Two questions about this power supply design

[FenderBender]

Hello all,

This is my second question this week. Sorry! I do try my best to give back.  ANYWAY, I have a lot of the parts for my power supply build, finally, but now I'm working on some of the other things that I never figured out.

The design I'm working with is on the bottom of this post. It's being altered slightly to give me a different range of current limiting, but otherwise, very much the same. My version is only going to go up to 2A or so. As you can see, this design says 5A. I have halved the current adjustment pot to 50k to give me a limit of up to 2.5A, which should be good. The design also calls for the MJ2955 pass transistor as the LM317 can only handle, by itself, 1.5A. Additionally, I have an LT1085-Adj voltage regulator that I bought for something else but never used. It is, as far as I can tell, a drop in LM317 replacement, and can handle up to 3A. So, the plan would be to drop the pass transistor and to just use a single LT1085. Simple enough?

My question in regards to that is: How does one just remove that pass transistor without messing with that whole current limiting feedback loop and current sensing section (or that's what it looks like). Is it easily done? Am I being stupid? 

---

My second question is: I need a very stable -5V supply as I need to pull the Adjust pin of the LM317 below ground to get down to 0V operation. I was going to use an LM7905, but I thought perhaps I could do better?

Before I go any farther, I should add that I am going to need an auxiliary +/-5V supply for my 7-segment display drivers (ICL7107s) which will be doing voltage and current readouts. So it would be nice if I could share the -5V rail between the two. (+5V rail will probably just be a 7805) But the -5V, I thought could be a voltage regulator with a shunt reference on the adjust pin?

I also happen to have a few LT1033s. Again, never used them. If you look at the datasheet: http://cds.linear.com/docs/Datasheet/1033fc.pdf

...you can see a schematic on page 7 for a high stability regulator. The only problem with this is that I think the minimum output voltage is -8.25V, if you add the internal reference with the external LM329 (7V). So although it would be more stable, I don't think I can achieve -5V with that. However...I'm sure I could achieve -5V with something like a 2.5V reference, right?

Really my question is: Can I use an external 2.5V reference and..How do I calculate the value of R2 in that circuit if I am using a 2.5V reference. If you plug 8.25V into that formula there, you get an answer of ~0ohms. But I'm not sure how that was derived...

Any help would be lovely! Sorry if this is long. Just wanted to give as much info as possible

Thank you.

[T4P]

Sticking with a clunky big power transistor is the best way, so you won't blow your LT1085
Here's a reason : IC's usually run on the edge
Big power transistors? Lots of headroom provided you saw the SOA graph
And they might probably survive a reverse load
What sort of power supply are you looking for?
Bench? If bench i suggest you ditch the LT1085 ... IC Regulators are slow no matter what
Happy catching weird transients some circuits might send back into the transistor

[FenderBender]

Thank you both. Yes, I figured that the current sensing part does actually need that pass transistor to function right.

And Dave, yea, I was just trying to see if I could simplify the design, but that's only one little IC soo...I guess that's nonsense.

I'll keep the power transistor. But I may just replace the LM317 with an LT1085 since I have it, and I've seen some tests which conclude that LT1085 does perform better than the LM317.

Either way. Thank you. Now onto that -5V issue...

[FenderBender]

Do you think a 7905 would be enough?

[efron]

The 7905 would be enough if you accept their specified regulation limits (see datasheet).

[FenderBender]

Thanks mate.