[Kevin Darrah] wanted to make a simple 3.3V regulator without using an integrated circuit. He wound up using two common NPN transistors and 4 1K resistors. The circuit isn’t going to beat out a cheap linear regulator IC, but for the low component count, it is actually pretty good.
In all fairness, though, [Kevin] may have two transistors, but he’s only using one of them as a proper transistor. That one is a conventional pass regulator like you might find in any regulator circuit. The other transistor only has two connections. The design reverse biases the base-emitter junction which results in a roughly 8V breakdown voltage. Essentially, this transistor is being used as a poor-quality Zener diode.
At normal temperatures, the pass transistor will drop about 0.7V so dividing the 8V across the “Zener transistor” by two gives 4V for the other device’s base. Drop the 0.7V and you wind up with 3.3V. [Kevin] does some quick tests and the performance really isn’t as bad as you might expect.
If you want to read about issues with Zeners, we posted about it. Note, though, that by that post’s definition, the Zener here is used as a reference, not a regulator. We also covered a video on Zeners.
The old ones look on and smile.
We did have linear regulation before the IC.
Yup. Shunt regulators weren’t perfect, but they did the job.
Well I thought the novel part of this was the use of a transistor as a zener diode. Admittedly, a poor zener and zerers are none too good anyway.
In years past I designed a large 5V regulated supply for a device that was produced using many 74xx chips. We actually used an array of transistors in the base of the pass regulator to reference (but not against a reverse bias junction) and do current sensing, etc. while having the transistors all at the same temperature. The current sensing resistor was a fat PCB trace.
zener is ok for a reference. i used one in an smps once, worked out great. just took a 555, and gate, opamp (in comparator mode, comparing the output with the zener), power transistor, and the rest of your typical smps components.
I think a reverse-biased B-E junction may actually perform better than a zener diode. True zener diodes are very noisy, so you have to filter the output of a zener reference with an R-C low-pass. I think that a reverse-biased B-E junction operates in avalanche mode, which may not have as steep a V/I curve, but is less noisy.
When I’m putting together quick-and-dirty linear regulators, I like using LEDs in forward bias. Different colors have different voltage drops, and they’re very clean.
Aren’t most “Zeners” actually more properly called avalanche diodes anyway?
I think all of the higher voltage ones are avalanche diodes, but now I’m not sure – I may have it backwards. And I also may have the other part backwards; maybe it’s the avalanche diodes that are super noisy. But I do know for sure that you can get a nice low-noise voltage reference from a forward-biased LED, because I’ve used that recently.
The higher voltage zenners are avalanche and have a positive temperature coefficient. The lower voltage zenners are actually a zenner and have a negative temperature coefficient. At about 5.6 volts a zenner is half avalanche and half zenner and the temperature coefficients cancel out.
So…could you just replace one of the transistors with an actual 8.1V zener diode (or whatever the nearest cheap voltage is)?
Yes, assuming you had one.
I would use a 4v zener instead and get rid of the voltage divider.
Neither would have great temperature stability, but I suspect the 8 V one would be better.
Yes, you can use a “normal” zener, but all avalanche devices (including “normal” zeners) are noisy and a small cap should be placed across them to reduce the noise. If memory serves correctly, the 6.8V zeners are the most stable of the lot Remember to look at stability across current ranges and across temperature ranges. Consult your datasheets.
Of course. But a 2N3904 is literally a couple of cents, and you probably already have a bag of them.
Sorry, not watching a YouTube video for what should be a schematic drawing.
We must fight the youtubefication of the web!!
And people who use WordPress on single page sites
i like videos where the creator puts relevant info/links ‘down below’ in the video description area.
unfortunately in this case, the links ‘down below’ are for financial support (donations), Twittering, and products+services for sale (shopping).
and that’s all okay because channels are funded in this way and questions end up in the comments sections. and video creators can include or not include whatever they [don’t] want. someone else can draw and share.
Learning about the skip-back, pause, and skip-forward keys for YouTube (‘J’, ‘K’, and ‘L’, respectively) has made my life so much better. I skipped past the babble to the schematic and got to the useful information (the schematic) in just a few seconds.
good method.
keys 0,1,2,3,4,5,6,7,8, & 9 are also for skipping around
he! that’s a nice tip, never knew that. Always used my arrow keys and spacebar, but this works much better as it forwards in larger steps.
Agreed. I usually have to skip posts where the only info is a YouTube link entirely because they are non-searchable bandwidth and time hogs and not very usable on my phone. On the rare occasion I have time to sit at a real computer other than at work I want to work on my projects rather than watch YouTube videos. The one exception is things like “how to do operation X with tool Y” where a visual demonstration is actually helpful.
I should be ummm contrary, and errr, record my ummmm comments on errrrr youtube and them ummm post the errr link here. and that wraps it up don’t forget to ummm subscribe and errr click my patreon link to errrr fund my meth habit.
IIRC Dave(@eevblog) has a video on this zener effect of transistors
Just Google “zener effect of transistors” and you’ll get lots of circuits. I like:
http://venerabelia.co.uk/science/transistor-as-a-zener.html
From the engineering and manufacturing point of view this solution is awful. For fun and learning experience it’s ok.
You would not use this solution for the following reasons:
– component count (a single ldo is just one component, the pick and place cost of single component is around $ 0.02-0.1, even if you use additional input and output capacitor, you cant beat it)
– board space (ex. NCP114 comes in 1mm x 1mm size and costs around $0.04 in quantities, but there are several others)
– safety features (over current limit protection, thermal shutdown, reverse current etc.)
– power consumption
– precision
just to name some