Linear regulated power supply design

[gareth33]

Hi all

I am in the process of learning about linear regulated power supplies. My goal is to design a high performance, low noise and ripple, 40 VDC power supply capable of supplying 1.5a of current to the load.

The PSU is to power a clone of the Bozak CMA 2DL that I am building. I am learning as I go, and reading every available resource on power supply design. I have put together a basic and preliminary schematic to start refining. I am considering using the LM117 HV (or LM317HV) as regulator, but am keen to see where I can improve on this circuit, whether through zener reference diodes, or possibly a discrete design, (series pass transistor design?)

I've started reading about LC Pi filters. Is it possible to strap inductors across the 4 filter caps? Is there an improvement in performance to be gained and is this even the way to do it?

You'll also see a random inductor preceding the rectifier bridge, I am thinking about including an AC line filter before the primaries of toroidal transformer (see below attachment).

I am not too concerned with cost, my goal is to make the highest performance PSU I can,  so I am looking for ways to improve this circuit, I have searched a lot on Google and the information is great, but I don't find many schematics beyond the simplest ones I find on Texas Instruments data sheets etc.   

 

[gareth33]

Possible AC line filter to go ahead of the transformer primaries.

 

[Martin Griffith]

470R, whats the voltage drop at 1.5A ? Putting 100nF across each diode would be recommended (a tip from Win Hill). Whats the pre LM317 Voltage

 

[gareth33]

Thanks for the tip re diode capacitors, I'd read this somewhere as well, but was unsure. 100nf ceramic caps, placed in parallel with each diode?

I am still defining all the specs, and what toroidal I will need, I have a rough idea, and how things are being affected by component choice, this is my next step.

The load circuit will only draw 1a maximum I believe, as part of my over spec-ing I figure providing 1.5a is a good safety margin should there be unexpected current draw. 

 

[Gus]

do a search for
"Linear regulated power supply design " National semiconductor
Texas instruments
Motorola
ST
etc.

You should find some of the older databooks and app notes.

You should find some good information about what you need

 

[PRR]

> 1.5a of current ... Bozak CMA 2DL

Seems like gross overkill. I bet 1/4A is ample.

 

[rackmonkey]

If you haven’t, check out PSU Designer II from Duncan amps. Handy app to have, although it’s limited to simple designs and isn’t a do-it-all tool.  Won’t help you with the regulated part, but you can check transformer current, cap charge/discharge, current vs load, ripple, etc.

http://www.duncanamps.com/psud2/index.html

 

[JohnRoberts]

> 1.5a of current ... Bozak CMA 2DL

Seems like gross overkill. I bet 1/4A is ample.

IIRC the old 10-2DL was all discrete so probably more current draw that modern op amp designs but agreed 1.5A seems overly generous.

How good do want the PS? Back last century when I was still designing phono preamps I rolled my own regulated PS using a TL072 op amp with some pass transistors.  http://www.johnhroberts.com/P100_PS.jpg  This was way way overkill for back then, and now.  :

The old National Semi applications handbooks had a lot of good notes on PS design.

Take care to pay attention to ground current flows.

JR

 

[gareth33]

Thanks for the replies, everyone.

> 1.5a of current ... Bozak CMA 2DL

Seems like gross overkill. I bet 1/4A is ample.

Yes, the CMA 2DL is class A discrete design, I think 1 a is the standard for these power supplies, I went and added 50% margin, which is surely overkill. Once I have had a closer look at actual current requirements of the unit, I will likely drop the PS to 1 amp, which will simplify things somewhat.

How good do want the PS? Back last century when I was still designing phono preamps I rolled my own regulated PS using a TL072 op amp with some pass transistors.  http://www.johnhroberts.com/P100_PS.jpg  This was way way overkill for back then, and now.  :

This is great, thank you, I would like to breadboard this at some point, though calculating the values of everything may be tricky and take me some time. What are IC 101, IC 2, IC 1? IC 3a and 3b is the TL 072? The input is AC?

Sticking to an IC based regulator may be a better choice for this project, though I would like to achieve highest possible performance. Grounding I am still looking into, I have a fairly good idea of how this will go.

Is the inclusion of inductors (chokes) a good idea? C - L - C - L - C? At the position of the 4700uF filter caps? Even though the LM317 has overload protection I may try to include some added short circuit and overload protection to the circuit.

Gareth

--

 

[JohnRoberts]

Thanks for the replies, everyone.

Yes, the CMA 2DL is class A discrete design, I think 1 a is the standard for these power supplies, I went and added 50% margin, which is surely overkill. Once I have had a closer look at actual current requirements of the unit, I will likely drop the PS to 1 amp, which will simplify things somewhat.

This is great, thank you, I would like to breadboard this at some point, though calculating the values of everything may be tricky and take me some time. What are IC 101, IC 2, IC 1?

This PS was for a tweaky phono preamp so IC101,1,2, were the rest of the preamp active electronics, while it had a discrete front end.

IC 3a and 3b is the TL 072?

yes... cheap, 3uV input noise, with decent gain bandwidth...(and several decades ago).

The input is AC?

yes an external wall wart, to keep hum fields outside the preamp chassis.

Sticking to an IC based regulator may be a better choice for this project, though I would like to achieve highest possible performance.

+1 a modern high performance regulator will probably be more than adequate. It might be worth resolving how much current you actually need, as that could limit your choices. My PS design was admittedly overkill, but that was my intent.  8)

Grounding I am still looking into, I have a fairly good idea of how this will go.

currents from charging reservoir capacitors need to be managed so they don't corrupt the audio ground.

Is the inclusion of inductors (chokes) a good idea? C - L - C - L - C? At the position of the 4700uF filter caps? Even though the LM317 has overload protection I may try to include some added short circuit and overload protection to the circuit.

Gareth

--

I have never use LC filters in PS so have no hands-on advice to share.

JR

 

[PRR]

> IIRC the old 10-2DL was all discrete so probably more current draw that modern op amp

Proposed 40V 1.5A is 60 Watts.

Most of the Bozak (I have hacked one) is jellybean transistors and resistors, that can stand at-most 0.3 Watts each.

There are not 200 transistors in that box. Barely 200 resistors.

3/4 will be run much-much less than 0.3W.

Also I bet the thing has huge PSRR built-in, and can eat crap without passing it to the audio.

When polishing a turd (or even a good shoe), don't start with an elephant turd/shoe when you only need a paperweight.

 

[JohnRoberts]

I wouldn't make bets about PSRR of a class A design, but 'class A' is just another way of saying discrete single supply design that generally has poor PSRR wrt local rail voltage.  :

You can always filter the output of a regulated supply if more is needed, and I repeat my caution about managing PS reservoir cap charging current in grounds.

JR

PS: I knew one of the engineers responsible for keeping those production lines running in the 70s, but original design was before my time consulting with Bozak, so I had zero involvement or reason to look under the hood. I recall that production engineer sharing a story with me about a component problem related to device HFE (IIRC?), the component was within its published spec, but the design wasn't happy (classic manufacturing engineering 101). 

 

[PRR]

> generally has poor PSRR wrt local rail voltage.

Thanks for reminding us.

The original Bozak engineers mostly were not Bozos, and were aware of this. Feeling peckish, I got the manuals.

The total consumption is given as 20 Watts, including PT regulator and pilot-lamp.

The individual modules have added R-C filtering where required. The power stages are bootstrapped/etc and have decent PSRR inherent.

The part which Gareth thinks needs a make-over looks like a fine audio voltage regulator, albeit with strong 1970 flavor. If it has failed, just fix it. I also note a 1970ish current-limit, D7 D8 Q3 R1 R3, which comes in at 0.6 Amps (some shy of proposed 1.5A). Going through the modules and module-list I get under 0.21 Amps @ 40V idle, 0.27A full test tone, total 8-11W, which is a fair fit with 20W total.

 

[Rob Flinn]

On the other hand if your LM317 is dropping a lot of volts across it then its 1.5A capability will be somewhat diminished.  Look at the data sheet.

 

[Tubetec]

Always wondered what a 'Bozak' was ,
Its refered to in 'Sure Shot' by the Beastie Boys ,'Too all the party people that are on my Bozak.....'
it all makes sense now .The rotary fader dj mixer was very much favoured by the New York Style HipHoppers back in the day ,I guess they just proved more reliable than linear when busting out the tunes on the street corner or in the park.

 

[JohnRoberts]

Always wondered what a 'Bozak' was ,
Its refered to in 'Sure Shot' by the Beastie Boys ,'Too all the party people that are on my Bozak.....'
it all makes sense now .The rotary fader dj mixer was very much favoured by the New York Style HipHoppers back in the day ,I guess they just proved more reliable than linear when busting out the tunes on the street corner or in the park.

Yup the 10 2DL with rotary faders was an iconic favorite of old school DJs. I suspect it was an accidental success, in a sparse product category.

I knew Rudy Bozak and really doubt he spent much (any?) time in clubs.  In fact he was more of a speaker maker than electronics guy. I designed some consumer electronics for him in the 70s as a outside consultant, but Bozak also had a significant commercial/install business in the past.

JR

 

[Tubetec]

Hmmm interesting ,
The hifi shop I worked at in the early 2000's took advantage of the reemergence of the Dj scene around that time ,initially it was just a few big old reliables like Technics ,Pioneer,Vestax ,but as time went on the market became flooded by newer players in the market ,well in one sense, old names with new product lines ,Stanton, Newmark and Behringer. For a while it was like the goose that layed the golden egg ,but that bubble burst too when all the bedroom DJ's realised that aside from playing at parties for their friends ,there really wasnt much money to be made from it . The smart guys had always dabbled in music production ,where a remix of a remix of a pinched sample could see you hit gold on the underground club scene,by the time the pre-cursor chemicals for MDMA (extascy) stopped being produced in quantity that was the end of the end of the good old days of rave and club culture here . Didnt take long for the Colombian marching power to fill the void and that turned all the happy huggy lovey dovies into greedy monsters.

 

[scott2000]

Hmmm interesting ,
The hifi shop I worked at in the early 2000's took advantage of the reemergence of the Dj scene around that time ,initially it was just a few big old reliables like Technics ,Pioneer,Vestax ,but as time went on the market became flooded by newer players in the market ,well in one sense, old names with new product lines ,Stanton, Newmark and Behringer. For a while it was like the goose that layed the golden egg ,but that bubble burst too when all the bedroom DJ's realised that aside from playing at parties for their friends ,there really wasnt much money to be made from it . The smart guys had always dabbled in music production ,where a remix of a remix of a pinched sample could see you hit gold on the underground club scene,by the time the pre-cursor chemicals for MDMA (extascy) stopped being produced in quantity that was the end of the end of the good old days of rave and club culture here . Didnt take long for the Colombian marching power to fill the void and that turned all the happy huggy lovey dovies into greedy monsters.

LOL..... You may have lost a few on that one..... it's an old school observation .........but not that old.......

edit-- and that last part happened well before the new millennia here.... it created the push for the anti-rave laws where I'm at..

that and the biggest church and the Christian coalition headquarters in sight of several late night venues....lol

 

[Gold]

It was mostly Urei and Bozak mixers in the NYC clubs in the 90’s. I worked in  the Gatien clubs (palladium, Limelight and Tunnell) and  helped maintain the  audio systems.

 

[PRR]

I knew that mixer (maybe a diff model with more mic less phono) as a film-mixer, late 1970s. There was not much around that was portable.