High Current (500mA) 48V supply?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Joined
Apr 24, 2007
Messages
17
Location
Leeds, UK
Hi, first post!

I recently built an 8 channel mic pre in which I'm using a 48v supply for phantom power and to power a relay for phase reverse on each channel, plus LED indicators for each function. The total draw with everything turned on should be 440mA. At the moment the 48V supply is sagging to around 20-25V with everything switched on. It's based on a pair of 7824s stacked up, run off an 18-0-18 transformer with the secondaries in series which gives me 56V rectified/unregulated (I tried a 24-0-24 but it came out at 84V, the regulators couldn't take it). I'm guessing this probably isn't a great design, although it's worked in the past at much lower currents with no problems.

Everything still works but I want it rock solid so at some point over the next month I'm going to replace the PSU board completely with a new design. Does anyone have any suggestions for a layout that will handle that kind of current with no sag? The 317 is obviously a popular choice but I'm interested in the TL783 since that won't involve floating the reference... Any suggestions are welcome! Also any advice regarding heatsinking and so forth.

I found this thread which seems to be a similar problem, yet that was solved by using a meatier transformer, and I'm pretty sure mine should be ok.

Thanks!

Daniel.
 
Hi Daniel,
I built a 48v supply recently to the same sort of spec as you need.
I ended up using a LM317 as I didn’t have a TL783.
Although from what I’ve read the TL783 is preferable. Something to do with it being short circuit protected.

I based my design around these examples:-
http://members.nuvox.net/~zt.robgrow/circuits/powersupply/phantompower.html

http://web.telia.com/~u31617586/Phantom PSU.jpg

I've a messy PCB layout if you'd like it, but it was a quick botch so u would defiantly be better off reading up and coming up with your own solution.


Farnell do TL783's by way.

Good luck
Paul
 
I have an off-the-shelf NOS power one linear, open frame supply that will do 2A @ 48 volts and has a sense input for remote racking, that I got on evilbay for 35$ a couple years ago...I think they are 60-70$ new.? that might be worth looking into
 
When you say you're getting 56V on the raw supply...is that with full load or without? A 36V transformer ought to be giving you about 50.9V Vpk output, ignoring diode drops, which suggests that your 56V is coming from the transformer being essentially unloaded.

Similarly, a 48V transformer (two 24V secondaries in series) ought to get you about 67.9V, ignoring the diode drops, not 84V. That would actually be a reasonable thing to do, as it gives you some leeway for drops in the line voltage.

My guess is that you've been measuring your unreg. voltages with no loads on them; when the load hits, they're sagging.

A TL783 regulator might be a good choice; two of them (one for the phantom, one for everything else) might be even better. If the one for everything else is expected to handle 400mA, then the dropout voltage will be 11 or 12V, depending on junction temperature. So you'll need an unregulated voltage of about 60V; since a 48V transformer gives you about 65.3V, loaded and taking diode drops into account, you have leeway for about 8.2% line voltage drop, maybe a little more if you keep the TL783 cool using a hefty heatsink. Under worst-case +10% high-line conditions, your unreg voltage will be about 72V; the everything-else regulator will be dropping about 24V @ .4A, or 9.6W. Heatsink accordingly.

So you should be okay with a 48V (2 x 24V) 1A transformer.

Oh, use a load resistor on the phantom regulator to keep at least 15mA flowing.

Peace,
Paul
 
Again I can warmly recommend my Excel spreadsheet for PSU calculations--see www.groupdiy.com/index.php?topic=21307.

After 20 seconds I got the following suggestions from it:
* 46 VAC transformer (so your 48 VAC should work fine)
* 4700 uF/100 V smoothing capacitor(s)
* 12 W maximum regulator dissipation (for 48 VAC transformer)

I do second Paul's suggestion to split the phantom power from relays and LEDs, though this is not super critical I'd say as you have a hefty RC filter for every channel.

In case it is not clear how to implement the TL783 properly have a look at the 48 V section of this: standard_linear_psu_r1.pdf--nothing more than what's in the datasheet though.

Oh, use a load resistor on the phantom regulator to keep at least 15 mA flowing.
I've found the TL783 to be quite happy with just the 8.6 mA from the 150 ohm + 5.6 kohm divider.

Samuel
 
Do any supplies on the market use a cap choke cap then a prereg then the reg?
 
Thanks for the replies, I think I've got some good stuff to go on now. The 56V was measured with the transformer hooked up to the rest of the PSU, but with no current drawn. I think I'll go back to my 24-0-24 transformer and do a new circuit using the 783, an enormous reservoir cap and a ton of heatsinking.

Unfortunately I can't really split the phantom off from the LEDs, since the whole lot is switched using an illuminated SPNO. Might think about a seperate supply for the relays (and their associated LEDs) though.

On the plus side it's actually working for the time being, just with slightly reduced phantom voltage, depending on how many mics I'm using. Tracked some drums last week though and it all sounded really good so I just need to sort the PSU to my satisfaction and then get onto another project.

Thanks again!

Daniel.
 
An enormous reservoir cap and a ton of heatsinking.
Just to avoid wrong conclusions: more reservoir capacity is not in any case better (despite numerous conflicting utterances you'll find on the internet). With too much capacity transformer efficiency is reduced and perhaps transformer stray field boosted (not sure about that one--left for future research). Going for 1 V ripple is a good compromise--this leads to the 4700 uF figure I posted.

Samuel
 
[quote author="Samuel Groner"]
Oh, use a load resistor on the phantom regulator to keep at least 15 mA flowing.
I've found the TL783 to be quite happy with just the 8.6 mA from the 150 ohm + 5.6 kohm divider.[/quote]

Okay. Even so, a 5.6k set resistor means you're dissipating about 0.4W in the resistor. That means it should be at least 1W for reliability, which in turn means you probably can't get a 1% resistor. So if you want precision you need to select from a bunch of 5% resistors. I prefer to use higher-value lower-wattage 1% resistors for the divider circuit and a separate load resistor, which can be 5% or 10%, no sweat.

Peace,
Paul
 
That means it should be at least 1 W for reliability, which in turn means you probably can't get a 1% resistor.
I use 0.6 W/1% which seems to work well (the rating is for 70 °C ambient temperature, so we're pretty far from exeeding any temperature limits). Agree though that sharing the dissipation won't hurt and I might adopt your suggestion for higher divider impedance in case I do another revision of my standard PSU board.

Don't have much trouble to get 1% parts even for 3 W in Switzerland..?

Samuel
 
Running relays and LEDs from a 48V supply is madness to me...
screwy.gif


Basically making the load variation colossal. 8 channels of mic power should typically draw in the region of 40ma. If adding the relays etc makes the draw 400mA, you've taken the rail of the supply which gives people the most problems and should arguably be the cleanest) and made its job 1,000% more difficult. -The LEDs will be converting 2V into (wanted) light, resistors feeding them will be turning 46V into (unwanted) heat. That's a 96% inefficiency...

Using a 2-pole switch and running the LEDs from a lower rail via the other pole is a better idea, but there are others... constant-current source was used by Harrison and Neve, and becomes VASTLY more efficient that plain old simple resistive heat-dumping as soon as more than one LED is turned on.

Keith
 
I do recognise that it's not exactly standard practice, but I had some 48V relays already and perhaps vainly decided I wanted to use illuminating pushbuttons. I couldn't find any that had a second pole to switch the LED so it all has to run off one voltage. I think I'm going to split the supplies for the relays and phantom to make things a bit more manageable.

In fairness the chances of having phantom power on every channel AND needing phase invert on every channel are, well, nonexistant, so I will never in the real world actually be drawing that amount of current, but I don't want my power supply to give up just because someone decides they want to make all the lights go on.

Thanks again everyone!
 

Latest posts

Back
Top