5-24-48VDC PSU Question

[syn]

Hi everyone
after a few days of trying to find out the solution I'll have to give up and ask, as i had no luck up to now...

What would be the way to drop from 38VDC to 5VDC in order to power up
some relays and leds (8 relays and 8 leds)?

~600mA 5VDC needed
Relays= 5VDC, 40mA, 0.2W (x8)
Leds= 5VDC ~20mA (x8)
Transformer 220/29.7 VAC(not loaded) 24VAC(loaded) - no C.T. (1.25A-30VA).
Schematic

Tried with 317- gets to hot to touch even with only one relay or one led attached to it,
Tried with simple voltage dividers, to hot to touch)...values calculated using Vout=I*R2=[R2/(R1+R2)]*Vin ohm's law, etc...
Tried with 7805- gets to hot to touch even with only one relay or a led attached to it.

The voltage (38VDC) for the 5VDC rail was puled from the same place at all times (from diode bridge(+) junction as shown in the schematic)...

So the problem is excessive heat that builds up quickly and only on the 5v (be it generated by 317, 7805 or voltage divider) (the circuit was changed every time accordingly to the data sheets...). Other voltages work as they should...(+24 and +48VDC).

Thank you very much

 

[JohnRoberts]

Well there are sundry high tech and low tech solutions.

If you have too much voltage you can allow the PS to be less effective.

For example;

for just that relay driver supply consider half wave rather than full wave rectification.

consider small R in series with diode before reservoir cap.

consider smaller reservoir cap so ripple reduces average voltage dissipation (note C must now tolerate ripple current).

consider power dropping resistor between unreg and regulator.
------
For more high tech approach:

DCC switcher.

Back slope (?) switch- this is not common (I haven't seen it used by anybody else) but imagine a semiconductor switch in series with secondary of the transformer (between diodes and reservoir cap) that just disconnects when you've charged your reservoir cap to your desired voltage.

I used this in the old Loftech test set ('80s) to get a high current 5V supply and low current +/- 15v without wasting a lot of power in 5V pass element. It worked nicely and you actually get more usable power from a given transformer that way (without a switcher) but you can encounter flyback inductance issues with some transformers and switching is not trivial so YMMV.

JR

 

[syn]

Thank you very much for your kind replays, this is what i came with (as it would be the easiest to implement). Would some one confirm that this would be ok?
~ 43.3ohm 6W from unreg to reg... (per led+relay)

 

[Jim Zuehsow]

The best way, and how it's done in the real world, is to use a buck step-down regulator to convert the voltage down to where you need it. All the other ways are much to lossy and archaic in this "green" world we live in today. Several of the big companies make these chips, including National and TI. They also have application notes on using their buck regulators.

 

[JohnRoberts]

[quote author="Jim Zuehsow"]The best way, and how it's done in the real world, is to use a buck step-down regulator to convert the voltage down to where you need it. All the other ways are much to lossy and archaic in this "green" world we live in today. Several of the big companies make these chips, including National and TI. They also have application notes on using their buck regulators.[/quote]

While it's pretty easy to use a modern canned switching solution, my little LF switch on the secondary is not lossy either, and doesn't generate HF switching noise (while there is a little inductive snap when it disconnects from the winding on the way up. If that's an issue you can only turn it on while it's on the way down (like lighting dimmers) but the control is a little more complicated if load varies much.

There's always more than one way to skin any cat,,, pick your poison.

JR

 

[Jim Zuehsow]

That's very true John. Many roads lead to the same place. I mentioned the buck regulator since it is a dirt cheap solution these days. Almost everything involved with telco today has one or more to provide the various supplies needed in the equipment. The scale of economy has driven the price of these so far down, it's hard to consider a different solution. Even Rupurt Neve is using DC to DC converters in all his very high class Portico equipment with no fear of switching spikes running amuck and causing problems.

 

[PRR]

> drop from 38VDC to 5VDC in order to power up some relays and leds (8 relays and 8 leds)? ~600mA 5VDC needed

38V-5V= 33V drop. 33V times 0.8 Amps is 26.4 Watts.

Your actual useful load is 5V*0.8A= 4 Watts.

You are throwing away 26 Watts to get 4 Watts. That's silly.

And the common regulator packages are rated 20 Watts on an infinite heatsink, less at high dropping voltage. (More like 1 Watt if you run them without a sink.)

You can buy 24V relays. Similar power, so the current is 5 times less.

You don't have to run LEDs at maximum current. For many uses, they will be plenty bright at 5mA. And you can dribble 5mA from 24V onto an LED with about a 5K resistor.

Eight 24V 10mA relays is 80mA.

Eight LEDs working at 5mA is 40mA.

Total 120mA.

Dropping 38V to 24V at 0.12 Amps is (38V-24V)*0.12= 2 Watts. The low-price regulators will handle that with a 2" square scrap of aluminum.

If you must use 5V power for these loads.... the Classic approach would be a 6VAC 10VA transformer, a bridge rectifier, and maybe 1,000uFd. That will give around 7 volts DC. Relays won't be too unhappy here.

JR has a clever trick. I've seen the idea in every lamp-dimmer (well, not salt-water dimmers) and TV set SCR power supplies.

The Modern approach is to use a DC-DC converter. As Jim says, they are everywhere, not precious, and well designed.

The get-it-done approach is to look behind your PC. The world is full of 5VDC 2A wall-warts. I have a dozen in my office. They powered USB 1.0 hubs, Palm Pilots, card readers, external ZIP drives, answering machines, and other garbage.

 

[CJ]

Or look inside the PC.
My brother put a bunch of banana plug sockets on a junked switcher, cool beans.
Free DC in mass quanities and voltages.

Don't know about the audio part.

 

[syn]

Thank you very much on your kind answers, think I'll take separate PSU route as I have all parts ready...

:guinness: :sam: to you all...

 

[JohnRoberts]

[quote author="Jim Zuehsow"] Even Rupurt Neve is using DC to DC converters in all his very high class Portico equipment with no fear of switching spikes running amuck and causing problems.[/quote]

Interesting, from a recent post in the Brewery it sounds like he's still a fan of (audio) transformers and wide bandwidth, but I guess that's why they make shielded inductors and transformers.

I remember having trouble sourcing mu metal in China back in the late '80s early '90s. I was dealing with big (70-100V) output transformers, coupling into small mic transformers. Surely it's better now (or I was just asking the wrong folks).

JR

 

[Guest]

[quote author="JohnRoberts"]

Interesting, from a recent post in the Brewery it sounds like he's still a fan of (audio) transformers and wide bandwidth, but I guess that's why they make shielded inductors and transformers.

[/quote]

I did not know he was a fan of transformers, I always thought he is a fan of optimal design for better sound...

Speaking of PSU, I recently had to make headphone amp's PSU much better than needed for headphones because of ground currents inside of the console...

 

[JohnRoberts]

[quote author="Wavebourn"]


Speaking of PSU, I recently had to make headphone amp's PSU much better than needed for headphones because of ground currents inside of the console...[/quote]

Common beginner mistake with console headphone amps is to not keep ground return paths isolated from other circuitry. Ground currents from low Z headphones can crosstalk and corrupt other nearby circuits. Even some well know console brands that should have known better have dropped the ball on this.


JR

 

[Guest]

[quote author="JohnRoberts"][quote author="Wavebourn"]


Speaking of PSU, I recently had to make headphone amp's PSU much better than needed for headphones because of ground currents inside of the console...[/quote]

Common beginner mistake with console headphone amps is to not keep ground return paths isolated from other circuitry. Ground currents from low Z headphones can crosstalk and corrupt other nearby circuits. Even some well know console brands that should have known better have dropped the ball on this.

[/quote]

Exactly. I thought I've corrected all mistakes in that TOA RX-216, but keep finding'em... :green: