I built a 4-way tube crossover for a friend. He’s experiencing noise like frying eggs and bacon when playing SACDs at a similar level to the music. The noise is present during quiet passages too. A large collection of SACDs is unlistenable.
There’s no noise when playing standard CDs.
Below is a block diagram of the system.
When playing SACDs there’s no noise when:
Is that what we’re hearing?
If the output of the SACD/CD player is causing audio band problems in the Absolare tube preamp:
Can anyone explain what the problem is and what we might do to fix it please?
There’s no noise when playing standard CDs.
Below is a block diagram of the system.
When playing SACDs there’s no noise when:
- The tube crossover is replaced with an Accuphase digital crossover
- The SACD/CD player is connected directly to the tube crossover.
Is that what we’re hearing?
If the output of the SACD/CD player is causing audio band problems in the Absolare tube preamp:
- how does the Accuphase digital crossover then get rid of those artefacts in the audio band?
- why doesn’t it do the same to the tube crossover which has the same architecture (grounded cathode amplifier stage followed by cathode followers)?
Can anyone explain what the problem is and what we might do to fix it please?
Attachments
Input stage
Hi. It has always been noisy.
As for the input impedance of the tube crossover, that is an excellent question. The balanced input goes through a Lundahl balanced to single ended transformer which feeds each of the 4 crossover sections.
Each crossover section has a Transformer Volume Control followed by a gain stage, a cathode follower buffer and the cathode follower filters.
I'll get back to you with the impedance of the Lundahl transformer.
Hi. It has always been noisy.
As for the input impedance of the tube crossover, that is an excellent question. The balanced input goes through a Lundahl balanced to single ended transformer which feeds each of the 4 crossover sections.
Each crossover section has a Transformer Volume Control followed by a gain stage, a cathode follower buffer and the cathode follower filters.
I'll get back to you with the impedance of the Lundahl transformer.
Attachments
Transformer Volume Control impedance
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]I'm guessing my problem might be that the Line Input Transformer is incorrectly terminated.[/FONT][/FONT][FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]Maybe supersonic frequencies from the DAC are being "demodulated into the base-band which means, causes noise in the audio band" by the incorrectly terminated transmission line consisting of the preamp source driving a balanced cable to the line input transformer.[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]The Line Input Transformer is connected to 4 Transformer Volume Controls which in turn are connected to the grids of triodes (very high impedance).
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]According to a source I found[/FONT][/FONT][FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]"the impedance a TVC presents to a source [/FONT][/FONT][FONT=Arial, Helvetica, sans-serif]is the reflected load in parallel with the inductive reactance of the autoformer. When set for no attenuation, the reflected load is simply whatever follows the volume control in the signal chain. However, that load is cut in half (resistance is doubled) for every 3dB of attenuation. At -20dB the reflected load is reduced by a factor of 100. At -40dB it is reduced by a factor of 10,000 which is 10,000 times the resistance!"[/FONT][FONT=Arial, Helvetica, sans-serif]
[/FONT]
[FONT=Arial, Helvetica, sans-serif]So, currently there is a very high impedance connected to the Line Input Transformers. I'll try putting 10k across them and see if that helps.
[/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]I'm guessing my problem might be that the Line Input Transformer is incorrectly terminated.[/FONT][/FONT][FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]Maybe supersonic frequencies from the DAC are being "demodulated into the base-band which means, causes noise in the audio band" by the incorrectly terminated transmission line consisting of the preamp source driving a balanced cable to the line input transformer.[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]The Line Input Transformer is connected to 4 Transformer Volume Controls which in turn are connected to the grids of triodes (very high impedance).
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]According to a source I found[/FONT][/FONT][FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]
[/FONT][/FONT]
[FONT=Arial, Helvetica, sans-serif][FONT=Arial, Helvetica, sans-serif]"the impedance a TVC presents to a source [/FONT][/FONT][FONT=Arial, Helvetica, sans-serif]is the reflected load in parallel with the inductive reactance of the autoformer. When set for no attenuation, the reflected load is simply whatever follows the volume control in the signal chain. However, that load is cut in half (resistance is doubled) for every 3dB of attenuation. At -20dB the reflected load is reduced by a factor of 100. At -40dB it is reduced by a factor of 10,000 which is 10,000 times the resistance!"[/FONT][FONT=Arial, Helvetica, sans-serif]
[/FONT]
[FONT=Arial, Helvetica, sans-serif]So, currently there is a very high impedance connected to the Line Input Transformers. I'll try putting 10k across them and see if that helps.
[/FONT]
But I thought the two crossovers were ok. I was wondering if the relatively high input impedance of the Absolare was an issue.
The crossovers could be OK. Certainly, there's no problem when feeding the SACD player directly into the crossover.
Thing is, if the problem is with the preamp that doesn't explain why when we feed it directly into a digital crossover, the problem disappears. If the pre-amp is putting noise in the audio band the digital crossover can't remove it.
If the preamp is passing supersonic noise from the SACD player straight through, the filters before the ADC in the digital crossover will certainly remove it.
Thing is, if the problem is with the preamp that doesn't explain why when we feed it directly into a digital crossover, the problem disappears. If the pre-amp is putting noise in the audio band the digital crossover can't remove it.
If the preamp is passing supersonic noise from the SACD player straight through, the filters before the ADC in the digital crossover will certainly remove it.
I was thinking more along the line that preamp can't cope with the noise. If the noise folds down into the audio band then it will make through the crossover.
With the same preamp but the Accuphase crossover, the noise is gone. Maybe the combination of ultrasonic noise and an unterminated transformer load causing some sort of current clipping in the preamp?
Does a 1 kohm series resistor between preamp and crossover change anything?
Does a 1 kohm series resistor between preamp and crossover change anything?
A problem within the preamp was my original hypothesis.
"the combination of ultrasonic noise and an unterminated transformer load" causing some sort of problem resulting in the noise we're hearing is my current hypothesis.
I'm going to try terminating the line input transformer correctly. If that doesn't work I'll give your suggestion a try.
"the combination of ultrasonic noise and an unterminated transformer load" causing some sort of problem resulting in the noise we're hearing is my current hypothesis.
I'm going to try terminating the line input transformer correctly. If that doesn't work I'll give your suggestion a try.
A terminating RC network on the input transformer would be my first suggested fix too. Without termination there is very likely a high Q resonance (from the trafo's leakage inductance) in the ultrasonic region where the DSD quantization noise is present, exacerbating the effect of said noise.
That's showing a termination that goes all the way down to DC. What I normally see is a series RC (in other contexts, called a 'Zobel network') which only provides damping where its needed. Which is in the ultrasonic range for a typical audio signal trafo.
Yes, optimization is by eye and ear. But simulation can help a lot in the initial stages if you know the trafo parasitics.
Yes, optimization is by eye and ear. But simulation can help a lot in the initial stages if you know the trafo parasitics.
That 1 kohm series resistor was just meant as an experiment rather than a fix. The extra series resistance aggravates transformer distortion, so it is much better to use a parallel damper network, if you can somehow figure out how to dimension it correctly.
Is your oscilloscope a digital model with a rudimentary FFT spectrum analyser built in? If so, maybe you can just measure the audible and ultrasonic noise along the signal chain to see where it goes wrong.
Is your oscilloscope a digital model with a rudimentary FFT spectrum analyser built in? If so, maybe you can just measure the audible and ultrasonic noise along the signal chain to see where it goes wrong.
No 50kHz fllter on ourput of CD/SACD player/DAC?
Testing has indicated that an unterminated line input transformer is not the problem and narrowed the problem down to the Esoteric CD/SACD Player/DAC and the tube crossover.
I’ve read that DSD noise-shaping moves quantisation noise from the audio band above the audio band and amplifies it.
I’ve also read that “if there is insufficient low-pass filtering at the output of the player, the HF noise can cause problems with some amplifiers, especially tube ones”
I gather this is from intermodulation effects but I'm not sure why tube amps should be more susceptible than solid-state amps.
My current theory is that Esoteric CD player has no 50kHz filter on the output and that the quantisation noise is generating intermodulation noise in the tube crossover's analog filters.
As mentioned, the manufacturer fixed this problem for us with the previous model with a brick-wall filter on the output.
Next step will be to use the FFT spectrum analyser on my oscilloscope to try and capture the HF noise on the output of the SACD player/DAC. If we can do that maybe we can get the manufacturer to fix the problem.
Testing has indicated that an unterminated line input transformer is not the problem and narrowed the problem down to the Esoteric CD/SACD Player/DAC and the tube crossover.
I’ve read that DSD noise-shaping moves quantisation noise from the audio band above the audio band and amplifies it.
I’ve also read that “if there is insufficient low-pass filtering at the output of the player, the HF noise can cause problems with some amplifiers, especially tube ones”
I gather this is from intermodulation effects but I'm not sure why tube amps should be more susceptible than solid-state amps.
My current theory is that Esoteric CD player has no 50kHz filter on the output and that the quantisation noise is generating intermodulation noise in the tube crossover's analog filters.
As mentioned, the manufacturer fixed this problem for us with the previous model with a brick-wall filter on the output.
Next step will be to use the FFT spectrum analyser on my oscilloscope to try and capture the HF noise on the output of the SACD player/DAC. If we can do that maybe we can get the manufacturer to fix the problem.
Last edited: