When developing quality devices that interface with telephone lines, sensitive audio measurements are sometimes needed. However, both frequency noise and distortion added by the test equipment—as well as low-frequency signals derived from the power line itself—must be minimized.
Usually referred to as "ripple," a better term for these low-frequency signals might be "hum." In these cases, remnants of the line frequency (50 Hz/60 Hz and harmonics) aren't always apparent on an oscilloscope, but they appear when making certain audio-analyzer measurements. Ripple usually implies an ac signal riding on a dc voltage. In contrast to the situation here, ripple can be fairly easily sensed. But remnants of the power-line frequencies may get into the output through other means, including magnetic, and it can be hard to prevent the disruption of sensitive tests.
For an instrument to properly simulate the telephone-line central-office (CO) voltage of 48 V dc, we need to avoid these unwanted frequencies. The most obvious solution is to connect batteries to get the needed dc voltage. But this approach is guaranteed to be bulky and not easily put into a box to become test equipment. Also, there would be limited adjustability as the batteries discharged.
Another method implements two +24-V switching supplies in series. This would shift most unwanted frequencies out beyond the audio range. Yet that leaves a supply connected to the power line. This means there's the possibility of injecting remnants of whatever might appear across the power line, for whatever reason.
Figure 1 shows a design that combines some of the best parts of these two approaches. It's compact, very adjustable, and built from readily available parts. Plus, it has no connection to the power line.
As can be seen, a push-pull circuit takes a 6-V dc battery voltage and steps this up using T1, a 6.3-V center-tapped power transformer. It's connected "backwards," so that what had been the ac side is now the output. U2 rectifies this output, and then U3 regulates it. The diode circuitry at U3 comes directly from the LM317 data sheet, for voltages over 25 V. Also, both R1 and R2 consist of two 20-, 1-W resistors in parallel. Different transformers will probably require different values for R1 and R2.
Some of the circuit's interesting features can be explained starting with the waveforms shown in Figure 2. These show voltage switching at the input and output of U1a, an ordinary 7404 inverter IC. Due to the nature of the 555 timer (U4), the duty cycle doesn't quite reach 50%, but it's close enough. Adjusting R6 ensures a switching frequency of at least 20 kHz—well above the audio range. Therefore, remnants of this frequency and its multiples won't interfere with sensitive audio measurements.
Once R6 is set, R4 allows a wide adjustment range for VOUT. With the components shown, VOUT can be varied from +28 to +68 V, given a +V = 6 V. As an added bonus, if the battery voltage drops as low as +3.0 V, VOUT can still be adjusted to +48 V, ±3 V.