Hi all diyaudio readers,
I am third year Audio Technology student and the time has now come for me to do my third year project. I am aiming towards building an arduino based audio analyser if it is possible, however before I go any further I need to establish the market / need for my project in the real world.
So I ask you all for your help and literally 5 minutes of your time, if you could do this survey it would be much appreciated.
DIY Audio Analyser Survey
also I would be very interested in your opinions and ideas in general on the subject of audio analysers.
Thank you for your time
Oberlix
I am third year Audio Technology student and the time has now come for me to do my third year project. I am aiming towards building an arduino based audio analyser if it is possible, however before I go any further I need to establish the market / need for my project in the real world.
So I ask you all for your help and literally 5 minutes of your time, if you could do this survey it would be much appreciated.
DIY Audio Analyser Survey
also I would be very interested in your opinions and ideas in general on the subject of audio analysers.
Thank you for your time
Oberlix
I've used many dedicated test sets for radio.
A DIY audio analyser would be nice.
An entry-level new audio analyser costs ~$5000.
In order to be worthwhile an analyser must be accurate and have a good resolution, and that means it must perform better than the DUT.
You can see the kind of performance required by looking at dScope and Audio Precision test sets.
It's possible, likely even, that the level of performance required is beyond that of a 3rd. year graduate student project. Since one of the things likely to come under test is a soundcard, performance should exceed that of a quality soundcard. Google 'audio widget'.
Some things missing from your list of tests are noise, IMD, crosstalk and jitter.
I'm not quite sure how an arduino will fit in, unless it's just providing a USB interface. My preference would be a PIC (google 'Building_a_PIC18F_USB_device') You almost certainly want to use a PC for its display capabilities. Organising a real-time display may require careful use of the USB bandwidth, and you will need some Windows or Linux GUI development skills, Visual Basic as a minimum.
Just getting a few measurements calibrated, up and running and showing an appreciation of the resolution required in a useful instrument would probably constitute sufficient to get a high mark for your project.
Good luck.
A DIY audio analyser would be nice.
An entry-level new audio analyser costs ~$5000.
In order to be worthwhile an analyser must be accurate and have a good resolution, and that means it must perform better than the DUT.
You can see the kind of performance required by looking at dScope and Audio Precision test sets.
It's possible, likely even, that the level of performance required is beyond that of a 3rd. year graduate student project. Since one of the things likely to come under test is a soundcard, performance should exceed that of a quality soundcard. Google 'audio widget'.
Some things missing from your list of tests are noise, IMD, crosstalk and jitter.
I'm not quite sure how an arduino will fit in, unless it's just providing a USB interface. My preference would be a PIC (google 'Building_a_PIC18F_USB_device') You almost certainly want to use a PC for its display capabilities. Organising a real-time display may require careful use of the USB bandwidth, and you will need some Windows or Linux GUI development skills, Visual Basic as a minimum.
Just getting a few measurements calibrated, up and running and showing an appreciation of the resolution required in a useful instrument would probably constitute sufficient to get a high mark for your project.
Good luck.
Counter Culture,
Thank you for your feedback, your perfectly write about the capabilities required and I will be aiming to get as accurate as possible within the time limit of my project. In terms of satisfying my degree criterion for the project module if I get an fft or Goertzel algorithm properly working I will of achieved this.
I'm intrigued with your preference for the pic 18, why would you side with these devices?
Thank you for your time
Oberlix
Thank you for your feedback, your perfectly write about the capabilities required and I will be aiming to get as accurate as possible within the time limit of my project. In terms of satisfying my degree criterion for the project module if I get an fft or Goertzel algorithm properly working I will of achieved this.
I'm intrigued with your preference for the pic 18, why would you side with these devices?
Thank you for your time
Oberlix
Why PIC?
I prefer to work directly with a chip, rather than a board. It's more flexible, and I happen to know that there's a working USB implementation available for download.
The arduino is promoted as a system for artists, designers, hobbyists. I don't have direct experience of it, but the experience I do have of systems intended to 'simplify' development is that what you learn about using them is not necessarily portable. Of course there are features of the PIC that mean that if you learn to program them in assembly language then you will have to learn a new vocabulary if you change to say, Atmega processors, but more of the experience will carry over.
Another problem with 'accessible' systems is that they often omit to provide access to all the capabilities of a chip. Even where this is not the case, facilities may be implemented in nonstandard ways which are of limited value in a professional environment.
There are other reasons which will not apply to you.
At home, I have a PicKit2 in-system USB programmer for PICs (which BTW is cheap).
I have an existing codebase for the PIC in assembler and C. This means that in most instances I don't start to write a program from scratch, I take a bit of working code and modify it.
I would still recommend that you learn to program in assembly language and C. C provides a level of abstraction with a minimal compromise of performance. Most C compilers allow the embedding of inline assembly language where maximum performance is required. Most compilers also produce assembly language as an intermediate step to machine code, and if inline assembler is not supported the intermediate files can be hacked. Assembly language provides the maximum performance and uncompromised access to the features of the hardware.
Of course there are advantages to systems such as the arduino. The learning curve is not as steep. Many of the features commonly required such as USB ports are already implemented in firmware, which can be a big time-saver. I don't implement floating point maths in assembly language, I use a C library.
There's no hard-and-fast right and wrong, it's a judgement call as to which route you choose.
I prefer to work directly with a chip, rather than a board. It's more flexible, and I happen to know that there's a working USB implementation available for download.
The arduino is promoted as a system for artists, designers, hobbyists. I don't have direct experience of it, but the experience I do have of systems intended to 'simplify' development is that what you learn about using them is not necessarily portable. Of course there are features of the PIC that mean that if you learn to program them in assembly language then you will have to learn a new vocabulary if you change to say, Atmega processors, but more of the experience will carry over.
Another problem with 'accessible' systems is that they often omit to provide access to all the capabilities of a chip. Even where this is not the case, facilities may be implemented in nonstandard ways which are of limited value in a professional environment.
There are other reasons which will not apply to you.
At home, I have a PicKit2 in-system USB programmer for PICs (which BTW is cheap).
I have an existing codebase for the PIC in assembler and C. This means that in most instances I don't start to write a program from scratch, I take a bit of working code and modify it.
I would still recommend that you learn to program in assembly language and C. C provides a level of abstraction with a minimal compromise of performance. Most C compilers allow the embedding of inline assembly language where maximum performance is required. Most compilers also produce assembly language as an intermediate step to machine code, and if inline assembler is not supported the intermediate files can be hacked. Assembly language provides the maximum performance and uncompromised access to the features of the hardware.
Of course there are advantages to systems such as the arduino. The learning curve is not as steep. Many of the features commonly required such as USB ports are already implemented in firmware, which can be a big time-saver. I don't implement floating point maths in assembly language, I use a C library.
There's no hard-and-fast right and wrong, it's a judgement call as to which route you choose.
Hi,
Suggested features:
1) PC/Mac connected "external USB sound card" for use with analysis software (RMAA etc).
2) Ideally powered fully from USB Bus with an option for quiet, linear external PSU.
3) Minimum 24/96KHz sample rate (192KHz preferred).
4) AD and DA with > 120dB DNR.
5) Analog stages build with measurements in mind, including calibration options, protection, BNC Connectors on Board etc.
6) Includes single channel Microphone preamp, Phantom Power Supply for Behringer ECM8000 Microphone and a single channel 1W amplifier to allow easy stand alone measurements of speakers from Battery powered laptop outdoors.
Ciao T
Suggested features:
1) PC/Mac connected "external USB sound card" for use with analysis software (RMAA etc).
2) Ideally powered fully from USB Bus with an option for quiet, linear external PSU.
3) Minimum 24/96KHz sample rate (192KHz preferred).
4) AD and DA with > 120dB DNR.
5) Analog stages build with measurements in mind, including calibration options, protection, BNC Connectors on Board etc.
6) Includes single channel Microphone preamp, Phantom Power Supply for Behringer ECM8000 Microphone and a single channel 1W amplifier to allow easy stand alone measurements of speakers from Battery powered laptop outdoors.
Ciao T
A DIY analyser has a different set of requirements from an AP or dScope. Those machines are doing double duty (at least) - they're used for both R&D and production test. The DIY analyser dispenses with the production test aspects and hence doesn't really need the level of accuracy those boys provide.
The Goertzel algorithm is a good one to use to find, e.g. how much 2nd/3rd/5th harmonic distortion a device is generating as FFT does lots of calculations for frequencies where no signal is expected to lie. The big disadvantage of FFT is the fact that its not real-time in most analysers. For DIY use I feel real-time feedback is much more useful and as an end customer I'd willingly trade this off for overall accuracy. Who really cares if the THD figure is accurate to 1% or 10%?
Just my 2p-worth
What I would like to see as DIY from such a device:
- Oscilloscope & spectrum analyzer functions for the audio band (up to ~ 200KHz)
- USB connectivity
- various input options: XLR analogue both for line level signals and microphones
(phantom power supply), SP/DIF, high impedance input (1 meg) for probes.
Basically an AP system 1 updated with USB ;-)
- Oscilloscope & spectrum analyzer functions for the audio band (up to ~ 200KHz)
- USB connectivity
- various input options: XLR analogue both for line level signals and microphones
(phantom power supply), SP/DIF, high impedance input (1 meg) for probes.
Basically an AP system 1 updated with USB ;-)
Opinions, Keep them Coming.
Hi,
Firstly counter culture thanks for your reply, I have worked with PIC's before and just happen to have a programmer sat on my desk. I have worked directly in HEX and other high level languages including C, as it happens the arduino language is C with a couple of specifics added in. Just to let you know arduino is pretty good about direct hardware interfacing and you can in fact interface with as far as I know all the features of the atmega chips.
I am currently siding with the arduino out of ease of implementation, let's say for instance I choose to use a DAC output (highly likely) most DAC's use SPI or I2S. The arduino has an SPI library built in which simplifies communication enormously, if not and I use I2S it is just a simple case digital in's and out's. In my case these little built in libraries are highly useful however I do understand in some cases they are not the most efficiently method. That is why if I were actually building this as a project I would definitely work on two versions an arduino or similar type for people wishing to either build themselves or easily modify. Or a dedicated chips for each function which work faster but are not easily modified. either way I would always make the source code open.
The USB interface as you so rightly say is a very useful feature especially for getting data out I have already used an arduino and python to read a corrupted bios of my main pc, just working on writing the new one to it with a nice check of each byte written.
Thank you once again for your time, I am definitely going to have to note PIC's as an alternate option in my project definition.
Thorsten L
I'm likely your ideas, most a little beyond the scope of my third year project but still duly noted. I think you point about USB 5v operation is definitely a possibility the only problem is it does become difficult to keep this applied when you take your point number five into consideration. So as you say the option for some measurements of a linear external PSU would definitely be need if we were to take accuracy and calibration seriously.
Which brings me to your point number 5 I personally think number 5 is the most important as in my mind, this is what really separates most of the professional solutions from the cheaper ones. Take for instance the Dscope series 3 a beautiful bit of kit but what really makes it special other than there very flexible software is there circuit design the input has a nice scope of 160dB ( dynamic) due to there switched impedance, switch attenuation and variable gain.
Calibration is also very important and its something I hadn't as of yet put much thought into though it has now gone on my list of things missed.
Your point 6 is an interesting one that I hadn't thought of, phantom power and a mic input stage were on my list but an amplifier for speaker testing wasn't. The reasoning for me of not adding a built in amplifier is because no amplifier is truly linear and putting something that generates a good amount of thermal noise etc into the same box as something you want to be precised seems a bit redundant to me.
However because you have mentioned it I would like to know other peoples opinions on whether or not they would want a built in amp in DIY audio analyser.
Thanks for your time
abraxalito
Thanks for pointing out something I had not thought of previously, the difference in spec for a professional unit and a DIY unit. However i find it interesting that you would trade accuracy for real time feedback. especially as all that real time analysis is in most cases is a glorified volt meter, you say your an end consumer what do current use audio T&M for?
SY & hidnplayr
So far I had considered a system including 200Ohm 600Ohm and 100K, now I understand the 1M would obviously have very little loading however does it not inherently add the possibility for RF interference?
Thank you all for your time
Oberlix
Hi,
Firstly counter culture thanks for your reply, I have worked with PIC's before and just happen to have a programmer sat on my desk. I have worked directly in HEX and other high level languages including C, as it happens the arduino language is C with a couple of specifics added in. Just to let you know arduino is pretty good about direct hardware interfacing and you can in fact interface with as far as I know all the features of the atmega chips.
I am currently siding with the arduino out of ease of implementation, let's say for instance I choose to use a DAC output (highly likely) most DAC's use SPI or I2S. The arduino has an SPI library built in which simplifies communication enormously, if not and I use I2S it is just a simple case digital in's and out's. In my case these little built in libraries are highly useful however I do understand in some cases they are not the most efficiently method. That is why if I were actually building this as a project I would definitely work on two versions an arduino or similar type for people wishing to either build themselves or easily modify. Or a dedicated chips for each function which work faster but are not easily modified. either way I would always make the source code open.
The USB interface as you so rightly say is a very useful feature especially for getting data out I have already used an arduino and python to read a corrupted bios of my main pc, just working on writing the new one to it with a nice check of each byte written.
Thank you once again for your time, I am definitely going to have to note PIC's as an alternate option in my project definition.
Thorsten L
I'm likely your ideas, most a little beyond the scope of my third year project but still duly noted. I think you point about USB 5v operation is definitely a possibility the only problem is it does become difficult to keep this applied when you take your point number five into consideration. So as you say the option for some measurements of a linear external PSU would definitely be need if we were to take accuracy and calibration seriously.
Which brings me to your point number 5 I personally think number 5 is the most important as in my mind, this is what really separates most of the professional solutions from the cheaper ones. Take for instance the Dscope series 3 a beautiful bit of kit but what really makes it special other than there very flexible software is there circuit design the input has a nice scope of 160dB ( dynamic) due to there switched impedance, switch attenuation and variable gain.
Calibration is also very important and its something I hadn't as of yet put much thought into though it has now gone on my list of things missed.
Your point 6 is an interesting one that I hadn't thought of, phantom power and a mic input stage were on my list but an amplifier for speaker testing wasn't. The reasoning for me of not adding a built in amplifier is because no amplifier is truly linear and putting something that generates a good amount of thermal noise etc into the same box as something you want to be precised seems a bit redundant to me.
However because you have mentioned it I would like to know other peoples opinions on whether or not they would want a built in amp in DIY audio analyser.
Thanks for your time
abraxalito
Thanks for pointing out something I had not thought of previously, the difference in spec for a professional unit and a DIY unit. However i find it interesting that you would trade accuracy for real time feedback. especially as all that real time analysis is in most cases is a glorified volt meter, you say your an end consumer what do current use audio T&M for?
SY & hidnplayr
So far I had considered a system including 200Ohm 600Ohm and 100K, now I understand the 1M would obviously have very little loading however does it not inherently add the possibility for RF interference?
Thank you all for your time
Oberlix
No more so than a scope- it's pretty standard in that world. The 1M isn't just for low loading, it allows the use of a 10x scope probe for extremely low capacitance and usefulness for higher voltages (old tube guy here).
Re: "However because you have mentioned it I would like to know other peoples opinions on whether or not they would want a built in amp in DIY audio analyser."
As a user of various Analysers here for tests, the facility of having a low impedance Output directly available can remove a requirement for additional hardware being needed.
A signal capable of driving down to 3 Ohms at a maximum of 4 Volts RMS would be sufficient to be able to then run standardised 1Watt/1Meter SPL measurements for 4 to16 Ohm nominal loads for example.
On another point regarding measurement bandwidth, the ability to detect/measure switching frequency leakage from Class D type designs could be of much use as well.
Mik
As a user of various Analysers here for tests, the facility of having a low impedance Output directly available can remove a requirement for additional hardware being needed.
A signal capable of driving down to 3 Ohms at a maximum of 4 Volts RMS would be sufficient to be able to then run standardised 1Watt/1Meter SPL measurements for 4 to16 Ohm nominal loads for example.
On another point regarding measurement bandwidth, the ability to detect/measure switching frequency leakage from Class D type designs could be of much use as well.
Mik
One of the features of the AP which I used the most when I was working as an audio designer in industry was the residual output, displayed on a scope. This gives real time feedback of the quality of the THD residual. So while tweaking a circuit (say the bias setting of an output stage) I was able to watch how the THD varied with setting. In a digital-only analyser with FFT, its likely this feature would be lost and its one of the must-haves in my book, which I'd willingly exchange for lower accuracy.
I currently design audio (dacs, source, amps, XOs) for fun, eventually to run a business in digital audio. I use T&M in two complementary ways - one for understanding a design and gaining insights into its operation, second to validate that the construction of a design is indeed correct.
audiomik
I suppose another advantage of having the built in amp is reference, you can measure and calibrate for the response of that amplifier. Meaning you can make your signal chain up to lets say a speaker transparent and effectively linear.
abraxalito
I can definitely see from a construction and set up point of view why you would want real time feedback, I do however have one question. You say you would look at the "THD residual" do you not mean THD+N ?
Oberlix
I suppose another advantage of having the built in amp is reference, you can measure and calibrate for the response of that amplifier. Meaning you can make your signal chain up to lets say a speaker transparent and effectively linear.
abraxalito
I can definitely see from a construction and set up point of view why you would want real time feedback, I do however have one question. You say you would look at the "THD residual" do you not mean THD+N ?
Oberlix
Check the XMOS web site, there is an excellent browser based but powerful o'scope based on one of their quad core processors. A lot of the work in hooking up analog in and out is covered. The processor is a hybrid micro-DSP so real time FFT's are no problem. I think it's the perfect platform for any DIY instrument. You could use it on anything that can load a web page, totally portable. This is not a first time project.
EDIT - I think these guys are in Wales and the guy that wrote the article is working for them (it also was a big student project).
Last edited:
abraxalito
That's what I thought you meant I was just making sure of the specifics, from what I have been shown THD+N is a very blunt tool and although a you highlight it has it's uses. I think the application of it has to be thought of as otherwise it can give very erroneous results.
Scott, thanks for that I had not come across them in my previous searches, definately got some scouring of there site to do.
Oberlix
That's what I thought you meant I was just making sure of the specifics, from what I have been shown THD+N is a very blunt tool and although a you highlight it has it's uses. I think the application of it has to be thought of as otherwise it can give very erroneous results.
Scott, thanks for that I had not come across them in my previous searches, definately got some scouring of there site to do.
Oberlix
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.