LM317's on positive AND negative rails....? Power supply design query

[Bunkey]

Hey, it's been a while...


So I'm designing something that makes noise. Amplifies it in fact. And drives a speaker 

I'm working out the power supply and could use a hand in busting my theory. (No I'm not trained).
Values are approximations.


My theoretical amplifier has a few different voltage requirements - which I'll call loads 1,2 & 3.

Load 1 requires a 18v regulated supply
Load 2 requires a 12v regulated supply
Load 3 requires the full 48v supply

I have a few LM317 regulators handy (Input Vmax = 38v)

I have a transformer with 2 secondaries (18v AC each). Which to my knowledge can be bridge rectified into either +/-24v DC rails with a centre tap ground, or connected in series and bridge rectified to form a single +48v DC supply.

Obviously the second approach would exceed the limits of the regulators and therefore can't be used, so +/-24v rails it is.

Can I configure my LM317's as below, treating the common ground as the Load 2 regulator's positive supply, to provide Load 2 with a +12v supply (grounded back to that common ground) - or would this approach mess with the common ground and/or create some sort of catastrophic loop in the space time continuum?



Common sense (and studying my ARCAM A60) would suggest to use an LM337 here instead to supply -12v to Load 2 (polarity doesn't matter in this case) but I don't have a 337 available.
I'm also interested in maintaining some sort of design symmetry, else I'd consider parallelling both regulators on the +ve rail - but aside from being aesthetically picky I'm unsure if this accentuated imbalance of loading between +ve and -ve rails would affect Load 3 and/or the supply smoothing of the reservoir caps C1+101.


I'd be very grateful if someone could throw me a life ring here before I destroy something 
Thank you.

[Bunkey]

...or would this be a better way of doing things?

[ElectricDruid]

The first way perhaps distributes the power across the two secondaries better (although that'll depend on what exactly the loads are). The big problem with it is that the "ground" points for the different parts of the design are at totally different potentials. Of course, you can design around that, but you'd better make damn sure not to forget and connect the ground of one bit to the ground of another bit.

The second way is much safer and easier to understand since everything is referenced to the same voltage. The downside in that case is that the lower of the secondaries is carrying the current drawn by all three loads.

HTH,
Tom

[danfrank]

Where is your -24 volts on the second picture of yours??
You could use the entire secondary and use it as a voltage doubler to get both +/- 48 volts and then use the appropriate regulators to get the +/-24 volts. Use dropping resistors to get the regulator in voltage down to around 40 volts

[Bunkey]

The first way perhaps distributes the power across the two secondaries better (although that'll depend on what exactly the loads are). The big problem with it is that the "ground" points for the different parts of the design are at totally different potentials. Of course, you can design around that, but you'd better make damn sure not to forget and connect the ground of one bit to the ground of another bit.

The second way is much safer and easier to understand since everything is referenced to the same voltage. The downside in that case is that the lower of the secondaries is carrying the current drawn by all three loads.

HTH,
Tom

*Edited*
Thank you.  I imagine this second approach is fine as long as the VA of that lower secondary isn't exceeded?

Where is your -24 volts on the second picture of yours??

There is none. I'm essentially just combining the secondaries in series to produce a rectified +48v rail at the top, then tapping that combined winding at the halfway point to provide a second rectified rail at +24v. Both are referenced to a rectified 0v ground (DC), which you could think of as being at the bottom of the winding as it's drawn here.

Of course, the secondary itself is AC and floating in this case (so the bottom of the winding is still -ve and the top is still +ve in terms of AC) but by anchoring the negative terminal of the DC rectifier to ground, all the potential of that AC swing is going to appear in the positive DC polarity; giving you a +48v and +24v rail, instead of swinging that 48v potential between a +24v and a -24v rail.
The potential is the same but where that potential appears just depends where you anchor the circuit.

[antonis]

My theoretical amplifier has a few different voltage requirements - which I'll call loads 1,2 & 3.
Load 1 requires a 18v regulated supply
Load 2 requires a 12v regulated supply
Load 3 requires the full 48v supply

And what are particular loads current requirements..??

[PRR]

...or would this be a better way of doing things?   https://i.imgur.com/aFt7IBXh.jpg 

I think there is a hidden short in this one.

[iainpunk]

...or would this be a better way of doing things?   https://i.imgur.com/aFt7IBXh.jpg 

I think there is a hidden short in this one.

yes there is actually in both, they can be easily corrected tho,


this configuration always leaves you with one supply only half wave rectified

the other is just a matter of removing the ground on the center tap of the transformer!!!!never mind

cheers, Iain

[Bunkey]

For what it's worth:

Load 1 = 18v / 16mA
Load 2 = 12v / 150mA
Load 3 = 48v / <20mA

...roughly




Looking at this again;

I think, for the first approach, the negative side of load 2 should actually be grounded to the -ve rail, like the LM317 supplying it - else the potential across that load 2 is going to be the sum of the -ve rail minus (or plus, since it's -ve) the regulated 12v output of that 317, instead of being the intended 12v output referenced to the same 'ground' as the 317. This coincidently happens to equal the required 12v in the case of a 24v supply but if the rail were to be slightly off (say -23.5v) then so would be the potential across Load 2 (11.5v).

For this first approach I'm concerned about drawing current from the common ground to supply Load 2 and destabalising the circuit somehow. From what little I've seen, I associate -ve rails with PNP devices and current in NPN devices generally flowing from +ve supply to ground; so whilst this layout makes sense in terms of polarity potential, it's a bit counterintuitive.
There's also the imbalance of current draw from the reservoir caps but I guess the whole pre-amp stage of the A60 I'm referencing is more-or-less built on a +ve supply to ground, ignoring the -ve supply, so a load imbalance between rails can't be too bad in practice..

this configuration always leaves you with one supply only half wave rectified

Why does this happen? I thought the rectifiers would just operate independently of each other at their respective supply voltages.

[Bunkey]

I think I'm just going to have to do the sensible thing and use an LM337 on the -ve rail to avoid any problems/oversights.

The rest of the first approach is pretty textbook from what I can tell.


*Edit* Scrap that, I want to see if this works - its interesting AF.

So, by reconfiguring the -ve side of load 2 as I said and as shown below, this will yield a +12v regulated supply from that Load 2 LM317, wheres an LM337 here would instead ground to common ground (like the first example before my correction) to yield a -12v regulated supply? Is that the only difference?

I'm deducing that current flows from the ground to the -ve rail regardless of whether I use an 'NPN' 317 or a 'PNP' 337; therefore this statement holds true and actually nothing bad is going to happen to the ground point because it's locked tight in to mother earth, yo




...or does the earth ground point not itself act like a current source and therefore requires the 337 configuration...?

I think I need to sleep on this one. It's been a long day, evidently.

[Bunkey]

Good morning,

Configuring an LM317 on the -ve rail like this would essentially turn it into a voltage inverter - The fact I can't find any material referencing its use as such suggests to me that it's not going to work...

Of course, I would try it but I'm poor as shit so I want to be sure this is going to work before I build it.

I can't afford to destroy my regulators.
I definitely can't afford to destroy my transformer!



Anyone?

[antonis]

You can implement two distinct bridge rectifiers for using LM317 dual supply..


For 48V regulated single supply, it should be better to implement something like "Phantom"..

https://sound-au.com/project96.htm

[Bunkey]

Thank you for the suggestion but I feel this is getting a lot more complicated than it needs to be.

I don't actually require a dual supply as such - The idea of configuring the transformer/rectifier for a +ve & -ve rail was just a simple way of limiting the potential across the LM317's to something they could handle. Load 2 is a heater filament so its supply polarity doesn't really matter, I just liked the symmetry of the design at the end of the day.

Gonna go back to the drawing board on this one...


I still can't work out why those rectifiers in the second approach don't operate independently of each other as Iain and Paul pointed out though? What am I missing?

[Rob Strand]

I think I need to sleep on this one. It's been a long day, evidently.

The way you have drawn this should work.  Be aware the 18V load does not share a common 0V with the 12V and 48V.  The 18V load would need to float.

I'm not sure you want to have ground as you have drawn it.  Maybe just wire the tx to the cap and have no ground.

[iainpunk]

For what it's worth:

Load 1 = 18v / 16mA
Load 2 = 12v / 150mA
Load 3 = 48v / <20mA

...roughly

that's quite handy to know, its worth a lot

this configuration always leaves you with one supply only half wave rectified

Why does this happen? I thought the rectifiers would just operate independently of each other at their respective supply voltages.

no. lets use peak voltages for ease of calculations
if you have both rectifiers with a ground connection, when the transformer voltage goes negative, the rectefier on the fop coil makes the voltage on the top most connection -0.7, which is the forward voltage of the diode.
that makes the voltage on the mid point 23.3v peak, and the bottom connection gets to 47.3v, which is way to high.

hope this makes sense

cheers, Iain

[Bunkey]

Hi Rob 

I've actually combined the two concepts into what I think might be the solution for this...

My instinct was telling me the smoothing caps in that first arrangement weren't going to smooth like I'd hoped, as the 48v load is single ended as opposed to being a push-pull type thing built around ground.
Then of course there's the dual rectifier half wave issue on the second approach (thanks Iain, I'll take a look at that just now).

So, applying the best of both:

Instead of having 2 +ve rails at different voltages with their own rectifiers, why not centre ground but look at it like a +24v supply with 0v ground for the regulators (plus associated C2 smoothing cap) and a seperate virtual ground at -24v for the 48v supply, keeping the whole 48v side of things floating with its own smoothing capacitors represented by C1..?

[Bunkey]

C1 = 100uF, accounting for the 48v / 20mA demand of L3.
C2 = 1200uF, accounting for the 24v / 166mA demand of L1 + L2.

I intend for Vripple be close to or less than 5% of the supply voltage for all 3 loads.

Using the formula C = I / (100 x Vripple)


(UK = 50hz 230v AC)

[PRR]

Do you really want the 48V grounded differently from the rest?

If you are so poor, put some thought into other circuits which will not need such an odd-lot assortment of voltages.

[Rob Strand]

One way to keep the full-wave rectifier & regulators and get a 48V rail is to use separate doubler.  It needs an extra cap.



D5, D6 though to +5V is like your existing 12V and 18V regs.

The "48V" part with a common 0V is the part from the 470uF, D7, D8 and the 220uF (ignore the following regulator).

The doubler pulls only current from one of the half-windings, which is OK if the current is low.  The input current is double the output current because of the doubling action.

What you gain is the common ground - if that's important.

IIRC there's a few ways to do doubling which preserves a common 0V like this.   The one shown does not load put a half-wave load on the transformer.

[Bunkey]

Do you really want the 48V grounded differently from the rest?

If you are so poor, put some thought into other circuits which will not need such an odd-lot assortment of voltages.

I really just want it to work in a way that keeps the circuit as simple as possible (ie. as few a part count and something which makes sense to me).
Whether the principle is conventional or not isn't a concern.

As for being poor as shit, that was a facetious comment 
Still bares some truth nonetheless.


I'm designing this power supply around a preconceived idea for an amplifier, therefore the voltages it puts out need to accommodate the larger picture of what it is I'm doing.

I'd like to have kept my cards to my chest to avoid raising further questions (or ridicule) but basically this is a low voltage tube pre-amp into an LM386 op-amp power amplifier, to partner a tiny 1x5 cabinet I built last year.






The pre-amp idea is based on 'starved plate' tube pedals that run on 12v supplies; however I'm trying to make it a little less 'starved' in the hope I can wring a bit more tone out of it. Admittedly it's not going to perform like a HT valve amp but then neither does a 1x5 cabinet perform like it's full-sized conterparts; the idea is a bit of a novelty but there's nothing stopping me from putting effort into that design and making it good in its own right.

So with that said, I'd like to try and make full use of the 48v DC potential this transformer has to supply the plate.
The alternative is just to keep the secondaries seperate and supply the plate with 24v instead  - Still twice what similar designs (like the Matsumin Valvecaster for example) use; but falling short of what I feel could be achieved with a little out-of-the-box thinking.

One way to keep the full-wave rectifier & regulators and get a 48V rail is to use separate doubler.

Voltage doublers aren't something I've come across before but I'm guessing it means double the current draw too?

With adding complexity I'm concerned about making the supply noisy but I will look into it for sure, thank you.