• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Schiit Freya S Preamplifier Review

majingotan

Major Contributor
Forum Donor
Joined
Feb 13, 2018
Messages
1,511
Likes
1,781
Location
Laguna, Philippines

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
That's correct. I adjusted the volume until I got the desired output, summed THD+N for each channel as expressed in SINAD and plotted it.
So that was in active mode? I wonder how passive performs at these lower volume levels. Given the other measurements, probably even better.
Is passive a standard ladder / voltage divider with consistent 10k input impedance and variable output impedance (so its max output impedance would be 2.5 kOhm at -6 dB where the 2 resistors are equal)?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,370
Likes
234,416
Location
Seattle Area
So that was in active mode? I wonder how passive performs at these lower volume levels.
It was. In passive mode you are just seeing the performance of the source then.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
Sure, but the question is, as you turn down the volume how much does the SNR drop?
With active circuits having fixed gain with attenuation, SNR typically drops 1:1 (more or less) with volume as you turn it down. For any given amount of attenuation, say -12 or -24 or whatever, I'd expect the passive to have a higher SNR because it's attenuating the full scale (signal+noise) instead of attenuating the signal against a constant noise level.
But that is just my guess/expectation; measurement will tell...
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
With passive attenuators the signal will get smaller.
The smaller the signal gets (distortion does not increase) the smaller the distance to the always present noise floor of the equipment after the volume control.
Basically when you attenuate 20db the S/N ratio also drops 20dB as the noise floor is constant.
In Amirs tests the noise floor of the AP.
 

stereo coffee

Active Member
Audio Company
Joined
Jan 2, 2019
Messages
143
Likes
44
With passive attenuators the signal will get smaller.
The smaller the signal gets (distortion does not increase) the smaller the distance to the always present noise floor of the equipment after the volume control.
Basically when you attenuate 20db the S/N ratio also drops 20dB as the noise floor is constant.
In Amirs tests the noise floor of the AP.

No, as the signal gets smaller with in the case of a passive, its higher resistance, thermal ( Johnson ) noise then contributes.
unless of course the DUT was immersed in liquid nitrogen, or Amir's equipment cannot measure Johnson noise ( distortion) contribution.

https://www.physics.utoronto.ca/~phy224_324/experiments/thermal-noise/Thermal-Noise.pdf
https://www.daycounter.com/Calculators/Thermal-Noise-Calculator.phtml

Measuring in passive mode the lower attenuation settings would , or should discover- worsening distortion, at lower levels.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
Imagine the full scale signal as a musical waveform with a tiny bit of ripple on it. The ripple is the noise at 120 dB smaller than the signal. A conventional active amp has a fixed gain ratio, so much of that noise ripple comes from the gain-bandwidth product, so imagine it as a fixed voltage level. As you turn down the volume, you shrink the signal while this noise remains constant, so the SNR shrinks more or less 1:1 with volume as you turn it down.

Now imagine that same full scale signal with the ripple. Instead of attenuating the signal independently of the fixed noise level, you pass the entire signal to a downstream voltage divider. The noise ripple is part of the signal, so it all gets attenuated together, which means the SNR stays the same -- except for whatever noise the metal film resistors of the voltage divider add. So if that noise is smaller than the noise that was in the signal to begin with, the SNR won't drop as much.

So the reason you may get better SNR at lower volume settings with the passive setup is because:

(1) The thermal noise of metal film resistor is probably less than the noise of the active amplification circuit. This seems reasonable, since the noise of most active amps is worse than 120 dB, and the noise of a 5k metal film resistor at room temp is about 120 dB, and that is the worst case -6 dB volume position. All the other volume positions of the passive will have lower resistances with lower noise.

(2) Thermal noise is just the theoretical minimum noise. All resistors have somewhat more noise than this minimum. Metal film resistors are among the best, quietest, closest to this minimum. Most analog pots of equivalent resistance are higher.

Anyway, that's my reasoning. Measurement will tell if it's right or wrong.
 

Snafu

Active Member
Joined
Dec 27, 2018
Messages
223
Likes
140
FYI Schiit is sending me another Freya S for testing....

Ask if You can measure their tube Freya. Would be interesting to see it measured with different tubes too.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,370
Likes
234,416
Location
Seattle Area
Ask if You can measure their tube Freya. Would be interesting to see it measured with different tubes too.
I am so far behind in testing electronics that I don't want to ask for more stuff. :)
 

stereo coffee

Active Member
Audio Company
Joined
Jan 2, 2019
Messages
143
Likes
44
Imagine the full scale signal as a musical waveform with a tiny bit of ripple on it. The ripple is the noise at 120 dB smaller than the signal. A conventional active amp has a fixed gain ratio, so much of that noise ripple comes from the gain-bandwidth product, so imagine it as a fixed voltage level. As you turn down the volume, you shrink the signal while this noise remains constant, so the SNR shrinks more or less 1:1 with volume as you turn it down.

Now imagine that same full scale signal with the ripple. Instead of attenuating the signal independently of the fixed noise level, you pass the entire signal to a downstream voltage divider. The noise ripple is part of the signal, so it all gets attenuated together, which means the SNR stays the same -- except for whatever noise the metal film resistors of the voltage divider add. So if that noise is smaller than the noise that was in the signal to begin with, the SNR won't drop as much.

So the reason you may get better SNR at lower volume settings with the passive setup is because:

(1) The thermal noise of metal film resistor is probably less than the noise of the active amplification circuit. This seems reasonable, since the noise of most active amps is worse than 120 dB, and the noise of a 5k metal film resistor at room temp is about 120 dB, and that is the worst case -6 dB volume position. All the other volume positions of the passive will have lower resistances with lower noise.

(2) Thermal noise is just the theoretical minimum noise. All resistors have somewhat more noise than this minimum. Metal film resistors are among the best, quietest, closest to this minimum. Most analog pots of equivalent resistance are higher.

Anyway, that's my reasoning. Measurement will tell if it's right or wrong.

The contributions in passive mode to worsening SINAD at lower levels , should be a fascinating composite of 1. Thermal noise, 2. Resistance material noise and 3. What very few ever discuss - Contact noise involving degradation of passing audio signals with contacts, worsening with use https://en.wikipedia.org/wiki/Contact_protection ,related then to the bathtub curve.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
The contributions in passive mode to worsening SINAD at lower levels , should be a fascinating composite of 1. Thermal noise, 2. Resistance material noise and 3. What very few ever discuss - Contact noise involving degradation of passing audio signals with contacts, worsening with use https://en.wikipedia.org/wiki/Contact_protection ,related then to the bathtub curve.
Edit: I updated the numbers below based on information from further discussion here.

1. Thermal noise - every resistor has this. Thermal noise at a 10k attenuator's worst case max volume is about 1.8 uV which is about -121 dB below a 2 Vrms signal, and 89 dB below a 50 mV signal.
2. Resistance material noise (above thermal noise) - potentiometers have more of this than metal film.
3. Contacts - lifespan of the relays shouldn't be an issue. Typical EMRs have a life of about 1M actuations, which is 27 years switching 100 times per day. If it uses SSRs the lifespan is effectively infinite. But Amir said he hears a mechanical click as he rotates the knob, so it is probably using EMRs not SSRs.

PS: I went back and studied the spreadsheet I made when I built my passive attenuator years ago. Consider how much SNR drops when you turn down the volume from full scale. Assume that SNR is 120 dB at full scale, and the active drops 1:1 as you turn it down. Assume you feed the full 2 V signal into the passive so the voltage divider reduces the volume for you. Also assume noise added by the passive depends on its output impedance, and the SNR depends on that against the (attenuated) output voltage. Now let's look at a few points along the curve as we turn down the volume:

Level Ratio SNR Active Passive (resistors, output impedance, output noise from active, output noise from passive, output voltage, SNR)
0 1.0 -120 (2.0e-6) 0-10000 10 2.00e-6 + 5.69e-8 2.0 -120
-3 .7079 -117 2921-7079 2073 1.42e-6 + 8.19e-7 1.416 -116
-6 .5012 -114 5000-5000 2502 1.00e-6 + 9.00e-7 1.0 -114
-9 .3548 -111 6542-3548 2290 7.10e-7 + 8.61e-7 .710 -113
-12 .2512 -108 7488-2512 1882 5.02e-7 + 7.80e-7 .502 -112
-24 .06310 -96 9369-631 591 1.26e-7 + 4.37e-7 .126 -107
-32 .02512 -88 9749-251 245 5.00e-8 + 2.81e-7 .05 -104


The passive attenuator has less noise at lower volume settings, due to using a smaller resistor in parallel. It outperforms the active circuit at -6 dB and the difference gets bigger as you turn down the volume further. I highlighted the -32 point since that is 50 mV output for a 2 V input, which is what Amir measures.

In summary, the active & passive are about the same from full scale to about -8 dB. Below that, the passive is quieter by a gradually increasing margin. Most actual listening happens below -8 dB, so the passive is a cleaner way to attenuate the signal at least on paper whether or not the difference can be heard.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
You are talking about the signal behind the device (be it passive or active) that really does not matter because after a volume control (being passive or active) there will always be amplification which always has its own fixed noise floor. Only when the gain of the last stages is controlled that noise floor would vary with the gain but in then would always end up in a noise floor.

The real question is (depending on the efficiency of the transducer, noise level in listening conditions and ear sensitivity (which drops as we age) whether or not the noise level reaches audible levels or not.

Both in passive and active pre-amp conditions the bottom noise depends on the noise level of electronics behind the volume control. That can only increase when the source (before the attenuator) has a worse signal to noise ratio. It can not become better.
 

stereo coffee

Active Member
Audio Company
Joined
Jan 2, 2019
Messages
143
Likes
44
1. Thermal noise - every resistor has this. Thermal noise at a 10k attenuator's worst case -6 dB is about -120 dB.
2. Resistance material noise (above thermal noise) - potentiometers have more of this than metal film.
3. Contacts - lifespan of the relays shouldn't be an issue. Typical EMRs have a life of about 1M actuations, which is 27 years switching 100 times per day. If it uses SSRs the lifespan is effectively infinite. But Amir said he hears a mechanical click as he rotates the knob, so it is probably using EMRs not SSRs.

PS: I went back and studied the spreadsheet I made when I built my passive attenuator years ago. If we consider how much SNR drops when you turn down the volume from full scale, assume that SNR is 120 dB at full scale, and the active drops 1:1 as you turn it down. Also assume the passive noise level depends on the resistor in series, not its output impedance (this will be less favorable to the passive since the lower volume levels use bigger resistors in series, smaller in parallel). Then let's look at a few points along the curve as we turn down the volume:

Active vs. passive should be equal at -6 dB. That's because active will drop about 6 dB and passive will have -120 dB of noise which adds to the -120 dB in the signal being attenuated, which doubles it, which is 6 dB worse. So they're both -114 dB.

At -12 dB the passive should be 4 dB quieter. The active will be 120 dB - 12 = 108 dB. In a 10k ladder at -12 dB, the resistors are 7500 and 2500 so the signal passes through a 7500 ohm resistor which has noise of -116 dB, plus the -120 dB in the attenuated signal makes -112 dB.

At -18 dB the passive should be 9 dB quieter. The active should be 120 - 18 = 102 dB. The passive resistors are around 8700 and 1250 so the signal passes through a 8750 resistor which has noise of -115 dB, plus the -120 dB in the attenuated signal makes -111 dB.

At -24 dB the passive should be 15 dB quieter. The active should be 120 - 24 = 96 dB. The passive resistors are around 9400 and 630 so the signal passes through a 9400 resistor which has noise of -115 dB, plus the -120 dB in the attenuated signal makes the same -111 dB.

Overall, the passive's biggest resistor worst case is 10 kOhm which has noise at -115 dB. So it is quieter than the active at every volume setting below -6 dB. And the more you turn it down, the bigger its noise advantage becomes.

That's only just fine if we propose to just listen to resistors, without any music signal present, which very few of us do. I am happy to call by to listen to resistors for a few hours on their own :cool:

But add a music ac signal and the performance related to signal integrity at lower levels, in passive mode is a completely different story

In a imaginary ideal scenario the attenuator would at lower levels simply reflect the same good integrity of the source regardless of level. This is invariably though not the case. We need to add reactance , and the effects of voltage drops as always lessening the experience with conventional stepped attenuator or potentiometer based passives.

But back to the imaginary ideal idea, if we could have a way of the higher resistance with passives being constantly compared with the source we would then have the perfect scenario of passive resistance being in harmony at all times with the source and the resistance presented as impedance for attenuation purpose,effectively disappearing with respect to signal integrity.

Measurement of passives and conventional preamps needs to change at ASR, to begin to identify the effect on signal integrity at low to normal listening levels, as this over time will lead to much needed improvement, where each product receives needed critique.

I consider the patent of Mary Hallock Greenewalt https://en.wikipedia.org/wiki/Mary_Hallock-Greenewalt is definitely on its last legs with regard to being acceptable to audio reproduction. Considering it was never intended from day one to pass audio signals, rather the rheostat assisted by Mary's feet the intended playing of a visual music she called Nourathar.

The future we need to extend effort to instead, is to improve the Voltage Controlled Resistor and its many exciting variants. Conventional pots though still have their uses for instance instead passing small levels of DC current to assist these new developments . In all my years with circuitry I have never had one single failure with a bourns 50k pot, so they are good IMO with controlling level with low level DC current.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
tenor.gif
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
You are talking about the signal behind the device (be it passive or active) that really does not matter because after a volume control (being passive or active) there will always be amplification which always has its own fixed noise floor. ...
Only when the gain of the last stages is controlled that noise floor would vary with the gain but in then would always end up in a noise floor.

The real question is (depending on the efficiency of the transducer, noise level in listening conditions and ear sensitivity (which drops as we age) whether or not the noise level reaches audible levels or not.

Both in passive and active pre-amp conditions the bottom noise depends on the noise level of electronics behind the volume control. That can only increase when the source (before the attenuator) has a worse signal to noise ratio. It can not become better.
If the SNR of the downstream amp is worse than the preamp, then sure, the preamp isn't the limiting factor. But the purpose of the preamp is to attenuate the signal without altering it. In my example, in both cases the SNR starts at 120 dB and worsens as you turn down the volume. The passive doesn't magically improve it, it just has less degradation.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
What is the difference between a potmeter going into an amplifier or a potmeter going into another amplifier ?

Why would degradation be less ?
 
Top Bottom