"linear"power supply "roundtable"

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Brian Roth

Well-known member
Joined
Aug 20, 2005
Messages
3,184
Location
Salina Kansas
Many of us prefer a "linear" power supply versus a "switcher" for audio applications, although the latter do offer some advantages, with newer designs running at frequencies well-outside the audio passband. However, that should be another thread here.

I'm a relatively newbie in this forum, but I did briefly skim the "power supply Meta" listings, but nothing jumped off the screen at me like what I want to discuss.

I have approx. 35 years of experience with linear supplies as used in pro audio gear, which would be a bazillion circuits designed by others, and a few jillion <g> power supplies that I had to concoct for my own designs.

Perhaps this has already been talked to death in other threads...tell me where, and I'll either add comments, or just shut up <g>.

In the meantime, I submit the following link. It's a scan from my 1982 National Semiconductor "Voltage Regulator Handbook", and discusses the CORE elements...power transformer, rectifiers, and "reservoir" capacitor.

(Sidebar: I found NO copyright info ANYWHERE in my paper copy ( a 200+ page book) which originally was distributed by National in 1982...I guess they were eager to share their knowledge. Hence, I don't see any reason why I can't share the document).

http://www.brianroth.com/library/national-ps-design.pdf

A couple of off-list discussions (here and elsewhere) tends to make me believe there is a lack of "the fundamentals" on this topic.

I would love to expand this discussion, and starting with the AC mains, then moving onwards.

These are a FEW topics that I *believe* that I know, and would like to discuss in depth:

I. AC Mains
A. Branch circuits
B. Proper wiring, switching, and fusing
C. Compensating for AC mains "sags"

II. Mains to power supply transformer
A. Iron core vs. torroids
B. Temperature rise in the tranny
C. Transformer regulation vs. load current

III. Rectifiers
A. Various topologies (half/fullwave, bridge)
B. Diode selection
1. PIV ratings
2. "Fast"/"soft" transfer curves
3. Forward drop, especially in low VDC supplies
C. "Specialty cases"
1. Voltage doublers/triplers/etc
2. High Voltage (tooob!!) supplies
a. Vaccum tube rectifiers for HV applications vs. solid state

IV. The "Reservoir" capacitor(s)
A. Avoiding "drop out sag" when feeding a downstream regulator
B. Best choices for component selection
1. ESR and other specs
2. Temperature ratings (sidebar: I automatically select 105 C parts)


Maybe I can "get this party started" <g> and we all end up with some wisdom.


Bri
 
Very good idea Brian,

An aspect we in DIY circles tend to sometimes overlook is compliance with electrical safety regulations. I think we should include this in the discussion and as a result come up with some guidelines.
 
Hi Brian,

I'd love to pick your brain on this stuff. Here's a few questions to start off:

What's the safest way to wire an A.C. mains switch to the front panel that will meet CE and other safety standards? What sort of shielded wire and voltage rating?

Is 10% a reasonable amount of margin for input A.C. sag?

Is it better to series connect regulators to drop voltages, or is it better to parallel off the raw D.C.?

Thanks,

Paul
 
AFAIK, the fuse and power switch in USA gear should be in series with the "black" wire...which translates to the "brown" wire in IEC.


Bri
 
Brian

I have worked on all kinds of power supplies 5V to 3 phase 470AC to 690DC for a 100 HP AC drive to gas laser power supplys TWT supplies.

There is a lot of ground to cover.

Here is a something that might help the lab.

You have a linear supply lamda, power one power mate condor etc. 723IC and NPN 3055 transistor types. The DC looks good on a meter but the system is not working correctly CPU resets, counts wrong, noise etc......
You use a scope or a DMM with the bar graph under the numbers ( it has a good use). You find alot of ripple.

Now your basic 732 Power suppply uses one or more NPN series pass devices what the designers often do is add a doubler (1 diode and a cap) to raise the base drive to the NPN pass section the collector is connect to the raw non doubled DC. I believe this is done because a diode and cap (and maybe some other small parts) is cheaper than a PNP pass that would not need the raised current drive the NPN needs in this type circuit. For the higher current ones often a NPN drives a bank of NPNs in a darlington.

What often happens is the cap in the X2 section goes bad often you change that cap and you fixed the supply If you have an actice load bank you can realy give it a test: however knowing basic Ohms law helps take the output voltage / rated current for the load R needed. After finding the R need calculate the power rating needed
power in watts(dc) = output voltage X current or current x current x R

A cool thing is a active load bank that can step the current draw this lets you watch the step response of the supply.

Brian is this the kind of stuff you are thinking about?
 
Iron core vs. torroids

Amp story I have posted this a few times. I have a laney vc30 amp that I changed the power supply caps to polypros of 1/2 the cap value of the stock Al ones.

The amp had a bad mechanical vibration and hum. I thought about it for some time and remembered the ESR of the polypros measure close to 0 ohms on a LC102(IIRC the stock 100uf Al can measured 1 to 2 ohms ESR at the output plate side of the power supply) I reasoned the diodes had high surge currents and the mismatch of the diodes turn on and off and conductace caused different current surges in the diode legs this caused a DC componet in the torroid power transformer I reasoned.


I added a 1 ohm power resistor between the diodes and first power supply cap as the "fix"

Two things here
Low ESR caps at the first stage of the power supply need to be thought about. A first stage very low ESR cap might be very bad with a tube diode circuit. FWIW the laney has Si diodes

Torroid transformers don't seem to work well with a DC componet

My Hafler DH200 has 30,000uf (from the late 70's early 80's)a side and no transformer mechanical hum to care about it has an EI core.

Brian I have that NS book somewhere(green cover?) It is a very good read and full of good information like you posted



Maybe some thing else to post about like zero crossing switching of power supplies.

edit changed diode to resistor
 
[quote author="Gus"]Iron core vs. torroids

Amp story I have posted this a few times. I have a laney vc30 amp that I changed the power supply caps to polypros of 1/2 the cap value of the stock Al ones.

The amp had a bad mechanical vibration and hum. (snip)


I added a 1 ohm power diode between the diodes and first power supply cap as the "fix"

Two things here
Low ESR caps at the first stage of the power supply need to be thought about. A first stage very low ESR cap might be very bad with a tube diode circuit. FWIW the laney has Si diodes

Torroid transformers don't seem to work well with a DC componet (snip).[/quote]

I guess you meant 1 ohm power resistor.

I've had similar disappointments with power toroids as far as the much-vaunted confinement of magnetic fields compared to E-I cores. Even with relatively conservative ratings I would see nasty pulsed fields that got into senstive loops in the circuit---in one case the even the output coupling caps were an important loop!

Toroids with simple tape-wound cores saturate pretty abruptly. Also, higher harmonics associated with the diode current waveforms may reach into the region of falling permeability of the core and that mag field component will not be confined as well.

I think your issue was less with a d.c. component and more with simply the charge current waveform---I don't think it would have gone away with perfectly matched diodes. The addition of resistance as described is what I ended up doing as well, and it helps a lot---but of course you have to account for the loss by specifying a somewhat higher voltage trafo.

I explored some active methods of doing the current limiting just for fun, and it gets complex. Then there is always "nature's way", namely a choke input, but one needs to be mindful of many issues there. And the choke adds cost, weight, takes up space, and if not toroidal itself radiates plenty.
 
[quote author="Brian Roth"]Yes, IMHO, the first "stop" for the AC mains cable MUST be the fuseholder. Anything else is folly....[/quote]
... or just use IEC Mains Inlets with integrated fuse holders ...
works best for me.
 
Ah yes a good thread!

Through the years I have found a few tricks:

Use a larger(current) than needed trafo.
use power Schottkys for rectification.
plenty of capacitance and bypassing.
mains filtering on inputs.
large gauge wiring throughout.
zener-follower discrete Vregs( a recent one.. but definately a good one!)
always shield the PSU from audio sections.

and last but certainly not least.. remembering that no matter how good your supply is, it still makes noise of some kind.

:thumb:
 
[quote author="Svart"]Ah yes a good thread!

Through the years I have found a few tricks:

Use a larger(current) than needed trafo.
use power Schottkys for rectification.
plenty of capacitance and bypassing.
mains filtering on inputs.
large gauge wiring throughout.
zener-follower discrete Vregs( a recent one.. but definately a good one!)
always shield the PSU from audio sections.

and last but certainly not least.. remembering that no matter how good your supply is, it still makes noise of some kind.

:thumb:[/quote]

Have you had a chance to play with the SiC schottkies yet? I'm about to for some tube stuff. They have too much voltage drop for low voltage supplies.
 
You know, I haven't had the chance yet since my partner decided that he disliked our boss more than he liked the job and left! :shock: I keep meaning to source some and get them in for testing but never have the chance to find them between interviewing and trying to keep up with the workload! Even my DIY is suffering.. this is not good.. :?

Have you sourced any yet? Where did you get them? preliminary testing or are you waiting to real world test them in something?


I'm really interested in these, just haven't had time.

EDIT: while the boss wasn't looking I took the liberty of trying to source some parts.. seems that Future electronics has some stock of the Infineon brand of SiC Schottky.. but the website wants you to buy them by the tube..
 
[quote author="Svart"]You know, I haven't had the chance yet since my partner decided that he disliked our boss more than he liked the job and left! :shock: I keep meaning to source some and get them in for testing but never have the chance to find them between interviewing and trying to keep up with the workload! Even my DIY is suffering.. this is not good.. :?

Have you sourced any yet? Where did you get them? preliminary testing or are you waiting to real world test them in something?


I'm really interested in these, just haven't had time.

EDIT: while the boss wasn't looking I took the liberty of trying to source some parts.. seems that Future electronics has some stock of the Infineon brand of SiC Schottky.. but the website wants you to buy them by the tube..[/quote]

I haven't bought any yet. Infine*n does seems to be the major supplier. They recently introduced some smaller ones with correspondingly lower capacitance which look interesting.

Since you probably have some decent apps at your work I suspect you could persuade the rep to get some samples.

They do seem to be by all accounts essentially perfect---while all standard schottkies have a little region of P-N that screws up the recovery behavior, according to Pease and others.
 
oh wow, damn I must have an out of date paper catalog because they aren't in there! I also searched very quickly online via part numbers but they didn't turn up... Guess I overlooked them :oops:

(edit) I see that the Infine*n part number called out on their webpage is different than the ones Digik*y has for Infine*n.. strange

I think I can talk the bosses into 10.. I've already put the bug in their ears about switching noise problems that we see with standard silicon..

Awesome.

I did a little more looking and a lot of companies have announced their SiC lineups.. but only about 2 or 3 have actually produced any..

Micr*semi has a nice announcement and even a link in their website that leads... to an empty page.

Vaporware for the EE world.



:mad:
 
[quote author="VacuumVoodoo"]And to be on the safe side: double fused double switched with an X1 cap across the mains.[/quote]

If you mean a fuse in the "neutral" line (USA usage), that is NOT permitted. Only the "hot" should be fused.

Bri
 
[quote author="Brian Roth"][quote author="VacuumVoodoo"]And to be on the safe side: double fused double switched with an X1 cap across the mains.[/quote]

If you mean a fuse in the "neutral" line (USA usage), that is NOT permitted. Only the "hot" should be fused.

Bri[/quote]

It's not IMO a bad idea to switch both sides of the line though, in case you are servicing something and the hot and neutral in a given building happen to be swapped.

Somewhere a while back I told a war story about a powered speaker that found its way to Keith Johnson, and had apparently a bracket misinstalled that managed to short neutral a.c. in to signal ground (Keith destroyed the evidence getting the thing apart). Just happened that Keith had a special balanced power system where it was not 0 - 120 but 60V either side of safety ground potential. It promptly took out his fancy D/A before blowing breakers. Keith concluded that we had a.c. neutral tied to signal ground by design :shock: and was ready to insist on a massive recall of product :evil:

Of course were neutral tied to signal common it would have been a very dangerous system indeed, given the incidence of miswired homes.

Needless to say, although we could never quite reconstruct the fault condition, we promptly implemented a new inspection and assembly procedure.
 
Back
Top