OK, Thorsten, I will give it up. Personally, I have found it difficult to make a very accurate RIAA EQ, consistently, due to component interactions, but I still think that it is important as a comparative reference with other phono stages. I find that 1% parts work well enough, if you dial it in properly. I will leave it at that.
The inverse network makes it easy. And I wouldn't limit it to frequent testing. If we are going to make a DIY phono amp, the network is trivial in comparison. One advantage of the network, is that it allows input levels from the generator of around 1V or so. With my inexpensive signal generator, the measured deviation of the meter is within 1 percent (less than 0.1dB) from 20Hz to 20kHz. This, in essence, calibrates the generator and meter together. One could make a compensation chart, if that's really a concern.
One advantage of the network, is that it's easy to also make an input divider for MC level inputs. I did find it necessary to build a simple Op amp buffer to drive an MC step up transformer. The signal generator alone couldn't do it at low frequencies. This shouldn't be a problem with a head amp.
Sheldon
Already posted an inverse RIAA 1/10th octave multi-tone file months ago.
BTW in their FAQ the AMR folks quote the implausible extreme of DNR hit you take with digital equalization. You gave a fairly brusque brush off of that AES paper where actually they presented a lot of data from a broad range of LP’s . In fact the crest factor of music changes very little w/wo RIAA, which means the headroom allowance for your A/D needs little or no adjustment and you get the benefit of “CD” like pre-emphasis of the highs. The loss at the low end is minimal, and I would make the argument that the lows are effectively oversampled in any case, but no need to go there.
You might argue both parties have something to sell, what else is new?
Last edited:
This was an attempt to shed some light on the variations of deemphasis curves used for vinyl playback....not to fuel a talk about how to make them or how narrow the tolerances need to be....
I do think its rather interesting that the dominating riaa curve that is the defacto standard in record playback, not always makes the record sound as intended...That alone makes it worthy of a debate...
I don't like the idea of variable tone controls...where you may or may not hit a correct setting..more the idea of switching in/out complete networks.. which in turn would be quite simple with my (oneshot Riaa)...
I figure that 3-4 fixed "type like" curves should do the trick...While it will not cover all.. it will for sure be much better than "just" a Riaa.
I do think its rather interesting that the dominating riaa curve that is the defacto standard in record playback, not always makes the record sound as intended...That alone makes it worthy of a debate...
I don't like the idea of variable tone controls...where you may or may not hit a correct setting..more the idea of switching in/out complete networks.. which in turn would be quite simple with my (oneshot Riaa)...
I figure that 3-4 fixed "type like" curves should do the trick...While it will not cover all.. it will for sure be much better than "just" a Riaa.
Scott,
True. Or maybe not. You may take a bigger hit if you record with some headroom. It would be nice to see some actual recordings compared... It can be quite illuminating to do that.
I did not "brusque brush off" the paper because they presented LP Amplitude statistics.
Yup. They do. Let the market decide what it wants.
Ciao T
True. Or maybe not. You may take a bigger hit if you record with some headroom. It would be nice to see some actual recordings compared... It can be quite illuminating to do that.
I did not "brusque brush off" the paper because they presented LP Amplitude statistics.
Yup. They do. Let the market decide what it wants.
Ciao T
I took your one line comment... "And they did so by comparing a theoretical 24 Bit recording of a unequalised LP followed by digital EQ to a theoretical 16 Bit recording of a pre-equalised LP." as some how qualifying their results. I saw plenty of data from real 24 bit recordings to support the basic claim, that you won't find real world LP's that as AMR states "The common RIAA equalisation requires low frequencies to be boosted around 20dB and high frequencies to be attenuated by around 20dB. To implement the equalisation in the digital domain, at high frequencies, one would have to remove some 40dB of dynamic range available from the Analogue-to-Digital Converter."
I see your point though, if all you get is a 16 bit result you have no interest. I think this process could be improved considerably by adding a single time constant to the preamp especially to quell concern over ticks and pops. If "using rare and primarily bespoke components" has to be in the equation then there's no point in discussing the two approaches is there?
I saw a picture of that PH-77 with it's valves and vibration dampers(?), I wonder what next years model will look like. OTOH I would think this stuff is more like Bristol Cars with the same clientele.
Last edited:
Scott,
Well, for analog filters I can just use a spreadsheet and enter the turnover frequencies and/or time constants and I have my filter. One more reason to stay analogue...
More seriously, is it really that difficult to generate the filter coefficients?
Ciao T
Well, for analog filters I can just use a spreadsheet and enter the turnover frequencies and/or time constants and I have my filter. One more reason to stay analogue...
More seriously, is it really that difficult to generate the filter coefficients?
Ciao T
I know Scott does knows how to, but this is a good opportunity for me to ask if there is a link to biquad filter treatment for the practical minded (and high-mathematics disabled).
Sorry for disturbing
Regards
George
Last edited:
Scott,
Well, for analog filters I can just use a spreadsheet and enter the turnover frequencies and/or time constants and I have my filter. One more reason to stay analogue...
More seriously, is it really that difficult to generate the filter coefficients?
Ciao T
Yes actually, there is no closed form solution for these kind of biquad filters to arbitrary accuracy. If you search for Bob Orban's solution you find one mathematically correct way. I did it using thermal annealing to best solution. It is amazing that you can ripple four or five numbers through a .wav and get .00006dB accuracy. The computing power has become irrelevant now you coud do brute force FIR filters (easily computed) at 64 bit floating point resolution in real time. Pretty huge ones too. All free software to boot, though I wish the Linux audio folks would get their act together.
I'm no big fan of the Pure Vinyl folks BTW. Their web site has a little too much of the "we're smarter than you, don't try this at home stuff". Again brute force computing power has made a lot of shortcuts and mathematical tricks irrelevant.
Those AMR guys that are pushing the PH-77 are lying, plain and simple. I haven't found any other reference on the web that supports their claims that -for example- DGG stereo records do not use RIAA; or that Decca FFSS are not cut using standard RIAA.
Nor i could find any of such curves. And i have years reading and reading about everything related to vinyl playback.
They want to push and sell their own preamp and they invent the nonexistent need for those curves, plain and simple. They even modified the RIAA pages on wikipedia to mislead people into believing that RIAA wasn't an international standard after 1958.
They don't published the curves implemented on the PH77. My bet is that they're RIAA with some tone compensations. It is true that Decca FFSS has different "tone" than London Phase 4 and that DGG tulip label stereo and DGG later stereo etc etc. But that's not because there was a "nonstandard RIAA" curve, but simply the mastering engineer's (or label's) tastes. All such records sound just fine using the standard RIAA curve. I have lots of them.
Last edited:
Me too. Except many mono records. Those do require different curves, of course.
Thorsten, i agree with most of what you write. But you should be very careful with believing what the AMR guys posted.
Now, again on "tone" differences between manufacturers: As i've said, it's both down to the mastering engineer's preferences AND also of the equipment's limitations.
For example the early Decca/London stereo FFSS records were half speed mastered. The head used by them (i'm 99.9% sure it was the "Telefunken-Decca ZS 90/45") had a high frequency resonance (i could only find one internet source and it claimed 8KHz). To control that resonance you use cutterhead feedback but this introduces other problems, so you want to use as little feedback as possible.
What the guys did at Decca from 1958 on, was to cut at half speed (yes, half speed mastering wasn't a later invention!!) so the resonance was now double the freq (probably 16KHz). At the same time this means that you have a problem with loss of bass frequencies (see source below)
Later (196
they bought the Neumann SX-68 head, they were called "FFRR" again, and they didn't need to cut at half speed anymore.Source: http://www.arsc-audio.org/journals/v18/v18n1-3p4-19.pdf
So this does account for the sonic signature of those records. It WASN'T because they "deviated from RIAA" and thus used "their own curve".
Hi,
Actually, most here know that I design for AMR. So some of what you object to I wrote and that after fairly lengthy and careful research (more than a quick google search).
The curves are not listed on AMR's with with time constants as these can be found in the various literature by anyone so inclined )just make sure to get the right sources, as said Peter Copeland has more right than anyone else I know).
We may debate of the source of why LP's from certain manufacturers have consistent deviations from a subjectively "flat" response, however I came up with something very close to the curves from literature purely by listening via RIAA EQ's Phono's and using an external digital EQ as "re-equaliser".
Both the shapes of the curves I arrived at by ear and my background pushed me towards the conclusion "different EQ".
That you have found no sources on the net does not particulary surprise me, though they do exist, but non in English, most of the people that knew anything are very long six feet under and not much of their knowledge made it into the public domain.
I have no interest in writing extensive treatises or in listing my sources (some explicitly asked to not be named, so take it or leave it, no skin off my nose.
Ciao T
Actually, most here know that I design for AMR. So some of what you object to I wrote and that after fairly lengthy and careful research (more than a quick google search).
The curves are not listed on AMR's with with time constants as these can be found in the various literature by anyone so inclined )just make sure to get the right sources, as said Peter Copeland has more right than anyone else I know).
We may debate of the source of why LP's from certain manufacturers have consistent deviations from a subjectively "flat" response, however I came up with something very close to the curves from literature purely by listening via RIAA EQ's Phono's and using an external digital EQ as "re-equaliser".
Both the shapes of the curves I arrived at by ear and my background pushed me towards the conclusion "different EQ".
That you have found no sources on the net does not particulary surprise me, though they do exist, but non in English, most of the people that knew anything are very long six feet under and not much of their knowledge made it into the public domain.
I have no interest in writing extensive treatises or in listing my sources (some explicitly asked to not be named, so take it or leave it, no skin off my nose.
Ciao T
I don't. I seldom visit this forum.
Not for the supposedly different curves for __Stereo__ Decca and DGG records. Please provide a reliable source. They're not on the "sound restoration" manual, AFAIK.
Everybody knows that different curves were used for mono records. I don't object to that.
Ok, so you don't want to list the sources. Then you can't support your claims. If you are a fellow DIY/electronics hobbists, you will understand the great benefit that we all could get if you would provide actual reliable sources of different EQ standards for **stereo** records. Note, i say "standards", not suggested curves determined by ear. I can also create a lot of suggested curves for different record manufacturers. That doesn't mean they don't stick to RIAA.
So please take advantage of this great oportunity to prove me wrong and give us the sources.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.