Keep in mind that you are likely using "ideal" capacitors and "ideal" inductors, which don't exist in the real world.

Yes, the simulated components are highly ideal, as I pride myself with getting just the "best" components I can get. *laughs* No, just kidding I knew about the different models, I just hadn't applied the knowledge to the simulation.

Add some parasitics to the inductors, too. 56 turns of AWG20 on a 31mm*18mm*6.3mm iron toroid would take about 100 inches (2.5 meters) of wire, for roughly 85m Ohms.

Yikes! 85 Ohms will totally destroy my output voltage, not to take into account two of them
This is what I get for buying stuff before I know what I need, I kinda wish I had bought a 12v transformer instead of the 9v I now have

Oh well, the theory about the inductors have been very interesting atleast, if not applied to this project it will most probably come in handy in a future project.

To improve load and line regulation to a fixed output 9 Volt regulator, try the following...

1. Add 1 ohm...limit inrush current. Check!.
2. Get rid of C4 See below.
3. Get rid of D4 and replace with a zener. Check!
4. Replace C2 with a 220u cap. Check!
5. Use 10u to 100u as the output filter. Check!
6. Put 0.1u across R2 for stability if required. We'll see

About C4, I get that it's interfering with the regulation, the way I understood it was that it lead any ripple(and sharp changes in regulation) directly to ground. This is a good thing for ripple, a little bit bad when it comes to sharp regulation changes.
But I'm wondering if the good stuff(very little ripple) is outweighing the bad stuff(slow reaction to regulation), I have not seen the latter in my simulations, but then again, in simulation everything is "stable"(ie. not changing). If I were to remove C4, I would have to compensate greatly in the "main" filter, probably ending up with 4 or more 4700uF caps(not that it's a problem, I'm just trying to be sofisticated about the ripple ).

I have made a chart showing the difference in ripple with different C4 values over a varying load, this is with 2x4700uF caps as the main filter(and every cap is now using a model with more real behavior). It shows very clearly that C4 is helping a lot in ripple rejection, but again, I feel I need more information about the ill effects before changing it out.

I should probably say that the load requirement for this supply will probably not exceed 100mA, as I am planning it as a supply for my guitar effect pedals.
My first incarnation on a breadboard is actually doing that duty already, and I have discovered that I might need a short protection, the 2n3904s I'm using as output stage(while waiting for bigger ones from mouser) is really not living more than miliseconds if I happen to short the output, so I'm fearing the worst when I replace them Atleast the transistor-god is happy with this much sacrifice

Have Fun

Will do, am and same to everyone else

-
Kolbjørn

 

You misunderstood me!
I wrote "85m Ohms" which is 85 milliOhms, or 0.085 Ohms. It will have just a slight effect.

 

You misunderstood me!
I wrote "85m Ohms" which is 85 milliOhms, or 0.085 Ohms. It will have just a slight effect.

Phew, I can deal with milliOhms, it's those pesky megaOhms that get you.

I got my packet from mouser today *celebrates*, lots of good stuff, mostly resistors in different values, but some different types of transistors and other snacks aswell.
Gonna be a busy night

-
Kolbjørn

 

I'm having a question I'm hoping someone have an opinion on.

As it stands now I am thinking of fusing between mains and the transformer, and one at the output, both of theese at 500mA.
Will this be sufficient? I know the one at mains is bigger than needed, but if something shorts it is very probable that we get more than half a amp running, and if not the internal fuse in the transformer should pop.

Both of the fuses can as far as I can see do with fast acting ones(aka. normal), there should not be a huge inrush on the primary, atleast not bigger than half an amp; Likewise on the output actually.

I have tried shorting the output with the new transistors in place(FJN965, 5A/0.75W), and they can cope with atleast 5seconds and still be alive; Before the 5 seconds are up the fuse should have popped, and we have prevented catastrophic failure.
I am thinking this is enough of a short-protection?

I must mention that during the 5 seconds of shorting the output Q1 got very hot(geee, who would have thought ), I couldn't keep my finger on it, but it survived, and with fuse in place it shouldn't get that long of a short.
I am happy with this though, as the 2N3904's popped in a millisecond(well, what did I expect, they are rated for 0.2A...), I don't think I need to build in a current limiter.

-
Kolbjørn

PS: Sorry about the messy post, tossing ideas around and discussing it with myself while writing

 

The mains side fuse should be rated for a higher voltage than your mains voltage. If it's rated for less, it may not be able to break the plasma arc that will form when the fuse blows, and your project will be consumed. 500mA fast should be fine.

If you simply used a commercially available regulator IC like an LM317, you would have thermal limiting built-in. There are also quite a variety of low-dropout regulators available nowadays with lower idle current and better regulation than the 117/317 series.

You should also test your circuit for stability over temperature.
The SPICE directive:
.step temp 20 120 20
will run the simulation multiple times, stepping the temperature from 20°C to 120°C by a 20° increment.

 

The mains side fuse should be rated for a higher voltage than your mains voltage. If it's rated for less, it may not be able to break the plasma arc that will form when the fuse blows, and your project will be consumed. 500mA fast should be fine.

Indeed, both sides will use 250v fuses.

If you simply used a commercially available regulator IC like an LM317, you would have thermal limiting built-in. There are also quite a variety of low-dropout regulators available nowadays with lower idle current and better regulation than the 117/317 series.

Aye, I could have done it that way, but I wouldn't have learned anything. If I get to improve my skills with projects they are worth doing. I could have ordered a switch-mode regulated 9v/1A wall power adapter for 3.5$ from ebay, which would have fit my application perfectly well, but wheres the fun in that

You should also test your circuit for stability over temperature.
The SPICE directive:
.step temp 20 120 20
will run the simulation multiple times, stepping the temperature from 20°C to 120°C by a 20° increment.

Oooh, more simulation directives, I like

I really appreciate the input, I have gained valuable knowledge here, and I suspect I will gain ever more before I have a stable 9v supply done

-
Kolbjørn

 

SgtWookie forgot to mention this one...

1. Replace the value of the load resistor with; {Rload}, where Rload is a variable of your choice.
2. Add the following ltspice directive; .step param R start_value end_value step_size. e.g. .step param Rload 20 10 1.
3. (2) steps the load resistor from 20Ω to 10Ω in 1Ω steps.

If you already know this... never mind.

Regards,
Ifixit

 

SgtWookie forgot to mention this one...

1. Replace the value of the load resistor with; {Rload}, where Rload is a variable of your choice.
2. Add the following ltspice directive; .step param R start_value end_value step_size. e.g. .step param Rload 20 10 1.
3. (2) steps the load resistor from 20Ω to 10Ω in 1Ω steps.

If you already know this... never mind.

Aye, already knew that, but good info to have in the thread in case someone stumbles upon it.

I've been a little bit away from the electronics world since the last update, partly due to being busy with work related things and partly needing a break. But now I am picking up the pieces again and storming forward.

I am thinking of ditching the 9v transformer for a 12v one, that will give me a little bit more headroom for filtering/regulation. Drawback will obviously be more heat as I need to drop the voltage alot more. Will have to think about the pros and cons of that approach.

Been doing some work on the box for the supply as well, grinding and drilling holes, I'll upload an image of the front test fitting I did.

-
Kolbjørn

 

Did some work in LTSpice and Eagle, and adjusted it for use with the 12v transformer, then proceeded to wire it up on a breadboard. I must say that I am totally impressed with the FJN965 transistors I'm using, I've had smoke coming off them when shorting the output, remove the short and they go back to working as if nothing had happened.

With the circuit attached I am getting about 1.2mV ripple at 500mA output(17 Ohm load). Most of the ripple is lost in the capacitance multiplier, C5 is actually lower value than I would like, but as it is at the moment it is the biggest ones I have below the 4700u ones. My inventory of components is also getting so big I'm having trouble filling my next mouser order over the free-shipping amount, so I'll just make due with what I have atm.

I note that upping C6 is having a good effect on the remaining ripple up to a certain size, and I'm guessing this will slow down the regulation if it's too big, is 1uF an ok size? Or even 10uF?

I see now that I have forgotten a 100uF cap on the output, oh well, back to redesigning

Images:

-
Kolbjørn

 

Uhhhmmm... For 500mA output you could use an LM317, that has very good ripple rejection.

 

Uhhhmmm... For 500mA output you could use an LM317, that has very good ripple rejection.

True, or even a 7809, but half the point of this particular project is learning something. Anyone can wire up a 3pin regulator, it's more fun doing it discretely and figuring out why it works(or fails to work...).

Just a few minutes ago I had serious oscillation, supplying one of my guitar effect pedals with juice I got a nasty hum out of my amp; Touching various components on the breadboard, or even moving my hand close to the breadboard would change the tune, make it stronger, weaker or remove it.

Turns out I omitted the capacitor on the output when I populated the breadboard. I tossed a 180nF cap there and it got dead silent. I should probably have known that omitting that cap could cause problems, especially on a breadboard with wires pointing in all directions. Another thing learned!

The reason for it being fused for 500mA is I'm feeding the circuit from a 12v/1A transformer, I feel 500mA is a safe number to land on, though it could probably survive more. The primary function for the supply will be driving guitar effect pedals, each drawing perhaps 10-20mA, 500mA will be plenty for a lot of pedals ^^ (and atm I just have two...)

-
Kolbjørn

 

Your voltage reference is an ordinary diode and the error amplifier transistor has a base-emitter diode so your circuit makes a pretty good thermometer. Its output voltage will drop when the parts get warmer.
It also has poor line regulation (its output voltage changes as the input voltage changes) because its reference diode is not fed from a constant current source.

 

Ok, if you want to learn the basics of voltage regulators, it's ok, but you could do that in the breadboard as an exercise. Voltage regulator ICs have a lot of advantages, like current limiting, thermal limiting, and much better voltage stability.

Download the "3-terminal regulator is adjustable" application note and datasheet for the lm317 from National's site, and take a look at them. You can learn something by doing that, and would be good to build as a final circuit, after you learned the basics. You probably want to do a search about sizing heatsinks, also.