Need help with interpreting turntable strobe
August Helmbright
the speed. Unfortunately, the adjustements with the slider bar are
rather coarse, and I can't get the speed to be perfect - at best I get
it ever so slightly slow or fast. In the locked center position on the
slider bar, the table is slightly slow.
Since I only use the turntable for input into a computer and process
the sound files with DC-V filtering software for making CD-Rs (I never
actually "listen" to records), I wonder if someone can confirm my
thinking on how to determine how much the adjust the speed with my
sound filtering software, as follows. The example below assumes a
correction for a turntable that is too slow. It also deals only with
33 1/3. If my thinking is right, the same process could be adapted for
45 RPM, or modified to correct a speed that is too fast.
a. There are 130 dots around the circumference of my turntable for
33 1/3 RPM on the 60Hz band.
b. While a record is playing, to make sure stylus / record friction
is taken into account, I use a stopwatch to time the amount of time it
takes for 130 dots to move past the strobe counter-clockwise, to judge
how much time it takes for the turntable to lose one full turn
relative to correct speed. While it is possible I could be off by a
fraction of a second due to measurement errors(holding my head still,
etc.), I suspect this tiny difference would be inaudible.
c. I convert the time it takes to gain or lose a turn to minutes and
decimals.
d. I calculate the number of revolutions that should have occured at
correct speed by multiplying the result in (c) by 33 1/3.
e. To the result in (d), I subract 1 to arrive at the number of
revolutions at the turntable's actual speed.
f. I calculate the percentage slow by dividing the difference
between (d) and (e) by (e). [Not by (d), because the "percent change"
to apply to the sound file with filtering software is based on
correcting the actual file, which is at (e) speed.]
g. I run the speed filter in DC-V and reduce speed by the percent
(in my practice it's about a quarter of a percent) obtained in (f).
A quarter of a percent slow doesn't sound like much, but it does have
the effect of turning A=440 into A=438.9, and I'd rather know that
when I make CD-Rs, they're correctly pitched relative to the source
(assuming the studio mastering and LP pressing was right).
Goofball_star_dot_etal
Helmbright) wrote:
>I have a new direct drive turntable with built-in strobe to correct
>the speed.
Your strobe and soundcard could be "wrong" . . . Get an old record
and cut a light radial groove with a sharp knife. Some prefer a
"locked groove", note. Count the number of samples between clicks on a
recording using your sound card. For your purpose you need to know
samples/ rev. and doing it this way you eliminate other possible
errors.
Direct drive you say. . . You don't happen to have a test record with
a nice little 300 Hz test tone, do you? (Said the spider to the fly )
August Helmbright
No I have no test records.
I don't understand what you mean by "counting the number of samples
between clicks." Could you give me an example? Also, wouldn't this
method be prone to rather large measurement errors? For example, what
if the radial groove isn't *exactly* straight along a diameter line?
Goofball_star_dot_etal
Helmbright) wrote:
>w...@needs.email.anyhow (Goofball_star_dot_etal) wrote in message news:<3fd612a9....@usenet.plus.net>...
>> On 9 Dec 2003 08:34:01 -0800, augusthe...@msn.com (August
>> Helmbright) wrote:
>>
>> >I have a new direct drive turntable with built-in strobe to correct
>> >the speed.
>>
>> Your strobe and soundcard could be "wrong" . . . Get an old record
>> and cut a light radial groove with a sharp knife. Some prefer a
>> "locked groove", note. Count the number of samples between clicks on a
>> recording using your sound card. For your purpose you need to know
>> samples/ rev. and doing it this way you eliminate other possible
>> errors.
>>
>> Direct drive you say. . . You don't happen to have a test record with
>> a nice little 300 Hz test tone, do you? (Said the spider to the fly )
>
>No I have no test records.
>
Not to worry, neither do I.
>I don't understand what you mean by "counting the number of samples
>between clicks." Could you give me an example?
Ah, you would need some sort of audio editor to do this such as
CoolEdit . I have heard that it is no longer available but there must
still be plenty of other shareware editors to be had. If you get stuck
you could leave a copy of the wav somewhere and I could count the
samples.
At 33 1/3 rpm the time taken for one revolution is exactly 1.8
seconds. The CD is recorded at 44100 samples/sec so I assume that you
record to .wav at this rate. With the aid of an audio editor you can
see the individual samples in the .wav file and can count the number
of samples between the recorded clicks ( with the aid of a cursor
read out) . At the correct speed this should be 1.8*44100 = 79380
samples.
Also, wouldn't this
>method be prone to rather large measurement errors? For example, what
>if the radial groove isn't *exactly* straight along a diameter line?
If you compare the circumference of the track to the pitch of the
scroll, you can see that the error will be quite small even if you
don't get the ruler quite through the centre. You could use the circle
right at the end of the record do do the same thing but the
circumference is smaller and perhaps the click less sharp, so the
uncertainty in the measurement will be a bit more.
August Helmbright
> >I don't understand what you mean by "counting the number of samples
> >between clicks." Could you give me an example?
>
> Ah, you would need some sort of audio editor to do this such as
> CoolEdit . I have heard that it is no longer available but there must
> still be plenty of other shareware editors to be had. If you get stuck
> you could leave a copy of the wav somewhere and I could count the
> samples.
>
> At 33 1/3 rpm the time taken for one revolution is exactly 1.8
> seconds. The CD is recorded at 44100 samples/sec so I assume that you
> record to .wav at this rate. With the aid of an audio editor you can
> see the individual samples in the .wav file and can count the number
> of samples between the recorded clicks ( with the aid of a cursor
> read out) . At the correct speed this should be 1.8*44100 = 79380
> samples.
>
of a second. I'm guessing I could record a few turns and carefully
mark 2 successive scratches with the "marking tool" in DC-V, then
double click inside the marks to select the portion of the wav file
between scratches. Surely, measuring for 1.8 seconds to within
thousanths of a second would be as good as measuring samples, which
isn't an option in DC-V. After all, one-one thousanth of a second off
from 1.8 seconds is far lower than I can even correct with the speed
filter, which only goes down to +/1 .0001.
Agree?
Goofball_star_dot_etal
Helmbright) wrote:
I am not sure. I don't know what a DC-V is. Give it a try. A few
points:
44100 samples/sec resolution is 44.1 times better than "one-one
thousanth of a second".
Resolution is not the same as accuracy.
You wrote:
"a. There are 130 dots around the circumference of my turntable for
33 1/3 RPM on the 60Hz band." Does this mean the strobe uses mains
(line?) frequency as a timebase? I would have expected a different
number of dots. . .
Do not trust a strobe based on mains frequency to make accurate
measurements.
Why does a direct drive require speed adjustment?
John's dad say : "Ratiometric measurement lot better than kick in
balls."
Alan Peterman
(Goofball_star_dot_etal) wrote:
>
>Do not trust a strobe based on mains frequency to make accurate
>measurements.
>
>Why does a direct drive require speed adjustment?
>
First of all, using a strobe based on main frequency is just fine. Why should
anyone care about being more accurate?? After all before the record was cut
there may be speed errors in the original tape recorder, or even the cutting
table. The accuracy of a strobe on the turntable is plenty accurate unless you
are trying to do some kind of scientific frequency measurement - not just
reproduce music.
Secondly, direct drive turntables nearly ALWAYS need to be checked for speed as
most of them run off DC servo motors. Some of them have quartz locked speed,
but most have a user fine speed adjustment (to correct for the above mentioned
recording chain errors - or even just to tune the music to a personal
guitar/piano/musical instrument) so since they CAN be easily adjusted, they NEED
to have away to check that they've been set nominally "right" when desired.
---------------------------------------------------------------
Alan Peterman al at scn.rain.com Tigard, OR
As I grow older, the days seem longer and the years seem shorter.
Goofball_star_dot_etal
<alno...@scn.rain.com> wrote:
>On Wed, 10 Dec 2003 17:43:25 GMT, w...@needs.email.anyhow
>(Goofball_star_dot_etal) wrote:
>
>
>>
>>Do not trust a strobe based on mains frequency to make accurate
>>measurements.
>>
>>Why does a direct drive require speed adjustment?
>>
>
>First of all, using a strobe based on main frequency is just fine.
Is this statement based upon synchronous motor clocks keeping good
time?
> Why should
>anyone care about being more accurate??
I don't.
After all before the record was cut
>there may be speed errors in the original tape recorder, or even the cutting
>table. The accuracy of a strobe on the turntable is plenty accurate unless you
>are trying to do some kind of scientific frequency measurement - not just
>reproduce music.
>
I just can't help myself.
>Secondly, direct drive turntables nearly ALWAYS need to be checked for speed as
>most of them run off DC servo motors. Some of them have quartz locked speed,
>but most have a user fine speed adjustment (to correct for the above mentioned
>recording chain errors - or even just to tune the music to a personal
>guitar/piano/musical instrument) so since they CAN be easily adjusted, they NEED
>to have away to check that they've been set nominally "right" when desired.
OK,. so they are crap.
Goofball_star_dot_etal
<alno...@scn.rain.com> wrote:
>
>First of all, using a strobe based on main frequency is just fine.
In the UK:
(3) For the purposes of this regulation, unless otherwise agreed in
writing by those persons specified in paragraph (2), the permitted
variations are -
(a) a variation not exceeding 1 per cent above or below the declared
frequency;
http://www.hmso.gov.uk/si/si2002/20022665.htm#27
This is not accurate enough to measure a turntable speed error of:
"(in my practice it's about a quarter of a percent)"
Message-ID: <12cbb4d6.03120...@posting.google.com>
henryf
frequency in the UK must be at least accurate within 1%. It
might always be very much better than that, for reasons other
than legal ones.
To put it another way, suppose a group of people were willing
to bet on the outcome of a mains frequency measurement. Based
on what you know of the accuracy of the mains frequency, what
odds would you offer someone who was willing to bet that when
a very precise and accurate meter was used to measure the
mains frequency, it would be within 0.01% tolerance?
Goofball_star_dot_etal
wrote:
>It seems that all you've established is that by law the mains
>frequency in the UK must be at least accurate within 1%. It
>might always be very much better than that, for reasons other
>than legal ones.
>
>To put it another way, suppose a group of people were willing
>to bet on the outcome of a mains frequency measurement. Based
>on what you know of the accuracy of the mains frequency, what
>odds would you offer someone who was willing to bet that when
>a very precise and accurate meter was used to measure the
>mains frequency, it would be within 0.01% tolerance?
>
Don't waste your money. I have measured it and posted the results in
the past. It was not within 0.01% tolerance short term.
henryf
within 0.01% about 12% of the time, and within 0.1% about 86%
of the time. What did your measurements show?
Goofball_star_dot_etal
wrote:
>Based on the US power grid, I would have guessed it would be
>within 0.01% about 12% of the time, and within 0.1% about 86%
>of the time. What did your measurements show?
I did not do any statistics but out of about four samples over an hour
or two I measured up to about 0.2% error. See:
http://www.wareing.dircon.co.uk/images/mains.jpg
Also a measurement from another source:
http://www.picotech.com/experiments/mains_voltage/results.html
I have read that the UK regs. used to be +/- 0.2% but it has been
changed to +/- 1% in line with the EU.
August Helmbright
watch counting strobe dots movement past the light) and the other with
a radially scratched record.
With the stop watch method, I ended up with needing to speed up the
wav file by 0.56% (.0056). My program (DC-V) won't allow a speed
correction more precise than four decimal points.
With the radial scratch, there was some difficulty in counting the
exact right point to mark for each scratch, since the wave forms
looked slightly different for each time around. Therefore, I marked
several in a row and measured the time (down to thousanths of a
second) between each one and averaged them. The result of this
analysis was that I needed to speed up the wav file by 0.44% (.0044).
At A=440, the difference between 0.56% and 0.44% correction equates to
0.528 Hz. Since both my methods contain some possibility of error
(measurement with the dots, correct interpretation of the wave forms
for the scratches), this didn't sound too bad to me. Maybe some folks
can, but - unless the tones are playing at the same time - I can't
really "hear" a differenct between 440 and 440.528.
I haven't made up my mind what to do exactly. I might just decide to
split the difference ("saving the appearances?") and correct by 0.5%.
That way, taking 0.44% as the low end of the range and 0.55% as the
high end, the most I can be off is 0.06% too fast (440.264 Hz.) or
0.05% too slow (439.78 Hz.). Those would be horrible differences if I
were trying to send a rocket to the moon, but I'm just trying to burn
CDs that sound right to my ears.
Goofball_star_dot_etal
Helmbright) wrote:
Just consider yourself lucky you did not post a 300Hz test tone. .
http://www.wareing.dircon.co.uk/images/moi-fmvt.jpg
henryf
happening with frequency control in the North American power
system:
http://www.ornl.gov/sci/btc/apps/Restructuring/ORNLTM200341.pdf
I found the conclusions on page 17 interesting:
"The preliminary investigation found that frequency is not
being controlled as tightly today as it was years ago.
Average frequency error doubled (from 0.002 to 0.004 Hz
standard deviation) ..."
Robert Gault
I don't know if anyone has suggested this, but why don't you disable the
slider bar and replace it with a multi-turn potentiometer?
It is almost certain that the slider bar is just a resistor. Replace it
with something that is similar in total resistance to the best setting
of the slider but with better resolution. If necessary, you can use a
narrow range high turn pot in series with a fix resistor.
If the regulator on the turntable is of high quality, you should be able
to match the speed such that A(440) is indistinguishable from a tuning fork.
Goofball_star_dot_etal
wrote:
>Here's a government document that describes what's been
>happening with frequency control in the North American power
>system:
> http://www.ornl.gov/sci/btc/apps/Restructuring/ORNLTM200341.pdf
>
>I found the conclusions on page 17 interesting:
>
> "The preliminary investigation found that frequency is not
>being controlled as tightly today as it was years ago.
>Average frequency error doubled (from 0.002 to 0.004 Hz
>standard deviation) ..."
>
The significance of this depends on how long the *Average * is
compared to how long August Helmbright spends counting dots. .
henryf
>
> I don't know if anyone has suggested this, but why don't you disable the
> slider bar and replace it with a multi-turn potentiometer?
>
> It is almost certain that the slider bar is just a resistor. Replace it
> with something that is similar in total resistance to the best setting
> of the slider but with better resolution. If necessary, you can use a
> narrow range high turn pot in series with a fix resistor.
>
> If the regulator on the turntable is of high quality, you should be able
> to match the speed such that A(440) is indistinguishable from a tuning
> fork.
>
Part of the problem is that the Stanton website doesn't have
schematics for their turntables. I saw a photo on the web
that showed the slider of a non-Stanton turntable plugging
into a control board that looked like it had a trim pot on it.
Who knows, maybe Stanton turntables already have an internal
trim pot that can be used to fine tune the speed? It would
make sense to me, but the Stanton web site is mum.
August Helmbright
The person I talked to by phone at Stanton said they have no fine speed adjustment.
August Helmbright
occured to me, simple as it seems once you think about it.
Upon further extremely close study of the waveforms resulting from the
radial scratches, I was able to position the curser such that it
seemed to my eye (at very fine resolution on the screen) to be in the
same position for each of 5 successive scratches, resulting in four
temporal measurements. In fact the largest differnce in the 4
measurements, showing that my eye for waveforms isn't perfect, was
0.0002 seconds. I averaged the 4 measurements and came up with 1.8081
instead of the correct 1.8 seconds, confirming what the strobe dots
show from a qualitative standpoint - that the turntable is slightly
too slow.
The amount by which I'm speeding up my wav files is 0.45% (0.0045),
which is the best the program allows in terms of decimal precision.
Assuming the 1.8081 is an absolutely correct measurement, the
difference between 0.45% and what is would be calculated based on the
scratch test results in a difference of approximately 0.0089 Hz
relative to 440 Hz. I'm satisfied I've gotten this thing much closer
to being right than will ever make a difference to my (or anyone
else's) ears. I'm also satisfied that the scratch test - particularly
based on averaging several extremely close measurements to minimize
the effects of small eyeballing errors with the waveforms - is much
more accurate that the stopwatch & strobe dot approach I had first
tried.
Goofball_star_dot_etal
Helmbright) wrote:
Thanks for your kind words. You would not believe how difficult it is
to get vinyl lovers to attack records with a Stanley knife.
Alan Peterman
wrote:
>Here's my report on using the 2 methods -- the one I suggested (stop
>watch counting strobe dots movement past the light) and the other with
>a radially scratched record.
>
>With the stop watch method, I ended up with needing to speed up the
>wav file by 0.56% (.0056). My program (DC-V) won't allow a speed
>correction more precise than four decimal points.
>
>With the radial scratch, there was some difficulty in counting the
>exact right point to mark for each scratch, since the wave forms
>looked slightly different for each time around. Therefore, I marked
>several in a row and measured the time (down to thousanths of a
>second) between each one and averaged them. The result of this
>analysis was that I needed to speed up the wav file by 0.44% (.0044).
I suppose I shouldn't mention this, but it is extremely unlikely that your table
ran at a constant speed over the 15-20 minutes that it takes to play a record.
In fact it's been shown that needle drag, which is proportional to groove
modulation affects the speed, and the modulation of course varies as you play
the record. In fact the drag is LEAST when playing the empty grooves, so when
you timed it using those grooves, chance are the record was playing a bit faster
than it was during music grooves.
If you want to see this effect, you can watch the strobe marks as you set the
needle down on the record.
The other major factor is the temperature change of the motor and circuitry
which is why the table usually take 3-10 minutes of operation before it becomes
relatively stable in speed. Plus any voltage variation in main voltage may
affect the speed, depending on the circuitry in the drive system.
And of course, as I mentioned, it's not likely the original recording table and
tape recorders were accurate to the degree you are attempting to be. But I
applaud you for your effort to make a silk purse out of a sow's ear!
August Helmbright
> I suppose I shouldn't mention this, but it is extremely unlikely that your table
> ran at a constant speed over the 15-20 minutes that it takes to play a record.
> In fact it's been shown that needle drag, which is proportional to groove
> modulation affects the speed, and the modulation of course varies as you play
> the record. In fact the drag is LEAST when playing the empty grooves, so when
> you timed it using those grooves, chance are the record was playing a bit faster
> than it was during music grooves.
>
> If you want to see this effect, you can watch the strobe marks as you set the
> needle down on the record.
>
> The other major factor is the temperature change of the motor and circuitry
> which is why the table usually take 3-10 minutes of operation before it becomes
> relatively stable in speed. Plus any voltage variation in main voltage may
> affect the speed, depending on the circuitry in the drive system.
>
> And of course, as I mentioned, it's not likely the original recording table and
> tape recorders were accurate to the degree you are attempting to be. But I
> applaud you for your effort to make a silk purse out of a sow's ear!
>
> ---------------------------------------------------------------
> Alan Peterman al at scn.rain.com Tigard, OR
> As I grow older, the days seem longer and the years seem shorter.
While you are correct from a technical standpoint, it is also true
that voices and musical instruments in live performance do not
maintain absolute accuracy in pitch either. Pitches change slightly as
instruments warm up - particularly noticeable if you're measuring
frequencies scientifically in woodwind instruments, where the bores
contract as the wood expands due to heat. That being said, the human
ear is very forgiving of small deviations, which is why the equal
tuning system used in pianos and other keyboard instruments (every
note is just a bit out of tune compared to the overtone series)
doesn't offend greatly.
Since I'm working for musical rather than scientific precision,
anything I can do to get *average* turntable speed as close as
possible to true 33 1/3 RPM (either mechanically or, as I'm doing,
using software to change the speed of a wav file) gets me well within
human precision, even for someone with a very good musical ear. If
it's scientific precision we need, sown to the smallest fraction of a
Hertz (thank God it's not - we'd all go nuts), then there's probably:
(1) no musical performance live that would fill the bill; and (2) no
method of reproducing a performance in recorded form that is exactly
in sync at all times with the original.
Svante
> ...., it is also true
> that voices and musical instruments in live performance do not
> maintain absolute accuracy in pitch either. Pitches change slightly as
> instruments warm up - particularly noticeable if you're measuring
> frequencies scientifically in woodwind instruments, where the bores
> contract as the wood expands due to heat.
I can't help myself here, I have to add that there also is an effect
of the CO2 in the exhaled air. The CO2 content goes up towards the end
of a breath, which reduces the speed of sound within the instrument,
and so the pitch drops.
http://www.speech.kth.se/qpsr/tmh/1996/96-30-037-043.pdf
...but, as stated in this article, it is likely that good musicians
compensate for these effects such that the effects are kept about as
small as what is audible.
So, it all comes down to what is audible. I seem to recall that a
frequency modulation of 0.2% was audible in the days of wow and
flutter measurements (correct me if I am wrong), that should give a
hint regarding the precision needed.