Recently I was working on 845 tube based SET amplifier. When I did listening test everything sounded good until I tried to play one of DSD music file I have. I did play this file many times before using my solid state amp (Bryston 4BSST). It sounded good without any unusual artefacts. But when I used my new SET amp, I immediately started to hear additional noise - more like low level static above usual noise floor. Initially I thought that there is something wrong with my DSD playback chain. So I switched back to my main amplifier, and sound was as clean as before. I suspected that I missed some kind of oscillation (even as square wave test looked Ok). So amplifier was back on my workbench. I did sweep tone test from audio frequencies to 10 MHz and didn't see any unusual resonances. This amplifier has no global feedback, frequency response is down about 1dB at 20kHz and there is very little output above 70kHz. So I brought my DSD DAC into my lab and started experiments with playback of real music files while watching amplifier using oscilloscope and spectrum analyzer. What I found was astounding - when I played DSD file there was as much output from DAC in 100kHz range as in audio frequencies. Obviously amplifier was not able to pass it, but what apparently happened was some of that noise (which is inherent for DSD processing) was wrapped back into audio frequencies due to low slew rate and thus extremely high IMD level at high frequencies. It manifested itself as low level static. I did test on the same music track downconverted to 24/88 PCM, and there was no unusual noise. There also was no noise when I played few 24/192 PCM tracks I have. But in that latter case I saw almost nothing on oscilloscope when I set 50kHz high pass filter in it, meaning that there was almost no real content in these files between 50 and 100kHz.
From all the above I see that many tube amplifiers are not suitable for direct DSD playback due to increased noise/static from transponded high frequency noise. SACD format is dead. But considering that we see more and more DSD downloads available from various sources one need to keep in mind that not all amplifiers can be used in DSD playback chain. As a work around analogue low pass filter should be used between DSD DAC and tube power amplifier to suppress ultrasonic noise. My DAC has several selectable internal filters for DSD, but lowest cutoff frequency is 90kHz and they it is not too sharp, so most of ultrasonic noise still passes through.
From all the above I see that many tube amplifiers are not suitable for direct DSD playback due to increased noise/static from transponded high frequency noise. SACD format is dead. But considering that we see more and more DSD downloads available from various sources one need to keep in mind that not all amplifiers can be used in DSD playback chain. As a work around analogue low pass filter should be used between DSD DAC and tube power amplifier to suppress ultrasonic noise. My DAC has several selectable internal filters for DSD, but lowest cutoff frequency is 90kHz and they it is not too sharp, so most of ultrasonic noise still passes through.
And any properly designed amplifier should have a low pass filter on the input to remove any possibility of slew rate limiting.
Also it is not so that the described problem can exist only in tube amplifiers, there are many solid state amplifiers with limited frequency response and missing input filters as well as tube amplifiers with wider frequency range than many solid state amplifiers.
Also it is not so that the described problem can exist only in tube amplifiers, there are many solid state amplifiers with limited frequency response and missing input filters as well as tube amplifiers with wider frequency range than many solid state amplifiers.
I agree. In my case DAC is Teac UD-501 and it does have filter - at 90kHz though. This is obviously not enough for tube amplifier. It looks like I need to add filter at amplifier to protect it from ultrasonic noise. I suspect here is a rare case when interconnect cables can make a real difference too. I am not surprised now that designers of Oppo 105 decided to apply filter with -3dB point at 35kHz and almost complete suppression above 50kHz. This probably should be a gold standard in DSD DAC design.
Last edited:
Valve amplifiers are generally more capable of coping with unwanted HF energy than many SS amps. You may have hit a bad combination of poor DAC and poor amp. Note that SET amps may have high levels of second-order distortion - this will be very good at transforming ultrasonic noise into audible noise.
You were working with one amp and only referenced a problem with one DSD file. This kind generalized, "revelations," are rampant in these forums.
As for the DSD file, what was the source material for creating the file? If it was a CD rip then you also also have what's known as a "dithering" component ( intentionally added noise to randomized the samples) that would not be audible from a CD player but would be part of the DSD file.
SACD isn't dead at all. Pleanty of great old and new stuff available for anyone looking for it. I have 2 SACD players and a still growing collection. If you cannot guarantee your DSD file source material isn't from a good HD master, then you shouldn't expect anything but garbage in and garbage out. And if it was an "up-converted" MP3, then... Please!
Try looking at a standard definition video on a HD TV. It is actually a worse picture than if viewed on an old SD TV.
And you've made a perfect case for using feedback. Nasties generated in the later stages of an amp getting cancelled out, better FR and a host of other betters at the expense of a few Db.
Last edited:
The source of DSD file was SACD. In this case it was "A distortion of love" by Patricia Barber. Of cause I tried other DSD files after that - similar problem with many. On some DSD files their audible noise level is high enough to mask additional static. As I mentioned already conversion to high resolution PCM, which implies 30kHz low pass filter, eliminates a problem. Using tone controls in my preamp to reduce treble helps too (and when I increase treble it gets worse). I will continue testing using different combination of DACs and amplifiers.
Resampled 2 channel analog from the SACD player to a new ADC .dsf sample file? Was it a downmixed surround sound from a Hybrid SACD?
So you have a SACD ISO burned to DVD that will play on a PS3 that then outputs a DSD stream to your Teac DAC?
No again. I used PS3 with ripping software to extract ISO image from real SACD. Then I used Foobar2000 to extract individual tracks into separate files using DoP packing plugin. Then resulting files were played using Squeezebox Touch with USB output enabled. Teac UD-501 DAC is connected to Squeezebox via USB. DAC recognizes audio stream as DSD64. When I did tests I used the same DoP files played via TEAC software player on Windows. Results were the same. Then I use Foobar2000 to create DFF files out of ISO image. Played them via Teac software player with exactly the same result. ISO image burned on blank DVD-R disk is recognized and played by Oppo player as SACD disk. So there was no error in digital conversion anywhere.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.