As I have said several times the infrasonic signal frequency modulates the wanted signal. Attached a couple of diagrams from a B&K research paper from the 70s. You can see the modulation.
Attachments
Last edited:
Hi Dan,
When I was reading about the RIAA Pre-Emphasis, I did wonder how the response could just keep going up and up. Obviously, it can't. The other side of that question, is what to do on the De-Emphasis on the replay side. I must admit that I've never tried to flatten the curve out in my home-brew phono stages because there didn't seem to be a "right" place to do it, and besides, that's way past what I can hear anyway, etc.
It's a shame RIAA didn't address this at some point. Or have they? I'm guessing that some people with "Golden Ears" might be able to hear the difference as it affects the the frequency response above 16KHz.
If someone was to ask you for advice on RIAA de-emphasis time-constants (or however it is most appropriate to phrase the question), what would you tell them?
Probably that people WAY over think gram amps, the curve on cutting is hacked about with to sort out the worst of the spring back issues and cutter head resonances, but is generally a bit random above about 18kHz anyway. Rolling off forever is not going to be very wrong, and is minimum phase.
In reality what happens is that you make test cuts and play them on a decent but not highly tweaky system and adjust and cut new tests until you get something that sounds good. This is typically done when doing the periodic service and any time you have to tear down the cutting head. Accuracy mostly plays second fiddle to sounds good and doesn't risk burning out the drive coils (A real issue when cutting hot, even with helium in the gap).
Then you cut a full test disk and make sure you can play it both on a decent system and on a crossley (The record has to track ok on the junk as well). Incidentally the crossley use case means that you pretty much cannot cut assuming a line contact stylus because the resulting HF will be unplayable on a cheap deck.
I would suggest that the usual 5534 with feedback network is about what the guys cutting are generally using, so if you actually want to hear what the studio is hearing.... Generally something in the order of an SME arm, Thorens deck nothing extraordinary (Personally I have a micro Senki, arm (Because it is what I had), but whatever works).
There probably are a few EMTs out there still in playback duty, but that is not the way to bet.
If you are doing archival stuff, if can be worth setting up a chain that has a flat pickup amplifier going straight into a soundcard with LOADS of headroom, that way your click detector is not fighting against 40dB of HF attenuation, and all the really weird curves from the early days can be trivially dealt with by a couple of biquad sections.
It is a weird reality of hifi that for the most part the guys making the recordings tend to be rather 'close enough' in their outlook, for tweakyness that would almost never occur to a working studio cat, you got to look at peoples replay chains.
Rumble on replay decks is generally lower then on the lathes, but a replay deck does not have to try to maintain constant speed while cutting, there is a reason the lathes go for HEAVY platters on machines that pre date direct servo drives, and a bearing that can take 60kg of platter is going to be noisier then one made for a relatively light replay deck. In any case lathes add rumble as part of the groove spacing optimisation anyway, so vertical motion in the bearing is of limited import.
My 'Corona' project is a lathe based on a 18n/m brushless direct drive servo motor from NSK, should come up to speed impressively quickly and with 2M counts per rev on the encoder the servo should lock HARD to the target speed.
In reality what happens is that you make test cuts and play them on a decent but not highly tweaky system and adjust and cut new tests until you get something that sounds good. This is typically done when doing the periodic service and any time you have to tear down the cutting head. Accuracy mostly plays second fiddle to sounds good and doesn't risk burning out the drive coils (A real issue when cutting hot, even with helium in the gap).
Then you cut a full test disk and make sure you can play it both on a decent system and on a crossley (The record has to track ok on the junk as well). Incidentally the crossley use case means that you pretty much cannot cut assuming a line contact stylus because the resulting HF will be unplayable on a cheap deck.
I would suggest that the usual 5534 with feedback network is about what the guys cutting are generally using, so if you actually want to hear what the studio is hearing.... Generally something in the order of an SME arm, Thorens deck nothing extraordinary (Personally I have a micro Senki, arm (Because it is what I had), but whatever works).
There probably are a few EMTs out there still in playback duty, but that is not the way to bet.
If you are doing archival stuff, if can be worth setting up a chain that has a flat pickup amplifier going straight into a soundcard with LOADS of headroom, that way your click detector is not fighting against 40dB of HF attenuation, and all the really weird curves from the early days can be trivially dealt with by a couple of biquad sections.
It is a weird reality of hifi that for the most part the guys making the recordings tend to be rather 'close enough' in their outlook, for tweakyness that would almost never occur to a working studio cat, you got to look at peoples replay chains.
Rumble on replay decks is generally lower then on the lathes, but a replay deck does not have to try to maintain constant speed while cutting, there is a reason the lathes go for HEAVY platters on machines that pre date direct servo drives, and a bearing that can take 60kg of platter is going to be noisier then one made for a relatively light replay deck. In any case lathes add rumble as part of the groove spacing optimisation anyway, so vertical motion in the bearing is of limited import.
My 'Corona' project is a lathe based on a 18n/m brushless direct drive servo motor from NSK, should come up to speed impressively quickly and with 2M counts per rev on the encoder the servo should lock HARD to the target speed.
Last edited:
Amplitude Modulation Effects or Dopplering?
Hi Bill,
The engineers at B&K are wizards!
You've convinced me that taming the infrasonics is very important.
What is your preferred turntable? Have you measured it and improved it? My everyday TT is a Denon DP-45 that just keeps rolling along.
Hi Bill,
The engineers at B&K are wizards!
You've convinced me that taming the infrasonics is very important.
What is your preferred turntable? Have you measured it and improved it? My everyday TT is a Denon DP-45 that just keeps rolling along.
I have two turntables. Neither are currently improved. Second brood is putting paid to that at the moment. All self inflicted though 
. I have a Roksan Xerxes with SME 309 (upgraded with damping trough from V) and a Kenwood KD550. The Kenwood has been in the family since new and I effectively inherited it, the Roksan I bought nearly 30 years ago now. I hate to admit the kenwood is probably the better turntable. Right now I am collecting parts for when I 'do' have time.
Regarding the vertical unwanted components. This is not really rumble, but IMO should at least be considered in the replay chain. I looked at the self devinyliser and was unsure if that adressed the real issue so looked further afield and discovered elliptic filters as used in mastering. Confusing name as they have nothing to do with what EEs would call an elliptic filter but were so called as Neumann lathes had a Goniometer on them to check for vertical signal levels. The ideal was a nice ellipse, hence elliptical filter. In fact the Neumann design was a first order cross feed, so reducing the signal to mono below a certain point.
I then discovered Wayne Kirkwood, who sometimes posts on here had developed a second order elliptical for 21st century mastering engineers. Overly complex and lots of opamps would offend lots of delicate souls so I figured I had to have one. Rather than cut and paste here is a link to the circuit and a description of operation. A Second Order Elliptic Equalizer for Vinyl Mastering - Page 5 - Pro Audio Design Forum
I already have a L-R to M-S conversion in my phono preamp so It's the core 6 opamps I would be building as time allows. There are much simpler ways to do this, but in for a penny in for a pound!
The reason I am building this is to just see if reducing cross talk below 150Hz gives any audible benefit. I would be very happy if I found it didn't but if you don't try you don't know.
. I have a Roksan Xerxes with SME 309 (upgraded with damping trough from V) and a Kenwood KD550. The Kenwood has been in the family since new and I effectively inherited it, the Roksan I bought nearly 30 years ago now. I hate to admit the kenwood is probably the better turntable. Right now I am collecting parts for when I 'do' have time. Regarding the vertical unwanted components. This is not really rumble, but IMO should at least be considered in the replay chain. I looked at the self devinyliser and was unsure if that adressed the real issue so looked further afield and discovered elliptic filters as used in mastering. Confusing name as they have nothing to do with what EEs would call an elliptic filter but were so called as Neumann lathes had a Goniometer on them to check for vertical signal levels. The ideal was a nice ellipse, hence elliptical filter. In fact the Neumann design was a first order cross feed, so reducing the signal to mono below a certain point.
I then discovered Wayne Kirkwood, who sometimes posts on here had developed a second order elliptical for 21st century mastering engineers. Overly complex and lots of opamps would offend lots of delicate souls so I figured I had to have one. Rather than cut and paste here is a link to the circuit and a description of operation. A Second Order Elliptic Equalizer for Vinyl Mastering - Page 5 - Pro Audio Design Forum
I already have a L-R to M-S conversion in my phono preamp so It's the core 6 opamps I would be building as time allows. There are much simpler ways to do this, but in for a penny in for a pound!
The reason I am building this is to just see if reducing cross talk below 150Hz gives any audible benefit. I would be very happy if I found it didn't but if you don't try you don't know.
Thanks Dan, There is a lot of "art" in what you do as well as the technical stuff dealing with these very complex and expensive machines.
My first deck (1966) was a BSR with a Sonotone 9TAHC - down at the Crossley end of the market! Most of us built Mullard tube amps from kits and used EMI 13.5X9 ellipticals (with the whizzer cones!) in very live boxes (also home built). This was definitely several up on a Dansette! Anyway, I loved it!!
When will your new system be ready? When it goes live, I'd like to hear the some of the records.
Going to be a while, lots of heavy mechanical stuff that I cannot get to the workshop to do at the moment (Machine tools, lathes included are generally much better when built HEAVY).
Got most of the bits, but there is a lot of steel to cut and software to write (Motor control stuff mainly) as well as a fair amount of electronics still to design.
Then there is the problem with sourcing lacquers at the moment (The Apollo Masters plant burnt down a few weeks ago and the one remaining company cannot come close to meeting demand), this IS going to put a crimp in new releases sooner or later.
I have some ideas, but nobody quite knows what the real solution to that is going to be.
Ah, Dansette, shudder!
Got most of the bits, but there is a lot of steel to cut and software to write (Motor control stuff mainly) as well as a fair amount of electronics still to design.
Then there is the problem with sourcing lacquers at the moment (The Apollo Masters plant burnt down a few weeks ago and the one remaining company cannot come close to meeting demand), this IS going to put a crimp in new releases sooner or later.
I have some ideas, but nobody quite knows what the real solution to that is going to be.
Ah, Dansette, shudder!
Those Sonotone 9TA's were and are light years ahead of anything Crosley.
Don't think they were crap - they performed quite well compared to their current MM offerings.
I had one, with the RIAA adapter plugs, back in the 1960's, tracking on an old Garrard at 2 grams into a Bell tube amplifier, and it satisfied me nicely.
Hi WOT, it's fun to reminisce about the old days. The UK recovered from WW2 much more slowly than the US and everything seemed expensive and with far less choice. The cost of the Sonotone was a bit more than my weekly food budget as an undergrad!
As I remember it, I think, I paid about $14.95 US for that Sonotone 9TA cartridge, including the magnetic RIAA adapters.
To me, at the time, it was a substantial cost, but I was young.
Nevertheless, it sounded wonderful to me.
And I give Sonotone high praises for producing such a nice piece of equipment.
However dated by now.
Slightly off topic, but you can approximate a 1st order rising slope up to a frequency fmax pretty well while limiting the boosting of ultrasonics above fmax a bit by this method:
1. Make a simple RC network of which the response starts rising wherever it needs to start rising and levels off at a frequency fL. fL must be somewhat above fmax, say 1.5 to 2 fmax.
2. Use the pole at -2 pi fL that you have created as the real pole of an odd-order low-pass filter and add stages that create the complex pole pairs.
For example, when fmax = 20 kHz, the response needs to rise from f = 1/(2 pi 75us), fL = 38 kHz, and the low-pass function is third-order Butterworth:
1. Make a network consisting of a resistive voltage divider and a capacitor, for example:
R1 in out 15.95744681kohm
R2 out ground 943.8323221ohm
C1 in out 4.7nF
This network will have a response that increases from 1/(2 pi 75 us) ~= 2122 Hz onwards and stops increasing around 38 kHz, but it will already be 1.0619 dB lower than the theoretical infinitely rising response at 20 kHz.
2. Add a second-order low-pass stage with a natural frequency of 38 kHz and a Q of 1, this realizes the complex pole pair of a third-order Butterworth filter.
1. and 2. combined will have a response that is only 0.09135 dB below the theoretical infinitely rising response at 20 kHz. Above 38 kHz, the response goes down with a second-order slope, so extreme ultrasonics get attenuated rather than boosted.
I was 15 save money from my parents for one years then in xmas I bought my first TT was a Pioneer PL-112d, I do not remember exactly the cartridge was a Shure 44 or 45, I diy a pre-amp with TBA231 and the amp with 2N3055, lol...nice memories, and of course diy speaker with Philips drivers, then later I got a SUV6 Technics amp and Jensen speaker, well the TT Pioneer last for 25 years and maybe more, when I left Italy 20 years ago was still working perfectly, I just change the belt 3/4 times in 25 years, what now amazing me is that I never had rumble problem with the PL-112 but with my new TEAC TN-300 yes, so I think the quality and care of old TT was superior in the past when listening music was just possible with vinyl, radio and tapes....all the best
I'm not surprised that your PL-112D when cared for and maintained, out-performs the new "junk".
In the past, bearings were designed and micro-polished with care, quality control inspections were rigid and strict, and company reputations were very important.
THAT WAS THEIR MAIN FOCUS - with many competing companies, they HAD to try their hardest to stay in business. And having their factories in-house, they could monitor things the "old school way".
There was none of that "farm it out to china and save money" stuff going on.
Advertising claims were truthful and accurate. - not designed to dazzle or impress a customer with meaningless junk-talk.
My older Kenwood turntable, not even a "top line" model, has a platter bearing assembly that simply amazes me. Absolutely no rumble, no bearing noise - when spun with a finger to a few hundred RPMs, it's totally silent - no "whirring sound" at all with my ear almost touching the spindle.
So at 33 1/3 RPM, I'm sure it's superior to most of the stuff out today.
Woth reading what the Korf audio guys are doing with accelerometers. Blog Table of Contents
I don't agree with all their conclusions but some very interesting findings.
I don't agree with all their conclusions but some very interesting findings.
Hi Tubewaller, I am wondering, probably need to dig more into this, why don't you just low pass R channel and feed this signal to a buffered Left (after the buffer clearly) and do the same for L to R? This way you should only need two low passes and the buffers.
Hi Gluca, All the LR filters (2 HP, 1 LP) are 4th order and need 2 opamps each. The beauty of the LR filters is the way the phases of the 2 signals add at the crossover points. The frequency response is flat and doesn't have a notch or hump. Incidentally, the filters consist of 2 identical Butterworth 2nd order filters (also called "sync" tuned). This does NOT produce the ideal maximally flat response you'd expect from a 4th order Butterworth (it was the Riley of Linkwitz Riley that discovered this clever feature).
Gluca, do you have simulation software? The easiest thing is to simulate your circuit and see if it does what you expect and want from it. Please share your circuit and the responses with us on this site. Thanks.
Gluca, do you have simulation software? The easiest thing is to simulate your circuit and see if it does what you expect and want from it. Please share your circuit and the responses with us on this site. Thanks.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.