A little while ago I had a peculiar day in the listening room. It led me to ask: Do preamps pack magic sauce?
First the background: I had brought my CD player project to one of Oslo's HiFi stores to test. The box packs 4x full single-ended mono channels (PCM1704, IVC+filter around OPA2132.) My digital filter sits in an FPGA that can send +/- L/R to either DAC channel. Two of the channels pass through a PGA2311 from TI/BB. I had the whole thing configured to give me one single-ended stereo output with volume control and one straight one.
Setting the PGA2311 at 0dB (2Vrms) we were unable to distinguish between it and the straight output. To get a good listening level from the System Audio speakers, the Hegel preamp attenuated both by about 30dB before the Hegel power amp.
However, when we connected the PGA2311 output directly to the Hegel power amp input and set the volume down by about 30dB the sound stage was completely altered. Words like "flat", "boring", "dull", "undynamic" etc. abounded. The guys in the store told me that this is not the first time a signal source with variable output falls short of a conventional preamp.
Why did we observe this? Do preamps come with magic sauce that adds "correctness" and dynamics to the music? I don't believe so. Or do preamps have a dynamic range far better than any single-chip volume control can sport? That is way more likely. The PGA2311 has a dynamic range of about 120dB, and playing at -30dB is something it is probably not too good at.
But how do preamps achieve this? Better PSRR? Extreme all-over dynamic range and low noise? Or a dynamic range that tracks the volume control so that the range is independant of the volume setting?
I have never designed a preamp myself, but I'm eager to understand the basic principles and how to apply them to my CD player project.
Cheers,
Børge
First the background: I had brought my CD player project to one of Oslo's HiFi stores to test. The box packs 4x full single-ended mono channels (PCM1704, IVC+filter around OPA2132.) My digital filter sits in an FPGA that can send +/- L/R to either DAC channel. Two of the channels pass through a PGA2311 from TI/BB. I had the whole thing configured to give me one single-ended stereo output with volume control and one straight one.
Setting the PGA2311 at 0dB (2Vrms) we were unable to distinguish between it and the straight output. To get a good listening level from the System Audio speakers, the Hegel preamp attenuated both by about 30dB before the Hegel power amp.
However, when we connected the PGA2311 output directly to the Hegel power amp input and set the volume down by about 30dB the sound stage was completely altered. Words like "flat", "boring", "dull", "undynamic" etc. abounded. The guys in the store told me that this is not the first time a signal source with variable output falls short of a conventional preamp.
Why did we observe this? Do preamps come with magic sauce that adds "correctness" and dynamics to the music? I don't believe so. Or do preamps have a dynamic range far better than any single-chip volume control can sport? That is way more likely. The PGA2311 has a dynamic range of about 120dB, and playing at -30dB is something it is probably not too good at.
But how do preamps achieve this? Better PSRR? Extreme all-over dynamic range and low noise? Or a dynamic range that tracks the volume control so that the range is independant of the volume setting?
I have never designed a preamp myself, but I'm eager to understand the basic principles and how to apply them to my CD player project.
Cheers,
Børge
after a quick look at the pga2311 data sheet I see that thd+noise is only graphed at 0 dB atten - not a test of attenuation linearity at all
this leaves open the question of mux switch Ron and parasitic C modulation with signal amplitude at lower attenuation levels
I would hope BB did a good job but this data sheet doesn't tell you much more than that the op amp isn't too awful
also the op amp buffer "load capacitance stability" spec is for only 100 pF load C - I would add a 50-100 series R to prevent a few feet of RCA cable from exceeding that Cload
other possibilities include intentional frequency response shaping by the pre and unintentional shaping from preamp vs pga buffer out Z interacting with cable and amp input Z - possibly a added buffer external to the pga, designed to drive cable load C would help
this leaves open the question of mux switch Ron and parasitic C modulation with signal amplitude at lower attenuation levels
I would hope BB did a good job but this data sheet doesn't tell you much more than that the op amp isn't too awful
also the op amp buffer "load capacitance stability" spec is for only 100 pF load C - I would add a 50-100 series R to prevent a few feet of RCA cable from exceeding that Cload
other possibilities include intentional frequency response shaping by the pre and unintentional shaping from preamp vs pga buffer out Z interacting with cable and amp input Z - possibly a added buffer external to the pga, designed to drive cable load C would help
I agree, it's a pity they didn't show the figures at lower signal levels. With the good 0dBm audio test, this pretty much means I can use the PGA2311 in the range where it makes sense to attenuate digitally.
Next I'll be looking at different topologies and checking if I should rather build a more conventional preamp with a digitally controlled resistor for the volume control.
Do you know of designs like that having been discussed on the forum?
The volume control isn't just nice-to-have. In my player there is an FPGA and four analog outputs. So what I want to build is a digital X-over. But not too many peopole (me included) have four-channel pres.
Børge
Next I'll be looking at different topologies and checking if I should rather build a more conventional preamp with a digitally controlled resistor for the volume control.
Do you know of designs like that having been discussed on the forum?
The volume control isn't just nice-to-have. In my player there is an FPGA and four analog outputs. So what I want to build is a digital X-over. But not too many peopole (me included) have four-channel pres.
Børge
borges
I am not familiar with any of the equipment you mention. However it is not unusual for tuners and other front end equipment to sound better when driven through a pre amp.
Without knowing your equipment it is difficult to be specific. However the normal reason in my experience for the pre amp "improving" the sound is simply that the pre amp offers a better match of input impedance for the front end equipment. The pre amp is then designed as a line driver to deliver the signal through the cable and to offer an output impedance that a power amp can match. I am guessing your pre amp is providing the right impedances for your front end equipment and for your power amp.
As an example of this a quad fm4 tuner really needs a pre amp with an impedance over 20k ( and preferably 50k ) If the input impedance of the pre amp is lower than 20k the fm4 tuner looses base and what base there is becomes muffled.
I also sometimes find that the pre amp can seem to add dynamic range if the power amp requires more voltage on the input signal than the front end equipment can offer.
Don
I am not familiar with any of the equipment you mention. However it is not unusual for tuners and other front end equipment to sound better when driven through a pre amp.
Without knowing your equipment it is difficult to be specific. However the normal reason in my experience for the pre amp "improving" the sound is simply that the pre amp offers a better match of input impedance for the front end equipment. The pre amp is then designed as a line driver to deliver the signal through the cable and to offer an output impedance that a power amp can match. I am guessing your pre amp is providing the right impedances for your front end equipment and for your power amp.
As an example of this a quad fm4 tuner really needs a pre amp with an impedance over 20k ( and preferably 50k ) If the input impedance of the pre amp is lower than 20k the fm4 tuner looses base and what base there is becomes muffled.
I also sometimes find that the pre amp can seem to add dynamic range if the power amp requires more voltage on the input signal than the front end equipment can offer.
Don
Don,
the PGA2311 datasheet states that it can drive a 600Ohm load directly. And since we are dealing with audio frequencies and relatively short cables here, transmission line effects and matched impedance shouldn't really matter.
The strange thing is that the PGA2311 seems to do a great job at 0dB. We could not hear any difference at all, and that was on pretty good equipment.
It was at -30dB that the dynamic range (presumably) was poorer. From that I believe that signal swing is a non-issue.
Do you know of volume attenuation / pre amp primers out there?
Børge
the PGA2311 datasheet states that it can drive a 600Ohm load directly. And since we are dealing with audio frequencies and relatively short cables here, transmission line effects and matched impedance shouldn't really matter.
The strange thing is that the PGA2311 seems to do a great job at 0dB. We could not hear any difference at all, and that was on pretty good equipment.
It was at -30dB that the dynamic range (presumably) was poorer. From that I believe that signal swing is a non-issue.
Do you know of volume attenuation / pre amp primers out there?
Børge
Borge
For volume attenuation I usually just use a series resistor and parallel resistor on the pre amp input.
Don
For volume attenuation I usually just use a series resistor and parallel resistor on the pre amp input.
Don
since the pcm1704 is 24 bits I would just attenuate digitally in the fpga, as long as you keep 24 bits of result there will be no practical "bit loss" (I hate that sloppy reasoning!)
Well, I had to go down by 30dB to get to the "comfort level" of the power amp and speakers, that does eat a fair chunk out of the 1704's dynamic range. And the voltage level in my active filter becomes unpractically low.
A few dBs I'd gladly add or remove digitally, but not 30!
Børge
A few dBs I'd gladly add or remove digitally, but not 30!
Børge
- Status
- Not open for further replies.