-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
DAC for industrial vibration data
- Thread starter cedric
- Start date
Not a log sweep.
Do a linearity sweep in Multitone using a 5Hz tone to see if there is something strange going on at low frequencies. Of course, you can also look directly at the 5Hz waveform to compare it to the test signal to see any obvious irregularities.
EDIT: Oh, and try using larger size FFT to help measure lower frequencies to help increase resolution.
restorer-john
Grand Contributor
Oh come on, you and I both know things get crazy when your ADC is DC coupled. That is the reason I have an AC and a DC input on my AudioLab device. Basically the AC input has a blocking cap up front of the internal ADC buffer/scaler, to prevent FR plots from being affected by fluctuating DC at very low frequencies in a linear (0.5Hz step) sweep. No different to the coupling switch on a CRO/DSO.
Outputs on even capacitively coupled devices are not 0V. There is always some offset from leakage, be it a fraction of a millivolt or microvolts. And that fluctuates in any amplifier, slowly.
No issues measuring ADI-2 Pro FS at 5Hz (measured by Cosmos ADC) at 48k. The level is steady during measurement, no variations.
- Joined
- Feb 23, 2016
- Messages
- 20,747
- Likes
- 37,568
Oh well, I don't know. Analog meter, digital meter, ADC meter, and Multitone software meter all show fluctuations below 10 hz. The recorded waveforms look clean and consistent, but lower in level the lower in frequency you go. Not inconsistent with a high pass filter that is 3 db down at 2 hz.
if you're talking about the level meter in Multitone, then I can understand why it might fluctuate at lower frequencies since it's updated a number of times per second. It simply catches a different portion of the waveform with a different peak value depending when it is computed.
restorer-john
Grand Contributor
Not a bad idea to have a selectable gate time or even a defeatable auto gate time like a frequency counter has (0.01/0.1/1/10sec etc) to get a stable number.
This discussion got me thinking about magic mushrooms, those drawn on an oscilloscope in x-y mode using L-R PC sound output. Unwanted filtering is a challenge for that application too apparently. Somewhere there was a list of compatible equipment. Some modify their boards to bypass the filter.
I'll put in some automatic interval selection to make the level meter more stable.
- Thread Starter
- #50
When I see one of these threads, I'm usually a little confused.
What are you trying to solve with a consumer product that couldn't be (better) solved with an industrial one?
Is it just a cost issue? If so, I get it. But you'll likely spend more (money) time than the cost difference using an ill suited product due to inefficiencies. You also may introduce possible points of human error. It seems like it introduces too many possible issues/what ifs, that a commercial product designed for this purpose doesn't have.
If you have many files to process, depending on what kind of input system you are using, you may be able to automate the process too. This could save a ton of time.
There are probably dozens of companies with collectively hundreds of technical sales associates who will happily point you towards the right product for your situation.
The NI cRIO/cDAQ stuff can be pretty economical.
Voltage out and multi i/o (includes analog out) modules. The specs or ranges of specs seem more varied on the Voltage Out modules, but the multi I/O gives you room to grow for future projects.
You'll need to put them in a chassis, the cRIO single slot is probably the cheapest.
Edit: Also, is there not a way to view/process the files you already have, or do some kind of file/data conversion process into the new system/file type? Honestly speaking, this (digital file to analog out to another systems analog in to....?) seems like not a great way to do what you seem to be trying to do.
This may be the best answer here (noting that others have accurately answered the posted question), you're looking through the question to the underlying problem.
NI gear is an option. So why an Audio DAC?
I'm exploring the audio option to see if it is a simple option. Our data is ~3-10,000Hz, so why not look at a product dedicated to producing these signals.? (Answer: maybe I shouldn't - low frequency may not be accurate, voltage peak may fluctuate with heat, ensure use of dedicated programs & drivers i.e. not WMP, fixed sample rate support. That said, it may be accurate enough, testing will determine.
Process the files in another system?
Yes, possible, and sometimes done. Each program we have can achieve different things. An option to move the data into the analogue acquisition unit broadens our capability.
Why go Digital to analogue to digital?
The system archive only accepts analogue input into a proprietary database. There is no way to move the data across digitally.
I will explore further the National Instruments path.
KSTR
Major Contributor
Any AC metering involves time constants, an integration time. It is chosen to be practical in real life where AC for us means frequencies above ~10Hz, or at least above 1Hz.
Cheap multimeters have a simple peak rectifier, better ones have true RMS rectifier but all have time constants.
You wouldn't use an AC meter to monitor the slow voltage drifts of, say, a powersupply, wouldn't you? Technically, it is still AC albeit with mHz or even µHz frequencies but to measure that with an AC technique you'd have to use time-constants of hours, literally.
KSTR
Major Contributor
Log-sweep + convolution is a method that is really robust and pretty immune to noise at any frequency which is why it is the best method (and industry standard) for generating an impulse response (with the DFT of which being the frequency response of magnitude and phase).
When discrete tones are used in presence of high noise, you'd normally do a lot of (time-domain) averaging to reduce/remove the noise. Works also very well for log-sweeps, of course.
The log-sweep + convolution method's robustness can be examined directly, just take a generated log-sweep (say, from REW), add any kind of noise in a WAV editor (even with discrete tones like mains hum) and use that as the "recording" fed into the convolver creating the impulse response.
KSTR
Major Contributor
I don't understand the question. DC is one and the same sample value (+dithering, maybe) sent to the DAC over and over.
If the meter has the capability, set it to DCV with a small aperture and look at max/min values after a minute or so. You'd have to be quite unlucky to not catch the peaks that way. If the DAC is DC coupled, I'd be surprised if it wasn't flat down to DC.
Hmm doesn’t the RME ADI 2 series of PRO DAC’s have options for DC blocking or not in the DAC section ?
Saw some other tread that it’s on by default, as it is meant for pro use you have the option to turn it off if you want to check something ?
@MC_RME might be able to answer directly about their DAC’s low frequency capacity ? Maybe they dealt with scientific use before ? If you work in the 0-100kHz range a modern audio DAC could be a cheap and versatile signal generator with the help of some pc software.
Saw some other tread that it’s on by default, as it is meant for pro use you have the option to turn it off if you want to check something ?
@MC_RME might be able to answer directly about their DAC’s low frequency capacity ? Maybe they dealt with scientific use before ? If you work in the 0-100kHz range a modern audio DAC could be a cheap and versatile signal generator with the help of some pc software.
- Joined
- Feb 23, 2016
- Messages
- 20,747
- Likes
- 37,568
Well at least down to 1 hz peak values max/min were consistent with what I'm seeing otherwise. Something like 7-8 db down at 3hz and less. Took the max value and adjusted it for peak vs RMS. Max and min values were pretty close other than being positive and negative values. A few millivolt offset.
- Joined
- Feb 23, 2016
- Messages
- 20,747
- Likes
- 37,568
Well yes, my question was more how were you getting the file. I could manually create one, and thought maybe you had software to make it simple.
KSTR
Major Contributor
In Adobe Audition
- new file, set sample rate, bit depth and number of channels
- create silence for some seconds
- apply gain/fade, set constant gain, set 0dB, set 100% (or whatever value) DC bias, absolute
I would think with other editors there will be similar ways to do it.
KSTR
Major Contributor
Yes, both DAC and Pro have options for DC handling with latest firmwares
- high-pass filter (not what we want)
- don't filter but protect (not what we want)
- don't filter, detect only, just displaying a warning
- Joined
- Feb 23, 2016
- Messages
- 20,747
- Likes
- 37,568
So created a DC wav file and the D10 Balanced does put out 5.92 volts DC. Which is peak voltage for a 4.18 volt RMS sine wave. I previously measured 4.18 volts on 1 khz.
The RME also puts out an appropriate level of DC.
So the last 40 posts or so were not really worth posting.
@cedric if you are still with us, at least a D10 Balanced Topping dac which is $139 if it hasn't gone up will work down to DC. I would guess the RCA output version D10 which is cheaper does as well.
Also thinking about this as well as upsampling your vibration file, you can in software speed it up. For instance you can speed it up 400% and all frequencies would scale. Perhaps that helps whatever device you intend to feed this signal to from the DAC.
Thanks to everyone who had suggestions and helped me figure this out.
PS : Thought I'd add for anyone wanting a DC file, I didn't have Adobe. What I did was create a 1 hz square wave in Audacity, and then snipped out the negative portions for 4 seconds worth. Then copy and added at the end until I had 32 seconds worth.
The RME also puts out an appropriate level of DC.
So the last 40 posts or so were not really worth posting.
@cedric if you are still with us, at least a D10 Balanced Topping dac which is $139 if it hasn't gone up will work down to DC. I would guess the RCA output version D10 which is cheaper does as well.
Also thinking about this as well as upsampling your vibration file, you can in software speed it up. For instance you can speed it up 400% and all frequencies would scale. Perhaps that helps whatever device you intend to feed this signal to from the DAC.
Thanks to everyone who had suggestions and helped me figure this out.
PS : Thought I'd add for anyone wanting a DC file, I didn't have Adobe. What I did was create a 1 hz square wave in Audacity, and then snipped out the negative portions for 4 seconds worth. Then copy and added at the end until I had 32 seconds worth.
Last edited:
Similar threads
- Poll
