mark@mips.COM (Mark G. Johnson) (11/06/89)
In article <9705@june.cs.washington.edu> dylan@cs.washington.edu (Dylan McNamee) writes: > >I have this burning desire to modify my CD player ... >I am going to use the same transformer, and split the supply from >rectification on down. Here's my question: The current power supply >goes like this: Transformer -> rectifier -> large cap (150uF) -> >regulator (78M15CT) -> small cap (47uF) -> 15V out. >If I want to improve the quality of this set-up, should I change the >topology at all? Can I just make the Caps bigger? Any other advice? Before jumping headlong into a project like this, it might help to enunciate the "theory" behind the idea that power supplies are an area of weakness in low cost CD players: (1) their poor regulation does a disservice to the analog stages, D/A and beyond (2) time-varying current demand from the electromechanical parts like the disc motor, arm servo, and focus servo manage to modulate the analog stages' power supplies and muddy the analog waveforms. The solutions to (1) are to build battleship-like supplies. (2) requires you to have _separate_ DC supplies for the electromechanical stages and the analog stages. They may perhaps share the transformer but not the regulator and final +/- NN volt outputs. ("Enid Lumley" would of course disagree, stating that they must be separated all the way back to the pole transformer outside your house). (1) Power Supply Rectifiers It's important to use diodes with a very high current rating, as they conduct on only a teeny tiny portion of the AC cycle. (If they didn't, then you'd have ripple going into your regulator and you don't want that.) I'd recommend you use a 4-diode bridge with at least a 10 ampere rating. (2) Post-rectifier, pre-regulator filter caps ("bulk decoupling") The more capacitance you can afford to use here, the easier time your regulator will have; it directly decreases ripple and thereby makes your regulator need to perform less "line regulation". Here's how to compute their effect. Say that your circuitry draws (worst case, peak) I amperes of current from the regulator. Presume a full-wave bridge rectifier that refills the capacitors 120 times/second. Say that you have chosen a post-rectifier, pre-regulator capacitor of C farads. Then the ripple voltage is given by Vripple = (I / (120 * C)). For example if you chose 5,000 microfarads for your supply that draws 300 mA (peak), then Vripple is 500 millivolts. Have a look at Senturia and Wedlock's textbook, section 8.4.2, figure 8.18. If you have a lot of "headroom" (i.e. there is a large difference between the transformer secondary voltage and the desired regulator output voltage), then you might consider a Capacitance Multiplier before the regulator. This is just a series-pass transistor with a large capacitance at its base; the effective filtering is Beta*C hence the name. You can find this circuit in all the cookbooks. (3) The regulator circuit itself This is a religious topic for many people. _Audio_Amateur_ for example maligns the one-chip regulator ICs such as 7815, preferring a discrete-part implementation. One thing to note is that practically _all_ regulators, IC and discrete, have a lower output impedance at higher current levels. If you want to get fancy you can obliquely refer to "the translinear principle" and amaze your friends. In practice this means, Put A Resistor Across The Output Of The Regulator. It'll increase the current in the regulator and make it work better, due to a higher Gm in the power semiconductor. Up to the limit of your transformer, increase the current as much as you dare (taking care not to overload things and causing a meltdown). It's usually the case that the circuit can tolerate 250mA more current; for a 15V supply that's a 60 ohm, 4 Watt resistor. If you're determined to use an IC, please choose the LM317/LM337 devices. They're much newer designs, have tons better performance than 7815/7915, and cost only a bit more. They're adjustable so consult the datasheet. Plus, "Mister POOGE" himself uses them in his Audio Amateur preamp design. (4) Decoupling capacitors on the regulator output These are examples of "if some is good, more is better" brute-force application. Be sure to check the final design with an oscilloscope to insure there's no instability or parasitic oscillation. (5) Other Single-point "star" grounding will help reduce ground-loops and crosstalk. Many people believe that there is a difference in the "sound" of different types of capacitors. However the only type of capacitor that comes in 5,000 microfarads is Aluminum Electrolytic. You may choose to parallel that with one of the in-vogue Sidereal Polyprolylene of Wonder Cap devices. I expect flaming tirades from pals SJC and MH_batcomputer. -- -- Mark Johnson MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086 (408) 991-0208 mark@mips.com {or ...!decwrl!mips!mark}
david@sun.com ("If you have to ask, then you don't have it.") (11/09/89)
In article <30864@obiwan.mips.COM> mark@mips.COM (Mark G. Johnson) writes: >It's usually the case that the circuit can tolerate 250mA more current; >for a 15V supply that's a 60 ohm, 4 Watt resistor. I guess it's obvious that you shouldn't do this unless you can mount the resistor where it won't heat up the CD player guts. >(4) Decoupling capacitors on the regulator output > > These are examples of "if some is good, more is better" brute-force >application. Be sure to check the final design with an oscilloscope >to insure there's no instability or parasitic oscillation. Be sure to a provide discharge diode so you don't mess up your nice regulator when you turn the power off. -- David DiGiacomo, Sun Microsystems, Mt. View, CA sun!david david@eng.sun.com