less hum: switching power supply?

Mon Jun 5 18:16:12 CEST 2000

P.S.  Although I work in support for legacy 3780-based equipment now, I once
worked as an engineer for a computer firm.  I used to work on Commodore
computers back in the 80's as well.  The biggest problem Commodore had with
the 64's was a poorly designed power supply.  The power supply used the
LM7805 (the later ones that could not be opened) on a very small heatsink.
These power supplys would rarely blow during constant use, but would almost
always blow every RAM chip in the box if power was removed after the
computer was turned off.  The reason had to do with the face that when the
computer was shut off, all the power supply could do was charge the cap on
the output.  Then when power was removed the transformer side voltage
dropped first so there was an effective reverse-voltage (cap on the output
having a higher charge than the one on the input).  The regulator might
survive this a few times, but eventually it would become more unstable.  

I proved this to the Commodore engineers in Shaumburg, who believed that
they were buying "inferior RAM chips" , and that is why the C128 power
supply was redesigned (not well, but better).  

The biggest problem with the 78xx regulators is that when they become
unstable, they may work MOST of the time and only go nuts once in a while.
The test I used to prove to Commodore that the problem was the PS/regulator
was to wire a 5.1V zener across the 5V line.  We then hooked a scope to it
as well.  We then simply turned the computer on and off.  This worked
correctly 9 times but on the tenth off-on trasition the Scope showed 8.5
volts on the 5V line, the Zener glowed brightly for a second and blew up,
and the RAM chips were destroyed. 

Changing the power supply over to an LM309 in a TO-3 case and that power
supply I used for the testing still works on my worbench at home today, and
I use it for all sorts of testing besides.  It is capable of forcing current
through bad circuits well so it makes diagnostics sometimes as easy as
"finding the HOT part".  

Jay Martin

-----Original Message-----
From: Jay Martin 
Sent: Monday, June 05, 2000 10:45 AM
To: 'Tim Ressel'; 'Roel Das'; Synth DIY
Subject: RE: less hum: switching power supply?

I beg to differ,

The 78xx series of regulators do NOT have built in protection for shorting,
overheating or reverse-voltage damage (reverse voltage can happen when power
is removed but an output capacitor still has voltage in it.)  I have had
many circuits (especially EPROM and RAM types) destroyed by an LM7805 that
decided to start spiking at 8 volts.  I have never had the problem with a
Zener-Bypass transistor design or with the LM/SK309 or similar TO-3 based
regulators.  The LM309 has survived shorts, over-voltage spikes, reverse
voltage and will go into shutdown mode if overheated, protecting the
circuits being regulated.

Jay Martin
ISI/CLEO Technical Support.

-----Original Message-----
From: Tim Ressel [mailto:Tim_R1 at verifone.com]
Sent: Monday, June 05, 2000 10:16 AM
To: 'Roel Das'; Synth DIY
Subject: RE: less hum: switching power supply?

There is nothing wrong with using 7815/7915's. In fact, they are quite clean
on
the output. The standard trick to clean up the output of a noisy switching
power
supply is to use a low-dropout regulator on the output. Use plenty of
heatsink,
and make sure your filter caps follow the 2000uF/Amp rule. And keep in mind
they
are only good for 1 Amp.

There are lots of books that explain about power supplies, including Barry
Klein's.

I have gone with a Power-One linear supply +/-15V at 4A. Problem solved.

Tim Ressel--Compliance Engineer
Hewlett-Packard
Verifone Division
916-630-2541  
tim_r1 at verifone.com                     

-----Original Message-----
From: Roel Das [mailto:Roel.Das at student.groept.be]
Sent: Monday, June 05, 2000 7:20 AM
To: Synth DIY; Jay Martin
Subject: Re: less hum: switching power supply?

Hello Jay, others,

I am VERY interested in anything that has to do with power supplies. I'll
start building my modular in 2 weeks (after I graduate, woohooee, finally),
and the first thing you need is a powersupply, i guess... My midi cv
convertor still needs a decent psu too, it's now running on a very basic
7815/7915 design, even halfwave rectifier... Might be good enough, but I
want better.

Thanks,
Roel

----- Oorspronkelijk bericht -----
Van: "Jay Martin" <jmar at intface.com>
Aan: "chris" <espace at e-server.net>; <synth-diy at node12b53.a2000.nl>
Verzonden: vrijdag 2 juni 2000 20:23
Onderwerp: RE: less hum: switching power supply?

> OK, I have to put my $.02 in here.  It has been my experience that the
> biggest problem with linear power supplies is that most of them try to
make
> the regulator do too much work, and all they use for filtering is
> capacitance.  If you want to make a "clean" linear power supply, all that
is
> needed is a true "pi" type filter after the rectification using both
> capacitors for Voltage storage and inductors for Current storage (and to
> keep the current and voltage in phase).  The regulator would follow all of
> this and would have a very clean input so it would be able to deal with
the
> rest of the ripple.
>
> Trying to use a switching power supply is probably the wrong approach.  It
> is very difficult to get rid of the RF that is introduced, and the
circuits
> can be downright dangerous to work on (RF has much more transient
energy!!)
>
> Something like this approach is what I'm trying to convey:
>
> >From full wave bridge >-----+---^^^^^^---+---^^^^^^-----> To regulator
>                             |     L1     |     L2
>                             |            |
>                           C1=          C2=
>                             |            |
>                             |            |
>                Ground >-----+------------+------->
>
> L1,L2 are inductors, value is probably not critical.
> C1,C2 are capacitors, I prefer at least 3300mF ea.
>
> OK, So it's not a "pi" because of the extra L2, but I forget the name for
it
> with the extra inductor.  Also the values can be calculated so that the LC
> of the network produces a proper phase correction..but don't ask me what
the
> formula is...I'd have to dig out the reference manuals and they are at
home.
> It usually ends up not being too critical anyway.
>
> Also over design the thing.  It has also been my experience that the noise
> ripple will go up dramatically the closer you get to the tolerance of the
> parts.  For example, if you are trying to draw 1 amp from a 1 amp power
> supply transformer..there will probably be noise no matter how much
> filtering you do.  A good rule of thumb is to design 3 times the value
that
> you plan to draw.  If you are planning on drawing an amp, use a 3 amp
> transformer and the appropriate support components to match.
>
> One final note, use the TO-3 based regulators instead of TO-220 based (SK
> parts instead of the LM78xx). the regulation is much cleaner on them.
> Another idea is to use a zener-to-transistor regulator.  If done right, it
> will regulate better than any of the IC based ones and even has some
ripple
> compensation in the form of a "super capacitor" (the capacitance actually
> gets amplified by the bypass transistor).  If anyone is interested in this
> type of design, request through the group and I'll dig out the schematics
> that I did years ago.
>
> Jay Martin
> ISI/CLEO Technical Support
>
>
> -----Original Message-----
> From: chris [mailto:espace at e-server.net]
> Sent: Friday, June 02, 2000 11:20 AM
> To: synth-diy at node12b53.a2000.nl
> Subject: less hum: switching power supply?
>
>
>
> hello friends on Synth-Diy !
>
>    i'm usually lurking around here, sorry for the sudden surprise
entrance.
>    in a quest to totally eliminate the hum in the audio circuits...
>    i've decided that switching power supply is the way to go to get rid of
> the big 60-cycle EMF.
>    could anyone point me in the direction of a small, lightweight,
> onboard-style switching power module: input 120VAC, output +/-15V @ 500mA,
> +5V @ 1000mA.
>
> christo
>