morillon
Major Contributor
- Joined
- Apr 19, 2022
- Messages
- 1,768
- Likes
- 403
-
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!
Fosi Audio Box X5 Phono Preamp Review
- Thread starter amirm
- Start date
Great to see this coming to terms (once real data is at hand). Won‘t spoil your discussion, but is the input capacitance missing in D.Self‘s examples? If taken into the calculation we would have a resonant circuit with increased impedance compared to just a shunted inductor.
The noise figures NF in the table are quite interesting. Except for the FET opamps they are, I would say, low at 2dB or so. NF describes the noise penalty of the amp, all extra noise sources within it considered. Or, as depicted in the table with theoretical noiseless amps, best to be achieved is NF = 0dB. With NF = 0dB all noise is generated by the cartridge alone. And that‘s not nil / nothing.
@pma, to consider the self noise of the cartridge would round up your highly appreciated contribution above, the measurement.
Now, as the penalty is that low, do we care?
It doesn't- see my data from earlier in the thread. But as an absolute number, the window will reduce the reported noise. For random-ish noise, non-window will give the most accurate quantitative answer.
It is taken into account in my spreadsheet calculator.
Do we want engineers designing phono stages to measure best with 470mH cartridges shunted by 47k resistors, everything else be damned? It seems to me that's where this argument will lead if you incorporate this kind of specificity into your performance test.
No, we want the designs that do not add more noise to 47k resistor (shunted by cartridge inductance) intrinsic noise. Translated, additional current noise that makes additional noise voltage across the 47k resistor is unwanted. The whole idea of that 47k “standard” load impedance is highly questionable. And the MC situation is not better, because one needs 20dB more gain compared to MM. Who else than engineers is supposed to design the audio circuits? Artists?
OP
No, but they clearly show the noise floor and the change with source impedance (admittedly, it's clearer in my graphic presentation, but Pavel's data are consistent with mine). If we use your standard of 1kHz signal, the increased noise floor will absolutely drop the SINAD (or S/N). I mean, the noise rise in the top 1-1/2 octaves is 20dB at 10kHz, well within the hearing range of my aged ears; even weighting it for the RIAA curve, it's very significant.
That covers the vast majority of MMs (inductances vary, but going from 470 to 1000mH won't change things much, 47k is an almost universal standard), so yes.
Thankfully an excerpt of D.Self‘s take on the topic was cited above. What I learn from that is the importance of the self noise of the cartridge, or signal generator, including its environment, the 47kOhm resistor and capacitance namely.
For an illustration I want to come back to @pma. An example was given by how much the output noise increases once a real pickup is connected instead of grounding the input. The cartridge is standard, the pre amp is FET based**. From the latter I might conclude that the contribution of input *current* noise is neglegible.
If it is so, the rise in noise is only due to the self noise of the cartridge and its standard 47k load. It cannot be attributed to the pre amp, especially not as some degredation of SINAD. Conversely the actual SINAD of every phono replay down the amplification links is limited to that of the—cartridge‘s self noise. Reminder: there is thermal motion of electrons in a conductor, that gets amplified in parallel to the signal, Johnson noise. We are just talking about it. There is thermal motion in the air, …
**the actual pre amp was briefly discussed on this bord some time back
For zero input current, yes. Which is why this is a needed part of the test since zero input current is not universally the case. Likewise, the noise of the 47k resistor (shunted by any load capacitance) will be present under real world use, so should be part of the consumer-facing measurement.
I'm highly amused at the irony of someone who insists on unrealistic (for the intended use case) tests and someone who insists on only realistic (for the intended use case) tests when it comes to power measurement have swapped sides here.
Can‘t do that, won‘t work. There is frequency dependency originating in the load‘s C also, to many variables. I agree, or better to say, I would rather chose a noise figure representing the contribution of input noise current. But then how to standardize for comparison or listing? Really, worst I would expect are 4dB of a penalty from to high of an i/c. OPA1642 as used here should be just on the sunny side of a compromise. Alas, we don‘t know for sure if the NE5532 could be „rolled“ accordingly, or is anyway the MM input.
There‘s more to gain from being that pedantic with clean records (and needle!)
And yet it's trivially done. At least two of us here do it routinely.
Please, don‘t let me beg for the results.
For the fun of it, and partly on topic, a fantastic rant on cartridges and their makers. Alas, in German only, and a photocopy, sorry.
I've posted them at ASR a few times, and copied one of the graphs earlier in this thread.
Thanks @SIY! Referring to your post 125 here I still argue that the Johnson self noise of cartridge plus 47k, plus input C is not considered. The graph only shows that there is additional noise, while it does not discriminate which portion is expected from the cartrige alone, and which is the interference of the amp‘s noise input current with the effective impedance of the cartridge.
I accept halfway the idea to test an amp against an expected input generator. But the generator‘s inherent Johnson noise must not be attributed to the amp. Example simple: a theoretically noiseless amp would be limited to 85 something dB of a s/n ratio when fed via a standard cartridge. Not fair. But it is possible to subtract the self noise of a known generator type from the noise reading of the amp. While current noise is well understood in this context, why is the generator’s Johnson self noise ignored—or even attributed to the amp’s current noise?
Anytime I‘m willing to accept a limited noise penalty anyway. Especially contemporary opamps at pretty low cost allow an overall performance one could only dream of a few years ago. The vinyl cutting machine, as far as I‘ve read today, shows no more than 85dB s/n ratio, let alone the analog tape back in the vinyl‘s heyday.
But that doesn't matter for the end user. This is the noise he'll get with that preamp as the load.
It's of engineering interest to separate those issues and that's easy to do. But that wasn't the purpose of the measurement- it was to demonstrate what that preamp will do in the real world.
Let’s get back to topic please. This is an Official Product Review Thread. Keep your comments and observations rooted in the product reviewed and or the Test results provided.
Thank you for your cooperation and understanding.
Thank you for your cooperation and understanding.
morillon
Major Contributor
- Joined
- Apr 19, 2022
- Messages
- 1,768
- Likes
- 403
This concerns precisely the fact that in the case of this fosi, the measurements concerning the mm part are possibly not so relevant...to be verified with another protocol, more relevant ...
;-)
(metrology always requires precautions...here it seems that one just needs to be added...)
;-)
(metrology always requires precautions...here it seems that one just needs to be added...)
Last edited:
KSTR
Major Contributor
The self noise of the cartridge is quite low as it around 1k Ohm. Consequently, the additional noise developed by noise current from the amplifier is also rather low.
The problem is the inductance which in itself is noise-less but isolates the low cartridge resistance for higher frequencies. The resistance cannot effectively short the noise contribution (self noise and opamp noise current-induced) of the 47k load resistor -- whereas a low-impedance source like the AP does short out the 47k wideband, that's why Amir's measurement does not show the real-world noise one will see with this product.
Similar threads
- Poll
- Poll
- Poll
