Oct 30, 2019

On Power Amplifier DC Offset…

By Bryan Geyer

Heads up, please!  It’s time to consider an important power amplifier parameter that a great many audiophiles overlook. It’s known as DC offset, and it has direct bearing on the dynamic range capability of your low frequency loudspeakers.

Fluke 87V Multimeter
When a high quality audio power amplifier is in a quiescent state (operational, but no input signal), its output terminals will normally measure nearly zero AC and zero DC volts. A tiny residual AC voltage (some ~ 40-80 microvolts) will reflect the sum of the internal noise generated within the power amplifier. And a small residual DC voltage (generally on the order of ~ 2 to 15 millivolts DC) will represent the amplifier’s implicit output imbalance; a.k.a. its DC offset. (In the case of vacuum tube power amplifiers there’s no DC voltage present because the output is normally isolated by a transformer. Similar DC isolation applies if the power amp is coupled to the load through a capacitor, an archaic compromise that’s taboo in modern hi-end power amps.)

The loudspeaker system connects directly to the power amplifier, so any offset that’s present at the amplifier output will also appear at the inputs of the speaker system’s passive crossover network, and then feed directly to the DC-coupled low frequency woofer. Higher bandpass speakers will be DC-isolated by the network’s coupling capacitors; it’s only the woofers that will see the DC offset.

When quiescent, the woofer rests in the neutral middle of its magnetic field. Optimally, the cone is not displaced—forward or backward—from that mid-point rest position until it sees an input signal. Of course, the presence of any DC bias will slightly offset that ideal mid-point rest position, and this shift could potentially impact the cone’s full range of linear excursion. If the offset is very slight, that impact will be entirely negligible. As offset increases, its influence can become significant. Any major offset (e.g. ~ 100-200mVdc) could measurably (audibly?) degrade the woofer’s dynamic range.

It’s commonly promulgated that some small (~ 10 or 20mVdc) offset is both harmless and inevitable. Since DC offset will increase as operating temperature rises, it becomes more difficult to hold offset within desirable limits in the case of class A (or partial class A) power amplifiers. They operate at higher chassis temperatures than class A/B amplifiers, so owners of class A (or pseudo-class A) power amplifiers should be especially vigilant. DC offset is a very easy measurement to monitor, and accurate DC millivolt reading multimeters are readily available. (Select a meter with input impedance ≥ 10MΩ that can read 600mVdc full scale, with 0.1mVdc resolution. Fluke’s products are highly recommended. Here is an excellent hi-end model: https://www.myflukestore.com/product/fluke-87-5-industrial-multimeter)

Modern solid-state hi-end power amplifiers typically apply symmetrical input stage circuitry using complimentary bipolar, Jfet, or MOSfet discreet devices that are carefully matched to minimize potential DC offset. It’s especially critical to minimize this imbalance at the input stage because, in a DC-coupled amplifier, that offset will then be magnified by the voltage gain (generally +23dB to +29dB) of the product. As a result, it’s only reasonable to tolerate some modest (e.g. ~20mVdc) output offset, and one supplier (Pass) of partial class A power amplifiers specifies a 50mVdc maximum offset limit for their respected “Point 8” series (e.g. X250.8) of pseudo-class A power amplifiers.

Many makers don’t disclose any DC offset specification, although they might maintain an internal screening limit that’s never published (hence not guaranteed), so some snooping could prove helpful. The company’s Service Department can sometimes be a good place to start when it’s a domestic operation. In some cases (e.g. Parasound Products Inc., of San Francisco), the company president, Richard Schram, is both technically savvy and personally accessible. I like that kind of company!

My recommendation is that you personally measure your power amplifier’s DC offset. Know what the offset is at moderate operating temperatures, in normal use, and know what happens to DC offset after you’ve pushed your power amplifier through a heavy listening session. If you are in the course of considering the purchase of a new power amplifier, research the DC offset specification. If it’s not published, contact the maker; seek reliable information. High quality audio power amplifiers with negligible DC offset are readily available, as are identical models with excessive DC offset. It’s up to you to discern (and reject) the lemons*. Leave the latter for the lazy buyers—the audiophiles that always evaluate everything exclusively by listening.

BG (October 23, 2019)

*Moderate DC offset, e.g. ~ 50-150mVdc, is extremely difficult to diagnose by ear. Imbalance of that nature becomes apparent only when the amplifier is approaching full output, a condition that invites other sundry (and more likely) imperfections. Those issues will normally mask the subtle evidence of moderate offset error, so aural screening—even for those who profess exceptional sensitivity—will probably prove ineffective. To be more precise, measure the DC offset.

3 comments:

  1. Why do you use a multimeter instead of an oscilloscope?. A multimeter measures Volts rms or peak or peak to peak which tells nothing about the symmetricality of the waveform. The ideal waveform should have a 0 volts offset which means the peak value of the top matches the peak value of the bottom. An multimeter cannot tell you this

    ReplyDelete
  2. Measuring DC offset, not the waveform.
    DMM gives perfect measurement for this, it's cheaper and easily obtainable.

    ReplyDelete
  3. DC offset is commonly measured when the amplifier is operating WITHOUT any signal input (input jacks shorted), so the only AC waveform appearing at the output will be hum + noise. The latter is of no interest when seeking to measure the amplitude of any DC offset that's present, so the optimum instrument of choice is a sensitive (0.1 Vdc full scale) DC millivoltmeter, e.g., Fluke's 87v multimeter.

    ReplyDelete

Thank you for your comment. It will be published after review.