2. analog
  3. 1337559389

Phono Section, Preamp, gain matching for noise

I set out to optimize my phono section and preamp for the lowest noise. First my equipment, VPI Classic 3 Dynavector XX-2 MKII, Avid Pulsus phono stage, Acurus RL-11 Preamp, Music Reference RM-9 MKII amp (Genalex Gold Lion KT77 and Russian 6922) and Sophia II speakers.
Having owned the RM-9 MKII amp for 18 years, I settled long ago on the “high gain” setting for 36 db, this being the lowest feedback setting sounds the best, especially with the Sophias. So that is one variable out of the way. With the RM-9 set to “High” gain I had plenty of gain at any setting since the Acurus has 17db of gain (BTW, I have also had the Acurus RL-11 for 18 years). So I had three options from the phono section 48db, 60db or 70db. At 48db the volume control was the traditional 10:00 – 2:00 range. At 60db 9:00 – 12:00 and at 70db and even smaller range.
I set the Avid Pulsus to 48db of gain, 100pf, 100R and started the test. I set up a sound pressure meter pointed 1" from the tweeter (since white noise from the phono mainly comes from the tweeter). Using a test record I set the 1Khz tone to 114 db (wearing ear plugs). At that same setting the noise from the tweeter when I picked up the tonearm was 60db. At this setting I get perfect 9:00 – 2:00 volume control on the Acurus preamp.
I then tried the 60db setting on the Pulsus, with the same 114db sound pressure, volume control reduced of course, because of the higher gain from the phono section, the noise form the tweeter went down to 52db when I picked up the tonearm. The even higher setting on the phono section produced similar results.
Therefore for my system, the 10:00 – 2:00 rule for preamps produced 8db more noise for the same sound pressure of 114 db. I was able to reduce that noise by 8db at 114db SPL by setting up my preamp for the 9:00 – 12:00 range, by increasing the gain on my phone section from 48db to the medium gain of 60db.

RM9 MK II Tube Amp
low gain 2 db damping 0.7db feedback 1 db Input for 125watts 1.2V
Med gain 32db Damping 2.5db feedback 14db Input for 125 watts 0.8v
Hi gain 36 db Damping 1.5 db feedback 10 db Input for 125 watts 0.4v

Avid Pulsus Phono Section (Solid State)
GAIN
48dB – 60dB – 70dB
INPUT RESISTANCE
100R – 300R - 500R – 1K – 5K – 10K – 47K
INPUT CAPACITANCE
100pF – 200pF – 500pF
DISTORTION
Less than 0.001%
RIAA ACCURACY
+ 0.5dB 5Hz – 70KHz (Neumann HF correction)
NOISE (A weighted)
MM –81dB MC –67dB (high setting)
MAXIMUM OUTPUT
18V RMS
MAXIMUM INPUT @1KHz MM
120mV
MC
4.8mV (MC high)
CHANNEL SEPARATION
< -85dB 5Hz – 20KHz

Dynavector XX-2 MkII output 0.28mv

Acurus RL-11 Pre-Amp (Solid State)
Frequency Response: 20Hz-20kHz
Total Harmonic Distortion: 0.06%
Signal to Noise Ratio: 95dB A Weighted
High Level Sensitivity: 60mV
Maximum Output Voltage: 8 Volts RMS
Voltage Gain as Full Volume: 17.7 dB
Input Impedance: 10K ohms
Output Impedance: 47 ohms
captain_winters
almarg's avatar
Ag insider logo xs@2x
almarg
9,670 posts
 
05-22-12: Dhl93449
Al: I think he is measuring mainly white thermal noise as opposed to 60/120/240 Hz hum (via the tweeter).
Hi DHL,

I realized that he is not measuring hum. Note that I said "even though no hum occurs." Ground loop effects can cause noise at high frequencies, as well as low frequency hum. And it is conceivable to me that the former could happen in the absence of the latter.

If high frequency distortion components and/or noise that are on the AC line were to couple through stray capacitances onto the chassis and/or the circuit grounds of each component, to a degree that is unequal in the two components, and if the components have their chassis and circuit grounds connected together (or connected together through a low impedance), and if 60 Hz and its lower order harmonics are not able to so couple (to a significant degree), that is what would happen.

See this reference, and keep in mind that the interface between the particular components that are involved is unbalanced:

http://www.jensen-transformers.com/an/an004.pdf

Note this statement:
The noises originating with the power line are generally described as EITHER [emphasis added] "hum", which is predominantly 60 Hz, or "buzz", which consists of a mixture of high-order harmonics of 60 Hz. These harmonics are the result of power line waveform distortion, which commonly reaches 5% THD and is caused by many types of non-linear power line loads. Because the human ear is much more sensitive to frequencies in the 2 kHz to 5 kHz range at these very low levels, buzz is usually more audible than hum, even though the hum level may be electrically larger.
RFI/EMI pickup in the cables is also a possibility, as I indicated.

Best regards,
-- Al
captain_winters OP
133 posts
 
Thanks for the responses, of interest, and an indirect question, was based on the SNR published for the Acurus RL-11 of 95db A Weighted. The Avid Pulsus on the other hand has a noise spec from -81db to -67db. I have plenty of gain so operating the Pulsus at 48db of gain, even though the XX-2 is 0.28mv, still gives me the “typical” preamp range of 10:00 – 2:00. If strictly going by that rule, I would have set up the Pulsus gain that way. If you look at the SNR in that configuration it is SNRdb = Psignaldb – Pnoisedb. SNRdb = 114db – 60db = 54db. In the Pulsus 60db gain configuration. SNRdb = 114db-52db = 62db. Looking at the specs, Pulsus being noise floor at gain, and Acurus being SNR, it is not apparent to me how you would surmise this result analytically, without actually measuring.
It looks like Dhl93449 hit it on the head when he said
“I found that the lowest noise (as measured by my Tektronix 7L5 spectrum analyzer) was found at the higher input gain settings, minimizing the output stage gain.”

A related question, has anyone actually measured their SNR using this method? I believe I have a quiet system and I also believe 62db is a good result, but what is a reasonable expectation?
almarg's avatar
Ag insider logo xs@2x
almarg
9,670 posts
 
The reason the SNR specs for the Pulsus (and most other phono stages) worsen as the gain setting is increased is probably that they are defined relative to a reference level at the output of the phono stage that is the same, regardless of gain setting. Therefore the measurement for higher gain settings will be taken at a lower input signal level, reducing SNR in the circuitry at the front end of the phono stage. That in turn figures to generally be the most critical point in the signal path with respect to overall SNR performance, since signal levels are lowest there.

In contrast, your experiments involved a fixed input signal level. So the SNR differences you obtained for the various gain settings presumably resulted from differences in the relation between signal levels and noise contributors at points in the signal path between the phono stage's gain adjustment provisions and the volume control in the preamp (inclusive). As you indicated, the net result of all of that is probably not analytically predictable.

Regarding comparisons between SNR specs for the phono stage and the preamp, while in general it can be expected that the SNR performance of a good line stage will be superior to that of a good phono stage, I would comment that a direct comparison between the numbers is pretty much meaningless in the absence of specified reference levels. Unfortunately, SNR specs are commonly presented without any indication of what signal levels they are referenced to. That is the case for these particular components, as can be seen here and here.

Regards,
-- Al
dhl93449
455 posts
 
Al:

I understand what you are pointing out, but higher harmonics of the 60 Hz fundamental will be significantly decreasing in amplidtude as you get to higher orders.

One way to determine this experimentally would be to observe the noise at a given volume setting (of the line amp), then disconnect the interconnects and replace them with (a) shorting plugs or (b) better yet shorting plugs with resistors simulating the source impedance of the phono pre-amp. If the noise levels drop considerably in this test, then perhaps you are right and the increased noise levels are due to ground loops or IC issues. If not, they are simply the thermal noise generated by the various stages of amplification.

Everyone tends to assume that line stage noise is insignificant compared to phono stage noise, but depending on the relative gain (and other issues such as source impedances) this may not always be the case.

Also, one needs to abandon rules of thumb based on the rotational position of gain controls. There is no reference standard here, and the gain vs position is highly product design dependent.

Regarding my spectrum analyzer measurements, I was looking at the spectrum from about 50 Hz- 20k Hz. I did not apply any weighting (A or otherwise) to my measurements as the Tek will not do this (it is a vintage 1980's SA but has a S/N/distortion capability of -130 dB).
dhl93449
455 posts
 
Captain Winters:

Re your question of acceptable levels, this is a bit subjective. For me, -60 dB would not be good enough, as I can easily hear this level from my listening position at the highest sound levels I listen to.

You can check out my post "Impressions of the PS Audio GCPH" under the amps/pre-amps section for the actual numbers, but they were on the order of -90 to -100 dB (unweighted), depending on gain, referenced to a 1 volt output. You can add about 6 dB to those numbers since my normal line input levels (to the line amp) is 300-500 mV at max listening SPL. Even in the worst case of GCPH input gain of 66 dB and MAX volume gain control, noise was at -75 dB unweighted. This level is almost inaudible from my listening position.

I emphasize "normal max" listening volumes, which for my JC2 is 9-10 on the volume pot. If I crank up the gain to 12 or full CW rotation (4-5) the -75 dB levels become very audible.
More to discover