'Number9' TT AC motor controller web site finally up
- Thread starter andyr
- Start date
Shuggie
Trade: Ammonite Audio
I use a Martin Bastin adjustable PSU with my TD-124 and got around the strobe issue by replacing the neon strobe with a KeyStrobe LED conversion kit which is completely independent of the mains frequency (it makes the strobe markings much easier to see, too).
Steve#9
Member
Sorry take so long to reply to your enquiry of my typing skills are not all they could be and I've been getting a lot of e-mail enquiries over the last week.
I've included two calibration systems into the TCI. The first is a fairly traditional strobe system consisting of three watt LED lamp in a flexible gooseneck fitting which serves as a cueing light and strobe light. The strobe disk is supplied which as only one circular pattern on it because the frequency of strobe changes from 50.000 Hz for calibrating 33 rpm to 67.500 Hz for calibrating 45 rpm. Having a single pattern on the disk makes it a little easier see what's going on. The frequency of the strobe light is accurate to 3 ppm (0.0003%). A mains powered strobe is at best accurate to around 400 ppm and commercially available LED key fob type strobe lights are generally accurate somewhere around 100 ppm, so that neither of them are capable of sufficient accuracy to even get close to the absolute 0.0006 rpm resolution of the TCI. There is no real set up involved, it is exactly same process as it used with any third party strobe kit.
I included the strobe system into the design to provide a means of calibration that many audiophiles are familiar with and comfortable with.But I doubt that many people would have the patience or even the spare time to get anywhere close to the absolute speed accuracy of the TCI. I haven't tried it myself but we're talking probably days!
In normal day-to-day operation the lamp functions as a cueing light, it fades up to full brightness at power up and slowly fades down to 10% brightness 2 min after the start button is pressed allowing sufficient time to find your glass of Scotch and return to your favorite listening position, after 18 min of playing time or until the stop button is pressed whichever comes first, the lamp will fade back up to 100% brightness.
The more accurate calibration system is very similar to the process used In the Phoenix Engineering product. A small neodymium magnet of approximately 2 g mass is attached to the underside of the platter and a Hall Effect magnetic sensor mounted on a strip of flexible PCB material sends out a pulse once every revolution. Because the TCI uses 32-bit technology and 2.5 ppm clock that gives us around 300 million precise datapoints for each 1.8 second revolution. We are able to use this data in a number of ways i.e. evaluating platter mass determine the appropriate way that a speed correction is applied to avoid hunting or overshoot.
Once the sensor has been located on the plinth, the TCI goes through an automated process starting and stopping the turntable a number of times until the 33 and 45 rpm speeds are calibrated and verified. This entire process takes roughly 5 min complete.
Steve
London Lad
pfm Member
Brilliant. Almost makes me want to buy a turntable!
yuckyamson
pfm Member
Hi Steve, thank you very much for this explanation. It's much appreciated. I have a subsequent list of questions but will send later on.
Again and as per my initial comment, hats off to you. I don't know how much I'd like the PS, how great it sounds to my ears or not (or say, compared to my 'geddon or a radikal or whatever) but it just sounds like a very well engineered product regardless and that you've put a lot into it. It's great to see. Great website too.
Would love to see pictures of the light if available.
yuckyamson
pfm Member
K so Steve, first of all consider me a "techno tard", (this can be verified by other people who have known me on this forum, btw)......
So in the strobe based-version:
1) Does one insert the strobe disc upon every start up to re-calibrate the speed? Or is it something you'd do once? Every few months? Every few years? etc.
2) Does the strobe lamp sit permanently near the TT?
3) Can you supply a pick of the strobe lamp?
4) If I'm reading you right, you adjust speed by eye (the strobe disc lines). My only question then is that a very slow drift, how long would you have to wait until you were within your tolerance range to even know how accurate you were getting?
5) Is this what you mean with the "patience or spare time to get close to the absoulte speed..." i.e. that the strobe is fine for adjustment but utimately, unless watched for a long period of time, not terribly accurate relative to how capable the system actually is, when self calibrated via method B?
For method B
....I think I understand it, like a radikal but you calibrate it once, which takes 5 minutes, and could be done if desired every year or so. But once calibrated it should hold uber-accurate speed for a long period of time unless damage or wear befalls the motor, the belt, platter, or bearing. Yes?
So in the strobe based-version:
1) Does one insert the strobe disc upon every start up to re-calibrate the speed? Or is it something you'd do once? Every few months? Every few years? etc.
2) Does the strobe lamp sit permanently near the TT?
3) Can you supply a pick of the strobe lamp?
4) If I'm reading you right, you adjust speed by eye (the strobe disc lines). My only question then is that a very slow drift, how long would you have to wait until you were within your tolerance range to even know how accurate you were getting?
5) Is this what you mean with the "patience or spare time to get close to the absoulte speed..." i.e. that the strobe is fine for adjustment but utimately, unless watched for a long period of time, not terribly accurate relative to how capable the system actually is, when self calibrated via method B?
For method B
....I think I understand it, like a radikal but you calibrate it once, which takes 5 minutes, and could be done if desired every year or so. But once calibrated it should hold uber-accurate speed for a long period of time unless damage or wear befalls the motor, the belt, platter, or bearing. Yes?
Steve#9
Member
No, the calibration process generally only needs to be performed once. Obviously if you were to make any change to the mechanical aspects of the drive system i.e. replace the belt, change the lubricant in the bearing, change motors, change sub platters or any other physical changes to the drive set up, you would be wise to recalibrate that point. I am still playing around with the algorithms within speed regulation code, one of the things I'm considering is a routine that tracks the speed over a long period of time, like six months or more, looking for any trend of slowing over time which would indicate belt wear or lubricant degradation and warn the user accordingly.
Yes, the automatic fading cueing light unlike the strobe function is something that is useful on a daily basis.
As I said earlier, strobe system was incorporated on the basis that many audiophiles are comfortable with the process, also some are horrified by the concept of attaching the neodymium magnet to the platter on the unfounded basis that it will cause an imbalance in the platter mass, which according to some would unleash chaos in the known universe. From an engineering point of view, a 2 g added mass on a 4 kg flywheel might cause problems at 20,000 rpm but at 33 1/3 rpm, I have my doubts.
It's theoretically possible to achieve similar degree of accuracy with the strobe given enough time and patience but is really only necessary to use this method if you have an objection to adding the 2 g magnet to the platter. That said, because of the 3 ppm accuracy of the strobe frequency you will get better accuracy in a short time than any other strobe system currently available at any price. For optimum results with the strobe, I recommend performing the calibration in total darkness.
If you are looking for absolute accuracy, the magnetic tacho system is the only way to go. Barring mechanical changes in the drive system as stated above, the TCI is absolutely immune to short-term drift unlike servo based systems like the Radikal. The TCI uses a 2.5 parts per million temperature stabilised clock with guaranteed long term drift specified at less than 0.025 ppm per year, at that rate (fun fact) it would take something of the order of 400 years for the TCI clock to degrade to a level of accuracy on par with the 100 ppm standard clock chips used in other devices.
http://9tci.com/accessories/
Steve
yy - I find it fascinating that you say you don't know how much you'd like the 'Number9' (compared to your 'Geddon)!
For a start - it offers 45rpm (I understood the 'Geddon - being merely a transformer - can't change the frequency of the power going to the motor)!
Secondly ... it enables you to adjust the speed the platter is rotating (to get it to exactly 33 or 45rpm). Again, your 'Geddon doesn't allow this.
The above are elementary, my dear yy!
Thirdly, you can alter the phase angle between the motor windings, to minimise the motor vibration.
This leads on to fourthly - as a result (of negligible motor vibration), you can increase the voltage going to the motor (the direct opposite of what a Lingo does!) ... to get an increase in SQ.
And fifthly, you can drive 2 motors - if your TT supports this. As a result you get a further increase in SQ (over just increasing the voltage to the one motor).
Andy
Pyramid
Member
Actually, I used an 8 bit µP, which doesn't matter. You can get any level of floating point precision you want (my HP scientific calculator uses an 8 bit processor); it's all in the programming.
The RoadRunner tach also has 0.0006172725 RPM of resolution at 33 RPM, even though the display is only to 0.001. The time base is also a 2.5 PPM TCXO.
It sounds like the two are very similar.
Congratulations on your implementation. It looks like you added a lot of nice features.
Pyramid
Member
To put this in perspective, 2 g is about the same amount of pressure exerted by the stylus, so it shouldn't represent any sort of balance problem.
Just out of curiosity, why such a large magnet? The ones we used were 1/10 of a gram and were easily detectable by a Hall sensor.
Pyramid
Member
What happened to these guys? Their website: www.9tci.com is off line (suspended).
sq225917
Bit of this, bit of that
Hey I noticed that a few days ago.
I've always had an issue with the claims we see for a gazillion data points per revolution. The reality is that with a sensor at a single point you have one data point and the rest is just interpolation. You can deduce nothing about intra rotation accuracy, only if its accelerating, decelerating or bang on speed.
I've always had an issue with the claims we see for a gazillion data points per revolution. The reality is that with a sensor at a single point you have one data point and the rest is just interpolation. You can deduce nothing about intra rotation accuracy, only if its accelerating, decelerating or bang on speed.
exactly my thoughts,
Steve#9
Member
I'm sorry but your post makes no scientific sense, could you expand?
sq225917
Bit of this, bit of that
Sure. With one sensor, you have one sample per revolution. So you know what average speed the deck is doing per revolution. You can make tiny adjustments to the speed based on whatever your interpolation limit is. But you have absolutely no way of knowing what's going on within each revolution. You could have God awful motor cogging 12 times per revolution, now that might average out to a decent accuracy per revolution, but with only one sample point you'll never know. You can only fix the average error.
For example, you could perform a FM demodulation on a full revolution and find out your deck is rotating at an average of 33.3rpm but actually has 12 lots of +/- 0. 2% deviation from this per revolution. You can't fix this with one sample point per revolution. Because you can't see it, all you see is the accumulated error per rev.
So for example if you had an optical sensor, say a gated disc with 333 markings you could compute a motor drive function that took into account the effect of things like cogging.
At the end of the day a heavy platter, rubber belt and weak drive system can perform this averaging for you mechanically. All lots of interpolation gives is the ability to correct in finer steps.
For example, you could perform a FM demodulation on a full revolution and find out your deck is rotating at an average of 33.3rpm but actually has 12 lots of +/- 0. 2% deviation from this per revolution. You can't fix this with one sample point per revolution. Because you can't see it, all you see is the accumulated error per rev.
So for example if you had an optical sensor, say a gated disc with 333 markings you could compute a motor drive function that took into account the effect of things like cogging.
At the end of the day a heavy platter, rubber belt and weak drive system can perform this averaging for you mechanically. All lots of interpolation gives is the ability to correct in finer steps.
Steve#9
Member
Thanks Bill,
That means a lot to me coming from you. When your product hit the market I was pleased to know that somebody else "gets it". I had hoped that I would be the first to produce a motor controller that wasn't based on "1974 Highschool Science Fair Project" technology, but you beat me to it.
I would very much like to have the opportunity to privately discuss some issues with you, both turntable and Cheddar related. I'll PM my contact details to you.
I hope to have the website back up on a less crappy server in a couple of days.
Thanks Again,
Steve
Steve#9
Member
Okay, I can see where you're coming from in some respects, but I'm still a little mystified as to what audible benefit applying speed corrections more than once per revolution could possibly have.
There is a finite limit to the number of corrections that can be usefully applied in one revolution depending on the mass of the platter, available motor torque and other nonlinearities in the system.
My calibration system has a routine that performs a series of tests to estimate the mass the platter, so that the controller knows to a reasonable degree of accuracy, the percentage of correction that can be applied in one revolution. Speed increments are only applied at a rate at which the moving mass can respond to in less than one revolution, any additional correction is handled in the next rotation. The program also differentiates between the time taken to increase the rotational speed and the time taken to decrease the speed for a given platter mass.
This process virtually eliminates any "hunting".
My controller also has the capability to reduce cogging vibration to extremely low levels. The amount of reduction varies according to the type of motor, the ubiquitous Premotec 110V motor responds very well to accurate phase angle adjustment with a very large reduction in cogging vibration, the 24V version of the same motor, once adjusted, has virtually no cogging vibration.
The increase in sonic performance attainable once cogging vibration and the various mechanisms applied to turntable design to mask and minimise cogging has to be heard to be believed.
I am planning to release my version of the Linn Radikal using the 24V Premotech motor early next year and hopefully my turntable shortly afterward.
Last edited: