Hi soongsc,
Why not PM GedLee to see if he has some insights he can give you in broad strokes? After that, you're really into a reading session to gather the fine details. At least what GedLee might be able to give you is enough for you to decide whether to spend the time researching this or not.
At least (from my understanding), GedLee has given Virtins his paper and maybe even some background materials, plus has an open dialog with them to ensure the added feature / function is an accurate representation of his work. In addition, I would expect that any initial errors that may have crept in will be corrected by Virtins. The application will no doubt be reviewed by GedLee before the release. He may even be using the Virtins applications to further his own work. That wouldn't surprise me in the least.
-Chris
Why not PM GedLee to see if he has some insights he can give you in broad strokes? After that, you're really into a reading session to gather the fine details. At least what GedLee might be able to give you is enough for you to decide whether to spend the time researching this or not.
At least (from my understanding), GedLee has given Virtins his paper and maybe even some background materials, plus has an open dialog with them to ensure the added feature / function is an accurate representation of his work. In addition, I would expect that any initial errors that may have crept in will be corrected by Virtins. The application will no doubt be reviewed by GedLee before the release. He may even be using the Virtins applications to further his own work. That wouldn't surprise me in the least.
-Chris
In general, there are two major aspects of testing in audio.
1. Metrics and data to determine what is more acceptable. This is what a lot of discussion surrounds, and from my understanding, the focus of GedLee metrics.
2. Data that will allow you to back trace a design to a specific part to make a design more ideal. This is the area I am focused on. I think most engineers are going to focus on as well.
Further looking at the data logger, does it record and save at full sampling rate or just the data that meet certain criteria? Can I export the recorded data to external file? Can I use the Signal Generator stimulate a DUT and log the response in the Data Logger?
1. Metrics and data to determine what is more acceptable. This is what a lot of discussion surrounds, and from my understanding, the focus of GedLee metrics.
2. Data that will allow you to back trace a design to a specific part to make a design more ideal. This is the area I am focused on. I think most engineers are going to focus on as well.
Further looking at the data logger, does it record and save at full sampling rate or just the data that meet certain criteria? Can I export the recorded data to external file? Can I use the Signal Generator stimulate a DUT and log the response in the Data Logger?
Last edited:
Hi Chris,
Thanks for your comments. Just to let everyone know, Virtins, Dr. Earl Geddes and I all worked the integration of Earl’s GedLee Metric into MI Pro collectively prior to it being released. Earl has a copy of MI Pro as well so the implementation of Earl’s GedLee Metric has already been vetted by him (Virtins would not have released it without such prior vetting having taken place first).
My posting # 180 on this MI Pro GB thread shows the link to another diyaudio thread where I had emailed Earl asking him to chime in, (and he did). So all anyone needs to do to see Earl’s thoughts on the GedLee Metric and its uses, etc. is to look at my posting # 180 and the URL link to my other diyaudio posting (and there you’ll also see Earl’s postings too). Earl’s comments there are less than 2 weeks old so they reflect his latest thinking on this topic.
All of the AES and other papers on the GedLee Metric are in PDF format and are freely available from Earl’s website per my posting # 178 in this MI Pro GB thread too, i.e., the URL to Earl’s website / papers. It is there one can read and learn more about what exactly the GedLee Metric is.
There shouldn’t be anything in the GedLee Metric’s implementation into MI Pro that needs to be fixed, etc., (that was the whole point of involving Earl directly and vetting the implementation by Virtins prior to the release of MI Pro v3.8.0.1 a little over a month ago).
Hope that explains how the implementation of Earl’s GedLee Metric was implemented within MI Pro.
Have a great day my friend!
V/R,
MI Pro
Exploring in search of the elusive "X Factor" that separates one audio experience from another is an interesting undertaking. I have dabbled in it for years. In pure objective terms I have found no "smoking guns" even with distortion measurement far beyond an AP555X. The electronics are pretty much flawless if used optimally. There are many ways to compromise a system however and some are pretty difficult to identify. A high impedance source driving an input with significant capacitive modulation with level for example. Or shared currents across grounds.
Speakers are less perfect and the Gedlee measurements are well suited to that task.
One aspect that is very difficult to measure is the change over time. Hearing depends a lot on what happened earlier and quantifying that is something most measurement systems have no way of doing.
I suspect the best place to look for insights would be the research behind lossy encoders. For them to work at the level they do someone figured out a lot about hearing.
Speakers are less perfect and the Gedlee measurements are well suited to that task.
One aspect that is very difficult to measure is the change over time. Hearing depends a lot on what happened earlier and quantifying that is something most measurement systems have no way of doing.
I suspect the best place to look for insights would be the research behind lossy encoders. For them to work at the level they do someone figured out a lot about hearing.
The Data Logger add-on software module in MI is not for logging the raw ADC data. It is used to provide long time data logging function for 196 Derived Data Points (DDP) and 16 User Definable Data Points (UDDP). A DDP is a single-valued variable derived from an oscilloscope frame of raw ADC data. Examples of DDPs are Max, Min, Mean and RMS values, SPL values in dB, dBA, dBB or dBC, Peak Frequencies, THD, IMD, etc. The Data Logger logs DDP values with time stamps in TXT log files. The maximum logging rate is the frame rate of the oscilloscope (not the sampling rate).
In MI, the raw ADC data can be saved as a WAV file or TXT file. The analysis results can be exported as a TXT files, or directly copy & paste into EXCEL.
Yes, the response of a DUT can be saved. In Device Test Plan, you can configure a sequence of steps to generate a stimulus to a DUT, acquire and analyze the response, save the response as a WAV file, save the analysis results as a TXT file, or save the graphs as a bitmap. All are configurable.
It seems Mi Pro does not provide the data analysis I may need, but if There is a way I can record and save time history of data at max sample, I can figure out how to do the additional analysis. So how how much time of data can be recorded at maximum sample rate can be recorded? Or is it just limited by disk storage space?
Hello Everyone,
I have received a couple of private messages today concerning the issue of the proper use of Dr. Earl Geddes’ GedLee Metric (Gm distortion) and what are examples of things it can be used to measure. To address this question I refer you to Dr. Geddes’ comments shown in posting # 2306 here: DIY Audio Analyzer with AK5397/AK5394A and AK4490 - Page 231 - diyAudio
Note that in the above posting Earl states that Gm distortion is appropriate for measuring headphones as well as electronics. An example of electronics can be seen in Joey White's posting on Total Harmonic Distortion at this URL: Hephaestus Audio >> Blog Archive >> Total Harmonic Distortion Please note in particular the below portion of Mr. White’s posting:
“…a metric has been proposed that seems to have very good correlation with subjective impression. This is the “GedLee Metric” by Earl Geddes and Lidia Lee of GedLee LLC. Here are the first and second parts of the relevant AES papers. There is a tremendous amount of momentum to overcome for a metric such as this to ever gain traction in the audio industry. Especially resistant will be any manufacturers that stand to have their specifications suffer by it. For example, low THD linear amplifiers that have made use of large amounts of global feedback, with little attention given to the linearity of the open-loop transfer function, may look poor in the light of this new metric.…”
Speaking for myself, I’m far more interested in using both the GedLee Metric and NCD to measure both headphones, in-ear-monitors (IEMs a.k.a. “ear buds”) and loudspeakers than I am electronics. I hope that the above helps to illustrate the types of things the GedLee Metric can be used for.
V/R,
MI Pro
I have received a couple of private messages today concerning the issue of the proper use of Dr. Earl Geddes’ GedLee Metric (Gm distortion) and what are examples of things it can be used to measure. To address this question I refer you to Dr. Geddes’ comments shown in posting # 2306 here: DIY Audio Analyzer with AK5397/AK5394A and AK4490 - Page 231 - diyAudio
Note that in the above posting Earl states that Gm distortion is appropriate for measuring headphones as well as electronics. An example of electronics can be seen in Joey White's posting on Total Harmonic Distortion at this URL: Hephaestus Audio >> Blog Archive >> Total Harmonic Distortion Please note in particular the below portion of Mr. White’s posting:
“…a metric has been proposed that seems to have very good correlation with subjective impression. This is the “GedLee Metric” by Earl Geddes and Lidia Lee of GedLee LLC. Here are the first and second parts of the relevant AES papers. There is a tremendous amount of momentum to overcome for a metric such as this to ever gain traction in the audio industry. Especially resistant will be any manufacturers that stand to have their specifications suffer by it. For example, low THD linear amplifiers that have made use of large amounts of global feedback, with little attention given to the linearity of the open-loop transfer function, may look poor in the light of this new metric.…”
Speaking for myself, I’m far more interested in using both the GedLee Metric and NCD to measure both headphones, in-ear-monitors (IEMs a.k.a. “ear buds”) and loudspeakers than I am electronics. I hope that the above helps to illustrate the types of things the GedLee Metric can be used for.
V/R,
MI Pro
Last edited:
The data acquired in the oscilloscope can be saved as a WAV file via [File]>[Save]. The max. frame width of the oscilloscope is 500s. That means that max. 500-second data can be saved in this manner. "Roll" mode can be used to partially update the screen for a feeling of a faster response when the sampling duration is long. You can also stream data continuously to hard disk by pressing the Record button at the upper right corner of the screen. During the recording process, screen will still be updated in real time. Maximum 2G bytes per file can be recorded. These data can be sampled at the maximum sampling rate allowed by the hardware.
The WAV files saved via MI, while fully compatible with any other wave file players or editors, contain additional information:
(1) ADC Range used during sampling [via "Input Voltage Range" selection in the second toolbar from the top]
(2) Sensor Sensitivity used during sampling (via [Setting]>[Calibration]>[Sensor Sensitivity]
(3) Measurement notes the user enters (via [Setting]>[Notes].
The former two will ensure that, when the saved WAV file is reopened via [File]>[Open], the data will be properly scaled and absolute values in engineering unit will be displayed correctly on screen.
When a sensor is chosen for a certain application, one has to make sure that the sensor’s specifications such as frequency response meet the requirements of that application, so that the measurement error is within the tolerable range even without any compensation or correction.
MI supports predefined and arbitrary FIR, IIR, FFT digital filtering in time domain. They can be applied to the raw ADC data sampled in the oscilloscope before the data are saved into a WAV file. These digital filtering methods can be used to correct the frequency response of the sensor in time domain if necessary.
Please refer to Post #153 Point 2 on compensation in frequency domain ONLY.
In MI, the acquired signal is processed in time domain first and then frequency domain. Therefore any processing performed in time domain will affect the signal’s presentation in frequency domain. But the reverse is not true. If the sensor’s frequency response is already compensated in time domain, then it should not be compensated in frequency domain again.
MI supports predefined and arbitrary FIR, IIR, FFT digital filtering in time domain. They can be applied to the raw ADC data sampled in the oscilloscope before the data are saved into a WAV file. These digital filtering methods can be used to correct the frequency response of the sensor in time domain if necessary.
Please refer to Post #153 Point 2 on compensation in frequency domain ONLY.
In MI, the acquired signal is processed in time domain first and then frequency domain. Therefore any processing performed in time domain will affect the signal’s presentation in frequency domain. But the reverse is not true. If the sensor’s frequency response is already compensated in time domain, then it should not be compensated in frequency domain again.
I am not sure how the "FIR, IIR, FFT digital filtering in time domain" works in MI, and I believe Point 2 in Post #153 was about loopback compensation, which usually does not include the sensor (for instance a microphone used for acoustic tests). I am therefore not quite sure how compensation for "non-flat" sensors works.
Here's what I have in mind: acquire raw time domain signal from the sensor --> FFT signal to frequency domain --> deconvolve signal from sensor transfer function in frequency domain to compensate the test signal for the transfer function for the sensor --> iFFT back to time domain signal --> save the compensated signal.
Can MI be set up to do the compensation for the sensor transfer function automatically?
Here's what I have in mind: acquire raw time domain signal from the sensor --> FFT signal to frequency domain --> deconvolve signal from sensor transfer function in frequency domain to compensate the test signal for the transfer function for the sensor --> iFFT back to time domain signal --> save the compensated signal.
Can MI be set up to do the compensation for the sensor transfer function automatically?
1- anatech
2- kevinkr
3- lemon
4- PH104
5- douggoldberg
6- bk856er
7- mlloyd1
8- tjencks
9- agdr
10- waltzingbear
11- Dennis Hui
12- jwilhelm
13- 1audio (Demian)
14- soongsc
15- jhenderson01075
16- jazid
17- lbud4
18- diyaudnut
19- SilentGrace
20- acg
21- Ghianni
22- etlasr
23- B&W_arthur
24- markus22ch
25- niner
2- kevinkr
3- lemon
4- PH104
5- douggoldberg
6- bk856er
7- mlloyd1
8- tjencks
9- agdr
10- waltzingbear
11- Dennis Hui
12- jwilhelm
13- 1audio (Demian)
14- soongsc
15- jhenderson01075
16- jazid
17- lbud4
18- diyaudnut
19- SilentGrace
20- acg
21- Ghianni
22- etlasr
23- B&W_arthur
24- markus22ch
25- niner
Frequency compensation in frequency domain (Spectrum Analyzer) in MI (refer to Points 2 & 3 in Post #153) can be used to compensate the nonflatness of the measuring device in magnitude frequency response measurement, and the gain & phase disparity between the two input channels of the measuring device in Bode Plot measurement. This includes the compensation for sensors such as microphones.
However, you are right in the sense that the above compensations do not applied to the time domain signal in MI. If you want to compensate the sensor transfer function in time domain in MI, you will have to apply a digital filter (FIR, IIR, or FFT) in time domain (Oscilloscope).
The FFT digital filter in time domain in MI modifies/compensates the magnitude frequency response only and introduces no phase shift (i.e. zero-phase). This method is similar to the method you suggested. But it does not compensate phase response. A zero-phase digital filter has its advantage in some applications, for example, when you want to filter out the fundamental and compare the phase of the residual harmonics to that of the fundamental.
The FIR digital filter in MI modifies/compensates the magnitude frequency response only and introduces a linear phase shift (a delay of [FIR order]/2/[Sampling Rate] seconds, the FIR order is always an even number in MI).
The IIR digital filter in MI directly uses whatever IIR coefficients the user enters through a text-based IIR coefficient file (*.IIR). You can use other FIR/IIR design software to design an IIR filter (including linear-phase or non-linear phase FIR) that meets your requirements and use it in MI.
Therefore, if you only want to compensate the magnitude frequency response of the sensor, the above FFT or FIR method can be used. If you want to compensate the phase frequency response of the sensors as well, then the above IIR method can be used.
Can someone summarize the pricing options, at different points in time the pricing seemed to have altered. I am looking at the document at the beginning of the thread and it says no discount if the quantity is below 42. So I am confused. Also do the discounts apply to the add on modules as well?
Hello soongsc,
If you go to the 1st posting in this MI Pro GB thread you’ll see under the “Bottom Line Up Front” (BLUF) area two URL links. The two URL links are *very* close to the BLUF paragraph (just under it) for I’m trying to make it very difficult to miss them.
The 1st URL link points you to posting # 74. That means for you, as # 14 on the MI Pro GB list, you qualify for a $30 USD discount price which means for MI Pro without any additional add-on modules, you get the USB dongle shipped to you for $209.90 USD, (again, this is clearly explained on posting # 74).
The 2nd URL link points you to posting # 143. This really doesn’t apply to you since you have signed up for the MI Pro GB and you’re # 14 on this list of people whom have signed up for it. Posting # 143 simply states that if you don’t participate in the MI Pro GB via this promotion and on 1 Aug 2018 or thereafter you decide you want to buy MI Pro, it would cost you for the same thing ~$240 USD with the USB dongle plus ~$100 USD more for the custom API add-on that allows MI Pro to remote control the RTX6001 to achieve autoranging and autoscaling functionality, (total ~$340 USD).
soongsc, you’ll be paying $209.90 USD for MI Pro with the USB dongle since you signed up for the MI Pro GP already, (assuming you don’t order also any MI add-on modules or the MI “full package”).
The discounts for MI add-ons are 10% off during this MI Pro GB period, i.e., normally ~$100 USD per add-on but with 10% MI Pro GB discount on add-ons they cost now ~$90 USD).
Finally, the MI “full package” cost now if one signs up for it as part of this MI Pro GP cost $449.95 USD but if one waits to buy it or on 1 Aug 2018 onward the same “full package” will cost ~$540 USD. The MI “full package” includes MI Pro + all the MI add-ons too (to include the RTX6001 API add-on as well).
While I can modify the 1st posting of this MI Pro GB still, I am not able to replace / update the Adobe PDF file, ergo the rational for the URL links under the BLUF paragraph.
All future updates are 100% free of charge for the same type of MI license you purchase, e.g., if you buy MI Pro alone you’ll get all future MI Pro updates for free, etc. Likewise, if you buy the MI “full package” you’ll get 100% of all future MI updates free of charge for life.
V/R,
MI Pro
If you go to the 1st posting in this MI Pro GB thread you’ll see under the “Bottom Line Up Front” (BLUF) area two URL links. The two URL links are *very* close to the BLUF paragraph (just under it) for I’m trying to make it very difficult to miss them.
The 1st URL link points you to posting # 74. That means for you, as # 14 on the MI Pro GB list, you qualify for a $30 USD discount price which means for MI Pro without any additional add-on modules, you get the USB dongle shipped to you for $209.90 USD, (again, this is clearly explained on posting # 74).
The 2nd URL link points you to posting # 143. This really doesn’t apply to you since you have signed up for the MI Pro GB and you’re # 14 on this list of people whom have signed up for it. Posting # 143 simply states that if you don’t participate in the MI Pro GB via this promotion and on 1 Aug 2018 or thereafter you decide you want to buy MI Pro, it would cost you for the same thing ~$240 USD with the USB dongle plus ~$100 USD more for the custom API add-on that allows MI Pro to remote control the RTX6001 to achieve autoranging and autoscaling functionality, (total ~$340 USD).
soongsc, you’ll be paying $209.90 USD for MI Pro with the USB dongle since you signed up for the MI Pro GP already, (assuming you don’t order also any MI add-on modules or the MI “full package”).
The discounts for MI add-ons are 10% off during this MI Pro GB period, i.e., normally ~$100 USD per add-on but with 10% MI Pro GB discount on add-ons they cost now ~$90 USD).
Finally, the MI “full package” cost now if one signs up for it as part of this MI Pro GP cost $449.95 USD but if one waits to buy it or on 1 Aug 2018 onward the same “full package” will cost ~$540 USD. The MI “full package” includes MI Pro + all the MI add-ons too (to include the RTX6001 API add-on as well).
While I can modify the 1st posting of this MI Pro GB still, I am not able to replace / update the Adobe PDF file, ergo the rational for the URL links under the BLUF paragraph.
All future updates are 100% free of charge for the same type of MI license you purchase, e.g., if you buy MI Pro alone you’ll get all future MI Pro updates for free, etc. Likewise, if you buy the MI “full package” you’ll get 100% of all future MI updates free of charge for life.
V/R,
MI Pro
Last edited:
Assume a typical loudspeaker scenario with a measurement microphone that does not have perfectly flat frequency response. The user wants to measure the impulse response of the speaker, and the data should be compensated for the non-ideal transfer function of the microphone.
Compensating the frequency response magnitude only will screw up the time-domain signal due to the missing phase compensation. From your description it seems the user would have to determine the IIR coefficients that correspond to the micorphone frequency response. How does that work? Wouldn't it be much easier for the user if the software would deconvolve the raw test signal from the microphone transfer function?
It would be less confusing if we did not have to go through various posts to piece things together. There is also further discounts beyond 42 which I think is in one of the PDF file, does this apply?
Hello soongsc,
In my posting # 196 I stated:
“…While I can modify the 1st posting of this MI Pro GB still, I am not able to replace / update the Adobe PDF file, ergo the rational for the URL links under the BLUF paragraph…”
So yes, it would be less confusing if I didn’t have to have URL links to two different postings, but I’m not allowed by the diyaudio forum to upload a newer / revised PDF file as per the above statement I made.
Likewise, in my posting # 74 that I just cited to you in reply to your question it states:
Let’s address some other questions that may pop up. For example, what happens if the number of MI Pro GB licenses sold ends up dropping the MI Pro per license cost below even the above cited incentivized model discounts?
For example, if one looks at my MI Pro GB posting # 1 @ GB for Virtins MI Pro for RTX6001 autoranging/autoscaling & for soundcard end users - diyAudio and downloads the PDF document in that posting and looks at paragraph # 5 of the which is a table showing the MI Pro GB pricing model, if in the event the pricing on a per MI Pro GB license goes down due to the increase in sales of MI Pro per this table, than the given MI Pro GB purchaser will get the lower of the two prices (in the favor of the purchaser thereby not putting them at risk of paying more and saving them $$$).
To illustrate the above paragraph’s point, let’s say for hypothetical purposes, that by the end of July 2018 we have 112 people whom have signed up for the MI Pro GB thereby putting the price per MI Pro GB license at $190 USD per license. That means each MI Pro GB buyer will end up paying only $190 USD per MI Pro license, regardless of when they signed up for the MI Pro GB, i.e., with the USB dongle included and free express registered worldwide mail shipping.
So the answer to the two questions you raise has already been provided to you as per the quoted comments I have previously made.
V/R,
MI Pro
Apologies if this has already been answered - the thread is quite long now!
Surely, we only need to sign up to the base MI Pro GB now? We can decide what add-ons to order when Virtins open the GB sub site, and order them then. So why do I need to decide that now and bung it on the email...?
Thanks!
Surely, we only need to sign up to the base MI Pro GB now? We can decide what add-ons to order when Virtins open the GB sub site, and order them then. So why do I need to decide that now and bung it on the email...?
Thanks!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.