February 2015
About three years ago, I put together a 1.5 volt battery eliminator using a wall wart feeding into an LM4120 regulator. My goal was to power the clocks I have around the house and save myself the aggravation of replacing the batteries all the time. The clock that I started with is a Howard Miller mantle clock that a company awarded to me for busting my ass for 25 years.
After installing the eliminator, I set the clock to the time of my crack atomic wristwatch and let'er go. The clock ran for almost three years with phenomenal accuracy, matching my watch within a few seconds (a hex on those who disbelieve this). The clock finally died — probably from exhaustion — having gotten no rest between battery exchanges.
Well, I thought, what are you waiting for. Get with it with the other cheap clocks cluttering up our house; so, I did. To my amazement, none of the clocks running on the eliminator could keep time anywhere near what could be termed accurate — no matter how much I adjusted the voltage (usually, the clocks ran fast).
So, what gives? Why does a battery work and my eliminator won't? Some wizard out there must know what the problem is, and be willing to share the knowledge with me.
Please log in to post an answer.
The LM4120 regulator only puts out a few mA, and this 1.5V application is at the bottom end of the part's voltage range specs. When connected to clocks that need a bit more juice, the regulator sags and puts noise on the DC supply, and that screws up the time keeping. Solution: Look for a beefier LDO, ensure the wall wart is putting out clean DC when loaded by the clock, and ensure the LDO's input voltage is sufficiently higher that the 1.5V LDO output.
I recommend looking at the LM4120 output with a DC coupled oscilloscope and checking for noise or voltage drift, either short-term or long-term. I think you will see the problem.
I used battery clocks in an amp hour counter device and found my clocks running slow, sometimes stopping. Maybe your problem is related to mine. It seems that the pulse current to run the ticking solenoid is relatively high. Adding a 100µf electrolytic plus a 1µf ceramic cap directly on my fake battery (wooden dowel with screws on the ends) fixed the problem.
First off, I suggest doing the following:
1. Measure the OPEN-CIRCUIT output of your eliminator before connecting it to the clock you want to run.
2. Connect the eliminator to the clock, then measure the voltage output again.
If the difference between the "no-load" voltage and "load" voltage is more than 0.5 VDC, it's entirely possible your wart isn't delivering enough current to properly operate the regulator. In this case, try a similar-voltage wart with higher current output (say, 1.5-2X of your current wart). This may solve the problem as wall wart outputs tend to droop severely once you approach their maximum current capability, resulting in severe output instability, (i.e., the regulator won't "regulate" well), increased ripple, noise on the DC output, and severely shortening the life of the wart (i.e., overheating and such).
If the "no-load/load" voltage difference is negligible (< 0.1 VDC), try adding a filter capacitor, (start with 470 µF electrolytic — watch the voltage rating of the cap!), paralleled with with a 0.01µF mica or polyester to filter out high-frequency hash that may be on the DC output feeding the clock. This may give you the stability you're looking for as most, if not all, warts are half-wave, unregulated types with very minimal filtering to begin with. Adding more filtering (larger electrolytic) and bypass (small value) caps to the DC output greatly improves their overall stability and cleanliness of the DC output voltage.
Finally, (as you already know), having a regulator between the wart and your device guarantees a rock solid DC source, as long as you don't pull too much current from the wart.
Without more information, it's difficult to give a definite answer, but here are three possible causes to check:
1. Is the output voltage correct? Check with a DVM; anything from 1.35 to 1.6 V should be OK for LCD or quartz clocks.
2. The power supply has excessive AC in the output, e.g., a bad capacitor. Though you could check this with an audio amplifier or oscilloscope, it's easier to just put a 500 microfarad or larger electrolytic cap across the output and see if that fixes the issue.
3. AC or RF leakage from the power supply, either from the mains or from a nearby radio transmitter, is making its way into the clock. The clock circuitry is very low power, so any AC could flip some flip-flops a few extra times per second. To check this, you could make a Faraday cage (e.g. window screening) around the clock and connect it to one side of the supply. This is to satisfy your intellectual curiosity, though it's probably not a convenient way to run a clock 😉
