| Raka |
Maybe you already know about my power supply for the Thorens TD160, with a signal generator from a dhaen pcb.
After so long time, I looked again at it, made some tests and find out that the main advantage was to get rid of the voltage divider 10K resistor, because it's a great noise maker. But I want to develop another ps.
I'm thinking about the second signal, now created with a simple cap, like in a standard synch motor.
My plan to get rid of it: From the output of the signal generator I take the standard signal, and also take other from a cap. So I have two signal 90º phased. Then apply a volumen control for each signal, I amplify with two GC and two output transformers (I can wind my own trafos here, so I'll adapt the impedances for the GC to be looking at a 8ohm impedance), and the trafo wouldn't be looking at the phasing capacitor. Besides, the second signal can be made as high (or low) in volts as required.
Does this look ok? |
|
|
| GUILHERME |
Hi.
I'm also doing a ps and have done extensive tests measuring vibration and speed in various possible configurations. I don't know about your previous ps, but if I understand you correctly, you want to do the phase shift before the power amps and then feed each motor phase seperatly. If thats so, I think it is a very good move. My tests show that the phase-shifting cap makes the motor more noisy than it needs to be.
Best Regards,
Guilherme. |
|
|
| Raka |
You can check the thread called "change in frequency mains " or something like that. You can find useful info there.
I'm interested in your tests, could you please detail?
Better than the cap shifting, I will invert the signal with an inverting output amp, so the shift is always 90º (provided resistive loads, or identical impedance loads, of course).
Have you found any suitable AC motor? |
|
|
| GUILHERME |
I'm using the tt's original AC motor. From the brand name it looks like something from the east.
| quote: | | I'm interested in your tests, could you please detail? |
This is a cut from a post I made at the vinyl engine forum a few days ago:
===============
I have been designing a motor controller 'a la VPI's SDS for myself and now that it is a working prototype I have taken the time to test its efects on the performance of the TT by measuring both motor vibration and platter speed. I'm preparing the data to present it in a website dedicated to the project, but in the mean time here's some thoughts based on those tests (figured this might be useful in this thread).
Regarding motor vibration:
I've found that removing the capacitor and feeding the motor with two separate phases will cause the biggest improvement, reducing vibration by a large factor. Next comes reducing voltage, although, from a certain point, the vibration will increase again. My motor (16V) will vibrate the least when fed with 14V even if it can mantain speed with only 6V. The icing on the cake is phase triming, which provides a smaller but still measurable improvement. My motor literaly feels like its stopped with all those above and a phase trimmed to 92.8 deg.
Regarding mains, I measured 4% THD and looking at the scope the peaks were actualy clipped. This, of course, doesn't help the vibration of the motor. I live in an EEC country, so this will probably be similar in other coutries.
[...] I think you should also consider this: mains also has low freq. "noise" (FM, actually) and that will certaily afect TT performance.
Just last night I measured platter speed and found a peak dev. of 0.02 RPM (0.06%) when using mains feed and 0.005 RPM (0.015%) when using my device (over a period of 5 min). This numbers, are, of course, influenced by the wow introduced by the electromechanical system of the TT itself, but that is the same on both measurements. I think my device's figure is actually being influenced by this and if I had a better TT I would probably get even better results (but this is just a guess).
And then there's the absolute mains freq. and platter/pulley tolerances, which, combined, usually make the platter turn at something else than 33.3 RPM. In my case, I was actually surprised by mains freq. ( I remember it hoovered arround 50.0 Hz. ), but that translated into a platter speed of 33.50 RPM. Again using my device, I was able to get 33.333 RPM (average).
===============
To this I should add that speed measurements were made with a record free platter and that taking the same measurements while playing a record worsened the numbers a bit, suggesting that stylus drag does influence speed stability.
Guilherme. |
|
|
| Raka |
Ah, so you discovered that the cap is the most important thing to avoid... interesting. How did you measure the vibration, just the fingers? I usually use my sthetoscope
I suppose that the deviation from the 90º is caused by the difference in the loads that the not-exact windings represent, was 92.8 the best result by far, or was small difference when compared to 90º?
I have a lot of suitable transformers, so I can try to invert a singal easily in a short time.
I also have a 1KVA 220/110V transformer, so I'll have the chance to compare what fdegrove suggested.
BTW, I've rewired my RB250, but my stylus left me and I'm still waiting for the goldring1042 to listen again to records.
I had a RPM from Pro-ject, and I know your tt ;) |
|
|
| GUILHERME |
| quote: | | How did you measure the vibration, just the fingers? |
:D Not really.
I used a non-magnetic piezo binded to the side of the motor so it is measuring lateral movements. The signal is amplified on the spot by a discrete in-amp and sent to the scope/fft.
Speed is measured with an optic sensor measuring platter lap times with a 10Mhz clock and sending results to the pc via rs232.
92.8deg is a small improvement when compared to no-cap and low-voltage, but still measurable and one that you can feel in your hand. I don't remember the exact number, but I can check it.
I do remember the improvement in vibration from going stock to using my device with all features running: -19 dB.
| quote: | | I had a RPM from Pro-ject, and I know your tt |
You should hear it while being driven by my device ;) .
Guilherme. |
|
|
| Raka |
Wow, nice equipment :bigeyes: , maybe better than my sthetoscope :clown:
19dB seems a big improvement, I think that this could be heard.
I don't have your equipment so I have to keep the analog track, and build an inverting stage with another GC. I think in a week, everything will be at test stage.
Are you planning to publish your design? |
|
|
| GUILHERME |
Yes, it will be public.
A few screenshots I've taken for my website...
Measuring vibration - as I described earlier, a piezo is binded to the side of the motor. As it moves, the piezo slightly bends and produces a voltage. That voltage is what you see on these pics.
This one is from using the stock trafo while playing a record:
You can see an initial "sine" caused by the power being fed to the motor; the second lower one is the system moving to rest - this would continue to decay if nothing else happened. But the second phase kicks in and so the process repeats itself.
If you look close at the peaks you'll notice a slight amplitude modulation. It's arround 5Hz in frequency. The pulley/platter ratio is 1:10 so a platter moving at 33.3 rpm has a frequency of around 0.5 Hz. So the 5 Hz. would be, more or less, the frequency of the rotating pulley. I think that modulation is caused by the pulley not being perfetly centered; as it moves, it very slightly changes the tension and thus causes the modulation.
Next pic is same record, but now with my device driving the motor:
I don't have a pic here, but if I add gain to take a closer look the wave is not perfect anymore but is mixed with spurious repeating noise together with a much more evident 5Hz. modulation. In other words, the tipical buzzing from the AC motors is so down that it is getting at the same level of other types of noises that have nothing to do with the fact of it being AC. I think you get my drift ;) .
Next pic is a bit large, so here's a link instead:
http://www.itonami.com/temp/t3.jpg
The yellow line is platter speed with the stock trafo, the red is using my device. Note that It's not at 33.3 on purpose, so the two would be close. The blue dialog is the software I made to take speed measurements; those stats are for the red line.
The peak stat is actually peak-to-peak, so to be comparable to the standard wow figures shown on datasheets it should really read 0.0034.
By the way, the manufacturer of this tt specs it with a wow/flutter of 0.08% (if memory serves me), that would give a 0.026 peak dev.
The measured frequency coming out of the supply is within +/- 50 ppm. so these deviations are still wider than those of the frequency itself. Therefor, I think they are not caused by the supply but by the rest of system. A better tt should, in principle, give even better numbers. The limit of the speed stability of the tt becames the tt system itself and is not swamped by the crappy mains feed.
Guilherme. |
|
|
| Raka |
Thanks for the pictures, they show an interesting work indeed. The difference between the piezo output is impressive.
I'm happy to hear that you design will be public, thanks in advance for sharing
I'd like to hear your comments about the differences between the standard AC and your supply.
BTW, I didn't know you had a website |
|
|
| GUILHERME |
I meant the screenshots are for my *future* website. :)
| quote: | | I'd like to hear your comments about the differences between the standard AC and your supply. |
Besides what I already said, I can add this: that ten-minute yellow line on the graph really doesn't say it all. While debugging this project I've done numerous test runs over several diferent days and I've found that mains does drift a bit with some days being nastier than others. In fact, I'm thinking of logging mains freq over a period of, say, one week just to put some numbers on this. Add 4% thd to this and some voltage fluctuations, and well...
The website I'm making really details all of this with several diferent psu configurations and comparisons among them. I'm planning of putting it online even before the psu project is completly finished.
I noticed your comment about trying a large trafo. I've also made some tests with one of those and I'd like to hear your comments once you've tried it so I can compare with my own impressions.
Best Regards,
Guilherme. |
|
|
| Raka |
| quote: | Originally posted by GUILHERME
I meant the screenshots are for my *future* website. :)
I noticed your comment about trying a large trafo. I've also made some tests with one of those and I'd like to hear your comments once you've tried it so I can compare with my own impressions.
Best Regards,
Guilherme. |
OK, until next week I won't be able to test it, since my td160 has no cartridge until next monday (Goldring1042). I also rewired my RB250Expressimo, so I hope I can hear the differences. I'll report as soon as I adjust the tt. |
|
|
| Raka |
Well, I've made some tests, just for the motor stopped, and running. With my power supply and the motor stopped there is no noise with my sthetoscope if I apply just the voltage required for a proper starting. If I run the motor, there is no additional noise but some mechanical noise, that I have to fix if I knew.
With the 220/120V 1500va trafo, there is the same result. surprisingly with a bit more voltage (120V) there is no noise, but the transformer is very noisy itself.
Not listening tests until tomorrow. |
|
|
| GUILHERME |
Raka,
I'm not sure I understand you. You talk about a buzz even when the motor is stopped (I read your other thread). Where does this buzz come from ? is there another transformer inside the tt ?
Regarding your test, again I'm a bit lost. When you said trying a large trafo like fdgrove suggested, I thought you meant just the big trafo between your wallsocket and the tt. But since you talk about changing the voltage, I'm wondering if you're using your psu's amp to drive the big trafo.
Guilherme. |
|
|
| Raka |
Sorry, It seems I didn't explain myself.
No, there is not another trafo inside, is just that the motor has a low buzz when is connected and not running. Yes, maybe it's a bit strange, but it happens. Reducing the voltage it's almost erased, and is comparable to the noise produced by the series resistor, so I think maybe (I'm sure) the cap is to blame.
Yes, I used the 1,5KVA trafo between the wall and tt, so the input to the tt is 120V. I said higher voltage because with my psu the voltage required is only around 90V, that is 30V less. Hence, I'm surprised that with the trafo giving more voltage, there is no noise in the motor. |
|
|
| GUILHERME |
I wonder if the buzz isn't being caused by excessive dc on your mains being fed through the phase with no cap. That would explain the similar results between your psu and the big-trafo - they both block dc. That would also explain the noise from the transformer itself.
Guilherme. |
|
|
| Raka |
| quote: | Originally posted by GUILHERME
I wonder if the buzz isn't being caused by excessive dc on your mains being fed through the phase with no cap. That would explain the similar results between your psu and the big-trafo - they both block dc. That would also explain the noise from the transformer itself.
Guilherme. |
Good point, but the motor also buzzes stopped when fed by the LM3886 amp+step-up-trafo, and also when fed by the big trafo... :xeye:
Anyway, my trannies use to buzz, and I haven't checked the DC on the mains. How could I block it? |
|
|
| Raka |
I've thinking a bit more (not too much :clown: ) about the phase shift, and maybe it's good to provide the means to fine tune the 90º. This way, I can try different angles to see which one optimizes the motor vibration.
It could be done with a simple delay circuit, for 5ms (if 50Hz) and a pot regulating the time constant. This is only an opamp and some resistors and caps. Simulates good in my PC.
Sum up : Signal generator (crystal locked) with two outputs. One directly to a GC and the other one delayed and sent to another GC. Both GC feed two stepup trafos (that I have here in my trash bin) if I can't find a low voltage alternative motor. So no cap at the motor terminals.
I haven't implemented yet any Wien Bridge, has anyone got good results with this? I mean regarding from distortion and estability points of view. My tt turns a bit slower, and maybe I need tu build a 52Hz signal generator. |
|
|
| analog_sa |
Hi Raka
It's great that you are getting seriously interested in the motor PS - you can get serious improvement by optimising it.
FWIW i would rather try building a high voltage direct drive circuit than mess with transformers. It seems the motors prefer a low impedance drive but you are probably better qualified to make this judgement. At least a lot of the bulk and added distortion of the transformers can be avoided. I am not too sure of the minimum voltage requirements of the Thorens but if you're lucky a bridged 3886 powered at the max allowed voltage may just be enough.
You may also find that a microcontroller is ideal for generating variable and repeatable phaseshifts between the waves. I use a 8515 AVR in a very simple setup.
The advantage of using a Wien bridge is that you need not worry about digital noise generated by the PS entering your system. It seems to be a real problem with Linn supplies.
Being at the moment in Zimbabwe I may not be able to follow this (or any other :)) thread for some time but certainly hope you'll build the very best Thorens possible :) |
|
|
| Raka |
Yes, now that Ive finished a good amp, I'm trying to improve my thorens, or at least trying to get the most of it.
I've been looking to bridge configurations, and a bridged 3886 is not enough voltage, as my motor needs around 90V. Besides, the load the amp will see, is not the usual speaker coil, but a 10K winding, so don`t know the impact will have this on the 3886 peace of life. Maybe I have to search or to design an specific design for this requirements.
The transformer thing is not such a big problem, since I can build my own tranies under specification. Knowing the load of the motor I only have to calculate it, and the 3886 will see 8Ohm load, without saturation.
I'd like to see your configuration, is possible to have the scheme? |
|
|
| GUILHERME |
Raka,
I would suggest you try to make frequency variable at least arround the 50Hz instead of going after variable phase. I say this because it seems you already got noise down to accetable levels and because the sonic benefits of getting platter speed right are more than just pitch and are greater than trimming phase (IMO, of course).
Guilherme. |
|
|
| Raka |
Thanks for your suggestion, Guilherme. I was also thinking about making provisions for fine tuning the speed, as with a Wien Bridge it seems not so difficult. It's a pitty I can't use the Dhaen pcb... it works fantastic but my tt has bit bigger platter than desired.
Now I'm spending time with the output trafos, and cannot simulate a good solution. Need to study a bit more. Any place to find info about transformers and saturation in step up condition? Should a smaller turn ratio give better results? |
|
|
| GUILHERME |
Sorry, the only info I can remember is from the book "The Art of Electronics". They show a circuit to do just what you're doing - to step-up a sine through an amp and then a trafo. The amp's feedback loop is taken partially from the transformer's output, so output impedance is improved. If you dont have this book email me privately.
Guilherme. |
|
|
| SvErD |
| I've been thinking about copying the Linn Valhalla output stage. |
|
|
| Raka |
| Do you have the scheme? |
|
|
| SvErD |
| quote: | Originally posted by Raka
Do you have the scheme? |
I do.
Send me a mail |
|
|
| Raka |
| This is more or less what I have in mind for the amplification of each leg: |
|
|
| Raka |
Today I will build the wien bridge, including some adjusting device for the speed (47÷55Hz), but I don't know how to define the lamp.
I've seen 10V/14mA and 28V/24mA so far, but here I can only find 28V 40mA. How can I choose the right lamp? |
|
|
| Raka |
Well, the oscillator is done and works beautifully. Even less noise than before (can't note if the motor is switched or not, with a sthetoscope). Distortion is less than 0.2%
I fine tuned the speed, and results that my TD requires around 52Hz to achieve 33*1/6.
Sound? Very good indeed. Now I have to work on the power stage, with phase control, to get rid of the cap.
Thanks Guilherme for your link. :nod: |
|
|
| GUILHERME |
I'm glad you found it useful. :)
Now that you can get really close to 33.33, how would you comment the sound ? do you think it's all just about pitch or do you notice other things sort of falling into place ?
You remind me that I really have to get back to my own psu, stuff it inside a box and call it finished.
I've been having a lot of fun with those measurements; the speed meter was improved and I'm actually measuring the effects of stylus drag on platter speed. Of course, that led to more experiments, turning it into a project within a project.
I ordered a wow&flutter meter to complement the experiment's data, once that is done I'll show you the results (and then get back to the psu). It's a subject I see mentioned a lot but without much (any) references to supporting data.
So much to do and so little time available, sometimes it gets frustrating :(
Best Regards,
J.Guilherme |
|
|
| Raka |
Well, now things are a bit more crisp, but this has to do with the higher pitch. Anyway, I tested a LP vs a CD, now with same speed and you know who won.
But this afternoon I played a solo piano work (Satie) and there is a not so light wow in sustained notes. I fine tuned the oscillator frequency (I use two pots to equalize the resistance values and 1% cap values) but no big improvings. Later, I compared this PS with the big trafo, and the wow was the same, so the LP pressing is to blame. Anyway, I had no time to do more tests, tomorrow I will play some other piano recordings to see what happens.
Antonio? Vivaldi? are you there? where is your l'estro armonico? in my LP? I hear somethin... you are there!
Sorry, I'm back to my tt, Vivaldi is calling me. See you tomorrow.
:clown: |
|
|
| GUILHERME |
Voices - you must try voices ! :D
When I'm spot on the speed they became more real, like the person really was there. Well, at least that's my experience. Try it on a well known record and see if you find any change.
J.Guilherme |
|
|
| Raka |
I'll use soprano voices this evening, thanks again for the suggestion.
What I've thinking about is to change the digital circuit of the dhaen pcb. I've studied the counter, and think how to increase the frequency a bit, as is a 8bit counter, there is space for fine tuning. Besides, I think a digital oscillator is more constant, but I'll test both.
How close do you get this improvement? Has to be 33.33 exact to get that thing you mention? |
|
|
| GUILHERME |
Hi.
When I started using my psu, the initial impression was the obvious change in pitch. Latter, in the many test runs I did, I started to notice the improvement on the 'presence' of the voices.
I don't know how close you need to get to hear it, I can only tell you how close I am. The best I got was 33.333 rpm (average over an entire side of the LP).
But this is an average, stylus drag and other factors will make this number worse when you look at a single lap, still, it gives an idea of where I am.
I do agree with you with going digital.
J.Guilherme |
|
|
| Raka |
You are doing an interesting work, indeed.
Well, I've tested my digital ps, and works well, and in 180 minutes, the lp spinned 100 times, so I think the speed is as close as I want to get.
But my final conclusion (final?, what is final? :clown: ) is that the digital has deeper and better bass. There is absolutely no wow in the piano recording I mentioned yesterday. Theorically, the wien bridge would do a great job, but is not so stable (the voltage amplitude isn't as constant as the digital oscillator), so despite it sounds good, is not so good as the dhaen one.
Now I'll work on the amp stage, if I managed to stop listening to the Firebird ;) |
|
|
| deduikertjes |
Well,
After reading all those interesting posts about a power supply for an AC synchronous motor I decided to build one myself.
So I made a wien-bridge oscillator, and gainclone power amplifier which I connected to an inverted mains transformer.
This set-up gives me a nice sinus which I can vary in voltage from 0 to 240 volts.
When I hooked this up to my turntable (Thorens TD160 MkII) the sinus stays nice but the results are horrible. I managed to build myself a nice wowing turntable. Back to standard mains give a much better result.
After a lot of searching and measuring I found that my gainclone amp is looking at an impedance (of the transformer) of 68 Ohms. Could this be the problem. Is the motor getting too little current to work well.
And if so, how can this be solved?
any help appreciated, MArco |
|
|
| AndrewT |
Hi Ded,
I tried that;- reduced mains (500mVac) into poweramp (100W into 4r) into inverted 12V 20VA transformer. Adjusted the gain stage to give 90Vac output to the Thorens motor (110Vac) and it ran perfectly. This was phase1 of a crystal referenced two speed electronic drive for the Thorens 150. Never finished the crystal part.
The main difference between your set up & mine was the oscillator.
Is that your problem?
I think the reflected impedance from the low power motor will not affect the quality of the drive signal. The impedance must be high due to the low current draw of the low power motor (150 has 8W motor). As long as the effective impedance is above the safe value for continuous use of the amplifier then I see no problem here ( note CONTINUOUS). |
|
|
| deduikertjes |
Dear AndrewT,
| quote: | | The main difference between your set up & mine was the oscillator. |
I don't think that's the problem. The sine is looking pretty good on the scope. Even after amplification and the transformer it's still looking good. I've a small amplitude variation of about 0.75 Volts on an average voltage of 95. The sine stays the same when the Thorens (TD 160 mkII with a 2.5 W motor according to the sign on it) is running.
| quote: | | The impedance must be high due to the low current draw of the low power motor (150 has 8W motor). As long as the effective impedance is above the safe value for continuous use of the amplifier then I see no problem here |
Well I'm afraid that the problem is here because of the perfect looking sine during operation. Is it possible that the motor can't draw enough current from the PS?
Things did improve by bettering the layout and adding a capacitor to counteract oscillation.
greetings, MArco |
|
|
| AndrewT |
Hi Ded,
did you keep the two phases (4 wires) of the motor supplied with the cap to one of them (2 wires)? |
|
|
| deduikertjes |
| quote: | | did you keep the two phases (4 wires) of the motor supplied with the cap to one of them (2 wires)? |
Yes, I did. I only bypassed the 10k resistor in the thorens.
greetings, MArco |
|
|
| deduikertjes |
Well,
I've solved the problem of the wowing turntable partly.
It proved that my gainclone amp which amplifies the signal of the sinus-generator was oscillating and losing a lot of power on that. Having that solved it seems that the amp has just enough power to drive the motor trough the step-up transformer.
What I want to do now is mto measure wow and flutter so that I can strive to the optimum.
So I've put a test record with sinus-signals on the turntable and recorded the signal with my soundcard on my harddisk.
Visisual inspection does not lead to a usable assesment of wow and flutter (it can't be seen). So some analysis on the signal has to take place. Does anyone have a clue how to do this. DOes anyone know (free) software wich can be usefull???
MArco |
|
|
| EC8010 |
| Wow and flutter is frequency modulation, so it gently changes the frequency of the replayed signal. If you were to multiply the replayed signal by a known constant frequency signal, you would get sum and difference frequencies. If you replayed 1kHz and multiplied it by 1kHz, then the sum would be 2kHz and the difference 0Hz. If you low-pass filtered, you would be left with 0Hz. However, wow and flutter will cause the 0Hz not to be 0Hz and that's how you measure your W&F. Dedicated audio test sets sometimes included W&F measurement. Bear in mind that if the test LP isn't concentric it will cause W&F. |
|
|
| deduikertjes |
Dear EC8010
thank you for your explanation. but...
Well, in the past I did some signal analysis on pressure levels underneath water waves. Then I performed a few tricks I dont understand nowadays. Apperently unlearning is as easy as learning. In short:
HELP :bigeyes: I dont understand:| quote: | | If you were to multiply the replayed signal by a known constant frequency signal, you would get sum and difference frequencies. If you replayed 1kHz and multiplied it by 1kHz, then the sum would be 2kHz and the difference 0Hz. If you low-pass filtered, you would be left with 0Hz. |
Multiplying 1khz with 1 khz gives me a sum of 2 khz and a difference of 0 Hz? How do I do this multiplication? The low-pass filtering I can get done. |
|
|
| EC8010 |
The sum and difference thing comes from the trigonometrical identities:
sin(A+B) = sinAcosB + cosAsinB
sin(A-B) = sinAcosB - cosAsinB
Adding the two identities gives:
sin(A+B) + sin(A-B) = sinAcosB
You recover sinA from the LP and multiply it by your local oscillator cosB, giving you sin(A+B) (sum frequency) and sin(A-B) (difference frequency).
(By the way, I'm not a smartarse - I had to look up the identities.)
You need a four-quadrant multiplier. Once upon a time, all broadcast vision equipment contained one from vision mixers to PAL coders. The ubiquitous part was the Motorola MC1496.
A quick search found:
http://www.analog.com/en/prod/0,,773_862_AD834,00.html
I'm afraid multipliers were always expensive. But you might be able to get a sample? Alternatively, it will probably be cheaper and easier in the long run to search ebay for a commercial W&F meter than building one from scratch. |
|
|
| valveitude |
I was originally going to post this on my turntable thread, but I think it would be of interest here.
I have been working on a design for a AC power supply that would have the following characteristics..
A) It needs to be stable
B) It needs to be clean
C) It needs to have a 2-phase output to eliminate the motor cap.
D) It needs to be easy to change the motor speed (frequency)
E) It needs to be easy to build
Getting 2 or 3 of the five requirements is pretty easy, but getting them all was proving to be frustrating until I came across the MAX 29X series of switched-capacitor low pass filters from Maxim. My thinking went in a new direction after that. These chips are a 8-Pole Low-Pass filter that turns a square wave into a low distortion sine wave. A .1% THD sine wave is what you get with no additional filtering, and it includes a uncommitted Op-Amp for an addition 2-Pole analog filter a cleaner sine. Or you can go hog-wild and cascade 2 of them for a –70dB Dist+Noise (the floor of the chip). All is not free however. In addition to the signal you want to filter, you need to provide a clock 100X the freq. you are filtering. It would seem a digital solution is in order. But wait…If you only need 1 frequency (and you know exactly what that is) no problem, but if you want multiple speeds and pitch control it gets a whole lot more complicated.
I’ve attached a schematic of my solution. It’s a analog/digital hybrid with analog on the ends and digital in the middle.
First the master clock. I’m using a simple op-amp oscillator with 2 timing components. The freq. is set by C1 and R1. Since all we need here is a pulse, you can use the crudiest op-amp you have on hand. This clock only needs to run 400X higher than the desired freq.. The values shown are for 24Khz, 400 times 60Hz. The stability of this oscillator is determined by the temperature coefficient of the timing components R1/C1. Mouser sells 30ppm/Deg. C ceramic caps for pennies, and a cheap metal film res. is around 20ppm. So without any temperature stabilizer your looking at a 50ppm Oscillator. Stabilizing the temp would be pretty easy with a LM34 based oven. The real bonus here is the ability to change speed by changing the value of R1.
The clock gets split into 2 paths. One goes to the cascaded D-FlipFlops (4013’s), A1 and A2. This sets up a “divide by four”, providing a clean 50% duty cycle signal at 100X freq. for the Max291’s clock. The other branch feeds to a bank of three 4017 decade counters. The first two form a “divide by fifty”. The 3rd one is the slick bit, it is setup as a “divide by four” divider, since the 0 thru 3 outputs sequences high, they are each high 25% of the time…in other words they are 90 deg. apart :D . Taking the count off of the 2 and 3 outputs gives me two signals with a 90 deg phase difference. These go to a flip-flop set up as a “divide by two” giving the signal its final division down to 60Hz, and turns the 25% duty cycle signal into a 50% duty cycle. This in turn is fed to the MAX 291 filters.
Not shown is the power section. It will be a chip amp driving step-ups. I also plan to implement some other features, but this should give you a good idea of how it works. Hopefully all the parts will get here in a week or so, so I can breadboard it.
If anybody sees something I’ve overlooked PLEASE let me know.
Casey |
|
|
| deduikertjes |
| quote: | | You recover sinA from the LP and multiply it by your local oscillator cosB, giving you sin(A+B) (sum frequency) and sin(A-B) (difference frequency). |
Yes, I understand that a hardware solution can be build or bought. But it should be possible to do all this in my computer once the signal from the LP is digitized. Or am I :crazy: ??
Anybody any idea's. It shouldn't be that difficult. Even commercial software vendors have developed that. :D
Any freeware solutions avaiable? Or suggestions how such thing can be written? |
|
|
| deduikertjes |
| quote: | | Yes, I understand that a hardware solution can be build or bought. But it should be possible to do all this in my computer once the signal from the LP is digitized. Or am I ?? |
Well, Using AudacityAudacity its possible to do al kinds of programming via the Nyquist prompt. I succeeded in multiplying my digitized 2000Hz wave with a pure 2000Hz wave. The result: (see image).
Anybody knows how to proceed from here?
greetz, MArco |
|
|
| EC8010 |
| I'm afraid I'm a computer hooligan, so I can't help you. |
|
|
| deduikertjes |
| quote: | | Well, Using AudacityAudacity its possible to do al kinds of programming via the Nyquist prompt. I succeeded in multiplying my digitized 2000Hz wave with a pure 2000Hz wave. The result: (see image). |
| quote: | | Anybody knows how to proceed from here? |
I thought of it myself (Yes I know I should have thought before I posted :headbash:
What I did: apply a low pass filter (100 hz) and voila, there is the wow and flutter signal.
So i continued by writing a Nyquist plug-in for Audacity to this automatically:
;nyquist plug-in
;version 1
;type process
;name "Wow and Flutter..."
;action "Distilling wow and flutter signal..."
;info "Frequency of the signal"
;control f "Frequency" real "Hz" 250 500 4000
(lp
(lowpass8
(mult s (hzosc f))
100)
110)
Name the above wow_and_flutter.ny, put it in the plug-ins folder and it works. Select a piece of recorded audio, select wow and flutter from the effect menu and the wow and flutter part of the signal is returned (see attached image).
This analysis is very dependent on having the frequency exactly right. So before doing this trick, determine the exact frequency of the recorded signal (still working to get that programmed as well). You can get a good estimate in the image spectrum menu of audacity.
Now starts the interpreting part :D |
|
|
| deduikertjes |
Well, I've been busy building all sorts of things (amps, turntable, arm, kitchen), but I got around to this again.
Earlier I mentioned I had a lot of wow and flutter using the PS. That was due to a cold joint. The mass (earth) of the oscillator was not joined to the mass of the amp. Having fixed that things improved. Result definately better than mains:D .
In my new turntable (look for posts in a few days), I can't change speed to 45 rpm by running the belt over a larger pully, so I decided that I want to have a second frequency generated, for higher speed.
I've worked out the right frequency, and build an oscilator for that. Altough the sine was lookuing good it dind't work. The motor just vibrated and didn't run. I suspect the phase shifting cap. I think it should have a different value for a different frequency. Does anyone know how to calculate this value?
thanx, MArco
By the way. I never figured out how to extract the frequency of the recorded sound I mentioned earlier:
| quote: | | This analysis is very dependent on having the frequency exactly right. So before doing this trick, determine the exact frequency of the recorded signal (still working to get that programmed as well). You can get a good estimate in the image spectrum menu of audacity. |
Cooledit can find it out automatically. |
|
|
| Mark Kelly |
| quote: | Originally posted by deduikertjes
I think it should have a different value for a different frequency. Does anyone know how to calculate this value?
thanx, MArco
|
The phase shift cap and the motor form a second order low pass filter, the phase shift of which will be 90 degrees at f = 1/2piSQRT(LC) where L is the motor coil inductance. |
|
|
| deduikertjes |
| quote: | | 90 degrees at f = 1/2piSQRT(LC) where L is the motor coil inductance. |
Thank you, that was what I was looking for. For the lazy ones my calculations for a typical old Thorens (TD160 motor):
It runs on 33 rpm at 50 hz with a 0.15 uF cap.
So I need 45/33 *50 = 67.5 Hz for 45 rpm
Feeding that to the formula twice (once for determining the inductance, once for determining the capacity) gives me a Capacity of about 8.2 nF.
Will see if that works.
greetz, MArco |
|
|
| AndrewT |
Hi,
from 150nF (=0.15uF) down to just 8.2nF for a small change in frequency does not sound right.
Use the first pass through the equation to determine the L value of the motor.
Then keeping the L value the same insert the higher frequency.
However, I would expect the error in phase to be of little significance between 50Hz and 67.5Hz. Could the cap be just a little too big for the higher frequency and prevent starting?
Try 120nF or 100nF, 8.2nF must be too low. |
|
|
| Mark Kelly |
| quote: | Originally posted by deduikertjes
Feeding that to the formula twice ... gives me a Capacity of about 8.2 nF.
greetz, MArco |
| quote: | Originally posted by AndrewT
Hi,
from 150nF (=0.15uF) down to just 8.2nF for a small change in frequency does not sound right.
|
It's 82nF, I suspect a typo.
If you manipulate the formula you'll see that F2/F1 = SQRT(C1/C2) so for F2/F1 = 1.35, C1/C2 must equal 1.822.
150 / 1.822 = 82.3.
82 is as close as you'll get (don't forget caps are usually 10% tolerance) |
|
|
| Mark Kelly |
I was wondering about how well the calculation fitted the usual range of cap values when I realised that this leads to an easy approximation.
1.35 is approximately the 8th root of 10 so 1.35^2 is approximately the 4th root of 10, which makes it equal to the cube of the 12th root. Confused yet? Here's the rule:
The ratio of the caps between 33 and 45 will be three steps down on the E12 values.
If you start with say 100 then you step back three to 56, that's the new cap value.
Works for 78 too, but not as well. 2.36 is close to the 3/8th power of 10 so 2.36^2 is nearly the 3/4 power of 10 so it must also be the 9th power of the 12th root of 10. New rule:
The ratio between caps for 33 and 78 is 9 steps back (or three steps forward and drop a decade).
If you start with 100 you step forward three to 180 and then drop a decade to 18. The approximation is well within the 10% tolerance of the caps.
For those who aren't familiar with them, the E12 values are 1.0, 1.2, 1.5, 1.8, 2.2, 2.7, 3.3, 3.9. 4.7, 5.6, 6.8, 8.2, 10. The ratio for each step is approximately the 12th root of 10, hence the name E12. |
|
|
| deduikertjes |
| quote: | | It's 82nF, I suspect a typo. |
Excuse me. It was a typo indeed.:blush:
I knew it was not a good idea to present the result of my calculations.
But then it leads to a mind boggling but very useful post about cap values and rpm ratios. Sometimes from something bad comes something good.
Thank you, MArco |
|
|
| Mark Kelly |
Also works when changing between 50 Hz and 60 Hz supplies as 1.2 is nearly the 12th root of 10, so:
To change between 50Hz and 60 Hz supplies, the ratio is 2 steps on the E12 table with the smaller cap corresponding to the higher frequency. |
|
|
| deduikertjes |
Well, I'm not being lucky.
My TT PS runs on a classic wien bridge oscillator with a lamp as stabilizing element.
Frequency is very stable, but the amplitude is varying (too much to my taste; after the step-up transformer I've a variation of about 2,5 Volts on 90 Volts on average).
As I've included a variable resistor to set the oscillation I've played around with it. In theory the wien bridge shows least distortion when it just oscillates. It might do, but it shows the biggest amplitude variations. When I crank it up to a larger amplitude just before the waveform starts to show distortion the amplitude variations are smallest (but still to big).
(I've included a few screen shots of the recorded waveform)
Does anyone know how to have this oscillator behave a little better
?
Again any help greatly appreciated.
MArco |
|
|
| EC8010 |
| Your problem is that Wien bridges only work well when the thermal time constant of the thermistor (or lightbulb) in the amplitude stabilization network is considerably longer than the period of the wanted oscillation. When it isn't, it starts trying to control the oscillation causing the effects you've seen. There are various ways of making Wien bridge oscillators work at low frequencies. The easiest is to follow the oscillator with a precision rectifier, then smooth it with a very long time constant (a few seconds) and apply that DC to a FET to use the FET as a variable resistor in the amplitude stabilization circuit. This generally means that you need four op-amps instead of one. An even more complex method exchanges the FET for a four-quadrant multiplier... |
|
|
| Mark Kelly |
As EC8010 said, your problem is the loop control.
If you do go for the rectifier approach, here are a few things that I have found improve the performance:
Use an LED / LDR pair instead of the lamp - the rectifier drives the LED, the LDR is the variable resistance. Perkin Elmer make a part called "Vactrol" which is a matched LED / LDR pair in a lightproof envelope. Get the low Tempco version (VTL5C1 I think.).
Use an elliptical filter rather than a standard one. The easiest way to describe the filter desired is that it's a high pass filter followed by a differential amplifier which amplifies the DC difference and rejects the AC common mode. If the diff amp is made variable you can tune it to achieve almost perfect rejection of one frequency (the ripple frequency).
By the time you've done all this you might as well junk the Wien bridge and use a modified Fraser phase shift oscillator to give you a quadrature tracking generator. The Fraser works by placing two all pass filters in series and then inverting the output of the last into the input of the first. The loop has 360 degree of phase shift at the 90 degree phase shift point of the all pass ilters.
Since the phase shift of the all pas is controlled by a single RC pair you can make the RCs variable with a double gang pot and match the caps and the thing will track with 90 degrees between the two intermediate outputs over the range of the RC adjustment. |
|
|
| deduikertjes |
Mark, EC8010 thank you, but your answers swept me off my feet a bit.
I hoped for something more simple.
If I'm not mistaken, my set-up used to perform much better (variations after the step up transformer only a few tens of a Volt).
When I heard a sort of wow in my turntable that I was not used to I started measuring. Then I saw the larger variations.
Could the cause also be that the lamp has deteriorated over time, or something stupid like that?
(or do I have better measurement apparatus nowadays so that I see variations that were not there before).
After all it should be possible to use a wien bridge to generate a sinus with enough precision to drive a turntable. Or shouldn't it?
greetings, MArco |
|
|
| Mark Kelly |
I used to use a Wien bridge in my first ever TT drive and it worked reasonably well. Maybe I used a lamp with more thermal lag than yours. One thing which I know helped is to use two op amps in the Wien bridge with the AGC (the lamp) between them. The design I used is here:
Note the low impedance around the lamp
Further info at
http://www.members.iinet.net.au/~qu...audio/TTPS.html
Just one thing - when the Wien bridge is not driving the TT output amplifier, how does it perform?. |
|
|
| EC8010 |
| quote: | Originally posted by deduikertjes
After all it should be possible to use a Wien bridge to generate a sinus with enough precision to drive a turntable. Or shouldn't it? |
Sadly, the answer is no. It is possible to add fixes to the Wien bridge to persuade it to work well at 50Hz, but by the time you've done that you could have built a more suitable oscillator. Like Mark, I favour quadrature oscillators. |
|
|
| AndrewT |
and your opinions on a divided down crystal oscillator?
Does the final 50/60/67.5Hz signal have sufficient precision to drive a power amp to drive the motor? |
|
|
| EC8010 |
Divided down crystal oscillators are obviously more accurate in terms of absolute frequency than any analogue oscillator. The thing is, you often don't need that accuracy. Turntable drive mechanisms (belt, idler) have a little slip and the slip varies with temperature, humidity, belt tension or idler pressure, so you usually need a fine tweak of frequency to get the speed exactly right. It is possible to vary the division ratio of the divider in a crystal oscillator (usually by changing the preload to a counter), but making the changes sufficiently small as to be almost unnoticeable (and usable) is quite hard, so a divided down crystal oscillator with a nice fine speed control is usually quite complex. I expect the best way to do it would be to program a PIC, but I'm not so hot on programming.
Edit: And the other problem is filtering that square wave down to a sine wave without the amplitude changing as you change frequency. |
|
|
| AndrewT |
| quote: | Originally posted by EC8010
.........Edit: And the other problem is filtering that square wave down to a sine wave without the amplitude changing as you change frequency. |
That's what I mean about precision.
What if there is a little/lot of distortion in the sinewave signal to the poweramp/motor? |
|
|
| EC8010 |
| quote: | Originally posted by AndrewT
What if there is a little/lot of distortion in the sine wave signal to the poweramp/motor? |
The motor doesn't like it and vibrates more than necessary. I know Mark prefers less distortion but I'm happy if it's below 1%. |
|
|
| deduikertjes |
Having read all the opinions and followed the interesting link of Mark I decided to try and tweak my wien bridge with another lamp.
| quote: | | I used to use a Wien bridge in my first ever TT drive and it worked reasonably well. Maybe I used a lamp with more thermal lag than yours. |
If I want to choose such a lamp I guess I've to choose one which is designed to run at a higher voltage. Or ain't it that simple?
If tweaking doesn't work I think I'll use a cheap MP3 player. After all these come for less than 10 euro's nowadays. That seems to be the cheapest and easiest way to get a good 50 Hz sine (though it's not that much fun as building a thing myself).
greetings, MArco |
|
|
| Mark Kelly |
| quote: | Originally posted by AndrewT
That's what I mean about precision.
What if there is a little/lot of distortion in the sinewave signal to the poweramp/motor? |
I will shortly be in a position to report exactly how important this is.
A bit of background: I recently designed a drive which is basically a digitally clocked DAC with variable clock frequency. To keep things very simple my DAC has only five levels - +1, x, 0, -x and -1. The level of third harmonic is entirely dependent on x. When x = 1/SQRT2 the third harmonic is suppressed (better than -70dB and ) for all other numbers it rises but in different phase according to whether x is greater or lesser than 0.707. Since the lowest harmonic otherwise produced is the seventh it is very easy to filter this to give good purity with variable 3HD levels.
I received the boards from the fab house today and found that they had made the wrong ones so it will be a week before I get to test the result. I think that if the motor has constant reluctance then zero 3HD is ideal. With a salient pole motor there might be some small effect counteracting the cogging due to salience which is what I want to test.
The octal counter that clocks the wave is duplicated with a 2 bit shift, creating a quadrature pair. BTW the clock divisions are around 1 part in 5000 so the output frequency steps are around 0.01 Hz at 50 Hz. The drive also has variable phase angle, master level control plus automatically switched levels for different drive frequencies and a timed "start up boost". |
|
|
| kevinkr |
| quote: | Originally posted by EC8010
Sadly, the answer is no. It is possible to add fixes to the Wien bridge to persuade it to work well at 50Hz, but by the time you've done that you could have built a more suitable oscillator. Like Mark, I favour quadrature oscillators. |
Hi EC8010,
Any possibility you could share your quadrature oscillator design? I'd be very interested in seeing it, as I imagine others here would be. Sounds like a great solution for a table with a low voltage synchronous motor like my TD125.. |
|
|
| EC8010 |
| quote: | Originally posted by kevinkr
Any possibility you could share your quadrature oscillator design? I'd be very interested in seeing it, as I imagine others here would be. |
You have mail. Or you would have if you allowed mail. I'll temporarily allow mail to me..... |
|
|
|
