How about less than imported folks in Zug who can afford it.
(ever been to Zug, Switzerland ?)
The FM223 does about 4 times the tag of Mr Loesch's phono stage.
I agree with the yawn. These magical vinyl fixers employing mysterious techniques have been around since before the CD was invented.
What's really revealing is when you use a software model of such techniques and listen to the audio that is being removed by such processing. No thanks, I'd rather hear all of the music instead of sacrificing some of it just to get rid of pops and clicks - there are better ways to avoid pops and clicks in the first place, and then you don't have to suffer from removing good audio along with bad noise in some kind of magic process.
How about less than imported folks in Zug who can afford it.
(ever been to Zug, Switzerland ?)
The FM223 does about 4 times the tag of Mr Loesch's phono stage.
Only Zurich, but quite a posh weekend.
Available in principle, an interesting concept. The idea of billionaires sitting around cleaning up their old collection of 78's is amusing. Here's a man that understands. http://en.wikipedia.org/wiki/Joe_Bussard
Last edited:
Great, more potted discrete modules with no schematics = unservicable when the manufacturer isn't around anymore (even with the schematics how do you open the module?) . Reminds me of a 1970's preamp JC would remember. At least it will make a great chassis for a Pass Pearl (or a Vendetta if we could find the Jfets) when it's had it's day. I suspect some-one would take over FM when Manuel retires as it looks like a great money maker - all the transistor matching in the world doesn't disguise the mediocre supporting passive parts and pedestrian power supply- at least the bling chassis looks like its worth getting it out of the pawn shop when the market sinks again. The phono stage looks like a deal compared to the FM power amps. No I can't afford them, thats why i'm on Diyaudio. I heard an FM system during an Asian visit. It was nice, but not amazing - at those prices it needs to be amazing.
Last edited:
In all fairness, then, people who do not appreciate advance digital reproduction should refrain from commenting on digital audio systems. Excellent digital reproduction has been available for decades, even though it has only reached consumer electronics fairly recently.
Rsdio, IF I could only hear truly excellent digital reproduction, more often, then I might agree with you. Sometimes, I have heard excellent digital reproduction. We had pretty good digital when my past business partner, Bob Crump would modify a digital processor originally designed by Damian Martin, a colleague, and here on DIY. I always regretted not buying one about 10 years ago, when one came up for sale, used. Still, it was not as good as Vendetta, and equally high quality playback. Bob and I both agreed to this. Lamm, a few years ago showed their tube amp and preamp with a really good French made digital playback, I DID think that it was remarkable, but expensive. It was actually better than the analog playback that they had.
A really good SACD, 24-96K (or higher), or whatever, I am up for, but I can't afford good digital, either. IF you, rsdio, can give me a significant tip as to what to buy that does not cost a small fortune, I would appreciate it.
A really good SACD, 24-96K (or higher), or whatever, I am up for, but I can't afford good digital, either. IF you, rsdio, can give me a significant tip as to what to buy that does not cost a small fortune, I would appreciate it.
Hi,
Correct, it fills the analog equaivalent of the lower bits with "fuzzy distortion", which depending on specific implementation may be semi-deterministic, signal dependent or fully random. So there are no longer discrete levels, mainly because if we WERE to see the discrete levels of the converter we would realise that our 32 Bit converter is actually only a 12 Bit one.
The above is not entirely true.
First, with dither and a converter that is capable to handle all the dynamic range desired by direct conversion (Multibit DAC, SAR ADC etc), we get all the signal with added noise (mathematically and practically), so dynamic range is compromised.
Second, dither can only help with quite quite low frequencies compared to the sample rate, higher frequencies are not handled as well and short "wavelets" at low levels are by far more grossly distortorted than they would be without dither.
In fact, apply dither to a sharp transient and you can literally SEE the fuzzy distortion on any decent 'scope.
Of course, we can argue if such fuzzy distortion is audible or not, sadly extant research seems rare and the results are not unambiguous.
BTW, I am fully aware that the converter chip industry and those hard-line orthodox priests at the AES (aka Congregation of Faith - Audio Chapter) all promote large amounts of dither as the cure all. Yet their promotions are singularly lacking in actual evidence (past maths that fails to take into account how people hear).
Ciao T
Correct, it fills the analog equaivalent of the lower bits with "fuzzy distortion", which depending on specific implementation may be semi-deterministic, signal dependent or fully random. So there are no longer discrete levels, mainly because if we WERE to see the discrete levels of the converter we would realise that our 32 Bit converter is actually only a 12 Bit one.
The above is not entirely true.
First, with dither and a converter that is capable to handle all the dynamic range desired by direct conversion (Multibit DAC, SAR ADC etc), we get all the signal with added noise (mathematically and practically), so dynamic range is compromised.
Second, dither can only help with quite quite low frequencies compared to the sample rate, higher frequencies are not handled as well and short "wavelets" at low levels are by far more grossly distortorted than they would be without dither.
In fact, apply dither to a sharp transient and you can literally SEE the fuzzy distortion on any decent 'scope.
Of course, we can argue if such fuzzy distortion is audible or not, sadly extant research seems rare and the results are not unambiguous.
BTW, I am fully aware that the converter chip industry and those hard-line orthodox priests at the AES (aka Congregation of Faith - Audio Chapter) all promote large amounts of dither as the cure all. Yet their promotions are singularly lacking in actual evidence (past maths that fails to take into account how people hear).
Ciao T
The new ESS 9102 a 32 bit ( 24 bit SPDIF ) A to D converter has the ability to impliment an RIAA biquad filter so now you can have a digital analog stage. Would you still need a low noise linear analog gain stage to raise a moving coil cartridge's microvolts before sampling to get sufficent dynamic range to encode enough bits?
ESS press release - no data sheet on the site yet
http://www.esstech.com/index.php?p=news_pr_2011
ESS press release - no data sheet on the site yet
http://www.esstech.com/index.php?p=news_pr_2011
Last edited:
Hi,
... is not a 32 Bit converter, or even a 24 Bit converter, it is 21 Bit, if we are lenient.
This was done long before that one, though not directly on the AD Chip AFAIK.
Having tried it I prefer to do the EQ in analog hardware and to convert to Digital afterwards, strictly for subjective sonic reasons. But I also like multibit DA's without dither and followed by tubes that produce a completely scandalous around 0.2% 2nd HD at digital full scale, so WTFDIK?
Well, let's think.
The A2D will likely need 2V RMS for digital full scale. If you directly connect a MC Pickup with 0.5mV @ 5cm/S your maximum peak output would be 2.5mV @ 25cm/S and thus around 60dB below full scale.
With a noisefloor of -127dB this would leave around 67dB of dynamic range, or 11 Bit, but remember that the lowest of these bit's would heavily "noiseshaped". So yes, around 60dB linear, low noise gain would be needed for a 0.5mV MC Pickup. I think a PGA2500 might be a good choice for this.
Incidentally, in my personal view, especially in the light of the extensive use of low bit converters for A2D and D2A I would suggest that storing the direct unequalised Data from the A2D and performing analogue RIAA correction after the DA converter would be best from an archival viewpoint, but practicability is very limited of course...
Ciao T
... is not a 32 Bit converter, or even a 24 Bit converter, it is 21 Bit, if we are lenient.
This was done long before that one, though not directly on the AD Chip AFAIK.
Having tried it I prefer to do the EQ in analog hardware and to convert to Digital afterwards, strictly for subjective sonic reasons. But I also like multibit DA's without dither and followed by tubes that produce a completely scandalous around 0.2% 2nd HD at digital full scale, so WTFDIK?
Well, let's think.
The A2D will likely need 2V RMS for digital full scale. If you directly connect a MC Pickup with 0.5mV @ 5cm/S your maximum peak output would be 2.5mV @ 25cm/S and thus around 60dB below full scale.
With a noisefloor of -127dB this would leave around 67dB of dynamic range, or 11 Bit, but remember that the lowest of these bit's would heavily "noiseshaped". So yes, around 60dB linear, low noise gain would be needed for a 0.5mV MC Pickup. I think a PGA2500 might be a good choice for this.
Incidentally, in my personal view, especially in the light of the extensive use of low bit converters for A2D and D2A I would suggest that storing the direct unequalised Data from the A2D and performing analogue RIAA correction after the DA converter would be best from an archival viewpoint, but practicability is very limited of course...
Ciao T
John,
Of course dither and the transient can both be averaged. But then the Transient is hardly a transient anymore, is it now?
My problem is that dither does it's "magic" only in one out of the several domains neccesary for music reproduction, it does so at the expense of the others.
So, just like I am not per se against negative feedback looped around multiple stages, I am also not against dither as such. However, I find myself in disagreement with using either method as a fundamental means to obtain certain technical specifications in preference to fixing the underlying issues (even though that would be more expensive).
So I am FOR maximising bandwidth and linearity before applying feedback, of course, if you do the job well you find local, stage by stage degeneration can do the job well and you do not NEED to use negative looped feedback.
And I am for having an ADC/DAC Core that produces the necessary resolution (at least 16 - 20 Bit) on a hardware level, available as distinct, discrete level on a single sample, not by using dither and noiseshaping. We can then apply dither, noiseshaping and other methods to improve measured performance further, but often we find we do not need to use them.
Once we have the choice of controversial design techniques as option, rather than as absolute necessity, we can select what fits our specific purpose best, heck, we can even make it user selectable and let the customer/user decide what they like better.
Again, that's just me though and WTFDIK.
Ciao T
Of course dither and the transient can both be averaged. But then the Transient is hardly a transient anymore, is it now?
My problem is that dither does it's "magic" only in one out of the several domains neccesary for music reproduction, it does so at the expense of the others.
So, just like I am not per se against negative feedback looped around multiple stages, I am also not against dither as such. However, I find myself in disagreement with using either method as a fundamental means to obtain certain technical specifications in preference to fixing the underlying issues (even though that would be more expensive).
So I am FOR maximising bandwidth and linearity before applying feedback, of course, if you do the job well you find local, stage by stage degeneration can do the job well and you do not NEED to use negative looped feedback.
And I am for having an ADC/DAC Core that produces the necessary resolution (at least 16 - 20 Bit) on a hardware level, available as distinct, discrete level on a single sample, not by using dither and noiseshaping. We can then apply dither, noiseshaping and other methods to improve measured performance further, but often we find we do not need to use them.
Once we have the choice of controversial design techniques as option, rather than as absolute necessity, we can select what fits our specific purpose best, heck, we can even make it user selectable and let the customer/user decide what they like better.
Again, that's just me though and WTFDIK.
Ciao T
Got any pics for us then? Presumably 'decent scope' here means an analog one?
This makes absolutely no sense. Would you say that Johnson noise in a circuit makes that circuit unable to "handle or average transients?" Dither is just a carefully-controlled, specific-amplitude noise that is mixed in with the signal ahead of the A/D quantization stage. I'll need to see that article to understand how it presents any more of a limitation on transients that resistor noise, plate noise, or anything else in the physical universe of non-ideal circuitry.
I am not a fan of SACD, at all. I've already mentioned that AES3 (a.k.a. AES/EBU) and S/PDIF are improperly-designed transports that belong to history, not the present.
My recommendation would be the Metric Halo Labs MobileIO ULN-8 (or LIO-8 if you will never record your own vinyl) paired with an appropriate Apple Macintosh computer and balanced preamp and amplifier electronics.
At that point, you will have to deal with a limited selection of good recordings at 24/96 and 24/192. Many recordings offered in 24-bit are actually conversions from SACD and show the intense ultrasonic dithering noise that is inherent to sigma-delta conversion, so I toss those out when I find them. When I do find a digital recording that is free from ultrasonic aliasing and excessive ultrasonic dithering, it's a beautiful thing to hear. It's still very much an industry where caveat emptor applies in full force. i.e. Many recordings are sold as "HD" and yet contain gross errors. I'm actually not too surprised that many people do not like "digital" - between SPDIF and bad conversions, there is a lot of crap out there.
Dithering of some kind is absolutely necessary. It's not an option. Read Vanderkooy, et al, JAES, Vol 40, No.5, 1992 May, for the details if you don't believe my brief summary.
The only time you can successfully get away without any dither is when other sources of noise are above the level needed. This happens quite often with 24-bit A/D because it's still basically impossible to design an analog front end with better than 140 dB S/N.
On that note, please explain how dither noise is any worse than Johnson noise, or similar, with respect to the effect on a transient. If you look at a transient on a 'scope, then add a certain amplitude of noise, how can the nature of that noise be significantly different from some other kind of noise of the same amplitude?
I imagine that you must be speaking about the massive dithering noise used in SACD and other sigma-delta converters. I'd like to point out that Vanderkooy, et al, have shown that 1-bit sampling can never be properly dithered, and thus I personally put 1-bit digital in a separate category from multi-bit and hybrid converters.
In other words, if you're talking about 1-bit digital, then I will join in with the critical comments. But don't throw the baby out with the bathwater: not all digital has excessive dither. Properly done, we shouldn't be talking about a situation where people are applying varying amounts of dither; in a correct system, there really is exactly one level of appropriate dither, and any more than that is evidence of bad design.
Do not forget that every A/D and D/A are in fact 'analog'. Both taking samples and re-creating analog output. Digits are derived from analog values and they convert into analog again. There is no magic in analog input stages of A/D and output stages in D/A. And you are able to produce BETTER analog input and output stages than those integrated in A/D D/A chips.
In a way, this is what I am trying to say. Perhaps the only difference is that I consider the analog stage inside the converter chip as part of the overall analog stage. Or, to put it another way, it's a bit specious to look at the S/N of the analog stage and then claim that a 24-bit converter is actually only a 12-bit converter. The analog S/N is an entirely different spec than the quantization level. With the exception of companding converters, range-shifting converters, or those 32-bit converters that fudge the numbers, I'd say that it's generally misleading to claim that a 24-bit converter is not really a 24-bit converter just because the analog stage has less than 144 dB dynamic range.
If one wants to complain about the S/N, and it's not actually quantization noise, then just say what you mean.
My immediate reaction is to wonder how many discrete 24-bit converters there are. Sounds rather difficult to build, but it probably wouldn't be too difficult with an FPGA and some discrete analog parts. It would make one helluva DIY project!
Max Headroom prefers TV : AES Oral History 096: John Vanderkooy - YouTube
(all cheese-heads should be sent to CA to learn to speak proper English)
- Status
- Not open for further replies.