MANUFACTURER'S SPECIFICATIONS
Power Response: 125 watts rms into 8 ohms, both channels driven 20 Hz to 20 kHz at or below 0.2% total harmonic distortion or intermodulation distortion from full power or below 117 V a.c. (typically over 150 watts).
Frequency Response: 10 Hz to 100 kHz .0.5 dB.
Power Bandwidth: 10 Hz to 52 kHz at 0.2% THD.
Damping Factor: 100 or greater. Rise Time: Better than 1.5 S.
Slewing Rate: Greater than 17 V/uS.
Signal-To-Noise Ratio: Greater than 100 dB. Input Impedance: 100k Ohms. Input Sensitivity: 1 V for 125 watts.
Dimensions: 18 1/2 in. W x 12 1/2 in. D x 7 1/2 in. H; 19 in rack mount kit available.
Weight: 58 lbs.
Price: $649.00.
The Model One is the first electronic product of the Epicure Corp. of Newburyport, Mass., which has been producing speakers since 1967. The unit looks to be quite solidly constructed (see Fig. 1) and has a front-panel appearance rather different from most power amplifiers. Nearly all of the front panel area is dark when the amp is turned off; when powered, the word Epicure appears in white at the upper right-hand corner. There are three red lights per channel that spell out left or right Current Limit, Voltage Limit or Hi Temp when these conditions occur; these lights are in two vertical columns.
The front panel has a push-button power switch and three push-button speaker switches, which permit any or all of three separate speaker systems to be selected. Epicure states in their operating manual that it is okay to have three 8-ohm speakers powered at once, as long as the amp isn't driven consistently hard enough to overheat and thereby trip the thermal breakers. There isn't much said about three 4-ohm systems in parallel, but it is doubtful that the amp would drive three inefficient speaker pairs to any truly satisfactory volume without causing current limiting, distortion, and/or thermal overload.
Fig. 1--Interior view of the Epicure Model One.
Fig. 2--Controls and jacks on side panel.
The access to the input and output connections is through the left side of the unit, behind a panel which slides forward and out; see Fig. 2. There are three pairs of output binding post connectors, two input phone jacks, two input level controls (10 k), power cord, line fuse, and two slide switches for testing the overload indicator bulbs. There are no connections on the rear of the unit, as that section of the amp is occupied by the large heat sinks.
The front panel is the mechanical base for the unit, having a large power transformer, two 20,000 uF/75 V filter caps, the seven indicator bulbs, and the speaker and power switches mounted upon it. Extruded corner pieces interconnect the rear panel to the front. The recessed left side panel with the a large power transformer, two 20,000 uF/75V filter capacitors, top and bottom cover plates serve to stiffen and strengthen the unit as a whole. One PC board per channel is mounted on an extended fin about 3 in. off the inner side of each heat sink. Four TO-3 power output devices, two plastic drivers, the bias regulator transistor, and a thermal cutout are all mounted directly on each sink. Each amplifier module thus formed has an RCA phono jack on the PC board and a Molex line connector to allow easy disconnection for servicing. All in all, it's a very nice package.
Circuit Description
A circuit diagram of one channel is shown in Fig. 3. The circuit topology of this amplifier, when compared to other amplifiers, is most like that of a Phase Linear 400 or 700, containing an NPN bipolar differential pair direct coupled to a second NPN bipolar differential pair. One collector of the second differential pair drives an inverting PNP stage that forms the predriver. The collector of the predriver drives the output stage plus base drive line and proceeds downward (schematically speaking) through one transistor bias regulator to the lower base drive line and the collector of a constant current source made of an NPN transistor. The output stage is a quasi-complementary emitter follower, consisting of plastic NPN and PNP drivers and four RCA 410 NPN output devices. When output voltage gets within a few dB of clipping, the first NPN turns off and the bulb driver turns on, illuminating the Voltage Limit indicator.
Fig. 3--Schematic of one channel and power supply.
This circuit is unusual in that it uses emitter degeneration resistors in both differential amplifiers, which reduces the low frequency open-loop gain and improves the linearity of these stages. Further, a capacitor connected between the emitter of the second differential amplifier provides some lead compensation in the overall open-loop shaped response.
VI limiting is accomplished in the usual way, by sensing current and voltage in the output drives and shunting away the driver base drive when voltage and/or current are considered excessive. The base-emitter junction of an NPN transistor is connected across the plus and minus current sensing resistors and is biased to turn a few dB below the point where actual current limiting takes place. When this transistor turns on, it turns on a PNP based to the positive supply voltage drive, which in turn illuminates the Current Limit front panel indicator. Output voltage of the amplifier is divided down to an NPN device based to the minus supply voltage. This device is normally biased on and thus keeps a second NPN turned off. The collector load of this second NPN is the Voltage Limit indicator. These indicators of voltage and current limiting are a good idea and do a good job of informing the user when the amp is being pushed. The circuit constants are arranged so that the relative intensity of the indicators is a good measure of the relative severity of the limiting. For instance, if the Voltage Limit indicator only lights, the speaker impedance is in the normal range of greater than 4 ohms and voltage peaks are being clipped. If the Current Limit indicator flashes, the speaker impedance is probably below 4 ohms and is unusually reactive. When both current and voltage indicators occasionally flash, the amplifier is being most efficiently utilized.
Listening Test
Overall impression of this amplifier is that it sounds about as good as anything previously tested as long as output is kept below clipping. Considerable listening was done with the Model One driving the reviewer's own equalized arrays, which are quite efficient and ordinarily tend to utilize the low end of an amp's power range. The Model One was considered excellent in this use, having very good definition, tight and solid bass response, and an open, airy high end, with a relatively low amount of grit and high frequency edginess or irritation. Measurements aside, this amplifier sounds very, very good.
Further considerable listening was done with Stax SRX Mk-II electrostatic phones, Sony C-500 mikes, and an Ampex MR-70 recorder for monitoring during live recording and later playback. The Model One was considered quite good in this use as well, though not up to a pair of specially modified Marantz 9 tube amps and an experimental class-A transistor design.
More limited listening was done on Magnepan 2167F speakers, which use a great deal more power. The conclusion here was that the Model One was again quite good for this use so long as it wasn't overdriven, a condition which was readily observable with the front-panel indicators. If the amplifier was overdriven hard at high frequencies, then the misbehaviors noted in the measurements taken later do become audible as a harshness not apparent in other amps that clip more cleanly with difficult loads.
Measurements
The Model One was first run at one-third rated power into 8-ohm loads for one hour with a test signal of 1 kHz. The unit operated without thermal cutout but did get hot, as would be expected.
Fig. 4--THD and IM versus power output.
Fig. 5--One-watt frequency response and THD versus power. (Note break in frequency
response curve from 100 Hz to 10 kHz.)
Fig. 6--50-Hz square waves with 8-ohm loads. Top, approximately 200 watts
(scale: 20 V/cm, 5 mS/cm); bottom, 3.12 watts (scale. 5 V/cm, 5 mS/cm).
Fig. 7--10-kHz square waves into, top, 2µF loads and, bottom, 8-ohm loads.
(Scales for both: 5 V/cm, 20 NS/cm.)
Fig. 8--Top, 20-kHz square wave, 8-ohm load; bottom, 20 kHz sine wave, 2 dB
overdrive into 8-ohm load. (Scales for both: 20 V/cm, 10 µS/cm).
Fig. 9--Effects of 1µF load on, top, 20-kHz square wave and, bottom, 20-kHz
sine wave (scales for both: 20V/cm, 10µs/cm).
Voltage gain was measured at 1 kHz into 8-ohm loads and found to be 32.5X and 32.0X for the right and left channels respectively. These gains translate to 30.24 and 30.1 dB. Input sensitivity for rated power into 8-ohm loads is therefore about 1 V rms.
Harmonic distortion at 1 kHz and IM distortion as a function of output power are shown in Fig. 4. (Note that this graph is plotted on log-log paper rather than the semi-log paper used in past reviews. This allows greater resolution in reading the distortion magnitudes.) Measured distortion is satisfactorily low for the Model One. THD vs. power and frequency is shown in Fig. 5 along with the 1-watt frequency response. The 1-watt distortion is dominated by 60-Hz line harmonics, which suggests that the output noise of the Model One isn't quite what it could be. The output noise is shown in Table I below.
The 20 Hz-20 kHz noise is mostly line harmonics, somewhat higher than with other amplifiers, and it could be heard on the reviewer's efficient speaker arrays but wouldn't be bothersome on less efficient speakers. It is believed that the problem is a ground loop in the way the amp is internally grounded and/or in the signal-lead dress near the power transformer of the shielded leads connecting the gain control pots to the amplifier circuit board inputs. It is entirely possible that later production units don't have as much output noise. Át any rate, figures more like 100-200 µV are achievable with other designs and should be the design goal sought after by new designs.
Scope photos of response to various signals and loads are shown in Figs. 6 to 9. Fig. 6 shows the amount of 50 Hz tilt at about 3 and 200 watts, which is a result of the low-frequency rolloff illustrated in Fig. 5. Also, the amplifier doesn't have any change in LF tilt or phase response as a function of power which is both normal and desirable. Fig. 7 shows the 10V p-p, 10 kHz, square-wave response for an 8-ohm and 2µF loads. Risetime is typical of most solid-state amps, although the ringing caused by the RLC buffing network is noticeably less damped than with other amplifiers.
Table I--Output Noise, µV
Fig. 8 is for a large signal 20-kHz square wave into 8 ohms and with a 2 dB overdrive beyond the onset of visual clipping with a 20 kHz sine wave. The slew rate for plus and minus transitions of the square wave isn't equal and recovery from slewing isn't as clean as the very best designs. The measured rise time for a 60 V p-p output into 8 ohms which is fairly close to rated power, was 3.6 µS for the plus transition and 6 KS for the minus transition which works out to a slew rate of 13.3 and 8 V/µS respectively. "Sticking," which was explained in the September issue review of Ampzilla, is rather severe on this unit as the waveform is distorted and visibly nonsinusoidal on the slopes. The last scope picture is for 20-kHz square and sine waves into a 1 pF load. These last pictures aren't as good as other fine amps that have been recently reviewed here, as this amp will not deliver much current (necessary for fast voltage change on a step transition) and its sine wave output "latches" into a nonlinear mode when the amplitude is raised much beyond 20 V rms. (Latching means that the output jumps up to a certain level and wave shape, as shown; when in this state, further input drive doesn't increase the output, and reduction of input amplitude holds this wave shape and level until suddenly the output collapses back to a lower level sine wave.) The particular test signals and loads used here are admittedly quite difficult, and the Model One's performance with them are a departure from what an ideal would be expected of or perfect amplifier. The whole question is, perhaps, academic, since the amplifier sounds extremely good, and that is what really counts in the final analysis. Damping factor was measured at about 100 from 20 Hz to 1 kHz, decreasing smoothly to about 17 at 20 kHz. Power at visual onset of clipping with a 1-kHz sine wave was 200, 175, and 100 watts for 4-, 8-, and 16-ohm loads respectively, with i both channels driven.
In summary, the Epicure Model One is a well-built, rugged appearing unit that sounds very good when not pushed hard into clipping at high frequencies. It would be an excellent choice for use where speaker efficiency is moderate to high and the user doesn't want to have "blow your head off" rock volume.
-Bascom H. King
(adapted from Audio magazine, Nov. 1975)
Also see:
EPI 601 Speaker Systems (Jan. 1973)
Epicure Model EPI-100 Speaker System (Equip. Profile, May 1970)
= = = =