just finished my new sv-psu. because i wanted to use the transformer flexible for different purposes, i gave him his own home. basically this is again a simple lm317-schematic, very similar e.g. to tangent's steps (starting from C4A/D1A). since i'm a beginner and since it is not so easy to de-/resolder those p2p-cards i had to use, i decided to make it modular. here you'll see the single modules (from right to left):
- the rectifier (4x byv29f low drop, fast switch, soft recovery diodes + 4x wima mkp .047µf "snubbers")
- smoothing (11x 470µf/100v rubycons + 11x 1µf/63v mks as bypass caps)
- regulator (1x lm317) with some post-reg capacitance (220µf/63v) followed by some descending arranged mkp bypass-caps. i separated the regulator-stage to the top, to use the encasing itself (proma cooling-rip case) as cooling device for the lm317.
between the smoothing- and regulator-stage there's another module:
- electronic ripple- and hum-reduction. it is based on a swiss project of thomas schaerer (
"Spannungsregelschaltung mit elektronischer Brummsiebung", same
babelfished). one pnp transistor (here: MJE2955) and one npn (here: BD239B) form a "complementary darlington stage" that allows a huge first order rc low pass filter (with r1 = 1k2, c5 = 470µf) via current-amplifying (if i got it right there..):
(c3 = 10nf ceramic, r3=470 ohm)
theoretically this will reduce the overall-noise of the circuit to the (very low) inherent noise of the lm317 itself. additionally you'll get a nice slow turn on effect. well.. there's one drawback at least: another voltage drop of 2v (this psu is really not very effective: 2v drop for rectifyer, 2v for lm317 and now...). with bigger transformers you could flip the coin and take it as a benefit, because you'll get lower temperatures on the regulator. skip it, if you need high voltages (with best regards to head-fier steinchen...
).
the transistor coolers i used most likely are oversized: they got not even warm with 250mA dummy-load. the compl. darlington is followed by another pair of 470µf rubycons with their 1µf foil partners.
i made some mistakes with parts-arrangement and therefore had to place the regulator-stage directly above the rectifier-diodes. an argument could be made, that there's too much rfi in that particular place. i'll test the acoustics. i plan to build a new regulator stage using the lm317a anyhow, as soon as parts arrive. most likely it will be placed above the transistors.
altogether i realized some basic psu-design ideas:
- use hand-matched discrete rectifyer diodes (i have to concede, that hand-matching with my cheap dmm was not very reliable...)
- use many smaller caps instead of few big ones
- use hi-spec parts (e.g. 105°c, low esr, oversized voltage caps)
- bypass electrolytics
hm... and i disfigured the case with my dilettantish freehand drilling of the mounting hole for the regulator. so further casework (backplate design) has to wait until better tools arrive.
has anybody ideas, how to fix cards inside the case? they tend to slide a bit for- and backwards now. also i'd appreciate some advice on how to cut cards.. what are you using for this purpose?
[more pics
here]
Linear Mode
