Hello to everyone.
This is a review and test of the Vincent SAV-C1 multichannel preamplifier and homecinema processor. It was released in 2003 and cost about $2300 at the time. A new version having a marginally modified main printed circuit board (PCB) and an added steel plate behind the front display and control PCB was released in 2004. The device was available until 2009.
Part I: Presentation
As I understand it, Vincent is the brand name that was used by the Chinese hifi manufacturer Shengya for international sale.
As you can see, the SAV-C1 has an attractive styling, with all-metal knobs and buttons (and it came with an all aluminum remote control unit). The input selector and volume control knobs activate some encoders which have a pleasant feeling. The overall fit, finish and built quality are very good, although the adjustments of parts are not at Accuphase level.
The back of the unit reveals plenty of connectivity:
The SAV-C1 gets 3 coaxial digital inputs and 1 Toslink optical digital input and is able to decode perceptually coded format up to 7 channels (Dolby Digital, Digital EX, Pro-Logic I et II, DTS, DTS-ES). It can also decode HDCDs.
On the analogue side, the SAV-C1 has a 7 channels RCA unbalanced analogue inputs (front, left, center, surround left, surround center, surround right and subwoofer), 6 stereo RCA unbalanced inputs and 1 stereo XLR balanced input. There are 1 set of 7 channels RCA unbalanced outputs (with double front channel outputs and 2 subwoofer outputs) and 1 set of 7 channels XLR balanced outputs.
There is also some video switching capabilities for composite, S-Video and RGB signals.
The digital processing and video switching are now outdated. The main purpose of this review is to test the SAV-C1 as a multichannel analogue preamplifier.
I was not able to find any service manual for this device, but I was able to locate some schematics for its little brother, the SAV-C2, which is essentially the same as the SAV-C1 save for the balanced input/outputs. With that comparison point in mind and a bit of reverse engineering, I was able to figure out the overall design and signal path.
The digital circuits rely on what appears to be an off-the-shelf small processing card on which we find ubiquitous Cirrus Logic chips: a CS493263 decoder, a CS8415 digital receiver, a CS4228 24 bits/96kHz surround sound CODEC (2 A/D converter channels and 6 D/A converter channels) and a CS4340 24 bits/96 kHz stereo D/A converter. The analogue conditioning built in this small processing card is nevertheless not used. Vincent has designed its own op-amp based post-DAC low pass filters and analogue output stages. The front left and right, center and subwoofer channels get the "deluxe" treatment with Burr Brown OPA2604, the surround left, right and center channels use NE5532. There is also an interesting additional circuitry on all channels. I will come to them shortly.
From then on, we find the input selection circuitry, followed by the analogue signal path. This is where things get complicated. Indeed, as is the case with many homecinema products, all the channels do not get the same treatment. In fact, there are no less than 4 different signal paths.
Notwithstanding the balanced input made of a Burr Brown OPA2134 used as differential receivers, the 2 front left and right channels (which are shared with the front channels of the multichannel analogue input) get the luxury of relay switching and an input buffer stage made of another OPA2134 op-amp. Then follow the volume control ICs, themselves followed by the gain stages made of a Burr Brown OPA2604 per channel, one of the two op-amps being used to invert the signal to get a balanced signal. But this balanced signal is not used directly at the output: the hot and cold signals are further buffered by another dual op-amp per channel: an OPA2134. The unbalanced outputs use only the hot output from this buffer stage.
The center channel get the same treatment, save for the input buffer stage: the signal coming from the digital to analogue circuit or the multichannel center analogue input drives directly the volume control IC.
The surround front, left and center channels are switched by complementary metal-oxide semiconductor (CMOS) switches without input buffering to drive the volume control ICs. The gain stages are the same as above, except that the balanced or unbalanced signals are directly taken at the output of each OPA2604, without buffering.
Last, the subwoofer channel is switched by relay, without buffer to drive the volume control chip, and the output stage is made of NE5532 instead of OPA2604.
Sounds overly complicated? Yes it is. I wonder if the design of one single properly designed signal path with a single-type op-amp inventory would not have been cheaper.
Anyway, the volume control ICs are 2 reasonably specified six-channels Mitsubishi M62446AFP. Each of these chips has also two additional signal processing paths to form bass and treble tone control circuits (which can be defeated). This capability is put in good use in the SAV-C1, which provides tone controls for the front left, right, center and surround center channels.
There is one last interesting circuit to be mentioned: besides the OPA2604 or NE5532 in the post-DAC filters and analogue gain stages, there are accompanying TL082 dual op-amps. These op-amps are used on each channel in a sort of noise-gate which brings the gain down about 20 dB 2 to 3 seconds after the signal drops under a threshold. This circuit makes sense in a homecinema product. When there is no sound effects or music and only dialogues are left, this circuit will cut any hiss from the unused channels of a homecinema installation during the quietest passages. How will this circuit behave from a performance standpoint? We'll see next...
Part II: Measurements
All measurements were taken with an Audio Precision System One+DSP SYS222A (nominal input impedance: 100 kohms per phase in parallel with 170 pF). Unless otherwise noted, all measurements were performed using standard Audio Precision tests.
A. Balanced in / Balanced out:
I will begin with a dashboard of the two front channels, at unity gain, 4 V in/4 V out balanced, to follow the practice set by Amirm. Please keep in mind that the volume control ICs have 1 dB step. So it is not possible to get at the output exactly the same voltage than at the input:
Well, this is not a pretty picture. Not sure what's going on here, but the performance is severely restricted. Floating the output of the generator does change nothing. Lowering the input to a standard 500 mV RMS level improves things:
At 2 V RMS, level we get that:
Here is a sweep of output THD+N versus input level amplitude in absolute value (0 dBV=1 V RMS):
So, at unity gain, the preamplifier works better at input level close to the consumer standard (-10 dBV=316 mV RMS). Interestingly, the curve goes down at about -53 dBV input level on the right channels: this is the effect of the noise gate circuit, which brings the level of the noise down. Similar behavior will be visible on some other sweeps below.
Other parameters I have measured in function of input level includes intermodulation test signals with the SMPTE (low frequency+high frequency tones), CCIF (twin high frequency tones) and, yes, DIM (Mati Ottala's coined Dynamic Intermodulation Distortion, ie high frequency sine + filtered square wave):
I have no experience with the DIM standard test signal. Only comparisons with other device will show if that measurement has some relevance. Anyway, it seems obvious that the SAV-C1 was designed with consumer level in mind, and not pro audio levels.
Next, we will see sweeps of THD+N vs frequency at a standard 500 mV input level and in two successive analyzing bandwidth: 20 kHz and 80 kHz:
There is no dependency to frequency, as shown by the flat curves, which is mostly noise-dominated.
With the usual definition of clipping as 1 % THD+N, the balanced outputs can put out about 11 V RMS, but as is shown above, they severely distort way before that point is reached. In any case, the maximum signal to noise ratio (in a 22 Hz to 20 kHz bandwidth) I was able to get referred to 11 V with the volume control set at maximum are 105.6 dB (left) and 105.4 dB (right) unweighted and 107.9 dBA (left) and 107.5 dBA (right) A-weighted. At least, the balanced in/balanced out signal path delivers in terms of signal to noise ratio.
The Audio Precision System One is equipped to test the common mode rejection ratio (CMRR) of balanced inputs. I measured about 43.3 dB (left) and 42.3 dB (right) across a 50 kHz bandwidth. Nothing to write home about, according to me. These results are mostly the product of imbalances of the resistor values at the input differential receivers. They seems to me consistent with the use of 1% tolerance resistors across the board.
B. Balanced in / Unbalanced out:
Overall, the performances are better than the balanced input to balanced output signal path.
With the usual definition of clipping as 1 % THD+N, the unbalanced outputs can put out about 6 V RMS, but as is shown above, they significantly distort way before that point is reached. In any case, the maximum signal to noise ratio I was able to get referred to 6 V with the volume control set at maximum are 100.5 dB (left) and 100.9 dB (right) unweighted and 104.1 dBA (left) and 104 dBA (right) A-weighted.
C. Unbalanced in / Balanced out:
That way of operation may be the favored one by those who own many analogue output sources, including a multichannel disc player or DAC, and use power amplifiers at remote positions from the preamplifier or active speakers. So, it make sense to test that configuration. Let's also take the opportunity to test a different signal path than the main channels to compare. Here we go:
There is a 0.5 dB difference in level in favor of the surround channel, which is unfortunate. There is a bit more power supply noise on the unbuffered balanced surround output than on the buffered balanced output of the front channel, but the overall THD+N remain the same.
I made two measurements of the same channels in another situation than unity gain, in this case with a 300 mV input (a typical input level for consumer products) and the volume control set to get 2 V output (16.5 dB of gain). I have decreased 1 dB the level on the surround channel with the volume control to get closer to the 2 V output target. That explains that the level is 0.5 dB higher on the front channel on the following dashboards:
The numbers are better when the preamp is driven at low level and a bit of gain is applied through the volume control.
The swept input level THD+N analysis clearly shows that the two types of signal path behave differently from each other, especially at high input levels:
The same discrepancies in the performance level between the two types of signal path can also be seen with the intermodulation tests:
On the other hand, the THD+N sweeps vs frequency are less revealing:
(To be continued in message #2)
This is a review and test of the Vincent SAV-C1 multichannel preamplifier and homecinema processor. It was released in 2003 and cost about $2300 at the time. A new version having a marginally modified main printed circuit board (PCB) and an added steel plate behind the front display and control PCB was released in 2004. The device was available until 2009.
Part I: Presentation
As I understand it, Vincent is the brand name that was used by the Chinese hifi manufacturer Shengya for international sale.
As you can see, the SAV-C1 has an attractive styling, with all-metal knobs and buttons (and it came with an all aluminum remote control unit). The input selector and volume control knobs activate some encoders which have a pleasant feeling. The overall fit, finish and built quality are very good, although the adjustments of parts are not at Accuphase level.
The back of the unit reveals plenty of connectivity:
The SAV-C1 gets 3 coaxial digital inputs and 1 Toslink optical digital input and is able to decode perceptually coded format up to 7 channels (Dolby Digital, Digital EX, Pro-Logic I et II, DTS, DTS-ES). It can also decode HDCDs.
On the analogue side, the SAV-C1 has a 7 channels RCA unbalanced analogue inputs (front, left, center, surround left, surround center, surround right and subwoofer), 6 stereo RCA unbalanced inputs and 1 stereo XLR balanced input. There are 1 set of 7 channels RCA unbalanced outputs (with double front channel outputs and 2 subwoofer outputs) and 1 set of 7 channels XLR balanced outputs.
There is also some video switching capabilities for composite, S-Video and RGB signals.
The digital processing and video switching are now outdated. The main purpose of this review is to test the SAV-C1 as a multichannel analogue preamplifier.
I was not able to find any service manual for this device, but I was able to locate some schematics for its little brother, the SAV-C2, which is essentially the same as the SAV-C1 save for the balanced input/outputs. With that comparison point in mind and a bit of reverse engineering, I was able to figure out the overall design and signal path.
The digital circuits rely on what appears to be an off-the-shelf small processing card on which we find ubiquitous Cirrus Logic chips: a CS493263 decoder, a CS8415 digital receiver, a CS4228 24 bits/96kHz surround sound CODEC (2 A/D converter channels and 6 D/A converter channels) and a CS4340 24 bits/96 kHz stereo D/A converter. The analogue conditioning built in this small processing card is nevertheless not used. Vincent has designed its own op-amp based post-DAC low pass filters and analogue output stages. The front left and right, center and subwoofer channels get the "deluxe" treatment with Burr Brown OPA2604, the surround left, right and center channels use NE5532. There is also an interesting additional circuitry on all channels. I will come to them shortly.
From then on, we find the input selection circuitry, followed by the analogue signal path. This is where things get complicated. Indeed, as is the case with many homecinema products, all the channels do not get the same treatment. In fact, there are no less than 4 different signal paths.
Notwithstanding the balanced input made of a Burr Brown OPA2134 used as differential receivers, the 2 front left and right channels (which are shared with the front channels of the multichannel analogue input) get the luxury of relay switching and an input buffer stage made of another OPA2134 op-amp. Then follow the volume control ICs, themselves followed by the gain stages made of a Burr Brown OPA2604 per channel, one of the two op-amps being used to invert the signal to get a balanced signal. But this balanced signal is not used directly at the output: the hot and cold signals are further buffered by another dual op-amp per channel: an OPA2134. The unbalanced outputs use only the hot output from this buffer stage.
The center channel get the same treatment, save for the input buffer stage: the signal coming from the digital to analogue circuit or the multichannel center analogue input drives directly the volume control IC.
The surround front, left and center channels are switched by complementary metal-oxide semiconductor (CMOS) switches without input buffering to drive the volume control ICs. The gain stages are the same as above, except that the balanced or unbalanced signals are directly taken at the output of each OPA2604, without buffering.
Last, the subwoofer channel is switched by relay, without buffer to drive the volume control chip, and the output stage is made of NE5532 instead of OPA2604.
Sounds overly complicated? Yes it is. I wonder if the design of one single properly designed signal path with a single-type op-amp inventory would not have been cheaper.
Anyway, the volume control ICs are 2 reasonably specified six-channels Mitsubishi M62446AFP. Each of these chips has also two additional signal processing paths to form bass and treble tone control circuits (which can be defeated). This capability is put in good use in the SAV-C1, which provides tone controls for the front left, right, center and surround center channels.
There is one last interesting circuit to be mentioned: besides the OPA2604 or NE5532 in the post-DAC filters and analogue gain stages, there are accompanying TL082 dual op-amps. These op-amps are used on each channel in a sort of noise-gate which brings the gain down about 20 dB 2 to 3 seconds after the signal drops under a threshold. This circuit makes sense in a homecinema product. When there is no sound effects or music and only dialogues are left, this circuit will cut any hiss from the unused channels of a homecinema installation during the quietest passages. How will this circuit behave from a performance standpoint? We'll see next...
Part II: Measurements
All measurements were taken with an Audio Precision System One+DSP SYS222A (nominal input impedance: 100 kohms per phase in parallel with 170 pF). Unless otherwise noted, all measurements were performed using standard Audio Precision tests.
A. Balanced in / Balanced out:
I will begin with a dashboard of the two front channels, at unity gain, 4 V in/4 V out balanced, to follow the practice set by Amirm. Please keep in mind that the volume control ICs have 1 dB step. So it is not possible to get at the output exactly the same voltage than at the input:
Well, this is not a pretty picture. Not sure what's going on here, but the performance is severely restricted. Floating the output of the generator does change nothing. Lowering the input to a standard 500 mV RMS level improves things:
At 2 V RMS, level we get that:
Here is a sweep of output THD+N versus input level amplitude in absolute value (0 dBV=1 V RMS):
So, at unity gain, the preamplifier works better at input level close to the consumer standard (-10 dBV=316 mV RMS). Interestingly, the curve goes down at about -53 dBV input level on the right channels: this is the effect of the noise gate circuit, which brings the level of the noise down. Similar behavior will be visible on some other sweeps below.
Other parameters I have measured in function of input level includes intermodulation test signals with the SMPTE (low frequency+high frequency tones), CCIF (twin high frequency tones) and, yes, DIM (Mati Ottala's coined Dynamic Intermodulation Distortion, ie high frequency sine + filtered square wave):
I have no experience with the DIM standard test signal. Only comparisons with other device will show if that measurement has some relevance. Anyway, it seems obvious that the SAV-C1 was designed with consumer level in mind, and not pro audio levels.
Next, we will see sweeps of THD+N vs frequency at a standard 500 mV input level and in two successive analyzing bandwidth: 20 kHz and 80 kHz:
There is no dependency to frequency, as shown by the flat curves, which is mostly noise-dominated.
With the usual definition of clipping as 1 % THD+N, the balanced outputs can put out about 11 V RMS, but as is shown above, they severely distort way before that point is reached. In any case, the maximum signal to noise ratio (in a 22 Hz to 20 kHz bandwidth) I was able to get referred to 11 V with the volume control set at maximum are 105.6 dB (left) and 105.4 dB (right) unweighted and 107.9 dBA (left) and 107.5 dBA (right) A-weighted. At least, the balanced in/balanced out signal path delivers in terms of signal to noise ratio.
The Audio Precision System One is equipped to test the common mode rejection ratio (CMRR) of balanced inputs. I measured about 43.3 dB (left) and 42.3 dB (right) across a 50 kHz bandwidth. Nothing to write home about, according to me. These results are mostly the product of imbalances of the resistor values at the input differential receivers. They seems to me consistent with the use of 1% tolerance resistors across the board.
B. Balanced in / Unbalanced out:
Overall, the performances are better than the balanced input to balanced output signal path.
With the usual definition of clipping as 1 % THD+N, the unbalanced outputs can put out about 6 V RMS, but as is shown above, they significantly distort way before that point is reached. In any case, the maximum signal to noise ratio I was able to get referred to 6 V with the volume control set at maximum are 100.5 dB (left) and 100.9 dB (right) unweighted and 104.1 dBA (left) and 104 dBA (right) A-weighted.
C. Unbalanced in / Balanced out:
That way of operation may be the favored one by those who own many analogue output sources, including a multichannel disc player or DAC, and use power amplifiers at remote positions from the preamplifier or active speakers. So, it make sense to test that configuration. Let's also take the opportunity to test a different signal path than the main channels to compare. Here we go:
There is a 0.5 dB difference in level in favor of the surround channel, which is unfortunate. There is a bit more power supply noise on the unbuffered balanced surround output than on the buffered balanced output of the front channel, but the overall THD+N remain the same.
I made two measurements of the same channels in another situation than unity gain, in this case with a 300 mV input (a typical input level for consumer products) and the volume control set to get 2 V output (16.5 dB of gain). I have decreased 1 dB the level on the surround channel with the volume control to get closer to the 2 V output target. That explains that the level is 0.5 dB higher on the front channel on the following dashboards:
The numbers are better when the preamp is driven at low level and a bit of gain is applied through the volume control.
The swept input level THD+N analysis clearly shows that the two types of signal path behave differently from each other, especially at high input levels:
The same discrepancies in the performance level between the two types of signal path can also be seen with the intermodulation tests:
On the other hand, the THD+N sweeps vs frequency are less revealing:
(To be continued in message #2)
Last edited:
) and later also 4 Dolby Atmos Speakers (ev. KEF Meta Q8), reflective... don't laugh... but it's the only way in this room. 5.1.4 in the End
