JNeutron nit picks a couple of things on the thread I linked, but people who buy unobtanium by the truck load are allowed some leeway 🙂
Whilst it might not make comfortable reading to some who believe in woo and foo over science and engineering I think he covers most issues (with measurements) very well. Build it the Whitlock way and all is good 🙂 . Oh and I approve of his vanity plate 'MR CMRR'
Whilst it might not make comfortable reading to some who believe in woo and foo over science and engineering I think he covers most issues (with measurements) very well. Build it the Whitlock way and all is good 🙂 . Oh and I approve of his vanity plate 'MR CMRR'
well no static multi-tone test will measure that surely?
Demian: Is there a time sync issue to worry about, or could you, in theory, burn a CD for a 'complete system IMD' test?
Demian: Is there a time sync issue to worry about, or could you, in theory, burn a CD for a 'complete system IMD' test?
This is a simple FFT of a continuous tone set. It won't get issues related to level changes and modulations from different power levels etc. All you need is a good FFT program and a file. However fixing what you find may be beyond what the DIY'er or even a speaker manufacturer can do. Changing cone materials, profiles etc. is not easy and usually becomes quite expensive.
This highlights something that is a puzzle- given the limitations of transducers how to the really subtle "errors" of electronics ever get through to be heard?
This highlights something that is a puzzle- given the limitations of transducers how to the really subtle "errors" of electronics ever get through to be heard?
Nothing stopping a sequence of pulsed duty cycle tests. Thermal effects are long in comparison, so gating shouldn't be onerous.
It could be possible to construct two multitones at the same rms level where one has a normal crest factor (~3) and one has a a very high crest at one point. If the time records are long enough (FFT big enough) the peak could be once a second or less, this might excite a thermal transient.
Would the effect of the transient be visible in a 1 second FFT? Long FFT's on a computer platform are subject to lost samples and it takes only one lost sample to bork the whole measurement.
I can control the phase relationships of the tones but if they are really unrelated harmonically I would guess you will periodically get that peak. The tone set above has an RMS value about 4 dB below it peak value.
I can control the phase relationships of the tones but if they are really unrelated harmonically I would guess you will periodically get that peak. The tone set above has an RMS value about 4 dB below it peak value.
I wrote some routines to craft multitones, it's not a closed form process you have to play around until you get what you want. Basicly I would take a multitone with an expected crest factor of ~3 and find the peak and then take a random tone and rotate its peak (phase) to be at the same point and iterate until the crest factor was what I wanted. I don't know if this does anything for thermal transient analysis.
No, it's a sort of current output. I think the question was just misread. Not sure what op-amp they are using for I/V.
Is there anything to preclude stepping power/pulse duration and integrating the trials vs. a long FFT?
I didn't say plastic molding is a "problem". I would not deny it can be a subjectively better solution for some contexts. I meant to say we hear more difference between molded or non-molded than cables themselves.
I don't really remember how we did, but I still remember when we cut the molded Monster power cables' plug and replaced with non molded one, we are surprised with the difference. We tried this because we knew that molded plastic plug of the line cables can affect the sound a lot based on the experience with the Mogami Bantam cables. Mytec PSU is no good? I don't know, and I don't really care these days. I just randomly use power cables that I happen to own, some are regular ones and some are paranoid version.
Well A-Precision does it very well with their equipment. (Multi-Tone tests). So, it is not like it has unsolvable issues. Its done. I had a virus in my A-P connected computer for a long time..... so was looking for alternative multi-tone test systems... all done as a PnP package. But I fixed my computer and can use the A-prec.
In my case I am not interested in measuring harmonics.
Think of this..... that grass-like, high density distortion products shows a much higher amplitude thanTHD or typical IM tests. For example, if a THD test shows -100dbv, but a multi-tone might show -80dbv at higher freqs.
What happens when that -80 goes into another piece of equipment which does same.... multi-tones plus the previous -80 level of distortion prroducts.
Thru an entire recording system... console, EQ, limiters, mixing, mic preamps...... and same thru the whole play back system.... RIAA, line and power amp. The accumulated total multi-tone of distortion products. -60dB?
And, S/N is computed from max rated output to a no signal in. Multi-tones from A-P will show a dynamic noise level shift to higher noise level. Both of these things occure in real operating condiions which we hear but isn't being tested ... Multi-Tones do a much better job of measuring what we hear.
THx-RNMarsh
In my case I am not interested in measuring harmonics.
Think of this..... that grass-like, high density distortion products shows a much higher amplitude thanTHD or typical IM tests. For example, if a THD test shows -100dbv, but a multi-tone might show -80dbv at higher freqs.
What happens when that -80 goes into another piece of equipment which does same.... multi-tones plus the previous -80 level of distortion prroducts.
Thru an entire recording system... console, EQ, limiters, mixing, mic preamps...... and same thru the whole play back system.... RIAA, line and power amp. The accumulated total multi-tone of distortion products. -60dB?
And, S/N is computed from max rated output to a no signal in. Multi-tones from A-P will show a dynamic noise level shift to higher noise level. Both of these things occure in real operating condiions which we hear but isn't being tested ... Multi-Tones do a much better job of measuring what we hear.
THx-RNMarsh
Last edited:
I guess it would be worse for the studio recording, and better for the 2 mic recorded classical music. People adds distortion intentionally. Do you know that no professional recording engineer uses measurement microphone nor precision equipment for the recording, and the most popular pro audio computer softwares have been distortion plugins? 🙂
For recording and mixing, the amount of the distortion is not that important, what is important is the quality of the distortion.
Last edited:
Richard:
Would you like to see end to end electrical input to mike (acoustic would be really hard although an electrostatic actuator is possible with measuring microphones) to acoustic out on headphones? Measurinf with speakers is a different issue.
There have been multitone tests for a long time. The NTI RT-1M and RT-2M boxes were really only used for that. They came with software that looked in the spaces between the tones. I think I still have one if you want it. I may still have the software even. The AP test was a fast production test as well. They are good production line tests because you can quickly see response errors and obvious distortion issues. Adding more tones actually reduces the SNR because of the peak to RMS and the shared energy. The 12 tones above have each to at about -10dB from the total. 20 tones would cost another -10 dB.
Would you like to see end to end electrical input to mike (acoustic would be really hard although an electrostatic actuator is possible with measuring microphones) to acoustic out on headphones? Measurinf with speakers is a different issue.
There have been multitone tests for a long time. The NTI RT-1M and RT-2M boxes were really only used for that. They came with software that looked in the spaces between the tones. I think I still have one if you want it. I may still have the software even. The AP test was a fast production test as well. They are good production line tests because you can quickly see response errors and obvious distortion issues. Adding more tones actually reduces the SNR because of the peak to RMS and the shared energy. The 12 tones above have each to at about -10dB from the total. 20 tones would cost another -10 dB.
Thanks PMA for your input, but I need to find whether they used an IC, and what IC for the I to V conversion, compared to the OPPO 105. Of course the 28 has improvements over the 18, but are they really that important subjectively? I seem to have a problem getting the 28 designers data sheet. Do they release it to everybody, or just specific engineers?
My concern is that when I worked for OPPO, I found that they built their I to V converter with the LM3562 under less than optimum loading conditions. This is what I told them that I was concerned about, but they ignored my input. Now you, PMA, has shown that the LM3562 may not be the ideal listening amp from your listening test (did I get this right?) so what would be better? I did find that a competing Asian company used the ES9028 with a DISCRETE OUTPUT. They said they were avoiding the IC problem. Now, what is the difference between the OPPO 105 and Benchmark OUTPUTS? What makes the Benchmark better? Is it because there is a gain scaling feature that allows for lighter loading of the IC in order to keep the gain within limits? The 9018 appears to lack this feature. I'm just curious.
My concern is that when I worked for OPPO, I found that they built their I to V converter with the LM3562 under less than optimum loading conditions. This is what I told them that I was concerned about, but they ignored my input. Now you, PMA, has shown that the LM3562 may not be the ideal listening amp from your listening test (did I get this right?) so what would be better? I did find that a competing Asian company used the ES9028 with a DISCRETE OUTPUT. They said they were avoiding the IC problem. Now, what is the difference between the OPPO 105 and Benchmark OUTPUTS? What makes the Benchmark better? Is it because there is a gain scaling feature that allows for lighter loading of the IC in order to keep the gain within limits? The 9018 appears to lack this feature. I'm just curious.
- Status
- Not open for further replies.