WolfX700 Measurements of Topping Pre90/Ext90 Preamplifier/ Extender

[Veri]

I believe the Oppo's volume control operates inside the DAC at 64-bit float. I am so not a technical expert on this, but my impression was that this implementation reduces the SNR loss from reducing the volume. I could be wrong, though.

My own experience, and my opinion from what I've read, is that if your Oppo-->power amp chain is giving you sufficient gain as-is, there's no benefit to putting a preamp in between them.

If gain is not an issue (and you don't need additional source inputs beyond what the Oppo offers), then adding a preamp becomes basically just substituting the preamp's volume control for the Oppo's.

Agreed. Oppo's implementation is probably way good already. Only get a pre-amp if you want some feature of it (remote control relay attenuation, inputs/outputs, ... otherwise probably not worth it. IMO.

 

[hmscott]

Theoretically, there could be a difference. With digital volume controls, as you turn down the volume SNR typically drops 1:1 with it. Turn it down 30 dB and you also get 30 dB lower SNR. Analog volume controls can be worse than this, or better.

For example, if I read Amir's Pre90 measurements correctly, it has an excellent SNR of 126 dB at 4 V output. But when you turn its volume down to 50 mV output, its SNR is 85 dB. In other words, reduce volume by 38 dB (4 V to 50 mV) and SNR drops 41 dB. Compared to other headphone amps and DACs, 85 dB at 50 mV is good but not great. The UDP-205 may be better or worse, which is to say, putting the Pre90 in between might make a difference, for better or for worse. However, in this case I doubt it would make a difference since the Pre90 happens to be close to 1:1 as you lower the volume.

True, the SNR is reduced - but if you have no noise injected, reducing it with good shielded cables, identify and remove ground loops, remove EMI / RFI induced through proximity to power cables, isolate devices emitting RFI / EMI, and if you can get all of those noise sources under control, there is only the inherent noise of the circuit - and the Topping products have that well under control.

With my most sensitive IEM's, and an ear tuned for noise reduction with my D90 MQA / A90, mitigations in place that have stopped the USB power wire borne noise (GPU/PC), use shielded RCA / XLR cables, I don't hear any noise floor at +0.0dB, -6.0dB, -9.0dB, -15.0dB - all the way to -30dB, there is no noise when I then bump up the A90 gain from L to M to H using D90 MQA in Pre mode.

As long as you have the noise sources under control you won't have any noise to increase.

 

[MRC01]

True, the SNR is reduced - but ... As long as you have the noise sources under control you won't have any noise to increase.

That depends on the application. For driving a power amp, I agree that noise is not likely to be audible even when you turn down the volume. Typical power amp gain and speaker sensitivity is such that you won't turn it down far enough to expose the noise level. However, noise could become audible in other applications.

 

[mocenigo]

Theoretically, there could be a difference. With digital volume controls, as you turn down the volume SNR typically drops 1:1 with it. Turn it down 30 dB and you also get 30 dB lower SNR. Analog volume controls can be worse than this, or better.

Not entirely correct. A digital volume control is just an array of resistors that get switched in or out of the signal path, but this switch is controlled digitally. An analog volume control (here I am referring to a potentiometer) is just a variable resistor on the signal path. So they do exactly the same thing and fundamentally in the same way, with the "digital" scheme suffering less from mechanical deterioration of the sweepers, and the analog volume control potentially performing better initially due to larger contact areas. So, yes, the analog volume controls could be worse or better, but never by a significant margin, until they start deteriorating significantly.

Of course, when a digitally controlled volume control deteriorates, it usually breaks for good, whereas an analog control still works to some extent.

And the IC "digital" attenuators are the same thing as the discrete ones, but one should pick those that offer the option to bypass the internal opamp for increasing the volume, or get those without.

 

[MRC01]

Not entirely correct. A digital volume control is just an array of resistors that get switched in or out of the signal path, but this switch is controlled digitally. An analog volume control (here I am referring to a potentiometer) is just a variable resistor on the signal path. ...

There is confusion of terms here. By "digital volume control" I mean attenuation in the digital domain before D/A conversion. In this case, the analog stage has a fixed gain ratio so when you turn down the volume, you're shrinking the signal against a constant noise so the SNR drops 1:1 with volume.

What you call "digital attenuation" sounds like a stepped attenuator in the analog domain. The main advantage this has over a potentiometer is not noise, but perfect channel balance at all volumes. Noise-wise, in both cases, the SNR also drops roughly 1:1 with volume as you turn it down.

Another alternative is an analog volume control that changes the resistors in the gain-feedback loop, so adjusting the volume changes the gain ratio. Low volume = high neg feedback, low gain, low noise. This seems like it would provide the lowest noise at low to moderate volume settings, because you're attenuating the entire signal + noise that comes out of the DAC, instead of attenuating the signal against a fixed gain.

 

[mocenigo]

There is confusion of terms here. By "digital volume control" I mean attenuation in the digital domain before D/A conversion. In this case, the analog stage has a fixed gain ratio so when you turn down the volume, you're shrinking the signal against a constant noise so the SNR drops 1:1 with volume.

What you call "digital attenuation" sounds like a stepped attenuator in the analog domain. The main advantage this has over a potentiometer is not noise, but perfect channel balance at all volumes. Noise-wise, in both cases, the SNR also drops roughly 1:1 with volume as you turn it down.

Another alternative is an analog volume control that changes the resistors in the gain-feedback loop, so adjusting the volume changes the gain ratio. Low volume = high neg feedback, low gain, low noise. This seems like it would provide the lowest noise at low to moderate volume settings, because you're attenuating the entire signal + noise that comes out of the DAC, instead of attenuating the signal against a fixed gain.

Ah ok, I misunderstood what you meant. Of course attenuation in the digital domain is a different thing entirely. I prefer it, and I trust that AKM in the AK4499 has done a good arithmetic job.

 

[MRC01]

Ah ok, I misunderstood what you meant. Of course attenuation in the digital domain is a different thing entirely. I prefer it, and I trust that AKM in the AK4499 has done a good arithmetic job.

Right, but when they do a perfect job, SNR drops 1:1 with volume as you turn it down from full scale to your listening level. Since SNR is typically measured at full scale, look for a device that has about 30 dB more than you need. In that context, a 120 dB SNR is really only about 85-90 when you turn it down to actual listening levels. So that "120 dB" spec is a bit misleading; it's not the overkill it sounds like, but depending on the application, may be barely enough to be transparent at actual listening levels.

 

[Red@]

The only devices that may have issues are the ones with small DC blocking caps at balanced output.
The ones that don't have caps will have zero issue no matter what output impedance is.
The ones that have 100uF or higher electrolytic caps will have zero issue.
Output opamp is not a concern, jellybean 5532 will drive 600ohm with zero issue. Let alone current lm4562, opa1612 etc.

Hello @JohnYang1997 can you please inform me if the pre90 would suffer from xlr conection from the Gustard x26pro ?

 

[Red@]

The op-amps has a little bias current in A class, about 0.5mA. At 3Vp-p and 2k ohm load, we have 1.5mA, so, will work in AB class. With 10kohm will remain in A class.
The sound is not just SINAD score for static measurements. It's about the transients, it's about the power supply response in fact. At low input impedance the sound will be transparent but in the same time is a "flat" because the power supply will not be just as fast.

excellent explanation. Thank you. I was hoping to buy a pre90 since my x26pro has a DC coupled output stage with 100ohm xlr ouput. But I didn’t not realise the impact on transient response.
very helpful thank you.

 

[Veri]

excellent explanation. Thank you.

I'm really not so sure that explanation is correct or factual, at all.

 

[gvl]

I'm really not so sure that explanation is correct or factual, at all.

But nevertheless it helped to realize the impact on transient response!

 

[Red@]

But nevertheless it helped to realize the impact on transient response!

Exactly