Sunday, May 26, 2013

When the LED is glowing orange


I've been repairing some old Clear-Com hardware -- our MS-200C Base Station, some RM-120A Speaker Stations, and a broken headset or two.  Maybe not the best time to work on that project, but I was cleaning up the booth, and there are issues that have been annoying me for just a little too long.

Both of the rack mount type units were ordered with the goosneck microphone option.  Which we never use at the theater.  Instead the microphones are taped over with bits of foam and masking tape and such in a vain attempt to keep them from picking up noise.

For whatever reason, they were also assembled so that; A) there is no switch on the microphone for a headset plugged into the front (which makes for a lot of bruised ears), and; B) the sidetone for Channel B on all units is trimmed all the way up against the stop -- meaning the moment you switch any unit to Channel B, the whole affair erupts into feedback.

That was also ameliorated in practice by messy wads of masking tape and hand-scribbled warnings.

After a lot of puzzling over the schematic I brought the headset microphone into the master microphone switch, added a different switch to turn on the goosneck microphone when needed, added a 1K resistor (the Rat Shack no longer stocks trimpots and Al Lasher's was closed) to trim the gooseneck's electret condenser to something closer to the sensitivity of the headset microphones, and cut away the circuit that dips the volume slightly whenever the microphone is activated, in order to use that part of the toggle switch to activate a "microphone is live" light.

I also pulled out the dead bulbs in the channel buttons.

But here is where I fell into that trap of a little knowledge.  I've become used to the concept that an LED is relatively voltage-agnostic.  The voltage drop across the diode remains the same regardless of supply voltage.  The only trick is calculating the correct current-limiting resistor.

I made up a couple of clever little indicator-lamp replacements with green LEDs, quarter-watt resistors and a bit of plastruct; shaped so they fit into the original holder of the indicator lamp.  Worked like a charm.  I even took pictures as I created the last of these modified indicators, preparatory to creating a new Instructable about them.  I finished all that and closed up the Base Station.

I was diagnosing issues with one of the Speaker Stations when I held down the "Call" button on one a little bit too long.  And that was when I noticed the green LED was glowing orange.  Which was really cool, and also useful; meant you could see the incoming page even when the channel indicator was already lit (in the original design, the incandescent lamps would get a little brighter when paged).

For about two tries.  Then there was a "pop" I could hear across the room, and no light at all.

So here's where I fumbled.  LEDs are current devices.  As I said, the voltage drop is always (roughly!) the same.  The series resistor (also called the "ballast resistor") is chosen to limit the current through the LED.  Theoretically, at any supply voltage (that is, any voltage above the minimum voltage drop to light the LED in the first place), the LED will behave the same.

Where Ohm's Law has its revenge is on the resistor.  As the supply voltage increases, since the voltage drop across the LED is constant, the drop across the resistor increases.  And since the current in the loop is constant, the wattage in the resistor increases.  Above 24 volts, the resistor is burning several watts!  Which means even if it is rated for it (my poor little ones weren't), the heat produced is going to cause some damage.

Plus you are wasting a bunch of power.

In practical terms, the resistor starts getting non-linear as it heats, the current in the loop goes up, and the LED fuses -- hence the pretty orange glow.  It jumps back a few decades in technology and becomes an incandescent bit of  metal inside an air-proof plastic shell.  And it is a short circuit -- if I hadn't found this problem on the bench, I would have been installing a fire hazard.

So how do I fix it?  Possibly, by purchasing proper indicator lamps.  That's the smart move.  Except, of course, that everything is closed for Memorial Weekend, and we go into tech Tuesday.

The way the boxes are designed, it isn't plausible to stick in a single voltage regulator and tap that for all the indicators.  It might be possible to stick relays along the indicator paths and use those to switch a lower, regulated power supply.  That's a bit messy, though, especially for the ones that are built into the switch caps.

I also considered using a zener shunt.  The advantage is that the zener would hold the voltage at the correct level regardless of what call buttons were on in the loop.

The zener, like most diodes, has a reverse breakdown voltage.  Difference is, it is designed to operate in reverse.  If you put a zener backwards across a higher-voltage power supply, the voltage measured across the zener will be the breakdown voltage.  Which for zeners is very carefully calibrated at the factory.

Of course, the zener would also fry in an instant.  You also have to limit the current on them!  Because of the design of the shunt, the total supply voltage is seen across the total circuit (zener and current-limiting resistor).  Which means, just like the LED, the wattage in the resistor goes up as difference between zener voltage and supply voltage increases.  Not a help.

But that led me to new calculations, and new experiments.  Instead of calculating a ballast resistor for a 20ma loop (the rating of the LED), calculate for 5ma (LEDs don't really dim, but they can be run at lower amperages for a barely visible change in output intensity).

Because of the paradoxical nature of the fixed voltage drop, as you increase the size of the ballast resistor you increase the voltage drop only slightly, meaning the wattage would go up; but you decrease the current in the loop by a larger amount.  So up until the LED no longer turns on, a larger and larger resistor will actually consume fewer watts!

And it turns out that with the right resistor value (5.6K, in fact), the LED lights just fine, the wattage is under a quarter watt, and a half-watt resistor barely warms to the touch.  And the performance of the device appears uncompromised.

(And, really, this is practically the same thing as an incandescent bulb; the turning of resistance into heat is just happening inside the filament and heating up the lamp, instead of happening in a resistor elsewhere in the circuit).

So I'm off to check the Rat Shack just in case they have the appropriate indicators, and to get some more 1/2 watt resistors, but this looks to be how I will finish this particular circuit.

(The one thing that continues to bug me is that due to the all-in-one indicator design on the Base Station, a remote station page on an already monitored channel isn't visible.  This would only come up, of course, if there was someone in the booth within eyes-shot of the Base Station but not on headset, and no other live headsets were in the booth area.  But it still annoys me.  And as tempting as it is to play with the forward voltage of LEDs to make a dual-color indicator, it would probably require changing some of the internal circuitry of the Base Station and I'm loathe to do that.)

No comments:

Post a Comment