Linear power supply - two windings or linear voltage regulator

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,

I am inexperienced and have found myself questioning the need for multiple linear power supplies, or a single power supply with multiple output rails.

If I needed 12V 10A to power a computer motherboard for a music server, then if you wanted add in an additional PCe 5V 2A Audio card. Provided it's still within the capacity of the linear power supply, then it would be much cheaper to use a linear voltage regulator to drop the 12V down to 5V... such as DIY KIT LT1083 High Power Linear Variable Regulated DC Power Supply Board Kit-in Amplifier from Consumer Electronics on AliExpress - 11.11_Double 11_Singles' Day

Does this work well? Why do many solutions involve the use of multiple separate linear power supplies, with separate windings, different rails and so forth? Other than the heat generation, is a linear voltage regulator a very cost effective solution?

Thanks
 
There are many diy psu based on simple computer psu, it delivers your mentioned voltages, plenty of current and clean as well.
 

Attachments

  • ATX_Bench_1.jpg
    ATX_Bench_1.jpg
    219.9 KB · Views: 399
Thanks amplidude - I have tried SMPS including those with low ripple (e.g. superflower leadex). But they are not a patch on the Paul Hynes linear power supply I have.

I am therefore posing the question of if you have a linear power supply which is able to provide sufficent current then is a linear voltage regulator from the output of your existing linear power supply a much cheaper way (and better?) way of powering smaller 5V components, rather than buying / building another linear power supply?

Thanks
 
Cost-effective? Yes. Noise-free? Not particularly. Will it matter for a PCe audio card? Dunno.

If it's a linear voltage regulator, rather than a simple buck converter, and it is supplied with very lower ripple to start with (so noise rejection doesn't matter so much) then the theory I'm testing would be that it is a good solution - but I have no experience with such matters... need an expert to validate my thinking.
 
AX tech editor
Joined 2002
Paid Member
Hi,

I am inexperienced and have found myself questioning the need for multiple linear power supplies, or a single power supply with multiple output rails.

If I needed 12V 10A to power a computer motherboard for a music server, then if you wanted add in an additional PCe 5V 2A Audio card. Provided it's still within the capacity of the linear power supply, then it would be much cheaper to use a linear voltage regulator to drop the 12V down to 5V... such as DIY KIT LT1083 High Power Linear Variable Regulated DC Power Supply Board Kit-in Amplifier from Consumer Electronics on AliExpress - 11.11_Double 11_Singles' Day

Does this work well? Why do many solutions involve the use of multiple separate linear power supplies, with separate windings, different rails and so forth? Other than the heat generation, is a linear voltage regulator a very cost effective solution?

Thanks

It would work, but dropping 7V at 2A means 14W heat. Where are you going to leave that?
A separate 6V supply followed by a 5V regulator will decimate that heat.

Jan
 
Does this work well? Why do many solutions involve the use of multiple separate linear power supplies, with separate windings, different rails and so forth? Other than the heat generation, is a linear voltage regulator a very cost effective solution?

The board you linked to has rectifier diodes on it. If you wanted to power the unit from an existing DC power source, then you would probably want to bypass the rectifiers and connect to the input filter caps.

Regarding what you suggest doing, to a first approximation the only downside is production of excess heat as has already been discussed.

Some addition thoughts:

Having separate transformer windings allows grounds for each power supply to be independently defined. Sometimes that can help reduce ground noise problems.

Also, when an additional load (in this case a the regulator you are considering) is added to an existing audio power supply, the new load can modulate the voltage on the existing power supply a little as more or less current is drawn. Some audio circuits may have some audible sensitivity to modulation of the power supply voltage, so its possible the new load could cause a reduction in sound quality that way.

Then again, for your particular use case it could be that production of some heat would be the only practical downside.
 
Thank you all, very helpful. 14w of heat could be dealt with. If I had a Streacom FC9 fanless case and a low wattage CPU then the heatsinks on the side of this case would have enough headroom to be able to deal with 14watts from a linear voltage regulator - I'd just take a feed the voltage regulator from the 12v DC in and attach it to the common ground.

Shunt regulator seems like a good (but inefficient method).

Is there such as thing as a DC/DC toroidal transformer? 12v primary to 5v secondary just to convert DC? No need for rectifiers etc... as only a step down in current is required?
 
If you need clean efficient power and have a lot of volts to drop, then the combination of buck converter into a linear regulator has a lot of advantages, more efficient than just a linear regulator, much cleaner output than just a buck converter. You can also arrange the linear regulator at the point-of-load to reduce IR wiring losses, which is facilitated by the smaller heatsink it needs if fed from the minimum input voltage.


For currents upto about 2A there are loads of cheap LM2596 modules available for the buck converter stage too, so can be very cost effective - not that these modules represent state-of-the-art, but in this configuration that's not so important.
 
...You can also arrange the linear regulator at the point-of-load to reduce IR wiring losses, which is facilitated by the smaller heatsink it needs if fed from the minimum input voltage...

True. Although might we worth checking the linear regulator data sheet to see if PSRR or other parameters vary with regulator voltage drop. If so, might be reasonable tolerating a bit more heat in return for somewhat improved regulation.
 
I'm going to use a 78S05 to build a simple, 5V 2A power supply for a pi
There are lots of other three pin, fixed 5 v regs out there.

I have the TO220 package, and will attach to a good heatsink, but I'm using much less than 12V, so I have a lot less power to dissipate.

If you're going to mount to a streacom case, I hope you plan to mount whatever voltage reg you pick directly to the case heatsink. I'd drill and tap a hole, but a hole with a nut and bolt works too. You're likely have to add an insulator between the voltage reg and heatsink.

Randy
 
Hi Randy - yep exactly my plan. I haven’t found much out about the series voltage regulator though, I don’t know what products would class as this.

If a 78S05 was used one thing that has crossed my mind is it’s robustness and what if it failed - if I had a £400 USB card being powered from it that would not be good. Any ideas how safe the 78S05 would be? Maybe this is why manufacturers go down the route of additional rails as it’s safer and provides better isolation / ground?
 
Disabled Account
Joined 2002
Linear can be done quite good with a good uLDO and the rightly chosen transformer. You can have the power, low noise and low heat when designing right. I finished a 5V transformer to 5V 3A DC low noise low heat linear uLDO PSU last week and it does not even become lukewarm when supplying an average of 1A with 1.5A peaks to the load.

Many are still stuck in LM317/337 and 78xx/79xx blues. All these are very noisy compared to modern regs and they need a minimum 3V drop out voltage which ends up in heat. Most combine this with a randomly chosen transformer that has a too high secondary voltage. In the past no one cared as people were accustomed to devices becoming warm/hot and the regulator just dissipated the excess voltage to useless heat. It worked right didn't it?

Please move on to the way better LDO regs that can be found nowadays and marry it to a rightly chosen Rcore or toroid transformer. They are way more silent than almost any switcher. The switchers that are silent are the ones followed by a low noise uLDO :)

Point is that the word "cheap" does not go together with set goals. It also costs time to calculate things right for the application.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.