistor as a Switch

jw0752 — Sat, 14 Jan 2017 06:38:26 GMT

Thank you for a great experiment and explanation.

John

RE: Transistors: The Transistor as a Switch

pettitda — Fri, 13 Jan 2017 19:45:26 GMT

I thought I'd start blogging about some of the things that I'm doing as I learn

Either you're a very quick learner, or you knew quite a bit about the subject beforehand. [View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/contentimage_5F00_3.png:16:16]

RE: Transistors: The Transistor as a Switch

DAB — Fri, 13 Jan 2017 18:42:57 GMT

Great post.

It reminded me that I need to look at the semiconductor effect using my new atomic model.

I may be able to explain why you see these effects based upon the transfer of photons between the atoms in the semiconductor crystal matrix.

Thanks for the reminder.

DAB

RE: Transistors: The Transistor as a Switch

D_Hersey — Fri, 13 Jan 2017 18:08:04 GMT

Are things improved much by going cascode by switching out the load with a current source?

RE: Transistors: The Transistor as a Switch

dougw — Fri, 13 Jan 2017 18:01:13 GMT

Useful article.

It is probably worth showing that the speed-up cap would be across the 10K base resistor.

RE: Transistors: The Transistor as a Switch

jc2048 — Fri, 13 Jan 2017 13:45:35 GMT

I got to wondering how good a spice simulation of a transistor would be - specifically, whether it might model the saturation.

Here is my circuit (without speed-up capacitor) simulated with TI-TINA, using the 2N3904 that comes with that package

[View:/resized-image/__size/620x305/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/5875.contentimage_5F00_183311.jpg:620:305]

[View:/resized-image/__size/480x234/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-8230db0e-0954-480d-877f-20c6665351b4/3731.contentimage_5F00_183312.png:480:234]

Compared with my trace, it's very similar. The saturation time is slightly longer, but presumably that varies between manufacturers, anyway. Similarly, the delay time and the rise and fall times look to be reasonably accurate (slightly less than I'm seeing with a real device), so the model includes the internal capacitances.