That’s correct Richard.
I don’t know how high distortion can be at 2.5Vpp across the cap. Maybe you can give a number.
From what I’ve seen so far, I can not expect anything less that 1.5Vrms at 10Hz from a 1mV per 1cm/s MM cartridge. So the successful step is to build a phono stage with an adequate OLM
“Sound better” may be attributed to some other factors as well
George
Sound better --- just cut in/out that 20Hz filter and listen or measure. Digital if you like. [The phase wont be heard but group-delay could be more important.]
I recall owning gear that had just such a LF switch garbage roll-off. But beyond just the RIAA --
Of course, there is also the fact of low freq garbage passing thru to the PA and speaker ---- even if the speaker could not reproduce it.... a PA amp'ed up/exercised with ultra low freqs and heat and extra excursion will drive up a comparative difference in sound.... [not only a preamp OLM issue].
THx-RNMarsh
Last edited:
It doesn't matter... the low freq TT/LP junk (non-music <20Hz) is enough to cause issues..... using the 1mv number at 10hz given.... with the typical gain in phono, line and PA, how many volts and Watts are exercised in the amp and speaker? Just do the arith.
Because of this LF junk, T.Holman found the OLM needed to be at least 35mv rms in the 3-4Hz range.
THx-RNMarsh
Because of this LF junk, T.Holman found the OLM needed to be at least 35mv rms in the 3-4Hz range.
THx-RNMarsh
Last edited:
No the driver did something. It's totally bizarre if and only if I drop a <3sec file onto Cooledit it opens and freezes, completely repeatable after reinstalling Cooledit. If I open a large file first no problem. I have regularly worked with HUGE sound files with no problems.
If you run an IIR filter from the back of a file to the front and the same one from the front to the back you get double the filter and linear (no) phase. This only works well with even order filters but should work well for 4 or 6 poles at 20Hz. I gets tricky in that IIR filters with poles very near the unit circle tend to have coefficients many orders of magnitude apart. Works fine in SoX and Audacity (in Nyquist) because they are FP. I hope to put all this in my article.
I emailed Gary Gallo and he was not aware that FIR filters can be minimum phase but yes they can it's just that the crossover and room equalizer guys dominate the conversations.
Last edited:
is there a some what simple iir implementation in HW (IC) to do the RIAA 20Hz cutoff without introducing audible group delay?
Use the AES figure for audible GD at low freqs.
Some similar info is here -- http://www.aes.org/e-lib/browse.cfm?elib=4441
THx-RNMarsh
Use the AES figure for audible GD at low freqs.
Some similar info is here -- http://www.aes.org/e-lib/browse.cfm?elib=4441
THx-RNMarsh
Last edited:
There are several options for huge real time FIR filters. I would assume one could account for the processing delay.
Although I think there is a lot of good in the Holman paper, his reasoning about cartridge loading per the manufacturer's recommendations as needing to be included in the RIAA response is a bit perverse IMO.
However, to what can be assented is that the preamp should allow accurate loading, and look like a resistance (usually 47k for MM) and have a well-behaved capacitance, one hopes a small one from the active device(s) at the input, so that adding the cable and tonearm capacitance to this does not exceed the recommended loading, and additional C can be added if required.
I wish Tom had included some values on his schematic. I'm somewhat dubious about the two-pole filter components right at the input. Holman also indicates a belief that bipolars are most suitable at the time of writing, but fails to discuss base current noise. It's interesting that he acknowledges cooled terminations (he calls it a synthesized input impedance via bootstrapping), although he doesn't use one and doesn't say whose product did, although mentions its implementation via an ancillary feedback stage. Van de Gevel points out that this is suboptimal in a Linear Audio article a while back.
Here Dick is a 34490 point 5 pole Butterworth @20 Hz FIR with a constant group delay of 182 msec. There are a lot of places where latency is an issue but I don't think putting on an LP just to listen is one of them and certainly with offline processing for later listening it's not a problem.
The scale here is dB and linear Hz. I have been playing with some brute force techniques for computing filters since the compute power is there. From the sourceforge VST convolver read me...
Performance is excellent, possibly the best available under Windows, and subject continual improvement. A stereo 65536-tap filter, the largest that makes sense when applied to a 44.1kHz source, executes at 40 times real time, representing a 3% cpu hit, on a 3.4GHz Pentium 4. Even on a 300MHz Pentium II, the reported cpu hit is about 30% when convolving with such a filter. So your old machine can be put to good use. Mixing channels results in some slowdown (six 65536-tap filters will consume less than 10% cpu on a 3.4GHz Pentium 4).
Arbitrary-length convolutions for unusual applications (1 million tap limit imposed only as a sanity check)
Multi-channel input and output, 8, 16, 20, 24 and 32-bit PCM and 32 and 64-bit IEEE Float
Mixing, scaling and delay of both input and output channels (eg, for "true stereo" convolution, or cross-talk cancellation)
Dither and noise shape of 8, 16, 20 and 24-bit output
Wide range of filter file formats accepted (Microsoft WAV, SGI/Apple AIFF/AIFC, Sun AU/Snd, Raw (headerless 32-bit IEEE float), Paris Audio File (PAF), Commodore IFF/SVX, Sphere/NIST WAV, IRCAM SF, Creative VOC, SoundForge W64, GNU Octave MAT4/5, Portable Voice Format, Fasttracker 2 XI, HMM Tool Kit HTK)
Sample encodings supported include unsigned and signed 8, 16, 24 and 32 bit PCM, IEEE 32 and 64 floating point, U-LAW, A-LAW, IMA ADPCM, MS ADPCM, GSM 6.10, G721/723 ADPCM, 12/16/24 bit DWVWk, OK Dialogic ADPCM, and 8/16 DPCM. Wavpack files are not currently supported.
Windows Media Player plug-in (DMO), DirectShow filter and VST plug-in interfaces
Several filters can be loaded at once. The first to match the playback format (channels, sample rate) is automatically selected.
EDIT - Forgot this example is for 96K at 24 bits with all filter coefficients kept that are over 1/2 an LSB at 24bits.
The scale here is dB and linear Hz. I have been playing with some brute force techniques for computing filters since the compute power is there. From the sourceforge VST convolver read me...
Performance is excellent, possibly the best available under Windows, and subject continual improvement. A stereo 65536-tap filter, the largest that makes sense when applied to a 44.1kHz source, executes at 40 times real time, representing a 3% cpu hit, on a 3.4GHz Pentium 4. Even on a 300MHz Pentium II, the reported cpu hit is about 30% when convolving with such a filter. So your old machine can be put to good use. Mixing channels results in some slowdown (six 65536-tap filters will consume less than 10% cpu on a 3.4GHz Pentium 4).
Arbitrary-length convolutions for unusual applications (1 million tap limit imposed only as a sanity check)
Multi-channel input and output, 8, 16, 20, 24 and 32-bit PCM and 32 and 64-bit IEEE Float
Mixing, scaling and delay of both input and output channels (eg, for "true stereo" convolution, or cross-talk cancellation)
Dither and noise shape of 8, 16, 20 and 24-bit output
Wide range of filter file formats accepted (Microsoft WAV, SGI/Apple AIFF/AIFC, Sun AU/Snd, Raw (headerless 32-bit IEEE float), Paris Audio File (PAF), Commodore IFF/SVX, Sphere/NIST WAV, IRCAM SF, Creative VOC, SoundForge W64, GNU Octave MAT4/5, Portable Voice Format, Fasttracker 2 XI, HMM Tool Kit HTK)
Sample encodings supported include unsigned and signed 8, 16, 24 and 32 bit PCM, IEEE 32 and 64 floating point, U-LAW, A-LAW, IMA ADPCM, MS ADPCM, GSM 6.10, G721/723 ADPCM, 12/16/24 bit DWVWk, OK Dialogic ADPCM, and 8/16 DPCM. Wavpack files are not currently supported.
Windows Media Player plug-in (DMO), DirectShow filter and VST plug-in interfaces
Several filters can be loaded at once. The first to match the playback format (channels, sample rate) is automatically selected.
EDIT - Forgot this example is for 96K at 24 bits with all filter coefficients kept that are over 1/2 an LSB at 24bits.
Attachments
Last edited:
I think I related the story of the JBL Pro development of a very high-order digital-domain electronic crossover for PA apps. Two million was spent before the prototype got into the hands of performers and their audio crews, and it was discovered that the ~100ms latency was unacceptable.
And that was back when 2 million was a lot of money.
Jeez I would have thought that was a no brainer, my guitarist friends can't stand 5ms when playing along with a processed track.
Right now my spare time is consumed trying to finish my article on this, but the possibility is certainly there.
BTW using chip amps in anger, a wonderful concept, hats off.
Last edited:
In the Apt preamp manual there's some more circuit detail. http://www.amplimos.it/images/apt-holman-preamplifier-service-manual.pdf
The proponents of digital-domain processing were selling the notion very hard to a bunch of traditionalists, and really didn't have a whole lot of experience themselves at that time. There was also a good deal of ignorance about what you had to do to convey high-speed signals on circuit boards.
So it goes.
Thanks. That's a very good manual! Quite a departure from the schematic shown in the paper. In particular, he has lost the input highpass filter.
However, except for the mention of ..."noise generated by the phono cartridge nor by the amplifier's own input impedance..." there appears to be no mention of parallel (i.e. current) noise. But he has migrated in this actual product to a JFET as the input device, whose gate leakage and its associated noise are negligible in this application. It is odd that the particular benefits of this are not discussed.
Yes, that's about the best manual I've seen. The new factors paper was written earlier than the Apt preamp, when Tom was at Advent.
I know him from being in EE classes together at the University of Illinois-Urbana. He's now at Apple. Here's the schematic for the Advent receiver.
http://www.davidreaton.com/PDFs/Advent_300_newer_Amplifier_schematic.pdf
That is close.... needs more attenuation at 10hz, however. GD is fine as a constant amount. What I was wondering is if there was a HW solution (without using a PC). Like a small card maybe with a dsp and supporting IC's with built-in memory et al. Small enough to put inside a phono preamp chassis. P-n-P. Is that practical to do? Maybe something from a mini-dsp used for speaker cross-overs exists which can be tried/used?
THx-RNMarsh
Last edited:
Possibly some 32bit Sharks could do it. One problem folks don't realize is that a Pentium processor which can do these things no problem is way more expensive than the DSP chips. In addition Intel keeps some of the pipeline instructions secret so you need to buy their compilers to optimize your code. Check out BruteFIR for some serious geek value.
Yes, I know Tom (distantly). I hope he's getting a pile of dough for leaving Audyssey for Apple.
Brad
- Status
- Not open for further replies.