There are several examples in there, some of them boiling down to an open-loop emitter follower tethered to a zener reference. One of the designs there, the 53V version, uses feedback and I like that one as a basis for a 9V supply.
I'm not a designer, just a hobbyist, but here is how I'd proceed from that 53V design:
(1) I like the idea of using the Darlington topology, with a cheap and readily available 2N3055 for the power pass transistor. I might also like this because I still have a few laying around.
(2) The other transistor can be a 2N3904 or 2N2222. At Ic=1A for the
2N3055, I'd expect no better than beta=30, so the Ic on this other transistor would need to be about 33mA. Since Vce might be anywhere from 2.5V to perhaps 6V or so, this suggests that the transistor should be able to dissipate at least 1/4 watt, itself. I think both of those mentioned above can do this ... but not much more. So watch this.(3) 1A is a lot of current. Radio Shack sells only two kinds of transformers (12.6V and 25.2V) and its their 12.6V ones that come in
1.2A and 3.0A versions. With the 12.6V transformers, you'll need to use a bridge rectifier and with an average current of 1A, you'd be better to select the 3A version since the peak currents will probably be in that area.(4) At 1A average load, a 12.6V transformer is going to need a pretty seriously sized filter capacitor to get up to and hold a supply rail for the circuit. The peak voltage would be about sqrt(2) *12.6V or
17.8V, less a pair of diode drops at high current, or about 2V less to say 16V or so. There are books with charts to help you figure out exactly what size the capacitor should be for some desired ripple (you can work it out with trig) but my rough guess takes about 1/3rd of the 60Hz time (say, 5ms or so) and then takes some desired ripple voltage (let's say 1V here) and calculates C=dt*I/dV, which would be 5ms*1A/1V or 5000uF. Radio Shack sells 4,700uF at 35V. That should be okay.(5) The feedback method shown in the 53V supply on the web page looks reasonable to me. It's basically a zener being compared to a portion of the output voltage by way of a transistor that then pulls the Darlington base lower (more current) if the voltage climbs much above the reference point. However, I'd just use a zener from 6.2V to 9.1V or so, not the 15V (of course.) Call it a 6.2V zener for now. Here again, the transistor can be a 2N3904 or 2N2222, I believe.
(6) The zener needs a current through it. I'd go with 1mA. The data sheets often spec things at 20mA (Izt) but 1mA has worked well in the past for me. You can redesign for larger Izt, if you want. This current establishes the Ic for this transistor comparator and that, together with the minimum estimated difference between Vin and the base of the Darlington pair provides the size of the collector resistor. With a ripple of about 1V, going let's say from 15V to 16V, the minimum Vin will then be 15V. The Darlington pair will require some 1.25V or so, so the base should be at 9V+1.25V or about 10.25V. The difference is then 4.75V. At 1mA, this suggests a 4.7k resistor in the collector. I'd go with even less... a 3.3k here, just to be absolutely certain that at least 1mA was available for the zener.
(7) The base of the comparator transistor needs to divide the output voltage down. And it would be nice to have some slight adjustment range, so I'd recommend a potentiometer in here, as well. With a beta of about 200 or so for the transistor, I'd estimate that a current through this divider that is about equal to the zener current should be okay... another 1mA or so. With 9V on the output, this suggests about 9k ohms total. A little less would be okay. In any case, this seems kind of convenient for a 1k potentiometer (common value) to give us some variable range for adjustment. The base voltage should be
0.7V above the zener voltage, so about 6.2+.7 or about 6.9V. Actually, because of the light base current, it will probably be a little less than this. This suggests that we want the grounded side resistor to be 6900 ohms (assuming 9k total) and the high side to be the remainder, or 2100 ohms. However, with a 1k pot in the middle, and set to the midpoint, we'd take about 500 ohms off both sides, so we'd like 6400 and 1600, perhaps. Using 5% resistor values, I'd select 6200 ohms and 1500 ohms and put the 1k pot in the middle of the two.(8) As shown in the 52V supply on that web page, a small cap would also be nice across the zener.
(9) On the output, we should have another filter capacitor. I'd use a large value, perhaps 100uF or so.
All this arrives at a design looking like:
,------, ,-----| |-------+----------+-------------+------------, | | |(+) | | | | | | | --- 4700uF \ | |
12.6V | AC | --- 35V / 3300 | | AC | | | \ |/c NPN | | |bridge|(-) | +-----------| 2N3904 | '-----| rect |--, --- | |\e | '------' | /// |/c NPN | |/c NPN | ,-| 2N3904 '----------| 2N3055 --- | |\e |\e /// Vout | --+-----, | | | | | | | | /---/ | +---- Vout \ | / \ --- 10uF | 1500 / | --- --- --- 100uF \ | |6.2V | --- | | | | | | | --- --- --- \ | /// /// /// 1000 /