20 led display

[chris p]

Hi y'all

I'm trying to build an extended peak display based around the LM3915 LED driver - this drives 10 LEDs in 3db steps. The datasheet for the chip http://www.national.com/ds/LM/LM3915.pdf shows how to link 2 of them together, as well as a precision half-wave peak rectifier. At 10V full scale, that gives me a peak display from about +21 down to -40dbu or thereabouts in 3db steps, which is what I'm after.

So far so good. My issue is that the power supply is the good old JLM 48V single sided supply, which means I need to manage a few things.

I can split the power supply using 20K and 10K resistors to generate 0V, +12V and +24V. I can power the dual opamp and the 3915s with 24V and 0V, biasing the non-inverting input of the rectifier opamp and the inverting input of the display linking opamp (the "ground" of the 10K resistor next to the 309K resistor) with 12V instead of the ground as shown in the datasheet.

My understanding is that this will also have the opamp outs biased with +12V, which I will need to remove using an electrolytic cap (say 470uf, 50V). Where I seek help relates to the place and manner of this. Am I right in thinking of a small resistor (say 51R) in series with the cap, just ahead of each 3915 input at pin 5? Or do I need to do something before the output of the rectifier opamp hits the input of the linking opamp, like in this sketch

[JLM Audio]

You will need a resistor to 0v ground for each LM3915 input to get its 0v reference and because its input is a 20k resistor you will not need any 51R resistor in series with the 470uF. Really if you use 20k to 0v from the input the cap in series could be 10uF.

Technicially it looks like you can run the whole thing on a 18v rail with a 1/2 rail (no 7805 needed) and no DC removing caps if you reference R-Lo to the 1/2 volts and move the resistor going to 0v from Ref Adj to 1/2v volts as then the 1/2 volts rail is treated like the 0v of a +/- power supply.

The Led intensity pot may have to go to the 1/2 volts rail as well but not sure about that.

You could run the whole thing on 24v with half rail but you would need to use high intensity leds to keep the LM3915 pakage heat mW within spec.

[chris p]

Joe Malone, ladies and gentlemen - the DIY hacks very bestest of friends. Thanks for the comnments.

Technicially it looks like you can run the whole thing on a 18v rail with a 1/2 rail (no 7805 needed) and no DC removing caps if you reference R-Lo to the 1/2 volts and move the resistor going to 0v from Ref Adj to 1/2v volts as then the 1/2 volts rail is treated like the 0v of a +/- power supply.

OK - I can see that. I'll breadboard something along the lines of this and see where the LED pot goes - I try it to ground first, and to 1/2V if that doesn't work.

[JLM Audio]

ok your first resistive divider will not work as the current of the led drive and other things will be up and down by huge amounts. If you were going to run dot mode which is more constant current three series resistors might work.

You will need something like our TREX kit or (LM7818 reg with its input feed from the 48v supply by a 47ohm resistor 1watt) to supply the 18v from the 48v.

The led drive will need to go to +18v as the 1/2volt point cannot support any current draw.

When testing you will probabily find that the bottom 10k for the 1/2v will need to be made smaller in value to adjust the 1/2v to the correct voltage. So using a 25 turn 20k trim pot as the voltage devider for testing would be a good idea.

[Thirteen]

ok your first resistive divider will not work as the current of the led drive and other things will be up and down by huge amounts. If you were going to run dot mode which is more constant current three series resistors might work.

You will need something like our TREX kit or (LM7818 reg with its input feed from the 48v supply by a 47ohm resistor 1watt) to supply the 18v from the 48v.

The led drive will need to go to +18v as the 1/2volt point cannot support any current draw.

When testing you will probabily find that the bottom 10k for the 1/2v will need to be made smaller in value to adjust the 1/2v to the correct voltage. So using a 25 turn 20k trim pot as the voltage devider for testing would be a good idea.

Sorry to jump off topic for a moment guys, but can I ask Joe if any of your power supply kits could be easily adapted to be +/- 15V and +5V with reasonable current on the +5V rail as well?

[JLM Audio]

Sorry to jump off topic for a moment guys, but can I ask Joe if any of your power supply kits could be easily adapted to be +/- 15V and +5V with reasonable current on the +5V rail as well?

Yes our Powerstation kit can do this fine and if no 48v is needed our AC/DC kit can too by leaving two of the 1000uF cap left off and jumpered them over to drop the 48v rail to be able to do 5v with minimum heat.

To keep current up and heat down use a 2 x 15vac transformer like our world power transformer kit.

[chris p]

Thanks again Joe - the fact that resistor dividers only work to split voltage when current draw is constant is a consequence of Ohm's law, yes? That should be a in the newbie forum, says I with embarrassment.

In my PCB design I now have the option of installing trimpots or resistors for the lower half of the 1/2V divider network as Joe suggests, as well as in the LED intensity and Vref adjust resistors. While I will use R values derived from trimpots when breadboarding, the PCB will be more versatile that way.

BTW, Thirteen, feel free to hijack away - the thread is really about designing power supplies, and comments /questions just serve to educate us all.