Simple Power Supply Stuff

[Bob Zimmer]

Wow!!  Thanks for all of the replies!!

With all the questions asked, maybe I should have added more info to the
original post?

This is the first "anything" that I've tried to design from scratch.  I
thought that I'd start with something simple.  I started by setting out some
basic design goals:

1) Power supply must fit into a 3.5" rack chassis along with my ASM-1.

2) The entire Power Supply should fit onto a 3x5" pc board.

3) It must provide +15V and -15V at 1/2A per supply.

4) It must power both the ASM-1 and a second analog circuit board with
   minimally: an additional LFO, Sample & Hold, Ring Modulator and
   possibly one or two waveshapers.

5) The design would use a 28 VCT transformer that I already have. It will mount
   easily on the 3x5" board with room for the remaining components.
 
6) It will use LM317/337's as they provide 80 dB of ripple rejection with two
   small caps added to the basic circuit Vs. 65 dB for the LM7815/7915.  This
   would allow me to reduce the size of the main filter caps to 1000 uf or
   1500 uf and still have the same quality on the output.  Did I spend
   too much time in National Semiconductor's Databook???

Design as tested:

The supply for +-15V uses the 28VCT 2A secondary by using the center tap as
the ground line and the other two wires connected to a 100V 2A Bridge
Rectifier. The positive output of the rectifier runs to the LM317 (+ Reg)
and the negative output runs to the LM337 (- Reg).  The main filter caps are
1500 uF 35V.  There is also a 1 uf Tantalum across each regulator output and
a 10 uf 25V cap in the adjustment line to get the 80 dB regulation.  Other
than that, there are two resistors per supply and one 15T trimmer. There is
a 240 ohm resistor between the output of the regulator and the adjust line.
There is a 2.2K in series with a 1K trimmer between the adjust line and
ground to trim the output voltage to 15V.

I don't have the ASM-1 finished yet to load the Supply, so I simply put a
pair of 10 ohm power resistors (connected together in series to create a 20
ohm load) across the output connections of each regulator (+15 and -15). The
10 ohm was the best choice that I could find quickly (from Radio Shack). My
math says:

 I = E / R
 I = 15V / 20 ohms
 I = .75A

This turned out well as I measured a true load of .73A or so on each supply.
The input voltage to each regulator at this load measures 19.3VDC which
should be just enough above the necessary 18VDC.  I also figured that:

    P = E * I
    P = 15V * .75A
    P = 11.25 Watts

I counted on each resistor dropping half of the voltage, therefore 10W
should have been sufficient.

I think that a real load will vary in current draw over time (depending on
levels, waveforms, etc.), but felt that the Power Resistors would be a fair
"smoke test" and allow me to check the heat sinks to see if they were
sufficient in size.


My question on Duane's comments:
  Whoops! I was under the assumption that if I had put in a pair of resistors
  across the output, then each would be dropping about half of the   voltage
& current.

  The 11.25W is definitely too much for one 10W resistor, but with the pair,
  I was expecting that each resistor would be dropping 5.6 Watts, which should
  be within their 10W rating.

  Why isn't this true??  I can't find an explanation in my electronics books.


My comments on Barry's Klein's comments:
  On the transformer voltages, agreed, but if I keep the total output of each
  supply to about 1/2A, shouldn't the 28VCT should work for this application?

  At the .75A load, I'm measuring 19.3VDC input to the LM317T regulator which
  should be be fine to deliver 15V (15V output + the 3V needed by the LM317).

  It's not that I was trying to get the regulators to shutdown from heat.  I
  was simply trying to run the Supply to verify that the heat sinks were
  sufficient by checking their temperature as they heated up.  I turned the
  supply off once I couldn't touch them any longer.


My comments on R. G. Keen's Post:
  Thanks much for the description of working temperatures!!  It is an area that
  confuses the heck out of me!!

  Yes, my goal is to put it in a box with little airflow. How did you know? ;-)
 
  It seems from all the responses that my little heatsinks will have to grow
  and maybe even get mounted on the rear panel.


My comments on Mark (MACHINE MEDIA)'s Post:
  What I think I found out is that the transformer is expected to deliver 28V
  RMS at the rated 2 Amp output.  If the load is less (I only plan on it
  delivering 1A), then the voltage is somewhat higher.  Such as it may start
  at 40V with no load, and drop from there as the delivered current rises.
  I am trying to allow for 18V to be available to the regulator at the desired
  1/2A output current.

  I will watch the input AC line voltage!  Never thought about checking it.


My comments on Scott Gavenhorst's Post:
  Completely agree on "Too hot to touch is too hot"!  I need to make it run
  cooler since I want to enclose it.  What I'm wondering is just how hot these
  regulators need to run and how big a heatsink is needed to get their
  temperature down. 

  I wasn't getting any smoke out of the resistors, but I did shut the supply
  down after having it under load for about 4 or 5 minutes.

  I found a book at Radio Shack on building power supplies that seems to be
  very good for the $4.95 price, but it just skims the surface on a lot of
  these calculations, especially on the math and temperature calculations.


My comments Gene (ASM-1) Stopp's Post:
  I think that Gene is saying that by using Power Resistors as a dummy load,
  I'm not simulating real conditions, therefore the regulators are working
  harder than necessary???

  I have been unable to find any 36VCT 3A transformers on the surplus market.
  All that I've found are new ones going for $30+ each.  Does anyone know
  of a source for a reasonable one?

Thanks again to all and sorry for the long post!

Bob

  >=== Bob Zimmer -- Philadelphia PA ===<
  >=== bzimmer at voicenet.com          ===<