Doshi Audio Evolution Monoblock power amplifier

"Do you swear to tell the truth, the whole truth, and nothing but the truth, so help your filaments?" asked the audiophile judge of the tube.

"Since I am the truth," respondeth the tube, "I have nothing to say that is not already declared by my sound."

"But I must have the truth, and without bias!" proclaimeth the audiophile.

"What good is a tube without bias?" answereth the tube.

Having uttered these words, the tube began to glow. It gloweth with a light so golden that the audiophile's eyes were forced to close and his ears to open. Images did appear before them, images so overwhelming in their brilliance that the audiophile was filled with wonder at the divine power of music, and questions about absolute truth lost their urgency (Revolutions, 3:33-37).

Nishith "Nick" Doshi, 54, whose given, Indian name signifies "midnight" (footnote 1), answered our audiophile judge's question differently, in the form of a parable couched as an email (footnote 2):

A long time ago, in a land far, far away (New Jersey), a young audio engineer built a phono preamplifier for the radio station that employed him. It was a sophisticated beast—solid state, with discrete, class-A drivers and servos, adorned with spectacular specifications. He took it to a friend's house, where it was trounced by a Counterpoint tubed phono preamp that used a slightly modified circuit from the 1954 Radiotron Designer's Handbook.

Lessons were learned. Doshi thought that the simple design got out of the way of the music more than the one with lots of active circuits. While the tubed design was noisier, and the RIAA tolerance did not approach ±0.0000001dB, it was more enjoyable to listen to and sounded closer to what he—Doshi—would have mixed had he made the recording himself.

Thus did he build his own simple tube phono and line preamplifier and discover that if you could use simple circuitry and get away with as low feedback as possible, you had a better chance of getting a musically satisfying result. The choice made the power supply design and decoupling the circuit from vibration and other detriments rather more important, however, which was not an easy trade-off.

"This philosophy has threaded itself through all my products," Doshi told me later on. "I am completely agnostic to the devices themselves. I use tubes and transistors where I find each to be most useful. For example, the Doshi Stereo amplifier is a hybrid design with a solid-state front end and a tubed output stage. It had to be that way to satisfy my design goal: to create a powerful (65W) amplifier using only 2 gain stages and slightly less than 7dB of feedback.

"The Evolution (EVO) monoblocks ended up as an all-tube design because that was the optimal way to get more than 150W of power using the fewest gain stages possible (within my design capabilities). That is all there is to it." Doshi Audio's Evolution (EVO) class-AB monoblock power amplifiers ($41,995/pair) claim to output up to 160Wpc into 4 ohms, the first 120Wpc in class-A. They also attempt to improve on the shortcomings of their predecessors, the V3.0 monos, which I reviewed in the November 2018 issue.

421doshi.3

Doshi explained during one of several phone and email exchanges: "My original goal was to build an amplifier with the least amount of amplification gain stages but that could have enough power and damping factor to drive a real-world loudspeaker with a demanding impedance curve (footnote 3). I wanted it to have a sense of life and effortlessness—an open sound that was quite musical and that you could enjoy listening to for a long period of time.

"I wanted to take what I had learned during the seven years that the V3.0 units were in production and improve [on] what I [had] learned to be the shortcomings of their design and packaging. We stayed very true to all of the original design concepts and philosophy, including the simplicity and elegance of circuitry, ultra–high-quality parts, and the coupling inherent in the ways the monoblock's 14ga, nonmagnetic stainless-steel chassis was designed and constructed. Then we enhanced its visual and performance characteristics by, among other things, enhancing the look of its front panel and top and adding a constrained layer to the top."

Doshi says that the EVO's new machined-aluminum, constrained-layer top provides more vibration isolation for the sensitive audio circuitry of the tube section in the subchassis mounted to it from below. Mechanical vibration may be generated internally by the transformer or externally by loudspeakers, passing trucks, walking humans or beasts, or what have you. A new aluminum top plate replaces the original Corian composite and allows for extra precision in the machining process. "The parts fit together much better, and the absence of mounting holes on top make the monoblocks more visually appealing," he said.

Doshi stiffened the chassis to reduce damage to transformers and circuit boards during transport. He added handles to the rear, which, in addition to making the amps easier to handle, protect input connectors and binding posts. Expensive Pelican shipping cases equipped with wheels and retractable handles replaced the previous generation's double-sided cartons.

Heightened worldwide technical standards mandated changes to power supply efficiency and noise immunity. Doshi took advantage of the opportunity to improve the power supplies in all his units. Implementing passive power correction—"a fancy term for adding capacitance and diodes"—enabled the EVO to utilize wall current morefficiently, increased the efficiency of its linear power supply, and improved its immunity to powerline-transmitted noise. Passive power correction also greatly reduced the noise the amp feeds back into the power line.

Improved capacitor and resistor quality and "slight" circuit changes increased the amplifier's bandwidth. Switching to ClarityCap CMR and CSA copper-matrix coupling capacitors, from British manufacturer ICW, increased dynamic linearity (footnote 4). The custom-designed power transformer, which is fed via Schurter KFA series IEC inlets, "huge" filter capacitors, "soft start," high-voltage circuitry, and five stages of regulation for the various plate, screen, and bias voltages, now has Schottky-type rectifiers. Driver circuitry improvements allowed for higher clipping levels in the output stage, with lower drive impedance, extending peak power and frequency response. Lower feedback increased bandwidth. Less mechanical transformer noise, added regulation in the filament voltage supply, and the addition of a "huge" high-voltage film capacitor to the plate supply (which lowered its impedance) quieted the S/N ratio by 6dB, Doshi said; if true, this means that the new amplifier is twice as quiet as the old one. Doshi added that constrained layer damping, "a bit more" damping to the circuitry, and quieter power supplies and regulators contribute to the sense of a lower noise floor.

For readers who simply must know what each of the four buttons that surround the front-panel screen does, or the exact procedure for manually biasing the amp, please check out the V3.0 review online or check the Doshi website. Suffice to say that, while I can't hammer a nail straight, I could adjust bias easily.

Tweaks for every taste
Doshi recommends adjusting bias to tune the sound of the EVO monoblocks to your liking. 190mV is the default factory bias setting, but Doshi considers anything between 170mV and 220mV acceptable. Going lower can produce excessively lean sound; going higher may fatten things up to a level most desirable to the evil witch who lured Hansel and Gretel into her gingerbread house in the woods. Even if the fabled witch didn't know that excessively high bias can shorten tube life, she hoped to visit that very outcome on H&G.

"We expect that the dealer will do the bulk of bias fine-tuning in the process of setting up the amplifiers," Doshi said. "It changes the way they sound quite dramatically, so why not use it?"

One major difference between my V3.0 and EVO experiences was that, for my V3.0 review, Doshi visited Port Townsend with his photographer daughter and used his own reference tracks to determine the optimal bias setting for my setup. Because the pandemic rendered a repeat visit impossible, determining the best bias setting was now totally up to me. "Dive right in!" proclaimed the witch, as she licked her lips with glee.

Tube rolling, which most definitely affects tone and character, is an approved option. You can switch each monoblock's four new-production Tung-Sol KT150 power tubes to KT88s or KT120s—but only in consultation with Doshi: Using the wrong tubes could void the warranty. Doshi also offers a NOS tube package ($500) that replaces the standard NOS Hytron 12SJ7GT with a Sylvania-made NOS input tube and the GE 12SN7GT driver tubes with two Sylvania NOS "Chrome Dome" tubes from the 1950s and '60s. I experimented with the NOS Tube Package that Doshi sent me and found its sonic ramifications to be of major consequence. More on this below.

One thing you don't have to worry about is which amp supports to use (footnote 5). Thanks to the EVO mono's intrinsic vibration isolation, Doshi recommends placing it directly on the floor. That's good, because the EVO monos did not fit on my amp stands, and my aftermarket footers were all shorter than the EVO's supplied feet.

When I told Doshi that my AudioQuest Niagara 5000 power conditioner has always improved the sound of amplifiers I've plugged into it, he gave the okay to use it with his amplifiers.

I faced a decision whether to use single-ended or balanced (XLR) interconnects. The EVO monos are internally single-ended but accept both types of input. When I reviewed the V3.0, Doshi told me, "Theoretically, the unbalanced input, which bypasses the balanced-to-unbalanced conversion section, should be the purest path. Nonetheless, I included the best balanced-input section I could because some users have made substantial investments in XLR cabling and want to use it rather than changing everything out wholesale."

Back then, I was using the dCS Vivaldi DAC; because it had a volume control and its single-ended and balanced outputs sounded quite similar, I was happy to connect it to the Doshis with single-ended cabling. Since then, however, dCS's Rossini DAC (footnote 6) has replaced the Vivaldi in my system. The Rossini, too, has a volume control and single-ended outputs, and can be connected directly to the monoblocks without a preamp in between. I've found, however, that feeding its output to the balanced-only D'Agostino Momentum HD preamplifier produces larger and weightier images, more and better-controlled bass, heightened color contrasts, and a greater sense of texture. So, I prefer to use the Rossini with the Momentum HD. But since the Momentum HD preamp only has balanced (XLR) connectors, I asked Doshi how I should proceed.


Footnote 1: Doshi explains that his name actually signifies "the start of a new day." Nishith is rooted in the Gujarati language and derives from ancient Sanskrit.

Footnote 2: Edited for dramatic effect.

Footnote 3: Such as, for example, my Wilson Audio Alexia 2s.

Footnote 4: Dynamic linearity is the precision with which the output tracks rapid changes in the input—especially changes in amplitude.

Footnote 5: Editor-in-Chief Jim Austin is no doubt incredulous at the thought that JVS didn't spend at least three days deciding which of 25 different equipment supports sounded best.

Footnote 6: The dCS Vivaldi and Rossini DACs both include volume controls, so a separate preamp is optional.

COMPANY INFO
Doshi Audio LLC
Charlottesville, VA
USA 22901
(917) 952-2758
ARTICLE CONTENTS

COMMENTS
JRT's picture

JA1, I always enjoy reading your measurements page(s), and I have one suggestion for measurements of any amplifier using an output transformer. You are probably well aware that at some combination of lower frequency and higher current the transformer core will begin to saturate and 3rd order harmonic distortion will significantly rise. My suggestion would be to include measurement(s) that better explore the high pass corner of the full power bandwith. That is intended as suggestion, and not any negative criticism of the measurements, explanations and other commentary you are already providing, all worthwhile. Thanks for what you do.

Jack L's picture

Hi

Besides nonlinear hysteresis transfer of the transformer core, high frequency distortion of any output transformer is also caused by its leakage inductance & winding capacitances, which in turn, change the loudspeaker loading impedance & phase angle.

More complicated will be: such leakage inductances of each half of the primary winding of a push-pull output transformer, like this $21,000 monoblock reviewed above, should be as little as possible. Otherwise, this will cause incomplete cancellation of 2nd harmonics, causing undue generation of higher order harmonics.

Such high harmonic distortion may be fed into the secondary winding & then into the loading loudspeaker by capacitive coupling unless there is an effective static shield between the primary & 2nd windings.

Ideally, lower plate resistance of the driving tubes, less will be such distortion at high frequencies. That's why triodes are always preferred to pentode/tetrodes which all get much much high plate resistance than a triode.

That's one of the reasons I always go for single-ended Class A power topology using triodes only - more output transformers friendly + better sound !

Listening is believing

Jack L

tonykaz's picture

Nice literature here, hell of a review, applause +++!

I look to have one of these designs, one day.

Nice Work,

Tony in Venice Florida

a.wayne's picture

This Doshi does measure well for a modern tooby amp ,, interestingly prices are clearly for limited sales numbers , 40K plus is a bit much for an unknown Brand, especially when considering its 3times that of the last VAC integrated tested ..

Regards

Jack L's picture

Hi

Agreed. In fact, some $200/output watt per JA's test report for 8ohm loudspeaker load is very expensive. 8-ohm load s the standard specified load impedance of any loudspeakers, right ?

The bias adjustment of the power tubes: It looks so similar to Dynaco ST-70. I did the same bias adjustment upgrade to my ST-70 by reducing the cathode to signal ground resistor (used to monitor the "bias voltage") from original 15.6R to 2.3R. Such reduction allows me to set the optimum voltage from original 1.56V to 107mV.
Yes, it pretty sensitive to adjust as JA pointed out in his test report. But I found it so helpful to improve the sound quality.

Why? The best sounding bias topology, IMO, is ground cathode. So by reducing the "bias monitoring" cathode resistor to practically MINIMUM would help to improve the sound.

$21,000 for 106Wrms/8-ohm is pretty expensive, my friend.

Jack L

Ortofan's picture

... (and sound quality) of this Doshi amp compare to those of the $11K McIntosh MC1502?
https://www.mcintoshlabs.com/products/amplifiers/MC1502

The MC1502 is the regular production replacement for the limited edition MC2152 - minus the fancier chassis with carbon fiber panels - but with identical specifications and sound quality.
Hi-Fi World tested the MC2152 and found that it could deliver 220W at the onset of clipping and 240W at 1% distortion.
https://www.mcintoshlabs.com/-/media/Files/mcintoshlabs/ReviewAwardDocuments/MC2152-and-C70-Hi-Fi-World-July-2019.ashx

The Part-time Audiophile reviewed the MC1502 and concluded that it provided "true high-end audiophile sound in all its refined glory, without any rare tube types, finicky setup, or usage idiosyncrasies."
https://parttimeaudiophile.com/2021/01/24/mcintosh-labs-mc1502-power-amplifier-review/

X