Russell McMahon wrote:

> You will note that I gave some efficiency figures at the bottom of the post
> and I made some practical comments about it working over various voltage
> ranges with various parameters. I also provided a specific set of resistor
> values in the text under which I did some measurements. (I said : "Try R2 =
> R3 = 1k, R1 = 3k3, R4 = 10k for starters".) I didn't put these on the
> diagram because, as I noted, this is not a nice circuit as it stands,
> different values will be required depending on load and Vin etc and, as I
> noted, I was offering it as a basis for possible development for those who
> may want to play with it.
> .
> Putting it more clearly, hopefully:
>
> I drew this circuit up essentially as shown.
> I then built it,
> This circuit is working on my workbench.
> The results given were measured with standard test equipment.
> I monitored waveforms with an oscilloscope.
> It oscillates very well when it is in oscillation.
> The circuit works over only a limited input voltage range.
> The starting voltage range is within but less than the operating voltage
> range.
> As noted, I do not suggest this as a final circuit.
> Something may be able to be made from this circuit.
> It MAY take a very simple change (made possibly with lots of development
> effort) to make it much more useable.

And a VERY functional way of designing something.
:o)


> As to mechanism of operation -
>
> > As Vin goes up when power is turned on, Q2 will go on via R1 and keep its
> > collector low, thereby preventing Q1 from ever going on.  If R1 is low
> > enough Vin will continue to drain thru L1 and Q2 forever or until
> something
> > fries
>
> The mechanism by which Q2 comes out of saturation and causes switching was
> briefly covered in my original text. This is a fairly common and very time
> honoured means of causing switching transition in simple low power, power
> conversion oscillators. As I noted, it relies on the beta (current gain) of
> Q2 and is therefore not a nice method.
>
> Again:
>
> - Q2 turned on by R1.
> - Q1 turned off by R3 and not enough voltage from Vin to turn it on due to
> D1,D2 and then R2/R3 divider.
> - Q2 collector current via L1 increases (i ~- Vt/L)
> - Q2 collector current driven by Iba2 x Beta q2 via R1.
> - As current ramps up a point is reached where Icq2 exceeds beta x Ibq2 and
> Q2 will start to come out of saturation.
> - At this stage L1 attempts to maintain the peak current flow while Q2 is
> reducing its ability to handle this. Vq2 will start to rise regeneratively.
> At this stage it still has drive via R1.
> - When (IF) Vcq2 rises high enough to start to turn off Q1 the process
> becomes fully regenerative and Q2 is shut off completely.
> This happens about when
>             (Icq2 - 2 x 0.6) x R3/(r2+r3)m = 0.6 volts
>
>  ie when Q2 collector voltage - the 2 diode drops of D1,D2  divided by R2,
> R3 reach 0.6 volts approx and turn on Q1.
> Clearly this is not a very nice process and the range of input voltages
> which it works over are affected strongly by component values and Q2 beta.
>
> Nothing will "fry" as long as the circuit is "designed". ie Icq2 max does
> not
> saturate L1. R1 is dimensioned to cause beta starving of drive to Q2 at
> designed peak current. Beta of Q2 is known (or its range).

> Having such a simple WORKING circuit to look at allows people to examine and
> understand the mechanism of oscillation. More complex circuits which have
> been "optimised" for practical use hide the core processes and are hard for
> beginners (or experts ;-) ) to understand.
> This circuit is fundamentally the same as the EDN circuit forwarded by Alice
> a few days ago but its operation is much easier to understand. For that
> reason alone it is potentially valuable even if it is never used in this
> form.
>
> Roman has suggested (offlist) an emitter resistor for Q2 to stabilise
> oscillation.

This would increase stability and provide some
relief from changes in Vbe and Beta with each
device.

> As shown there is NO explicit voltage regulation but this could easily be
> added in rudimentary form with a zener from Vout to left hand end of R2 or
> (probably better here) a zener across the output. Power output can be
> adjusted by varying R1 to limit Icq2 peak appropriately.

> This is the key point.
> The rise in Vcq2 due to inductive ringing of L1 causes Q1 to turn on and
> regenerative turnoff commences.
> Very nice and square waveform at Cq2 over a limited Vin range.
>
> > At most one single brief pulse will be delivered to the output
> > each time power is applied.
>
> This is probably true for Vin outside the working range
> The circuit is dependant on Vin being neither too high nor too low.
> I can think of several possible ways of improving it but they will have to
> wait a while.
>
> Roman may turn it into a work of art in the meantime :-)

Sorry I've been away for a few days riding my
motorcycle but I am keen to have a fiddle. :o)
Apart from the emitter resistor already mentioned,
I can see some benefits from adding a capacitor
B-E on Q1, this could help tune the circuit when
matched with the performance of the inductor at
set currents, and may help oscillation over a
slightly larger current/voltage range than you
are getting now.

I think the main issue will be regulaton, and how
cheaply and simply it can be added, since you have
already proven that the circuit works and has an
acceptable efficiency for a quick nasty prototype.
-Roman

>     ---------------------------------------------------------------
>
>                               Name: boost1.gif
>                Part 1.2       Type: GIF Image (image/gif)
>                           Encoding: base64

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu