[TzeYang]

lawlz, they're acting as diodes.

EDIT: Oh wait, you could just use regular diodes like 1N4148, why use the transistors?

 

[tangent]

Small-signal transistors acting as diodes are less leaky than generic diodes. That's valuable here because they're supposed to be reverse-biased under all normal conditions. I suppose there must be a trade-off...lower maximum forward current, perhaps?

 

[rds]

How critical are the values of c10 and c8?
I'm looking at using 150uF 35v caps for c7-c11. I'm just a little iffy on c10 and c8.

 

[JamesL]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif Anything concrete?

Sorry, concrete? I was just wondering what you meant by an improved design..
It sounded like you had something on mind that you were working on.

 

[tangent]

Quote:

Originally Posted by rds /img/forum/go_quote.gif How critical are the values of c10 and c8?

Not very. I'd go smaller rather than larger. The main thing is to use caps with the expected case sizes, so they sit down properly on the board.

Quote:

Originally Posted by JamesL /img/forum/go_quote.gif I was just wondering what you meant by an improved design.

Oh. Well, all the Wishlist items above will get done before the next version. Between v1.0 and now, I have:

1. Renamed Q3 and Q4 to be in D series, to hopefully avoid more confusion on this topic
2. Made the test point holes bigger, to allow Keystone wire loops
3. Moved diodes and associated caps farther away from the board edge
4. Added DO-201 pads around TO-220 bridge diode pads, as in STEPS v1.2
5. Fixed spacing between D1 and its bypass cap

Relative to STEPS v1.2, obviously the regulator is a vast improvement. I've also gone back to the big 6-pin line chokes I used back the STEPS 1.0 days -- I've got the space again.

I'm dithering on whether to nix the 35 VA transformer option. Unless you change the CCS biasing, the circuit as designed can only put out about 330 mA, so there's no way to actually pull that much power from this supply. Even 25 VA is overkill, but the recommended Talemas use the same pin spacing for them as for the 15 VA ones. 15 VA is plenty big for this.

What's staying my hand from deleting the 35 VA option is that other manufacturer's transformers may not be as compact, and it's nice to have even a thin excuse to keep parts away from the transformer. I'll probably only nix it if I actually need the room again.

 

[TzeYang]

How much current that the regulator can handle is dependent on the quality of the pass transistor?

EDIT: Knowing that the super regulator design exists, I wondered why Amb went with a much more traditional sulzer like basic discrete regulator (Sigma 11) instead of following to something similar like Walt Jung's super regulator design

 

[rds]

A hole for 7.5mm diode caps would be nice.

 

[1UP]

Hi - is this a dual supply? What would be the main differences between this and the Sigma 22?

 

[amb]

Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif EDIT: Knowing that the super regulator design exists, I wondered why Amb went with a much more traditional sulzer like basic discrete regulator (Sigma 11) instead of following to something similar like Walt Jung's super regulator design

Not to thread-crap, but σ11 is the single rail rendition of the σ22, and you should read the headwize thread which discusses the thoughts behind the design. An all-discrete circuit with high output current capability was a requirement from the onset. IC-based solutions would have been impossible to meet all of the original design goals stated in the thread.

Btw, compare Erno Borbely's EB703/259 PSU to σ22. Borbely's PSU isn't auto-tracking like the σ22, but there are some topological similarities.

 

[tangent]

Quote:

Originally Posted by TzeYang /img/forum/go_quote.gif How much current that the regulator can handle is dependent on the quality of the pass transistor?

No, the limit on output current is set by the CCS that keeps the pass transistor "open". (D2, R5, R6, Q1) The maximum output current is that CCS value (~5.5 mA) multiplied by the pass transistor's hFE. The 330 mA number I gave above is actually the minimum output current, being calculated with the minimum hFE for a D44H11. You can lower the R6 value to increase the current limit, but you don't want to make it any bigger than you have to, because the error amp has to sink away all of this current when there is no load on the regulator. The less current the error amp has to sink away, the better it will perform.

Anyone interested in how this regulator works should first read my Op-Amp Based Linear Regulators article, which introduces the concepts behind this piece by piece. To dive deeper, there are links at the end of that article to the original Jung articles.

 

[rds]

I can't figure out how you fit the ELF850T on the board. Did you move the pins?

 

[tangent]

I'm not sure I understand the question.

Are you saying that the part you bought doesn't fit on the board? If so, can you take a picture to show it?

Or are you asking how I managed to get back to the larger filter choke? The answer there is, it's a bigger board than the STEPS, so I had more room.

 

[rds]

My ELF850T has 6 pins with wider spacing than the 4 pins provided on the pcb. I ordered the one listed in the parts guide afaik.
Maybe there are different pin versions of the ELF850T?

 

[tangent]

Crap....I misunderstood my own notes. The YJPS 1.0 boards are using the same chokes as STEPS 1.2, not STEPS 1.0!

You can adapt the choke you have to the board. The middle two pins are just for stability. Just match the outer 4 pins on the choke to the 4 pads on the board. You'll have to rig some way to keep it stable if you'll ever subject the joints to stress...ship it, travel with it, etc.

I'll be fixing this for the next version of YJPS. The space is there for the bigger choke.

 

[error401]

Mentioned it in the other thread, just a quick note here...

The transistors aren't listed in the parts guide. Doesn't seem to make sense for that to be intentional...