OK I need to give more throught to the friction measuring then. If nothing else would allow for an interesting datapoint. As I mentioned before my Roksan, although having a very good bearing is not a good candidate without some external drive to get it to speed (or finger) and not sure how the rosewood veneer on MDF would react to multiple sloshings.
All my other projects are DD based* so coast down will be dominated by the motor assembly. I need to think about what I can do.
*trying not to bid on a kenwood KD-5070 on ebay at the moment. I'm intrigued by it as rather than the usual 6 or 8 pole 12 slot motor its a 20 pole 30 slot affair. In theory this should give much reduced torque ripple effects and be a good basis for experimentation in motor control. Oh and the motor should drop into my KD-550 if it is good.
All my other projects are DD based* so coast down will be dominated by the motor assembly. I need to think about what I can do.
*trying not to bid on a kenwood KD-5070 on ebay at the moment. I'm intrigued by it as rather than the usual 6 or 8 pole 12 slot motor its a 20 pole 30 slot affair. In theory this should give much reduced torque ripple effects and be a good basis for experimentation in motor control. Oh and the motor should drop into my KD-550 if it is good.
Yes, part of the art is to minimise equipment and £££ junked by the test 😱 DD can be good because cutting power generally means free spinning. But platter inertia tends to be lower. SL-1200s are good though, and have been shown to work on this method. Ideally a free spindown time of about 1 minute is needed to get reasonably accurate result. Much of the variance comes from having to decide exactly when the stop time is, timing when it has exactly come to rest is not so easy and needs to be done with care.
LD
Hi Bill,
Great that you are willing to replicate my tests.
I'll try to answer your questions as good as I can
1) You will need the 12inch adjust+ disk. I'm not shure whether the LP can be bought without the accompanying software.
If not, you could probably buy a spare LP using my registration.
2) It would of course be nice to see the results for both MM and MC, but I can imagine that this makes things even more complicated for you.
On the other hand, LD warns for ruining the Cart while playing wet.
No Idea how the odds are, but I'm happy to say that my MC is still working.
3) You could definitely use a PC card for real time recording with programs like Arta or the likes. The PC card should be able to sample at 192Khz and have a large dynamic range, preferably over 100dBA.
4) The length of the 1Khz 0dB test tone on the Adjust+ LP is 1 minute.
I took two different types of recordings, the first with a filter bandwidth of 1.5 Hz. In that case I could average 32 FFT samples to smoothen the spectral curve.
The second set of recordings were made with 0.37Hz filter bandwidth, to better see the location of the resonance peak and the IM products around the 1Khz reference tone resulting from the Arm/Cart resonance.
With this narrow filter bandwidth I could only average 11 FFT's, resulting in a spectrum that is far less smoother. So it is exchanging smoothness for spectral resolution.
I also made 1.5Hz recordings of an endless silent loop on the outer side and on the inner side of the LP, having a ratio in diameter of almost 2:1, to find out the difference in noise behaviour.
Last but not least, I made a 1.5Hz filter bandwidth recording of the recording chain with the arm/cart in the air, but platter running, to know the level of this "base noise" in order to be sure it does not influence the measured noise from the LP.
This "base noise", when lying -12dB or more below LP noise, can be left from the equation in that case. This roughly means that the SNR of your Riaa amp should be something like -75dBA, ref 0.5mV at 5cm/sec.
If not, recorded noise from the LP should be compensated for this in later calculations.
5) the Fluid I used is called: Knosti Disco.Antistatic. This fluid is sold for using in LP cleaning machines and comes with the Knosti LP-Cleaning set.
Compound of the fluid is not specified on the bottle, but there is definitely some kind of alcohol in it.
I flooded the LP when playing wet in such a way, that the tracks on the LP were no longer visible and I had to use a scope to see where and when the track started.
6) As you will notice, you will have to do quite a lot of planning in advance, since once played wet, the LP is no longer virgin. So in principle the Dry/Wet tests can only be done once, although I noticed no change at all after the LP had been dried again.
Ears can fool you, but a recording with a digital scope when properly done can't. In my case I used a Picoscope 5243.
Hans
P.S. You can add a decade to your estimate.
A few comments on the nature of stylus-groove friction and its relation to surface noise.
µ, stylus-groove friction coefficient constant, is really an average of a continuously time varying quantity. There is a random variance and standard deviation wrt time. Sort of µ = µk + µv(t), where µv(t) is effectively a dynamic noise element averaging to zero, and µk is a steady state offset or average, effectively a constant.
Conceptually, this is nicely covered and illustrated in Variation in Dry Sliding Friction Coefficients with Velocity, J. Moran & T Sucharitakul, ISBN 978-1-61804-295-8 http://www.inase.org/library/2015/books/bypaper/MMMCE/MMMCE-20.pdf
Whilst µ doesn't generally change much with velocity, its standard deviation does. ie friction noise increases the faster a record rotates. "average friction coefficient was relatively constant with velocity as expected. However, the standard deviation of the friction coefficient changed dramatically with sliding velocity...."
Stylus groove friction force might be modelled like the attachment below. An average force determined by µ, labelled 1, superimposed with noise with peak levels labelled 2. Ignore the scales, it's from a different paper, but same principles apply.
Surface noise only happens when random variation in stylus friction force exceeds a certain threshold, perhaps near peak level (2). In that condition, stylus loses contact with the groove and moves away from at least one of the groove walls momentarily - this is the only way noise can be generated by friction. Below that threshold, the stylus stays in contact with both groove walls and there can be no surface noise. Variations in friction force alone do not move the stylus, it is not free to move, it is retained by groove walls and cantilever pushing down. Unless friction force exceeds a threshold needed to maintain stylus contact with both groove walls simultaneously.
How often critical friction force threshold is breached and surface noise occurs depends on both the average value of µ and its standard deviation.
Both µk and µv(t) depend on groove surface and stylus profile. Threshold for noise depends on cartridge and arm trackability performance. So noise varies from record to record, setup to setup.
Whilst µ doesn't generally change much with record velocity, ie across its surface, its standard deviation does. ie friction force variation increases the faster a record rotates. "average friction coefficient was relatively constant with velocity as expected. However, the standard deviation of the friction coefficient changed dramatically with sliding velocity...." So noise typically varies across the surface, if such noise is there at all.
As to variation in skate force, ie stimulus for cart-arm resonance and headshell instability, this is related directly to variance of friction force, ie its standard deviation. Unlike surface noise though, there is no retention threshold, all variation in friction force shows up as variation in skate force. This is why the LF resonance is such a good marker for stylus-groove friction variance.
It is somewhat complex, but common sense, and yes certainly not 1:1.
LD
Attachments
Last edited:
Just to get a feeling how reliable this test is, I tested 3 things twice:
Arm in the air : resp 53.95 and 56.06 se.
Silent track: resp 51,29 and 52,51
0dB 1Khz : resp 49.75 and 50.92.
Variation between measurements is ca 1.5sec or 3%, so you will have to take many recordings to get a reasonable accurate result and then what does it tell me in the end ?
LP noise when corrected with an anti riaa curve, shows that there are two different noise sources at work. White noise and 1/f noise with a corner frequency somewhere around 250 Hz.
Compared to the outer track, the inner track produced roughly 6dB less 1/f noise but also 6dB more white noise on the Adjust+ LP, so the relation to friction seems at least to be very complex.
Hans
I'm not much help with a lot of the experimental design, but as far as declaring a stop-point for a spin-down, I'd suggest putting a brightly colored sticky-note flag on the outer circumference of the TT platter and use a (mostly overhead, whatever works for you) webcam to go frame-by-frame.
Inexpensive, and far more precise than trying to do it real-time with a stopwatch.
Inexpensive, and far more precise than trying to do it real-time with a stopwatch.
I've got an SL-Q33 which is the same green motor as the 1200. Looking around at threads on here seems you can get the spin down to around 45 seconds. Take the magnet out and that should give a bit more. By reference the roksan has about a 20 minute run down with no load.
As for timing I was thinking of monitoring the FG feed as that is around 300 pulses per revolution and I need to test that as part of some other testing I want to do.
Silly question, as friction is related to velocity can we not start of from a non-standard velocity? 100RPM to 20RPM would seem to be better than 33 1/3 to zero assuming the needle will stay in the groove without too much VTF?
(can you tell who, as a kid used to go 78 to 16 on their parents SP-25 for the funny sounds it made).
As for timing I was thinking of monitoring the FG feed as that is around 300 pulses per revolution and I need to test that as part of some other testing I want to do.
Silly question, as friction is related to velocity can we not start of from a non-standard velocity? 100RPM to 20RPM would seem to be better than 33 1/3 to zero assuming the needle will stay in the groove without too much VTF?
(can you tell who, as a kid used to go 78 to 16 on their parents SP-25 for the funny sounds it made).
LD,
I'm sorry, we are simply living in two different worlds "and never the twain shall meet".
Hans
Yoof assumes I have a webcam 😛
Hans: Thank you for the information. You are of course right it will take a lot longer than my estimates. I've had a miniDSP waiting for me to drill 2 holes to fit the VOL-FP for nearly 6 months! But this is interesting and I have no preconceptions as to what answers I will get. I am also mulling over a cheapy ultrasonic cleaner. I can't afford the 60kHz units that some say are the only way to get things clean. But people report amazing reductions in surface noise after an ultrasound bath, yet haven't measured before and after.
Hans: Thank you for the information. You are of course right it will take a lot longer than my estimates. I've had a miniDSP waiting for me to drill 2 holes to fit the VOL-FP for nearly 6 months! But this is interesting and I have no preconceptions as to what answers I will get. I am also mulling over a cheapy ultrasonic cleaner. I can't afford the 60kHz units that some say are the only way to get things clean. But people report amazing reductions in surface noise after an ultrasound bath, yet haven't measured before and after.
You've taken what I posted out of context, Hans. There are many irreducible causes of surface noise. But that arising from friction, typically crackle-pop, only happens when friction instantaneously exceeds a threshold whereby stylus momentarily loses contact with at least one groove wall.
LD
Well no, you have obtained another thoroughly bizarre result, Hans. Typically spin down time for a 1.6kg solid platter might change from c 45s to c 24s, and then accuracy is quite reasonable.
So, once again, your result is just tough to explain, Hans. Don't take my word for it, check out Pardee's paper for confirmation, and JVC published figures for friction coefficient which tally.
Unless you have some super inertia platter or have achieved friction coefficient well below published and accepted norms, there must be some mistake, again.........
LD
Static and dynamic friction are two different things. I remember as a young kid watching Mr. Wizard on the telly explaining the difference with models he made. Can you imagine today a kids show with that kind of content.
Hi Scott,
Yes, static and dynamic friction are two different animals.
To my humble opinion in case of Needle versus LP wall we only have to do with dynamic or kinetic friction (as mentioned in #261).
Therefore I have a serious problem with "Below that threshold, the stylus stays in contact with both groove walls and there can be no surface noise".
When the stylus stays in contact with both groove walls there is of course friction. Friction resulting in energy consumption, partly converted into heat and into a mechanical movement of the cart, being the surface noise.
When the needle is working like a madman on a 0dB 1Khz track with 70G acceleration max, LF noise is the same as with a blank trace with no acceleration at all.
This makes the assumption that the needle is loosing the contact with the wall every now and then even more unlikely to my opninion.
Hans
This only confirms the suggestion that in my world other physical laws might be at work, I cannot explain it in a different way.
And as a matter of fact, spending $ 33,- for an AES paper, is a bit too much asked.
Hans
P.S. I already mentioned a number of postings ago that I use a heavy peripheral ring (2140 gram) to keep LP's flat on the platter of my VPI Scoutmaster.
This peripheral ring acts more than anything else as a flywheel.
A spare 10 mins of my life I'll never get back with excel suggests your stylus-groove friction coefficient (µ) might be c 0.02, Hans. That would be about an order of magnitude lower than published, peer reviewed figures range of 0.2 - 0.55 from JVC, which range is supported by Pardee's 1981 paper, and my own measurements in dozens of tests.
It would be truly remarkable if you could rock up with your 1st such measurement and obtain such a result! However, it would be consistent with your other remarkable results here suggesting ultra-low friction in that setup and with that record, if we have faith strong enough to believe............
To estimate your stylus-groove friction coefficient I assumed platter inertia equivalent to a 1.6kg uniform solid 15cm radius disc, 2g VTF, 14cm radius for the test track and 33rpm. As close to Bill's uniform spherical cow as we will ever get, probably 🙂 This is estimated because you didn't tell us anything specific, Hans. But even if all of the estimates were reasonably wrong in the same direction, your friction coefficient is still lower than ever seen or published IME, Hans.
I strongly suspect something is wrong, again........or quite remarkable entirely by accident. It is well worth exploring IMO, just in case it is the latter, and there is something to learn, and put in a bottle 😕
So how about starting by revealing the dimensions, mass and distribution of your platter, VTF, and radius of the test track, Hans ? And how you overcame the problem you said you had before which prevented your platter from switching to freewheel after drive?
LD
PS: posts crossed, just saw you use a 2kg peripheral ring which I think explains much - will post an update LD
Last edited:
Well, that explains the bizarre result, but also reveals that your spindle bearing friction torque is unusually high, about x4 normal, which might well be a sign of poor health or overload.
As Bill posted, your freewheel stopping time is unusually short given such high platter inertia, for whatever reason. Perhaps you have not slipped the belt, or the spindle bearing is unhappy ? Your spindle bearing friction torque is about x10 bigger than torque arising from stylus-groove friction, hence your bizarre result, Hans. That is the abnormality, for sure. Remove the peripheral ring...........
For a moment there I was starting to believe in miracles........🙄
LD
Last edited:
What if Hans' TT has sick spindle bearing syndrome, as it seems to, and consequently he is observing spindle bearing noise rather than surface noise as the noise floor ? I think that might fit with the weird observations? Wet/dry playback tests might then not alter noise floor due to masking, and LF from spindle noise might swamp variation in skate force as stimulus for cart-arm resonance, masking a major marker of wet/dry playback tests ?
I feel a sense of harmony about this.......3 for the price of 1, it has to be something like this.
Freewheel rundown time for a platter with that high inertia and normal spindle bearing friction torque should be c 180s, says a swift calculation. Not c 50s.
LD
Last edited:
I don't have the feeling that too much can be wrong with the spindle, but I'll checkt it.
I think that might fit with the weird observations? Wet/dry playback tests might then not alter noise floor due to masking, and LF from spindle noise might swamp variation in skate force as stimulus for cart-arm resonance, masking a major marker of wet/dry playback tests ?
I feel a sense of harmony about this.......3 for the price of 1, it has to be something like this.
Freewheel rundown time for a platter with that high inertia and normal spindle bearing friction torque should be c 180s, says a swift calculation. Not c 50s.
LD
I have never cleaned or oiled it in 10 years.
Remember that I calculated a 63dBA dynamic range with my TT, quite close to the 66dBA theoretical maximum.
And it would be quite odd if spindle noise would have exactly the shape of Arm/Cart resonance.
As a matter of fact, I never mentioned a problem which prevented my platter from switching to freewheel after drive.
And of course the recorded times are all with belt removed, nothing else but the spindle causing friction.
Hans
Shape of the resonant response is a natural phenomenum; all that's needed in stimulus from spindle friction noise, such as noise in the sub-audio band near the resonance, which seems very plausible.
Interesting to note in that EMT article I linked that they used wet playback for evaluation of TT performance.........that wouldn't be for the fun of it. And how could we forget that we still can't be certain about absolute 0dB level in your test, Hans?
OK, I could never see why it shouldn't be straightforward to slip the belt on your TT. But freewheel spindown time is a healthcheck for spindle bearings. Perhaps to do with overload in this case?
Spindle friction torque should not totally swamp stylus-groove friction torque, as it does here. The test at hand will struggle for accuracy, but you have bigger fish to catch if that is that case.
LD