Has anyone tried the new Spatial Audio Raven pre amp?
- Thread starter FT251
- Start date
Designed by Lynn Olson; I'm using very similar (the same?!) circuit for output stage of my new DAC, just with the step down OPT (finemet core OPT)
A great reading, Raven preamp design description included: http://www.nutshellhifi.com/triode1.html
Few years ago with the support of my friend I completed his Amity mono blocks. Amazing amps!
I have zero doubt Spatial Audio Raven is a great preamp!
A great reading, Raven preamp design description included: http://www.nutshellhifi.com/triode1.html
Few years ago with the support of my friend I completed his Amity mono blocks. Amazing amps!
I have zero doubt Spatial Audio Raven is a great preamp!
Attachments
Last edited:
Hi folks, yes, Don Sachs and I designed the current Spatial Audio Revelation Series: the Raven preamp and Blackbird PP300B power amp. I gotta step out now, but I'll answer any question you folks have about differences between the 25-year-old articles in the Web page and what's out now.
The picture is a few years old, showing the Karna four-box amplifiers, and the Don Sachs DS-1 preamp. The original Karna amplifiers, shown here up and running, had insanely complex grounding thanks to the four-box format. Don managed to get them down to two chassis with greatly improved power supplies, which were not available when the original Karna amplifiers were built.
In the last couple of years, Don and I designed the Revelation Series of electronics for Spatial Audio, based on the Raven and Karna amps on the Nutshell Hifi web page. The original power amps used 5687/7044/7119 input tubes, which were dictated by the off-the-shelf interstage transformer I used.
We now have custom-designed interstage transformers that allow us to use any tube we want, with the current lineup being 6SN7 (balanced) input, triode-connected KT88 (balanced Class A) drivers, and push-pull 300B's running in Class A, delivering about 20~24 watts from 20 Hz to 40 kHz, with zero global or local feedback.
I can assure the readers that IT-coupled Class A 300B's in balanced mode sound very different from 300B SET, and also very different from classical push-pull pentode amplifiers. Passive phase splitting (via transformer) removes most of the "push-pull sound" of old-school Dynaco or Mullard circuits, (which have active phase splitters). Think SET directness and immediacy, but without the 2nd-harmonic colorations.
In the last couple of years, Don and I designed the Revelation Series of electronics for Spatial Audio, based on the Raven and Karna amps on the Nutshell Hifi web page. The original power amps used 5687/7044/7119 input tubes, which were dictated by the off-the-shelf interstage transformer I used.
We now have custom-designed interstage transformers that allow us to use any tube we want, with the current lineup being 6SN7 (balanced) input, triode-connected KT88 (balanced Class A) drivers, and push-pull 300B's running in Class A, delivering about 20~24 watts from 20 Hz to 40 kHz, with zero global or local feedback.
I can assure the readers that IT-coupled Class A 300B's in balanced mode sound very different from 300B SET, and also very different from classical push-pull pentode amplifiers. Passive phase splitting (via transformer) removes most of the "push-pull sound" of old-school Dynaco or Mullard circuits, (which have active phase splitters). Think SET directness and immediacy, but without the 2nd-harmonic colorations.
Last edited:
The usual knock about interstage transformers is questionable transient response, but modern computer optimization allows us to have performance not possible in the Fifties or Sixties. Here's an unretouched square-wave photo from an analog scope (taken by Don Sachs):
Last edited:
But it's not all smooth sailing, or these things would be much more common. One of the nicest things about SET amplifiers is the varying current demand imposed on the B+ power supply is simply the DC polarizing voltage plus the music signal itself. This means you could listen directly to the B+ high voltage supply if you had an isolating transformer ... you'd hear some hum and buzz, of course, but music would be there, and reasonably undistorted, too.
This is NOT true for a balanced circuit. In principle, the demand on the B+ should be constant. This is the usual assumption in engineering textbooks ... a Class A balanced amplifier has steady demand on the supply with no variation. However, there's a giant assumption built in there ... that vacuum tubes are distortionless. Unfortunately, all amplifying devices have distortion, whether they are triodes, tetrodes, pentodes, bipolar transistors, or MOSFETs.
The distortion present in the balanced amplifying elements means they behave something like a balanced diode bridge, or AM modulator. The audio signal that appears on the B+ supply, though reduced 30 to 35 dB due to common-mode rejection, is quite distorted. This means the power-supply distortion of the output stage MUST NOT appear in the preceding stages. This mechanism, of distortion transfer from output to input, is another reason that push-pull amplifiers can have an irritating and fatiguing distortion.
The solution is full isolation of the output stage from the input and driver sections. The B+ regulators in the Blackbird have isolation of greater than -130 dB from input to output ... and that's just one regulator. For modulation noise to traverse from 300B regulation to the input 6SN7, it has to pass through a cascade of three regulators: the 300B regulator, the driver section regulator, and the VR tubes that shunt-regulate the input tubes. In addition, the input tubes operate in balanced mode, which provides another 30 to 35 dB of isolation.
By contrast, a typical push-pull amplifier of vintage design will only have simple RC (resistor-capacitor) filtering between input and output sections. The RC filter (mostly) removes hum and buzz, but when the amp clips, the whole amp clips all at once, like a guitar amp. That multisection clipping is great for a guitar amp, but not good at all for a hifi amp. A typical "audiophile" amp of modern vintage will have regulation for the input section, but not much else, so when the output section folds down, the regulator will fall out of regulation with a sharp transient.
The Blackbird (and the Karna) have full isolation from stage to stage. Each stage even has its own star ground. When the output section changes modes from Class A to Class AB (with loads lower than 2 ohms), or Class AB2 (gross overdrive), there is no effect on the input or the driver. They behave like the 2-watt driver amp that they are. Surprisingly, this has an effect at much lower levels, including background listening. Not quite sure why that is.
This is NOT true for a balanced circuit. In principle, the demand on the B+ should be constant. This is the usual assumption in engineering textbooks ... a Class A balanced amplifier has steady demand on the supply with no variation. However, there's a giant assumption built in there ... that vacuum tubes are distortionless. Unfortunately, all amplifying devices have distortion, whether they are triodes, tetrodes, pentodes, bipolar transistors, or MOSFETs.
The distortion present in the balanced amplifying elements means they behave something like a balanced diode bridge, or AM modulator. The audio signal that appears on the B+ supply, though reduced 30 to 35 dB due to common-mode rejection, is quite distorted. This means the power-supply distortion of the output stage MUST NOT appear in the preceding stages. This mechanism, of distortion transfer from output to input, is another reason that push-pull amplifiers can have an irritating and fatiguing distortion.
The solution is full isolation of the output stage from the input and driver sections. The B+ regulators in the Blackbird have isolation of greater than -130 dB from input to output ... and that's just one regulator. For modulation noise to traverse from 300B regulation to the input 6SN7, it has to pass through a cascade of three regulators: the 300B regulator, the driver section regulator, and the VR tubes that shunt-regulate the input tubes. In addition, the input tubes operate in balanced mode, which provides another 30 to 35 dB of isolation.
By contrast, a typical push-pull amplifier of vintage design will only have simple RC (resistor-capacitor) filtering between input and output sections. The RC filter (mostly) removes hum and buzz, but when the amp clips, the whole amp clips all at once, like a guitar amp. That multisection clipping is great for a guitar amp, but not good at all for a hifi amp. A typical "audiophile" amp of modern vintage will have regulation for the input section, but not much else, so when the output section folds down, the regulator will fall out of regulation with a sharp transient.
The Blackbird (and the Karna) have full isolation from stage to stage. Each stage even has its own star ground. When the output section changes modes from Class A to Class AB (with loads lower than 2 ohms), or Class AB2 (gross overdrive), there is no effect on the input or the driver. They behave like the 2-watt driver amp that they are. Surprisingly, this has an effect at much lower levels, including background listening. Not quite sure why that is.
Last edited:
That's why the photo of the B+ power supplies of the original Karna amplifiers (on the bottom of the rack) has that "Tokyo at Night" appearance. Each B+ power supply section has its own damper diodes, which are single, not dual diodes. Technically, they are hybrid bridges, with high-voltage Schottky diodes on the ground side, and damper diodes on the high side.
The Blackbird uses isolated windings on a common power transformer, followed by an array of B+ regulators, which are electrostatically screened from the audio side of the chassis. In this picture, the audio section is on the left, while the B+ section is on the right, with an isolation screen between them. The VR tubes (on the right) are providing 210 volts of shunt regulation for the 6SN7 input tubes.
The Blackbird uses isolated windings on a common power transformer, followed by an array of B+ regulators, which are electrostatically screened from the audio side of the chassis. In this picture, the audio section is on the left, while the B+ section is on the right, with an isolation screen between them. The VR tubes (on the right) are providing 210 volts of shunt regulation for the 6SN7 input tubes.
Attachments
Last edited:
A most interesting summation of some of prescient points of your design Lynn , I would comment at the risk of stating the obvious that whilst one might get away with utilising decent Iron , that the higher the quality of IT and OPT transformers the more one benefits from this type of circuit particularly when doubling or quadrupling up on traditional SET output tubes such as 45 , 2A3 , 300B etcetera in the quest for more output power , which as you say brings its own set of problems , a friend in Japan builds something along these lines , however not quite as advanced as the Blackbird , which I would love to hear some time.
The goal was to get distortion of each stage as low as possible while avoiding feedback, either locally or globally. This means the signal path is a one-pass straight through, with no secondary loops for feedback or DC servo stabilization. Since there's no feedforward correction (a good idea with transistors, or especially with MOSFETs), inherent distortion must be as low as possible, and intermodulation between stages minimized.
The lowest distortion requires matched pairs operating symmetrically, with errors summed out between stages, which is the primary function of the interstage transformers. Conventional amplifiers pass on 2nd, 4th, 6th, etc. harmonic errors from stage to stage via RC coupling. This crossmodulation results in multiplication of harmonics, first described by Norman Crowhurst in the late Fifties.
In the Raven, Blackbird, and Karna, even order harmonics are summed out in the primary of each transformer at each stage. Odd harmonics remain, but they are 30 to 35 dB below the even order ones, and the plates of the preceding stages see an 2~3X easier load than RC coupling, so distortion is further reduced. In practice, residual 2~3% gain imbalance between the pairs results in even-order harmonics having the same magnitude as the odd order harmonics, so for once, a little bit of dynamic imbalance works in our favor.
Don and I tried active loads for the input tubes, but we were not satisfied that the current sources were not adding traces of solid-state distortion. Building a clean current source is not trivial, and they add an element of unreliability if the high voltage transistors break down.
P.S. For those of you who are traveling, the Raven and Blackbirds will be exhibited at the Spatial Audio room at the upcoming Dallas show. I won't be there, but Don Sachs will be, along with the Spatial team. The latest Spatial speakers will be on demo, with Lampizator providing the DACs.
The lowest distortion requires matched pairs operating symmetrically, with errors summed out between stages, which is the primary function of the interstage transformers. Conventional amplifiers pass on 2nd, 4th, 6th, etc. harmonic errors from stage to stage via RC coupling. This crossmodulation results in multiplication of harmonics, first described by Norman Crowhurst in the late Fifties.
In the Raven, Blackbird, and Karna, even order harmonics are summed out in the primary of each transformer at each stage. Odd harmonics remain, but they are 30 to 35 dB below the even order ones, and the plates of the preceding stages see an 2~3X easier load than RC coupling, so distortion is further reduced. In practice, residual 2~3% gain imbalance between the pairs results in even-order harmonics having the same magnitude as the odd order harmonics, so for once, a little bit of dynamic imbalance works in our favor.
Don and I tried active loads for the input tubes, but we were not satisfied that the current sources were not adding traces of solid-state distortion. Building a clean current source is not trivial, and they add an element of unreliability if the high voltage transistors break down.
P.S. For those of you who are traveling, the Raven and Blackbirds will be exhibited at the Spatial Audio room at the upcoming Dallas show. I won't be there, but Don Sachs will be, along with the Spatial team. The latest Spatial speakers will be on demo, with Lampizator providing the DACs.
Last edited:
In practice, the hum-injection circuits for the 45 drivers proved unnecessary and were removed from the amplifiers. Note local sub-star grounding, which has been retained in the Blackbird amplifiers. (Drawing error: the phase/mute switch should be on the primary side of the input transformer, not the secondary, where a little DC from the grids would go through it.)
The current-production Blackbird amplifiers use precision high-isolation regulators for each section, along with balanced, floating filament regulation. The tube lineup is now 6SN7 input, triode-connected KT88 driver, and 300B output, with Monolith power and output transformers.
The current-production Blackbird amplifiers use precision high-isolation regulators for each section, along with balanced, floating filament regulation. The tube lineup is now 6SN7 input, triode-connected KT88 driver, and 300B output, with Monolith power and output transformers.
Last edited:
For anyone who will be at the SW Audio Fest in Dallas in a couple of weeks, I will be there in the Spatial Audio Lab room and we will have the Raven and Blackbirds driving their open baffle speakers. So if you want to hear the circuits in action and happen to be in Dallas.... Drop by and listen and have a chat.
cheers from Canada
cheers from Canada
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |