MPS-1 Measurement and Analysis - Part 1
I did some quick spot-checking on the MPS-1 amplifier gain structure to ensure it could be properly driven with a wide assortment of preamps or receivers. For unbalanced, I found the Voltage Gain (Av) to be 27.1dB while for balanced it was about 26dB. The Signal Detect Threshold of the "Music" trigger is about 1.7mVrms, which should work fine with most preamps/receivers including the Emotiva DMC-1. I found the 12V trigger to operate without issue for all of the receivers and processors I had on hand.
Editorial Note on Balanced vs Unbalanced:
I measured the voltage gain of the MPS-1 via balanced and unbalanced inputs and found the following:
- Unbalanced: Av = 27dB
- Balanced: AV = 26dB
- Typical THX Product
- Unbalanced Av = 29dB
- Balanced: Av = 23dB (6dB lower than balanced)
It's a standard practice in most consumer and pro audio equipment that the gain structure of unbalanced to balanced is about 6dB less for the latter for power amps. The preamp therefore usually has a +6dB boost via the balanced outputs to compensate. I was a bit puzzled that the MPS-1 didn't follow this convention. However I didn't find it to be much of an issue, especially since the balanced outputs were about 3dB hotter than convention. It took about 1.65Vrms to drive the MPS-1 to full power. Had the MPS-1 followed standard convention of Av = 23dB some preamp sections of receivers such as the Yamaha RX-V2500 may not have been able to drive it to full power since that receivers' preamp outputs clip when output signals exceed 2Vrms. My personal criteria for amplifier gain structure is that it should be able to hit full power when driven with 2-3Vrms. The MPS-1 fell well within this criteria.
Check out our Balanced vs. Unbalanced Connections Article for more information.
Editorial Note on Measuring Digital / Switching Amplifiers
The switching process of digital and multi-rail amplifiers, such as the Emotiva MPS-1, adds fast rising edges at the switching frequency to the audio output signal. These fast edges are of no consequence to the typical load, a loudspeaker, but present a difficult signal for measurement instruments. The fast switching edges present high energy content and will introduce slew rate limiting when presented to the input stage of most measurement instruments.
When stressed by these fast edges, the analyzer input amplifier will usually slew rate limit and will not be able to function effectively in its normal mode. Auto ranging will be affected and the signal under test will be misrepresented to the following measurement circuits.The result is that noise and distortion measurements of switching amplifiers with almost any analyzer without preconditioning will yield inaccurate and unpredictable results.
The solution to this problem is to precondition the signal before presentation to the analyzer. The way to do this is in the form of a low pass filter that will soften the fast edges while passing the primary audio signal intact. The best approach to this filter is a passive design, as it will handle the fast edges properly, is relatively low cost, and will not require power. A well-designed passive filter will also not compromise the audio signal passing through it by adding noise or distortion as an active design might. A passive design is also necessary to handle the wide dynamic range of signals that are normally presented to an analyzer. Any active filter has a limited amplitude range of operation. Thus, an active design would require an input attenuator and variable gain to accommodate the wide range of possible signal levels to be analyzed. Including such capability in an active design would be virtually duplicating the complete front end of the analyzer, an impractical approach from a cost and application point of view.
The Audio Precision AUX-0025 Switching Amplifier Measurement Filter is a dual channel multi-pole LRC passive filter that provides the necessary attenuation of out-of-band signals and reduces the steepness of the fast switching edges .We currently don't have this device and must conduct any measurements on digital and switch mode amplifiers with no passive filter. As a result, our distortion and SNR measurements may yield worse results than representative of the products under test. While the MPS-1 is not a "digital" or classic switch mode amplifier, we do believe its unique power supply topology, wide bandwidth design, and limiter circuit may affect the results of our tests and may decisively revist these measurements in the future with the addition of a LPF. Therefore we caution the reader to understand the limitations of our tests but will publish them nonetheless for a basis of which to compare once we have the proper filter networks in place. We also caution readers of other review publications to be mindful of this when evaluating their published measurements of these types of amplifier topologies, especially Class D / PWM type designs.
Power & Distortion Measurements
The frequency response was smooth and extended to a -3dB point of 125kHz.
At 1 watt into 8 ohms, the MPS-1 exhibited impressively low distortion
(9.178+89.984)dBv = 99.2dBv or 100*alog^-1(-99.2/20) = 0.0011%
At 175wpc into an 8ohm load, we see the distortion rise, but again the amp switching effects contaminate the distortion measurement slightly. FFT analysis reveals (50.546+31.462)dBv = 82dBv or 100*alog^-1(-82/20) = 0.008% THD.
At 255wpc into a 4ohm load, we see the distortion rise, but despite the lack of the proper LPF implemented on our test setup, we still observe commendable results. FFT analysis reveals (30.097+47.28)dBv = 77.37dBv or 100*alog^-1(-54.85/20) = 0.014% THD.
I was able to achieve a maximum unclipped power of 175wpc into 8 ohms with Vin=1.65V Vout=37.39V unbalanced and about 255wpc into a 4 ohm load with Vin = 1.42V and Vout = 32V. Once I exceeded these measurements, the clamping circuit kicked in rounding off the signal and dramatically increasing distortion. Based on this, I would rate this amp to be a 175wpc into 8 ohms and 270wpc into 4 ohm, not 200wpc / 300wpc 8/4 ohm respectively like Emotiva rated in their literature.
When I informed Emotiva on my findings, they suggested that my test conditions were preventing me from achieving their published ratings due to line voltage sag. I respectively disagree since I was only testing one channel and monitored my line voltage to be a constant 124Vrms throughout the entire test. In addition, this was the first amp I tested that did not meet its power specification. However, please note this is also the first amplifier I tested with a thoughtful limiter circuit which does make it a bit tricky to accurately test.
I asked Emotiva to furnish their power measurements for this review as displayed below.
From Emotiva Engineering Labs:
According to Emotiva, this measurement was taken at 4 ohms (1% purely resistive load) and with a line voltage of 120V ac (1KHz input frequency). In this graph you can see they achieved 300W at 1% THD. Though my contention here was they were using an automated script in Audio Precision to find maximum power of the amplifier whereas my tests were using continuous tones. What likely occurred here is their measurement captured the power of the amplifier before the clamping circuit kicked in and limited the power output. This is a common measurement some publications use which I don't really like for that very reason. I like seeing steady state full bandwidth power response to see what the amp is really doing.
Despite that I was unable to measure their claimed power, I still believe maximum power ratings should be specified as unclipped ( usually 0.1% THD or less) not 1% which can clearly be audible and seen on an oscilloscope as a clipped waveform. For more information on our abilities to discern audible distortion, we recommend reading our Human Hearing Article Series.
In any event, I never found the Emotiva MPS-1 to be lacking in power or dynamics. I suggest not getting too caught up in power games since in most cases only a fraction of the amplifier power will ever be utilized under real world listening conditions.
