AC line fluctuations

[TimFox]

I am close to finishing a long-running project:  a preamplifier (including RIAA equalization) using 7586 Nuvistor triode vacuum tubes.  I tested prototypes for the stages using laboratory power supplies with good results.  Unfortunately, I had naively looked at power-supply rejection only at 120 Hz (easily obtained with filter capacitors on the DC lines).  Since the individual stages (CC, CC, CF per stage) are DC coupled, the rejection at frequencies below audio is not good enough for an unregulated power supply.  Looking into that problem, I found that the DC fluctuations are caused by short-term fluctuations in the AC line voltage.
Specifically, I measured the nominal 120 V 60 Hz line, measured with an Agilent U1252B 50,000-count DMM at 7/sec display rate, in TRMS AC voltage mode.  Logging max, min, mean over 240 sec measurement time, I found 1.47 V pk-pk, or 1.2% of 121.6 V mean value fluctuations.  This corresponded roughly to the fluctuations of the DC line, monitored on an analog CRO, where the characteristic time of the fluctuations was roughly 0.5 sec.  I moved the voltmeter from the workbench to an outlet immediately adjacent to the breaker box (with no other loads) and saw 1.1% pk-pk fluctuations on a slightly higher average value.  These measurements were done on a cool day with the central air conditioning off.
My question is:  is this level of short-term fluctuations normal for the residential AC line in an American city?

[floobydust]

Yes, distribution voltage moves around as grid load get switched on/off, several %. Frequency moves a tiny amount <<0.1Hz.
It seems worse during weekdays when big businesses are running.
Large voltage jumps occur when transformer tap-switching happens, when a utility makes voltage adjustments.

If your preamp offset is upset by a few % change in DCV, I'd say you will have to add a voltage regulator to B+, or a servo to keep offset from moving around. Don't forget tube transconductance drifts with age too.

[TimFox]

Yes, I am now adding active voltage regulators to both B+ and B- lines.  In order to stabilize against very low frequency fluctuations, the regulator must use a DC reference, in this case a TL431. 

[Vtile]

Yep. The Euro standard allow iirc. 5% error (+-) from nominal voltage. From memory.. assume the US standard allow similar movement as the electrical engineering tend to like internationally equivalent standards. Starting from Giorgi unit system. :p Like said the frequenzy is much more controlled in fraction of percentage.

[Vtile]

If i'm reading correctly my old source (70 yrs old electrical physics brickwork) you can use neon-lamp as a voltage reference. Just food for thoughts and sharing odd pieces of knowledge.

[TimFox]

Yes, neon lamps work as voltage references, and I have a good collection of actual reference tubes (0A3, etc.).  The current drain here is a bit much for a shunt regulator, so I am using series-pass regulation to minimize the voltage drop from the raw supply.
I was expecting 5% or so voltage variation, but was surprised to see it fluctuate so quickly (< 0.5 sec).  Although each three-triode stage is internally DC coupled, there is a coupling capacitor at the input to each stage and at the overall output.  However, since those coupling networks must pass low frequencies, they pass the ball-park 2 Hz waveforms nicely.

[Vtile]

Cool. The short term transient/fluctuation limit system is not more than archaic in distribution systems and is still heavily based on the percrption of indascent bulb flicker .

[tooki]

I am close to finishing a long-running project:  a preamplifier (including RIAA equalization) using 7586 Nuvistor triode vacuum tubes.  I tested prototypes for the stages using laboratory power supplies with good results.  Unfortunately, I had naively looked at power-supply rejection only at 120 Hz (easily obtained with filter capacitors on the DC lines).  Since the individual stages (CC, CC, CF per stage) are DC coupled, the rejection at frequencies below audio is not good enough for an unregulated power supply.  Looking into that problem, I found that the DC fluctuations are caused by short-term fluctuations in the AC line voltage.
Specifically, I measured the nominal 120 V 60 Hz line, measured with an Agilent U1252B 50,000-count DMM at 7/sec display rate, in TRMS AC voltage mode.  Logging max, min, mean over 240 sec measurement time, I found 1.47 V pk-pk, or 1.2% of 121.6 V mean value fluctuations.  This corresponded roughly to the fluctuations of the DC line, monitored on an analog CRO, where the characteristic time of the fluctuations was roughly 0.5 sec.  I moved the voltmeter from the workbench to an outlet immediately adjacent to the breaker box (with no other loads) and saw 1.1% pk-pk fluctuations on a slightly higher average value.  These measurements were done on a cool day with the central air conditioning off.
My question is:  is this level of short-term fluctuations normal for the residential AC line in an American city?

FWIW, I haven't found the U1252B to be particularly talented at peak min-max logging. Despite having identical min-max specs as the Fluke 87V (1ms regular min-max, 250us peak mode), the 87V catches more peaks. (Real-world example when I was testing audio amp current draw: the 87V would catch peaks about 25-30% higher than the U1252B, e.g. 1.2A versus only 900mA.) I suppose an oscilloscope would do far better than either, of course.

[TimFox]

It was straightforward to use a true-differential CRO in AC coupled mode on the rectifier filter output.  I used the DMM to see if there was a comparable fluctuation on the AC line.  Using a digital scope on the line through a transformer didn't give enough resolution for a valid measurement.