Vladimir Shushurin, president and
chief designer of Lamm Audio Laboratory, says that the M2.1 hybrid
mono amplifier is one of his best creations. It’s a
Class-A/AB design rated at 200 watts in Class AB into an 8- or 4-ohm
load, and it is said to handle impedances as low as 1 ohm.
Lamm also makes the Model M1.1, which
is the same size as the M2.1 but is $400 more expensive.
Although the M1.1 delivers only 100 watts into an 8- or 4-ohm load,
it does so in pure Class A. (And, like the M2.1, it can drive
loads down to 1 ohm.) The DM1, a dual-mono version of the M2.1
rated at 125 watts/channel into 8 ohms, will soon be available for
about $8,700. Another Lamm Audio Laboratory component is the
L1 line preamp ($6,990). Under the Lamm Industries banner are
two mono tube amps, the push-pull ML1 ($9,990 each) and the
single-ended ML2 ($14,645 each).
M2.1 switches to Class AB at a
relatively high power level. This ensures that it will be in
Class A during most of your music listening, even if you use
low-impedance speakers. But it also gives the amp very high power
dissipation – more than 200 watts – at idle. The resulting
heat is handled by six large, finned sinks, three on each side.
The M2.1 is big and beautiful, with
nothing on its front panel except lettering for the company name and
model number, a red power-on LED, and a pair of rack handles.
The on/off switch is on the rear (presumably, Shushurin wants you to
leave the amp warmed up, despite its heat dissipation).
Another rear-panel switch sets the M2.1 for a high-impedance load (8
to 16 ohms) or a low-impedance load (1 to 6 ohms). Also on the
rear panel are a balanced input, accessible via a Neutrik XLR
connector or a pair of Esoteric Audio phono jacks (also usable for
inverting or noninverting input from an unbalanced source); two
pairs of Esoteric Audio five-way binding posts for speaker
connections; two remote-control jacks, which enable daisy-chaining
two or more units together for common control; an IEC AC cord
connector; an AC line fuse; and a binding post for ground
connection. A pair of handles on rear panel, a nice touch,
greatly assists you in handling and moving the amplifier.
Most of the circuitry, including the
tube (a 6922 dual triode) that marks this amp as a hybrid, is on a
large board attached to a horizontal sub-chassis. I noted many
high-quality parts, including Dale metal-film and PRC wire-wound
resistors, Electrocube and Roederstein film capacitors, and Cornell
Dubilier switching-grade electrolytic capacitors. On the underside
of the sub-chassis are a pair of 39,000-microfarad/75-volt main
filter capacitors, a power-supply and system-control board, and the
power transformer. That transformer, a 1,700-volt/ampere
toroidal unit made by Plitron, is isolated in a metal sub-enclosure
that measures 3 ¾ x 7 x 7 inches. Construction and wiring
quality is first-rate. An especially elegant touch is the use
of Lemo CAMAC connectors for the cables that connect the input jacks
to the main circuit board.
Measurements
Total harmonic
distortion plus noise (THD + N) as a function of power into 8-, 4-,
and 2-ohm loads is plotted in Fig. 3. Figure 4 shows SMPTE IM
distortion for the same load conditions and for a 4-ohm load with
the amp’s rear-panel loading switch set for high impedance,
instead of the recommended setting.
In Fig. 5, THD + N is
plotted versus frequency for different power levels; note how little
it rises at high frequencies, a desirable but very rare
characteristic. A spectrum analysis of the harmonic distortion
residue of a 1-kHz signal, at a power level of 10 watts into 8 ohms,
is presented in Fig. 6.
Commendable is the
absence of any higher harmonics above the third. All in all,
the M2.1’s distortion performance is very good, especially in view
of the low amount of feedback employed.
Common-mode rejection ratio (CMRR)
for balanced input is shown in Fig. 7 for both of the amps I tested.
The curve shapes may differ, but each amplifier did well on this
test. That’s particularly true at high frequencies, where
many amplifiers’ CMRRs decrease.
The M2.1’s damping factor is
reasonably constant with frequency. You can see this clearly
in Fig. 8, but it can also be inferred from Fig. 1, which
demonstrates that output level at all frequencies drops uniformly as
the load impedance decreases.
Dynamic power with an 8-ohm load was
233 watts (corresponding to clipping headroom of 0.66 dB) and did
not sag at all during the 10-millisecond test period. With
lower load impedances, there was some sag over the burst interval,
however: Output went from 264 to 253 watts into a 4-ohm load, from
484 to 462 watts into a 2-ohm load, and from 800 to 684 watts into a
1-ohm load. Dynamic headroom at the beginning of the
tone-burst test interval was 1.2, 0.83, and 1.2 dB, respectively,
for the three loads. Clipping power at about 1% distortion was
280 watts for an 8-ohm load, 310 watts for a 4-ohm load, and 500
watts for a 2-ohm load, corresponding to clipping headroom of 1.5,
1.9, and 1 dB.
Voltage gain was 31.8 dB, with
negligible difference between unbalanced and balanced input.
Input sensitivity for 1 watt into 8 ohms was 72.7 millivolts.
Output noise for unbalanced input was 655.9 microvolts, wideband,
and 223.7 microvolts, A-weighted; for balanced input, the results
were 655.3 and 253.4 microvolts, respectively. Input impedance was
40 kilohms at 1 kHz. With the M2.1 warned up enough to
stabilize, DC offset at the output was within 10 millivolts of zero.
When it was cold, AC line draw was a little more than 5 amperes and
dropped to 4 amperes once the amp stabilized, which is consistent
with Lamm’s claimed levels for power output in Class A.
But let’s look at that question
another way. In Class-A operation, every output device in an
amplifier is always conducting current, even when there’s no
signal – hence the high idling current. As a signal is
applied, current through one output device will rise while current
through its push-pull or complementary mate will decrease until the
signal reverses polarity. When the current through either
device reaches twice the idling current, natural circuit action
reduces the current through its mate to zero, beyond which point the
amplifier’s operating mode changes from Class A to Class AB.
In the Lamm M2.1, setting the load
switch for high impedance also sets the output stage’s idling
current to 1.5 amperes. (To simplify things, let’s consider
the six paralleled MOS-FETs in each half of the output stage as two
single devices.) If you drive one output device to the point
where it’s conducing 3 amperes of current and the other device’s
current falls to zero, the entire 3 amperes of current will pass
through the 8-ohm load. At that point, peak output voltage
reaches 24 volts (3 amperes x 8 ohms). And 24 volts into 8
ohms equals 36 watts – exactly what Lamm says the M2.1 can deliver
in Class A with an 8-ohm load.
Up to this point, the M2.1’s output
stage is drawing constant power from the power supply. Past
this point, because the power-supply rail voltages are considerably
higher than +24 and –24 volts (+69 and –69 volts, in fact), the
mode of operation shifts to Class AB but output power continues to
increase up to the point where the output waveform starts to clip.
In my lab, with an 8-ohm load and the Lamm amplifier’s load switch
set for high impedance, AC line draw stayed at a steady 4 amperes
until output power reached 30 to 40 watts. At higher output
levels, the line draw started to increase, indicating that the amp
was no longer operating in class A. Although I did not repeat
this test for 4-, 2-, and 1-ohm loads (or for the low setting of the
impedance switch), I am confident the results would bear out
Lamm’s claims of Class-A power output of 36, 18, and 9 watts,
respectively.
Use and Listening Tests
I’ll cut right to the chase: The
Lamm M2.1’s sounded absolutely terrific! Their ability to
resolve sonic detail was amazing. They provided an excellent
sense of air and space and a very specific soundstage –
miraculously, without the edginess that too often accompanies such
fine resolution. Their portrayal of dynamic shadings was
outstanding. When things got loud, they got loud naturally and
didn’t sound strained. I found myself wanting to play things
louder than I usually do with most other amps. Bass extension,
sense of pace and rhythm, and slam were as good as I’ve heard in
my system. Overall, the M2.1s were among the very best
amplifiers I have heard in my listening room. I believe these
sound qualities correlate with the M2.1’s high output-stage idling
current, its low order of distortion production, and the way its
distortion and damping factor stayed relatively constant with
frequency.
The only negatives to the M2.1 are
its high power consumption and high cost. An AC line draw of 8
amperes is definitely not trivial, and $15,000 per pair isn’t
cheap! Ah, but really good things are seldom cheap. The
Lamm amps performed flawlessly in my lab and in my listening room.
If you get the idea that I am enthusiastic about them, you’ve got
it right!
Circuits Highlights
Signals flow into the M2.1 throughout
high-speed, unity-gain buffers designed for video instead of the
usual op-amp input buffers that use feedback to achieve unity gain.
Separate buffers for the input signal’s positive and negative
phases provide a high input impedance and a low output impedance to
drive the circuity in the input stage proper. This is
differential cascode amplifier, consisting of a pair of P-channel J-FETs
whose drain outputs drive the emitters of a pair of PNP transistors.
A multi-transistor current-mirror circuit is used here as a current
source for the Lamm amplifier’s J-FET input devices.
Next in the signal chain is the 6922
dual triode. It serves as a the last voltage amplifier (LVA)
stage, which provides the voltage swing for the output stage.
The grid of the first tiode section is driven from one collector
output of the cascode input amplifier. This section’s plate
is coupled to the second triode section’s cathode and grid
circuit, in an arrangement similar to that called a mu follower.
Even though the output impedance of this stage is fairly low, on the
order of several kilohms, Lamm did not deem it low enough to drive
the input capacitance of the output stage. Therefore, the
output of the LVA tube stage feeds a drive stage, which consists of
an NPN bipolar transistor loaded by an NPN-transistor current
source. Negative feedback (6 to 7 dB of it, according to Lamm)
runs from the output of the driver stage to the inverting input of
the input differential cascode. (There is no negative feedback
around the output stage or overall feedback around the amplifier.)
The feedback loop is a bit out of the
ordinary. The input buffer is coupled to the input differential
cascode via four resistors. Two of the resistors form a
voltage divider from the positive-phase buffer into the noninverting
input of the differential amplifier. The other two resistors
form a feedback voltage divider. One of these (which would be
grounded in a conventional design) connects the negative-phase
buffer to the inverting input and serves as the shunt feedback
resistor; the other resistor, for series feedback, connects the
input with the output of the NPN driver stage.
M2.1’s output stage comprises six
pairs of complementary MOS-FET power transistors. The input to
this stage is capacitor-coupled from the emitter of the driver
transistor via separate capacitors to the gate circuits of the N-
and P–channel MOS-FETs. A voltage divider and bias-spreading
regulator at this stage’s input keep its quiescent idling current
stable and virtually eliminate DC offset in the output signal.
The resistors closest to the MOS-FET gate drive lines in this
voltage divider are bootstrapped from the amplifier’s output so
that the coupling capacitors will see a high input impedance.
This enabled Lamm to use relatively small, high-quality film
capacitors.
A hybrid design like the M2.1, whose
solid-state stages are in full operation while its tube stage warms
up, needs some way to prevent large output swings during start-up.
In the M2.1, an output speaker relay remains open long enough for
the amplifier to reach stability. This relay also opens if the
AC line voltage drops below a set threshold level or if there is
excessive DC at the amplifier output. Furthermore, if the
output relay opens for these reasons, the output stage’s idling
current is cut back to zero. In addition, a resistor in series
with the power transformer’s primary reduces in-rush current when
you power up the M2.1; after a suitable delay, the resistor is
bypassed by another relay.
The M2.1’s front end (everything
except its output stage) is powered by a regulated supply that
delivers +125 volts and –125 volts, via dropping resistors.
This supply also feeds three-terminal regulators to provide
+13
volts and –13 volts to power the input buffers. The output
stage’s supply voltages and quiescent idling current depend on the
setting of the loading switch on the rear panel. When this
switch is set for low impedance (1 to 6 ohms), the voltages are
about +53 volts and –53 volts and idling current is about 2.1
amperes. When it’s set for high impedance (8 to 16 ohms), voltage
increases to +69 volts and –69 volts and quiescent idling current
drops to 1.5 amperes.
An unusual circuit, between the power
cord and the power transformer’s primary, is intended to eliminate
small DC components that may arise from waveform asymmetry in the
incoming AC. Eliminating this DC keeps the toroidal power
transformer in the middle of its linear magnetic operating region,
helping to reduce mechanical hum. This circuit is, effectively, a
full-wave rectifier bridge whose two DC outputs are shorted
together; it eliminates the DC while AC proceeds to the transformer
via two large capacitors that bypass the bridge.
ML3 Signature
