Matrojig, a measured jig with everything built-in

[xmag]

I have developed a jig (called Matrojig) that I perhaps to interest one or the other. I have a complete wiring diagram and a diagram of how to pull all the cables etc. Unfortunately I do not have time to post pictures of the project but you can get links to it. Everything is in Swedish but there is a translation function on the website that can be seen at the top right. At the bottom of the link are new chapters of construction, function, calibration and a little about how to measure:

UPDATED:
New address, brand new Matrojig

https://www.exmag.se/250109-matrojig-s-hogtalarmatning/

https://www.exmag.se/

Special box to keep everything in one place:

Old Matrojig;
Data and other information in the thread applies to the older version below which is no longer used.

Unfortunately I can not answer questions in detail because my English is not what it should be.

 

[Antonio Di Motta]

Is an interesting project for beginners who want to try their hand at the world of audio measurement. Unfortunately, the components used are basic and do not perform well.
I have created a similar tool that also includes the processing part see the link :

https://www.avnirvana.com/threads/make-a-good-measurement-system-with-rew-and-raspberry.5592/

This measuring instrument is currently undergoing a major overhaul and I have made a dedicated high performance sound card which is currently in its third version, furthermore the audio amplifier is based on the lm3886 integrated with much better performance than the tda2003 and dual power supply + / - 20Vdc.

Anyway congratulations because despite everything it is a good start ...

Greetings
Antonio

 

[xmag]

Is an interesting project for beginners who want to try their hand at the world of audio measurement. Unfortunately, the components used are basic and do not perform well.
I have created a similar tool that also includes the processing part see the link :

https://www.avnirvana.com/threads/make-a-good-measurement-system-with-rew-and-raspberry.5592/

This measuring instrument is currently undergoing a major overhaul and I have made a dedicated high performance sound card which is currently in its third version, furthermore the audio amplifier is based on the lm3886 integrated with much better performance than the tda2003 and dual power supply + / - 20Vdc.

Anyway congratulations because despite everything it is a good start ...

Greetings
Antonio

I have studied your project! Good work. My project focuses on being simple and cheap but still good. The components are not as bad as you might think. Matrojig can measure himself through REW. Not so much an idea to lay out straight curves on frequency response and phase etc. However, the system's own distortion can be laid out:

Brus = Nois..

The peak at 30hz is something where I live that disturbs.



Has worked for a long time to reduce distortion etc. to low levels. This is possible if you balance the start levels in the sound card and the amplifier. For some time to come, I will produce a circuit board and a slightly stronger amplifier.

 

[xmag]

Back:

The washer and screw cover a hole I had a component in which I then removed. The pictures are on the prototype. The potentiometer is for adjusting the level of the Reference Channel in relation to the signal input.

 

[trobbins]

The impedance measurement configuration seems prone to changes from input balance pot setting, and that the short-circuit and reference resistances do not include the measurement leads to the speaker (ie. the loopback occurs in your box and via switch contacts) and possibly does not include the sense resistor?.

 

[xmag]

The impedance measurement configuration seems prone to changes from input balance pot setting, and that the short-circuit and reference resistances do not include the measurement leads to the speaker (ie. the loopback occurs in your box and via switch contacts) and possibly does not include the sense resistor?.

Everything works great!
Reference resistor calibration must connect an accurately measured resistor around 10 ohms at least 5 watts on the speaker connector.
Input balance pot setting goes back to "Line In R" before the impedance measurement. That potentiometer must not be moved more than + - 1dB, for adjustment of the same level in impedance measurement, after the calibration.
It is the person who builds and owns Matrojig who decides whether cables to measuring objects should be included in the calibration (I hope I understood you correctly).

If something is wrong, it's good for me to know so I can fix it quickly.

 

[xmag]

Missed this: "include the sense resistor?."

Well, sense resistor is included in the impedance measurement. Without it, the impedance measurements would be completely wrong. My measurements of known elements show that it works as it should.

Here is a stripped image (potentiometer, resistor and switch removed) on impedance measurement setting :

 

[Antonio Di Motta]

I have studied your project! Good work. My project focuses on being simple and cheap but still good. The components are not as bad as you might think. Matrojig can measure himself through REW. Not so much an idea to lay out straight curves on frequency response and phase etc. However, the system's own distortion can be laid out:
View attachment 49871



Brus = Nois..
View attachment 49872


The peak at 30hz is something where I live that disturbs.



Has worked for a long time to reduce distortion etc. to low levels. This is possible if you balance the start levels in the sound card and the amplifier. For some time to come, I will produce a circuit board and a slightly stronger amplifier.

From my point of view, to check if your system works correctly you should at least post a graph generating with the REW generator a fundamental level (1Khz) set at 2.83Veff equal to 1W on 8ohm and check the harmonics produced.

Your graph does not display the fundamental level and therefore only shows the system noise without being able to calculate the THD. As for the impedance, I believe that the sense resistance is too high for use with a power amplifier, in my project, I use a resistance between 47ohm and 33ohm.

Also you should post at least three impedance graphs detected on an accuracy resistor of 10 ohm, 100ohm and 300ohm at 0.1% to verify the accuracy of your system.

Greetings
Antonio

 

[xmag]

From my point of view, to check if your system works correctly you should at least post a graph generating with the REW generator a fundamental level (1Khz) set at 2.83Veff equal to 1W on 8ohm and check the harmonics produced.

Your graph does not display the fundamental level and therefore only shows the system noise without being able to calculate the THD. As for the impedance, I believe that the sense resistance is too high for use with a power amplifier, in my project, I use a resistance between 47ohm and 33ohm.

Also you should post at least three impedance graphs detected on an accuracy resistor of 10 ohm, 100ohm and 300ohm at 0.1% to verify the accuracy of your system.

Greetings
Antonio

Sorry, I did not write that the picture is an old picture I took to show the principle. I use 10 ohms for R-sense. I have measured noise at several different levels. The graph I posted above is measured at -30dB (the level I use to measure in the near field). If I lower more than -35dB, the noise does not drop more, then I am down at the noise floor. I have worked a lot to bring down the noise and THD. In the beginning I had a noise of about 0.05% with the same level as in the graph above. I added a filter to the power supply and lowered the level into the final stage and got significantly lower noise and THD. I only use RTA to measure distortion and noise. SPL can not measure this with because it has a rising noise floor and is not accurate. https://www.avnirvana.com/threads/r...s-what-interface-do-you-like.9471/#post-73717

I have measured the impedance many times with all possible values of resistance I measure. The measurement result is always exactly below a few hundred ohms.

In any case, it is fantastically easy to use and measure.

 

[trobbins]

Your explanatory and setup documentation is quite extensive. That advantage can be sometimes glossed over in pursuit of a quick check of some aspect - I have sometimes glossed over REW help and other notes only to be pointed to the relevant doc section by John M.

It's great to see some customised setup interfaces coming out for particular applications. I made a custom interface rig for my impedance measurements (which are not typically speaker related) - it is very simple and just uses the soundcard's headphone output and standard 2 inputs, so the interface is just passive, but has short shielded leads/plugs to suit insertion in to the three front panel ports of my soundcard, and has the sense resistor and connections within a metal box to assist shielding and avoid flexing/displacement of parts and connections for measurement consistency.

 

[xmag]

I did a jig for Speaker Workshop a few years ago. It has an even more comprehensive construction description and I called it Mätro. Then I was away from the hobby for about 10 years and when I returned I discovered REW. I immediately built a testjig for measuring REW. I received requests to build for others as well, so I developed Matrojig. The difference then and now is that many more people wrote on hifi forums and similar forums. Today, the response to, for example, a new stuff is very calm. When I showed my jig Mätro about 15 years ago, it was a completely different speed. Seems like we have entered a time of forum death.

As for Matrojig and that there are some doubts about the measurement (see above) I will show more measurements when I get time.

I also used to make microphones. There are some descriptions on my website about it.

https://www.xmag.se/mm-1-rca/
https://www.xmag.se/mm-1-xlr/

 

[Antonio Di Motta]

Sorry, I did not write that the picture is an old picture I took to show the principle. I use 10 ohms for R-sense. I have measured noise at several different levels. The graph I posted above is measured at -30dB (the level I use to measure in the near field). If I lower more than -35dB, the noise does not drop more, then I am down at the noise floor. I have worked a lot to bring down the noise and THD. In the beginning I had a noise of about 0.05% with the same level as in the graph above. I added a filter to the power supply and lowered the level into the final stage and got significantly lower noise and THD. I only use RTA to measure distortion and noise. SPL can not measure this with because it has a rising noise floor and is not accurate. https://www.avnirvana.com/threads/r...s-what-interface-do-you-like.9471/#post-73717

I have measured the impedance many times with all possible values of resistance I measure. The measurement result is always exactly below a few hundred ohms.

In any case, it is fantastically easy to use and measure.

I agree with you that the system works, but I really don't think it has the performance it claims in terms of distortion when the TDA2003 datasheet reports at least twice the value.
Then the distortion introduced by the sound card must also be added to the distortion of the amplifier, so I think it is not very true. However, I confirm that the system you created works and is simple, I do not think it can be considered a measuring instrument, especially for the medium-low quality of the individual components.

Greetings

Antonio

 

[xmag]

I agree with you that the system works, but I really don't think it has the performance it claims in terms of distortion when the TDA2003 datasheet reports at least twice the value.
Then the distortion introduced by the sound card must also be added to the distortion of the amplifier, so I think it is not very true. However, I confirm that the system you created works and is simple, I do not think it can be considered a measuring instrument, especially for the medium-low quality of the individual components.

Greetings

Antonio

I have only carried out my measurements and I have posted the results. I do not know if REW results are less true but I think it measures exactly what it sees. But I say it again that I measured with a very low output signal and I will post for measurements later with other levels. I have several times measured circuits that are significantly better than data sites say and vice versa. The "low quality" I use I think you mean TDA2003. I'm planning a new jig that will contain a stronger amplifier, but it may take a while before I finish it.

 

[trobbins]

Antonio, I'd suggest your comments about impedance measurement performance are somewhat misguided. The REW calibration process takes in to account the actual resistor and gain values of the test setup (if a buffer amp is used or not), and measurement accuracy is primarily determined by the accuracy of the reference resistor value used (to its actual value) and any variations in parts (such as temp coefficients of of external jig parts and any gain setting parts internal to a soundcard - both from the output and the two required input channels). The REW measurement is also based on tracking a sinewave, like a lock-in amplifier type measurement system, so harmonic distortions introduced by the soundcard output and input channels, and any external buffer/gain amps, and the DUT itself are primarily rejected.

I'd suggest the main second-order measurement tolerance effect is related to reconnecting any interface module and to probing the DUT. REW calibration recommends doing short-circuit and reference and open circuit measurements using the target benchtop connection setup (eg. flexible leads to crocodile clips or whatever). There may also be measurement noise that can be seen for frequencies higher than circa 50kHz and less than a few Hz, which are likely caused by frequency response roll-off of the soundcard output and input channels and any external amp (as REW assumes the reference resistor has zero magnitude and phase shift across the measurement frequency range).

Imho, doing a repeat measurement on a known test resistor - both before and after any DUT measurements - is a valid verification that is worthwhile, as it confirms the frequency response is not polluted, or the absolute resistance value has not varied.

Wrt speaker impedance measurements, I'd be far more concerned about any influence that speaker resonance build-up of deflection, and changing room modes have, given the time that REW takes to sweep through portions of frequency that may be affected by such modes.

 

[Antonio Di Motta]

Sorry to contradict you, but my observations in terms of ADC and DAC converter performance are by no means misleading.
Specifically, relative to the impedance measurement, the noise level of the system assumes considerable importance, intended as the sum of the entire DAC + Analog Amplifier and Buffer + ADC chain.
Noise negatively affects the measurement of very low impedance values, vice versa the balance between your ADC channels together with the input impedance value of the input buffer alters the measurement of very high impedance values. (it is not known that the noise is only at high frequencies, there is also the noise at low frequencies below 100Hz typically introduced by power supply problems)
Precisely for this reason the latest version of REW has introduced a new calibration procedure that minimizes the defects of sound cards, the fact remains that the more performing the sound card, the better the final result. I collaborated with John by carrying out a series of measurements in order to validate the current calibration procedure.
For the other measures, frequency response, RTA, distortion measures it is very important the linearity of the sound card the number of conversion bits the sampling frequency and also the distortion level. In short, it seems completely obvious to me that the performance of a measuring instrument is dependent on the quality of the individual modules that compose it. On one thing I can agree with you, specifically the impedance measurement, the distortion levels are not that relevant to the accuracy of the measurement.

Then everyone is free to think what they want, the world is beautiful because it is varied !!!

Greetings

Antonio

 

[xmag]

Sorry to contradict you, but my observations in terms of ADC and DAC converter performance are by no means misleading.
Specifically, relative to the impedance measurement, the noise level of the system assumes considerable importance, intended as the sum of the entire DAC + Analog Amplifier and Buffer + ADC chain.
Noise negatively affects the measurement of very low impedance values, vice versa the balance between your ADC channels together with the input impedance value of the input buffer alters the measurement of very high impedance values. (it is not known that the noise is only at high frequencies, there is also the noise at low frequencies below 100Hz typically introduced by power supply problems)
Precisely for this reason the latest version of REW has introduced a new calibration procedure that minimizes the defects of sound cards, the fact remains that the more performing the sound card, the better the final result. I collaborated with John by carrying out a series of measurements in order to validate the current calibration procedure.
For the other measures, frequency response, RTA, distortion measures it is very important the linearity of the sound card the number of conversion bits the sampling frequency and also the distortion level. In short, it seems completely obvious to me that the performance of a measuring instrument is dependent on the quality of the individual modules that compose it. On one thing I can agree with you, specifically the impedance measurement, the distortion levels are not that relevant to the accuracy of the measurement.

Then everyone is free to think what they want, the world is beautiful because it is varied !!!

Greetings

Antonio

For my personal part, no one needs to be sad. The reason is that it's not me who designs the measurement graphs, it's my jig and they look like they do no matter what anny think, it's just that.

Must add one thing, I measure without load. With load, the graphs may have looked in other ways.

 

[trobbins]

Antonio, whilst the OP has used a soundcard that 'is not the greatest', it appears to me that the jig is effectively soundcard agnostic, so the aspect of ADC and DAC performance is not a direct reflection of the capability of the jig to 'perform'. People will continue to make the most out of the soundcards they have, and many have made attempts to improve on the soundcards themselves (including Behringer's range) - my particular card is the EMU0404 and it has one input channel with known noise issues, but that hasn't stopped me from making excellent use of the soundcard for impedance measurement. So suggesting a jig is just for entry level and can't perform well is a bit rich imho - perhaps it would have been more useful to indicate that the soundcard may end up being the limiting aspect to performance.

It also appears that you bring some past baggage in to your comments with respect to earlier versions of REW's impedance measurement performance - again, that is not a direct reflection of the capability of the jig.

Referring to REW's impedance measurement accuracy - that is a many facetted topic and often quite linked to the particular parameter target being measured and the particular part that is being tested. There appear to be so many applications for this tool, and whether a particular soundcard or jig makes any difference probably needs a better awareness of the wide range of applications that could be assessed. John was able to provide a frequency sweep capability down to 0.1Hz for me when I was exploring the inductive behaviour of valve amplifier output transformer primary winding impedance as it was insightful to see just below the pre-existing 2Hz lower limit. John has just indicated he should be able to extend the displayed parameter resolution from 3 to 4 digits, which sort of enters the world of 'accuracy' given that the pre-existing circa 1% resolution may be viewed as for 'beginners' (sic) who want to try their hand at the world of parameter measurement. Another aspect is that a frequency sweep of impedance may well allow irregularities at a particular frequency (like at mains harmonics) to be 'filtered by eye' and to have no consequence at all (depending on the application) for the user - I certainly use that eye filtering capability when using the 'poor' input channel on my EMU0404, even though some find that such irregularities must be exorcised from a measurement system before they can go public with a spectrum plot result.

 

[xmag]

I promised more measurements.
Measurement line: REW - SOUND CARD - POWER AMPLIFIER - SOUND CARD - REW

Red = Noise
Peak 30Hz = Noise disturbance in my home
Peak 50Hz = The plastic box does not fully protect against net noise

-5dB = is close to the maximum level of the power amplifier.

Normal measurement is made at this level

Normal measurement is made at this level

Near field measurement (approx. 10mm) is made with approximately this level

 

[xmag]

Spectrum:


No signal in at all l:

Test tone = 1kHz:

Measured with too high a level to show what it looks like. It can handle -5dB but not -4.5dB. At -5dB there is a sharp limit!

 

[xmag]

 

[xmag]

About the measurements:
Not bad job of cheap stuff

 

[xmag]

Some test measurements of a 5 inch element:

Frequency curve with impedance:

Impedance and phase:

RTA measured distortion for with all parts included.

Waterfall with burst decay


The text "50c" at the top of the pictures means that I paid 50 cents for the speaker element. Bought 40 pieces ...

 

[Antonio Di Motta]

I enjoyed making a direct comparison between the distortion measurements between the Matrojig system and the VT board sound card. I took the chart of the Matrojig system at -5dbfs compared with the Vt board at -3dbfs.

Matrojig system

VT Audio Board ver. 2.0

 

[Antonio Di Motta]

with this I certainly do not want to despise the Matrojig system in the slightest, but it was only to make a comparison. I avoided putting the Matrojig graph at -4.48dbfs because it was already in saturation ...
From the two graphs you can see very well the difference in level of the harmonics with respect to the fundamental.

Greetings

Antonio

 

[xmag]

with this I certainly do not want to despise the Matrojig system in the slightest, but it was only to make a comparison. I avoided putting the Matrojig graph at -4.48dbfs because it was already in saturation ...
From the two graphs you can see very well the difference in level of the harmonics with respect to the fundamental.

Greetings

Antonio

Hi Antonio!
I just want to show that with very low costs you can build a jig that can measure almost all speakers. I have no problem with there being even better stuff. I also do not think that those who want to build a Matrojig have such problems, I think they just want to be able to measure their speaker at home.
Your unit of measurement seems to be very good and very advanced. It may deter some from building it but then they can build the Matrojig instead.