Hello,
I'm using a double DDS generator to run my Airpax/Premotec motors. For that I'm using LM3886 based amp modules, 2 channels. That is not a straight choice I think, I had some ringing problems before I've inserted 2 antiparallel 47uF caps at the output of the chipamps, but I always wanted to try TDA7293 because I saw some others use those without problems.
Now I need a 3 channel amp (also building a 3 channel DDS gen) and there is no ready made 3 channel module, so I've decided to make one. I have power supplies, so did not put it on the board, but plan to use a simple Graetz and 4x 22mF caps.
So, if you have experience with these chipamps, would you have a look?
As I know the feedback resistors has to be placed the closest possible (some even solder on the IC legs) and obviously in/out has to be separated. Any other thing to watch?
Thanks,
JG
I'm using a double DDS generator to run my Airpax/Premotec motors. For that I'm using LM3886 based amp modules, 2 channels. That is not a straight choice I think, I had some ringing problems before I've inserted 2 antiparallel 47uF caps at the output of the chipamps, but I always wanted to try TDA7293 because I saw some others use those without problems.
Now I need a 3 channel amp (also building a 3 channel DDS gen) and there is no ready made 3 channel module, so I've decided to make one. I have power supplies, so did not put it on the board, but plan to use a simple Graetz and 4x 22mF caps.
So, if you have experience with these chipamps, would you have a look?
As I know the feedback resistors has to be placed the closest possible (some even solder on the IC legs) and obviously in/out has to be separated. Any other thing to watch?
Thanks,
JG
Attachments
Okay. You're using direct digital synthesis and LM3886 modules as motor drivers, is that correct? If so, is this for a TT or what? A robot?
Also, don't know much about Graetz rectifiers, if that's what you're referring to?
Also, don't know much about Graetz rectifiers, if that's what you're referring to?
Correct. The Airpax/Premotec/Philips sync motors You can find in Rega, Systemdek and several other turntables. Those are operated on 1 main and 1 capacitor shifted (90 degrees in theory) phase. To generate both phase accurately and independent of mains brings great improvement.
Now, I want to experiment with 3 phase motors and I need a better driver amp.
Now, I want to experiment with 3 phase motors and I need a better driver amp.
C10 & C11 do not look right. For high frequency stability, you should have 2.2 Ohm resistor in series with 0.1µFd cap to ground at the output of the chip amp instead of two 47µFd caps in series.
The purpose of these caps may be similar to the Zobel network you may refer to (?). Due to its inductive nature, the motor could be high impedance as the frequency goes up. The anti-parallel 47uF caps are based on experience. I saw it in a Voyd power supply, tried and it does work well. Will try without it initially.
There are even ready made, excellent 3 phase drivers on this site, but all has PWM output like what you linked and I want to avoid that. That would be simple, the amplification would be simple H bridges. Even more simple, I was playing a lot with a Yaskawa 3 phase motor driver which works very well, but, the switching there is worst, it is in audio spectrum. These are good to read, but not good for a TT motor drive.
DDS sine waves are typically a stepped approximation. IOW, they have quantizing distortion. You could use techniques as are done with audio dacs to deal with that however. One question is how many bits of digital approximation to a sine wave to you need. 12, 16, 24 bits? Then there is the question of what sample rate is wanted to produce the close-to-analog sine waves?
IOW, do you want a 16/44 audio dac channel to drive each motor?
IOW, do you want a 16/44 audio dac channel to drive each motor?
DDS generators end with DACs, but in extreme high f comparing to the f we generate here.
The way I got into DDS generators is that I was using for years already in radio applications. There the DDS generators represent much higher class comparing to, -for example- a pll based signal generators. On the other hand, of course, a bad implementation can ruin performance. There are good reasons why DDS is popular in instrumentation and higher class rf circuits. It also has to pay back it's generally higher cost there.
In this application, to filter the 20MHz clock (PLL off of course) is easy since it is several decades away from the audio spectrum and also because we are not speaking of on/off switching like in case of PWM, but only a very fine step on the generated sin curve. The distortion you speak about on an 50Hz signal, clocked out with 20MHz is ... small. On the other hand, filtering out the harmonics of a PWM drive, clocked close to the audio spectrum is not easy and the remaining unwanted components are in a very different level. Also, you have a lot less steps in PWM, that is a DAC at the end also and if we translate the error to quantization distortion, that would be decades higher.
I know this is a HUGE overkill for a motor drive, but in this hobby it is not about rational solutions 🙂 The efforts needed to put in comparing to results is an exponential curve. For sure, I can not imagine to use this almost $100 IC in a commercial consumer equipment ... but, I have no intention to mass produce it at all.
No, I do not want an audio dac (3 channel) to drive one motor. Why would I want when there are ICs with all the digital signal generation, DACs etc. already built into one package and I just need to tell it what to put out on its output + the sampling freq is a lot higher too.
I do not mind to discuss DDS generators, but if someone have info on the TDA7293 use I would appreciate. At the weekend I plan to finalize the pcb design and order it.
The way I got into DDS generators is that I was using for years already in radio applications. There the DDS generators represent much higher class comparing to, -for example- a pll based signal generators. On the other hand, of course, a bad implementation can ruin performance. There are good reasons why DDS is popular in instrumentation and higher class rf circuits. It also has to pay back it's generally higher cost there.
In this application, to filter the 20MHz clock (PLL off of course) is easy since it is several decades away from the audio spectrum and also because we are not speaking of on/off switching like in case of PWM, but only a very fine step on the generated sin curve. The distortion you speak about on an 50Hz signal, clocked out with 20MHz is ... small. On the other hand, filtering out the harmonics of a PWM drive, clocked close to the audio spectrum is not easy and the remaining unwanted components are in a very different level. Also, you have a lot less steps in PWM, that is a DAC at the end also and if we translate the error to quantization distortion, that would be decades higher.
I know this is a HUGE overkill for a motor drive, but in this hobby it is not about rational solutions 🙂 The efforts needed to put in comparing to results is an exponential curve. For sure, I can not imagine to use this almost $100 IC in a commercial consumer equipment ... but, I have no intention to mass produce it at all.
No, I do not want an audio dac (3 channel) to drive one motor. Why would I want when there are ICs with all the digital signal generation, DACs etc. already built into one package and I just need to tell it what to put out on its output + the sampling freq is a lot higher too.
I do not mind to discuss DDS generators, but if someone have info on the TDA7293 use I would appreciate. At the weekend I plan to finalize the pcb design and order it.
DDS is somewhat different in RF applications as opposed audio frequencies. For instance, is close-in phase noise an issue for your application? Its can be minimal for oscillators down around 5-6MHz. OTOH, some much higher frequency RF clock devices have low jitter, but its more like far-out phase noise in most cases.
For driving motors at a few 10's or 100's of Hz it seems like you would be looking at something more like audio dacs.
For driving motors at a few 10's or 100's of Hz it seems like you would be looking at something more like audio dacs.
I do not know what do you want to say with that. Why would an audio DAC be better in the app in any aspect? Lower sampling rate would increase phase noise, except in a special case when the desired freq times x is exactly the sampling rate, but that will never happen. The higher sampling rate would get closer to the desired output with lower noise in freq domain too. Also, to feed an audio DAC you need to compute the signal. Why when all this is available in an ic with better results?
But maybe you want to open a thread on DDS because no one wants to help now on the TDA7293 specialities what I need 🙂
But maybe you want to open a thread on DDS because no one wants to help now on the TDA7293 specialities what I need 🙂
According to theory and pretty close to that in actual measurements, phase noise increases by 6dB with every doubling of clock frequency (although there is a sweet spot at around 5-6MHz and extending pretty much up to 10MHz or so).
Close-in phase noise is commonly dealt with as a separate issue, since it tends to cause certain effects that are different from those of far-out phase noise.
None of this is to say that it matters at all for your application. I don't know if it does or not.
Last edited:
For the dac, In dBc/Hz yes, but calculate the time for the same (or 6dB higher) dBc/Hz for 44k and 20M sampling. Anyway, the main source for the phase noise will not be the base dac clock intoduced noise.
The difference between an ideal sine wave and what the dac output is getting smaller and smaller with increasing resolution in time and amplitude. In time domain this difference makes phase difference. Less with higher resolution. Look at it as a system, not only the dac part. The only excemption is when the sampling rate divided by the desired frequency is an integer. That case the dac clock will contribute in the output phase noise, but that only exist in theory.
The difference between an ideal sine wave and what the dac output is getting smaller and smaller with increasing resolution in time and amplitude. In time domain this difference makes phase difference. Less with higher resolution. Look at it as a system, not only the dac part. The only excemption is when the sampling rate divided by the desired frequency is an integer. That case the dac clock will contribute in the output phase noise, but that only exist in theory.
There is well established industry as well as controllers along with the math and related software. All one needs is DSP/RISC chip feeding phase values into DAC-s and linear amps instead of PWM sections of such chip. Most even works with standard tacho-s. Since the phase shift in motor is constant, it could be as simple as a set of counters and sine table ROM with DAC. VCO|/PLL for speed control, analog or otherwise to one's liking. Something like that will usually suffice, with few tweaks
https://www.homemade-circuits.com/wp-content/uploads/2014/12/3phasegeneartorUsingTransistors.png
https://www.homemade-circuits.com/wp-content/uploads/2014/12/3phasegeneartorUsingTransistors.png
Someone here needs to buy and test this ThinGap Nema23 motor with their sine wave driver. 8 poles, built in encoder, etc.
https://www.ebay.com/itm/204061302561
Expensive at $1000, but cheaper that I would have thought to get a ThinGap motor!
https://www.ebay.com/itm/204061302561
Expensive at $1000, but cheaper that I would have thought to get a ThinGap motor!