From 30V down to 3.3V

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Rochey

Well-known member
White Market Member
Joined
Jul 2, 2004
Messages
2,697
Location
Dallas, TX - Formerly UK
Folks,

Since we released the digital controlled front end, some people have been asking about integrating a 4 switch front end into the 1176 design.

Having looked at the 1176, it looks like there's a +30V supply, and a -10V supply.
The MSP430's that we use run from a maximum 3.3V supply. As such, some circuitry on the board needs to bring down the 30V, to a more sensible 3.3V.

This thread is about the power supply. I'll start a thread regarding the switch in the future.

So, here are the options I've chewed on to solve this power supply problem.
  • One big linear regulator, that takes 30V and brings it all the way down to 3.3V
  • A power resistor in series with the regulator, so that some of hte power is dissipated in teh resistor
  • A small switch mode power supply circuit

The easiest is one great big linear regulator, however, dropping 26.7V accross it, at 40mA max (4 LED's, each driving 10mA) will generate 1.06W of heat. In a T0220 package (large) this will increase the temp by 20C or so. In a smaller TO-92 package (that's used for my LP2950-3.3 regulator) it'll increase the temperature by 150C!

Using a switch mode supply would be the most environmentally friendly way of doing it. However, most Pro Audio folks I know don't want SMPS's in their system, if they can avoid it, and it adds significant complexity and difficult to source components (and maybe some SMT work too!)

So the conclusion I came to was to use a series resistor in front of the LP2950-3.3 to dissipate some of the power.

I've created a small excel file that I used to calculate the value of the series resistor. It can be found here: http://www.tendolla.com/designgallery/main.php/v/tools/Power+dissipation+in+regulators+with+series+resistors.xls.html

I make no guarantees for it's accuracy.

My current requirements will be between 10mA (one LED on) and 40mA (all LED's on).
by toying with the resistor value in the second window, at maximum current draw (40mA), my LP2950-3.3 will increase in temperature by 35C if I use a 470Ohm series resistor. My resistor will need to dissipate 0.95W. (within the 1W spec)

Any holes in my theory here? My ego won't be hurt - I promise :)
 
Why not an auto transformer at the 30VDC AC supply transformer taps?
http://en.wikipedia.org/wiki/Autotransformer

What is the AC voltage to the DC supply?  Calculate the ratio (stepdown) figure out the needed core size and inductance and...(stuff in books)

This is if the supply can handle the extra load.  How many amps at 3.3VDC needed?
 
Hello Gus,

as it's a retrofit into an existing design, I know for certain that there's 30VDC in the schematic. (from Gyraf's page)
I don't want people to have to change their transformer in the design.

The requirement is very low - I estimate a maximum of 40mA, as the two digital IC's are in the uA range.

 
3.3VDC at .040amps = .132 watts.

A 48VAC CT transformer?

Or just drop in a small 120/240VAC to 6.3VAC transformer to regulator circuit.
http://www.hammondmfg.com/160.htm
http://www.hammondmfg.com/166.htm
http://www.hammondmfg.com/266.htm
etc.
 
Although not the most environment friendly solution in terms of energy saving, this is probably the most sensible solution. You could increase the resistor value to 560R, the dissipation would still be less than 1W in the res, and the LM317 would still have 4V between input-output. You would need a cap at the regulator input.
 
Hello Gus,

thanks for your suggestion. I went hunting around the usual sites for a small transformer like that, and found that they are around $5 a piece. Not bad.

I think if I had a lot more digital circuitry in the system (and power draw), I'd be very tempted.

I am scratching my head about the relays that these will eventually drive. However, I think I should be able to have a separate 24V regulator for those relays later on.
 
If you don't want to be wasteful, you could use some "voltage-halvers" by flipping some voltage-doublers.
 
Hello John,

I need to control the LED's from the micro - they are each connected to a separate GPO.

I could use transistors to switch the 30V through each LED, however, that's more complexity, more parts.

My aim is to find the simplest, easiest to assemble, smallest PCB size solution
 
What's your budget?

For $15, you can use a 24-pin DIP, EC4A17H.
Outputs 3.3V @ 1A, from an input voltage range of 18-36VDC

landingpagecinconconverter.jpg
 
Rochey said:
I could use transistors to switch the 30V through each LED, however, that's more complexity, more parts.

It's also lower noise. I'll bet that switching 40mA worth of LEDs on or off from the +30V-rail of a 1176 will be audible.

Put the LEDs in series, connect a CCS at the top or bottom of the string, add a shunt transistor (BJT or FET, both have pros and cons) across each LED. Not a zero-parts fix, but very doable for a handful of LEDs.

JDB.
[cascodes tied to the VCC of your microcontroller make for cheap and easy level shifters]
 
jdbakker said:
Put the LEDs in series, connect a CCS at the top or bottom of the string, add a shunt transistor (BJT or FET, both have pros and cons) across each LED. Not a zero-parts fix, but very doable for a handful of LEDs.

you always make me thinking a little harder than what I was planning to do.
I haven't looked at CCS's in a LOOOONG time, and I'm not sure I understood them the first time.

From what I can understand, a CCS will present a constant load on the 30V rail. (not sure if I understood that correctly from all the sources I was reading)
the CCS at the top of the string of LED's (in series), then I control a path to ground by using switching transistors at each point between the led's.
I can see how that could work if I was making a bar-graph or something, (i.e. LED1, or LED1 and LED2, or LED1,2 and 3, etc. But how to I control each of them separately?
 
nah, configure a LM317 to be a CCS.  check the datasheet for the configuration.  I've used them for years powering LED strings and high power LEDs. 

 
Dunno if your board needs a shared ground..?
But if not, couldn't you make a 24 V regulation first (easy from 30 V, and could be usefull for those "relays that these will eventually drive" you're scratching your head about)?
From there, you get 6 Volt difference between +24 and +30 DC... Easy to drop down to 3.3V with your integrated regulator.

You may need opto couplers to drive the relays.
If your 3.3V regulator needs 3V more than its output voltage, I guess you can use 23V instead of 24, as 24V relays usually works fine with a tiny less voltage....

Axel
 
Rochey said:
I can see how that could work if I was making a bar-graph or something, (i.e. LED1, or LED1 and LED2, or LED1,2 and 3, etc. But how to I control each of them separately?

Let me paint you a picture.

led-chain.png


When IN1 is logic high (3V3), a negligible current flows through the collector of Q1 and base of Q4 (tens of nA for modern small-signal transistors at temperatures that are comfortable to humans), Q4 is off for all practical purposes, and all but maybe a few uA of I1's current flows through D1. When IN1 is logic low (0V), enough current flows through R1 and Q1 to turn Q4 on. Now (almost) all of I1's current flows through Q4, D1 is off. Repeat for all other LEDs, you can switch each on or off independently.

Select the resistor value so that Q4/5/6 hFE times resistor current is much larger than I1, but have resistor current much smaller than I1. This is because resistor current ends up flowing through Q4/5/6 base and emitter; for longer chains this means that LEDs higher in the chain get more current depending on how many lower LEDs are switched off. I'd say 10k is a good starting point.

JDB.
[picking BC557C's or other high-hFE transistors for the PNP parts doesn't hurt, although I suspect any modern part will do]
 
I'll apologize in advance if I missed something.

I admit I have nowhere near the experience of any of you, but it seems to me Gus idea is best... a separate trafo that can do your WHOLE circuit. Mainly because you don't know what power trafo each unit will already have, so you don't know if it is already overtaxed or whatever.  I don't know if the 1176 uses much current but it seems good engineering to make the add-on board self sufficient, since you're driving LEDs and relays.

Have you looked at Pulse's encapsulated trafos?  They have a small (less than 1.5 inches square) trafo that is 2x115 primary : 2x 6VAC sec, capable of anywhere from 83-466mA (depending on model). This was from a quick look on digikey and www.pulseeng.com. The 83mA is $1.87 at digikey (not normally stocked, MOQ 300, and not dual pri), but I can't imagine the higher amp ones would be much more, direct from Pulse. And I'd think with a half amp you could drive a boatload of (lower voltage) relays. 

 
JDB, interesting, I haven't seen common-high base used a whole lot in the wild. 

I used a similar concept for driving a set of small motors in a robot, I used FET drivers though.  Since they had logic inputs I configured them to work so that the outputs were all high and the legs would toggle low to turn the motor a direction.
 
Yes, JDB has it nailed...

The only difference in what I would suggest is if one of the leds is always on, or always comes on first, like in a meter, you can make it actually turn on the current source, so when no leds are on there is no current draw.

I am messing with some small mosfets for high side switches in a battery powered design to save control current.

But that is the general concept.. also with that much PS voltage you could probably get away with a resistor instead of current source to save a few cents. The change in LED current from 1 to 3 on won't be much if visible at all.

JR

 
Back
Top