Fosi Audio Box X5 Phono Preamp Review

[SmokedBurger]

I just purchased this preamp because of the amazing measurements which leads me to wonder, how necessary is a subsonic filter? the RIAA curve does boost the bass by 20db but how present is rumble in say, a technics turntable? @amirm would you say that this not having a subsonic filter is a problem?

 

[Heinrich]

The 1612 has very low input noise voltage (about 1nV/rt Hz) but high noise current making it unsuitable for MM duty. The 5532 is well suited to MM duty.

The 1612 is suitable for most MMs also, as far as the Texas Instruments datasheet holds. Just hoping they have learnt about audio in the last decades, though! But that‘s an academic discussion anyway, neglecting the surface noise of even the best, new (!) disks.

In case I revive my once fetishized vinyl machine, I would have some options. … … in every case I would be hard pressed to not buy the X5. It would be impossible to justify a DIY solution, starting from the housing down to re-re-recalculating, and simulating all alternative solutions, passive, active, opamps, discrete … . Going for example given the cheaper X1 isn‘t as nice, versatile.

For the sake of just rationality, with a very minor solder job I could adopt the otherwise simply perfect X5 to every situation I can think of. It is well understood that people don‘t like it for that. But once mission has to be declared accomplished, or failed, sorry.

(Rain dancing works, you just have to do it long enough and then it even pours down in the Sahara.)

 

[Heinrich]

I just purchased this preamp because of the amazing measurements which leads me to wonder, how necessary is a subsonic filter? the RIAA curve does boost the bass by 20db but how present is rumble in say, a technics turntable? @amirm would you say that this not having a subsonic filter is a problem?

Good question, but subsonic addresses disk warble, there‘s no relevant connection to a specific platter. The rumble filter is a dropped standard, it is dropped.
Rationale: most speakers are ported today, hence need a speaker (!) specific subsonic anyway (some other discussion), and riaa must not interfere with that

 

[SmokedBurger]

Good question, but subsonic addresses disk warble, there‘s no relevant connection to a specific platter. The rumble filter is a dropped standard, it is dropped.
Rationale: most speakers are ported today, hence need a speaker (!) specific subsonic anyway (some other discussion), and riaa must not interfere with that

I see! So if using a ported speaker or even a subwoofer with passive radiators, i should not be worried about the x5 not having that feature?

 

[morillon]

I see! So if using a ported speaker or even a subwoofer with passive radiators, i should not be worried about the x5 not having that feature?

The "passive radiator" ones have specific problems... ;-)

 

[mike70]

I just purchased this preamp because of the amazing measurements which leads me to wonder, how necessary is a subsonic filter? the RIAA curve does boost the bass by 20db but how present is rumble in say, a technics turntable? @amirm would you say that this not having a subsonic filter is a problem?

I use vinyl since 70s ... and I never needed subsonic filters with the right cartridge in the right tonearm (compliance matched).

With warped records you will suffer of "woofer pumping", but I replaced the warped records in my collection and that's all.

 

[antcollinet]

I use vinyl since 70s ... and I never needed subsonic filters with the right cartridge in the right tonearm (compliance matched).

With warped records you will suffer of "woofer pumping", but I replaced the warped records in my collection and that's all.

It can be quite difficult to eliminate tonearm/cartridge resonance.

My own personal example, though the whole thread is more interesting.

 

[guy48065]

This *is* using a linear PSU. AC transformer is external which reduces magnetic coupling. The AC waveform is then internally rectified and regulated to DC.

More specifically, an external Linear DC PSU.
IMO AC should be far away from a phono preamp.

 

[mike70]

It can be quite difficult to eliminate tonearm/cartridge resonance.

My own personal example, though the whole thread is more interesting.

i don't said i "eliminate" the resonances ... surely they're there, but controlled without warped records, at least to my eyes the woofers doesn't "dance". That's all.

 

[Heinrich]

I see! So if using a ported speaker or even a subwoofer with passive radiators, i should not be worried about the x5 not having that feature?

My player is, presumably ;-) a Denon DF37, that has an active (electromagnetic) damping of the tonearm in place. Shure used a brush for damping the fundamental resonance of tonearm vs pickup at like 10Hz or so. German Dual used a passive, tuned counterweight as „antiresonator“ etc.
All ported speakers go nuts below their tuning frequency. With some that occurs at even 70Hz, some go way deeper to 30Hz (Revel). Sure enough a proper filter never made it into amplifiers until today.
More so, a proper filter could extend and strengthen the bass to a new level. Today we use bass management for that purpose.

Point is, one cannot expect, or should we, a revolutionary solution from an RIAA pre. That path has fallen out of time …

 

[guy48065]

I'm hung up on the lack of a HP filter too.

Meh.
I would welcome a switchable subsonic filter but if it were full-time there would be just as many complaints that it's coloring the sound. To be effective, such a filter needs to have a very steep roll-off below 20Hz. To do it right would bump this amp into a higher price category.

 

[Skeeter]

TI are in the business of selling opamps so I have no doubt they will suggest the OPA1612 is good for MM. The input noise current is 1.7 pA/rt Hz so it is 4x worse than the NE5532/4. The OPA1612 makes a great MC front end amp and line level amp when fed from volume lots of not more than 10k, but it will be 4-5 dB worse in MM applications than the NE5532/4. I’ve used the 1612 - very nice opamp, but not for MM.

 

[morillon]

Meh.
I would welcome a switchable subsonic filter but if it were full-time there would be just as many complaints that it's coloring the sound. To be effective, such a filter needs to have a very steep roll-off below 20Hz. To do it right would bump this amp into a higher price category.

It would cost them practically nothing... especially if it was switched internally.

 

[KSTR]

An good working rumble filter is neither simple to design nor cheap to realize, sadly. Since much of the rumble is usually out-of-phase and most audio content on vinyl is mono down low, a "differential" filter is the way to go, additional to a standard high pass. Conceptually, you convert from L/R to M/S, then high-pass the S-channel and apply the same phase via a proper allpass to the M-channel, then convert back. In the actual circuit of course you try do avoid the explicit L/R<-->M/S conversions.
And because of the low frequencies, the filter capacitors tend to be large, even with rather high resistance values.

 

[Heinrich]

... very nice opamp, but not for MM.

That isn't all the math. Input noise is the generator's resistor self-noise, also noise contribution from the feedback path (needs to be blanced for lowest distortion), current noise multiplied by resitances, voltage self-noise of opamp, then summed up all squared, taking the root while frequency weighting acording to RIAA filtered amplitude response, A-weighted taking the special sensitivity around 7kHz into account.

We shouldn't be that easy on the trigger in dismissing a design we do not even have a schematic for.

It would cost them practically nothing... especially if it was switched internally.

You should tell them.

 

[morillon]

That isn't all the math. Input noise is the generator's resistor self-noise, also noise contribution from the feedback path (needs to be blanced for lowest distortion), current noise multiplied by resitances, voltage self-noise of opamp, then summed up all squared, taking the root while frequency weighting acording to RIAA filtered amplitude response, A-weighted taking the special sensitivity around 7kHz into account.

We shouldn't be that easy on the trigger in dismissing a design we do not even have a schematic for.

You should tell them.

you are right..might as well encourage them not to consider it in a future version ;-)

 

[antcollinet]

TI are in the business of selling opamps so I have no doubt they will suggest the OPA1612 is good for MM. The input noise current is 1.7 pA/rt Hz so it is 4x worse than the NE5532/4. The OPA1612 makes a great MC front end amp and line level amp when fed from volume lots of not more than 10k, but it will be 4-5 dB worse in MM applications than the NE5532/4. I’ve used the 1612 - very nice opamp, but not for MM.

OK - school time for me. Why is the 1612 worse for MM cf MC?

I've always understood that the higher gain needed for MC makes it more challenging wrt to noise, not less.

Also - what is a volume lot?

 

[Skeeter]

That isn't all the math. Input noise is the generator's resistor self-noise, also noise contribution from the feedback path (needs to be blanced for lowest distortion), current noise multiplied by resitances, voltage self-noise of opamp, then summed up all squared, taking the root while frequency weighting acording to RIAA filtered amplitude response, A-weighted taking the special sensitivity around 7kHz into account.

We shouldn't be that easy on the trigger in dismissing a design we do not even have a schematic for.

You should tell them.

I am certainly not dismissing the Fosi. I would expect that to achieve the lowest noise, the OPA1612 would be used as the MC head amp (I have used it in that function) and the NE5534/2 in the MM stage. Without seeing the circuit, I am presuming that is in fact what they have done. The OPA1612 is noisy as an MM amp because the input noise current is high.

 

[Skeeter]

OK - school time for me. Why is the 1612 worse for MM cf MC?

I've always understood that the higher gain needed for MC makes it more challenging wrt to noise, not less.

Also - what is a volume lot?

Both MM and MC have specific challenges related to noise because one is a low source impedance sensor (MC) and the other is a high source impedance sensor (MM). There are two input noise mechanisms in an amplifier: input noise voltage and input noise current. For high-impedance sources like an MM cartridge, you want to use a part that has LOW noise current because the noise current flows through the source impedance creating an additional noise voltage that is added to the device's native noise voltage. The OPA1612 features a noise current of 1.7pA per root hertz (pA/rt Hz) which is high. It does however feature very low input noise voltage. So, if you feed it with a low source impedance device like an MC cartridge, you get very little noise current x source impedance contribution, and because you have low noise voltage (1nV/rt Hz) it is a perfect match for MC. However, if you feed it with an MM cart, the source impedance is about 20k in the upper half of the audio band. The 1.7pA/rt Hz noise current flowing through this gives rise to a significant additional noise voltage. A much better part for MM is the NE5534 with an input noise current of 0.4 pA/rt Hz (so about one-quarter of the OPA1612) or a JFET input opamp like an OPA1642. The OPA1612 is a great part for source impedances up to about 2.5k Ohms which is what you get with a 10k pot when set to the electrical mid position if fed from a low source impedance - this is typically what you would see in a line preamp taking in CD, DAC or a phono preamp.

To get low input noise voltage, opamp designers run the input device emitter currents at high levels and the transistors are large. Unfortunately, this has the side effect of increasing the noise current - i.e. there is a trade-off. For this reason, if a phono designer is looking for good noise performance, they have to match the type of device being used in the amplifier with the type of source.

Here is a ~27 minute video discussing some of these aspects in amongst other things (see from 3.45 to 13.00)

 

[antcollinet]

Both MM and MC have specific challenges related to noise because one is a low source impedance sensor (MC) and the other is a high source impedance sensor (MM). There are two input noise mechanisms in an amplifier: input noise voltage and input noise current. For high-impedance sources like an MM cartridge, you want to use a part that has LOW noise current because the noise current flows through the source impedance creating an additional noise voltage that is added to the device's native noise voltage. The OPA1612 features a noise current of 1.7pA per root hertz (pA/rt Hz) which is high. It does however feature very low input noise voltage. So, if you feed it with a low source impedance device like an MC cartridge, you get very little noise current x source impedance contribution, and because you have low noise voltage (1nV/rt Hz) it is a perfect match for MC. However, if you feed it with an MM cart, the source impedance is about 20k in the upper half of the audio band. The 1.7pA/rt Hz noise current flowing through this gives rise to a significant additional noise voltage. A much better part for MM is the NE5534 with an input noise current of 0.4 pA/rt Hz (so about one-quarter of the OPA1612) or a JFET input opamp like an OPA1642. The OPA1612 is a great part for source impedances up to about 2.5k Ohms which is what you get with a 10k pot when set to the electrical mid position if fed from a low source impedance - this is typically what you would see in a line preamp taking in CD, DAC or a phono preamp.

To get low input noise voltage, opamp designers run the input device emitter currents at high levels and the transistors are large. Unfortunately, this has the side effect of increasing the noise current - i.e. there is a trade-off. For this reason, if a phono designer is looking for good noise performance, they have to match the type of device being used in the amplifier with the type of source.

Here is a ~27 minute video discussing some of these aspects in amongst other things

(see from 3.45 to 13.00)

Thanks for the very clear explanation. Better than I could have hoped for my lesson learnt of the day.