Issues with Emu tracker and THD measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
fotios said:


If you remember, in one post i reffer that Virtins (please don't take it as advertisement, i fear maybe missunderstand me all of you in the end) recently presented a cheap and nice USB interface such as a DSO of 8bit at 100MSa/s. Because its price was only 170 euro, i asked the company for its compatibility as well with the FFT of MI in which included also a scope. The answeer that i got was, that it is better for FFT analysis the use of a proffesional sound card. The reason is the dynamic range according to formula: DR = 20*log (bit depth in decimal / 1). For example, an ADC with 8bit res = 1byte = FF hex = 256 Dec... according to the above formula a 8bit ADC has DR = 20*log(256/1) = 48dB. For a 16bit ADC accordingly the DR = 96 dB and for a 24bit ADC the DR = 144dB.
The sales conductor said me, that an 8bit independently if its sampling rate it is 100MSa/s or 2GSa/s, due to its limited dynamic range can't gives reliable results in THD, THD+Noise, SNR, IMD, NL measurements.
I have checked just now that your HP54522A has an ADC of 8 to 10bit resolution. Maybe is there some secret of your nice device?
Also i have checked the Pico Scope site, and they reffer the same as Virtins, that their better device for FFT analysis was the 216 of 16bit (800USD).
Really i am confused after your last post. What can i do? To by or not to by this ADC of Virtins?

Regards
Fotios

You are correct. The shorthand for converting number of bits to dynamic range is to multiply 6dB x # bits. This is the maximum amount possible, as no converter actually produces the FULL theoretical capability. There are problems with non-linearity, distortion, and noise. The actually capability is called "effective bits", so it's common for an 8 bit scope to have 7.1 effective bits, and so on.
There is no trick in the 54522A scope, but it is useful for looking at noise from digital electronics and correlating it to the source. You are limited to the 8 bit dynamic range, but you can see much higher in frequency and it's useful to see if there are FFT peaks that correlate to known clocks or sample rates far above audio frequencies. It's really more suited to people that build their own A/D's or D/A's. It's a full featured scope with GPIB that sold for $8K only 10 years ago.
So for every job the right tool. Unless you have a use for a 40MHz BW 100MS digital scope, don't buy it. It won't help you measure distortion of audio projects. It may help you look for oscillation and so on for your analog projects, but if that's all you need, I'd prefer to use a nice 100MHz analog scope instead. I only mentioned the 54522A to highlight the FFT capability. You can see a DC-500MHz FFT spectrum in real time, which is pretty cool for that kinda price.
Bob
 
After the latest reports, we can result in some basic conclusions:
1. The bit depth it is the important thing and no so much the sampling rate. For true measurements, a depth of 16bit is the least needed as can gives a DR of 96dB. Of course there are cases, such as a S/N ratio C-weighted measurement which in most audio devices can exceed the 100dB, therefore only under a depth of 24bit--> DR=144dB can be executed any measurement.
A sampling rate of 48KSa/s it is more than enough for true results. Also an averaging of 10 frames it is also enough, as well a FFT size of 32768pts. Of course the increase of the last three, can gives by far better results due to random events occured during measurement but... are these true? OK, OK, we can do the trick to cause impression in people :D
2. There is not offered - at least from my searching - on the market any ADC/DAC usb interface (yet the very expensive NI DAQs) with depth above 16bit. I don't have any information about proffesional grade devices such as Audio Precision or Sound Technology what kind of ADC/DACs those include. Maybe the same that are found in the proffesional sound cards... and the only difference in their reliability is found at their more precise analog I/O buffers than sound cards. If i had the time, sincerely i would like to modify the I/O circuitry of my EMU, but inside there is a real confusion of interconections. Very difficult thing.
3. By general confesion, the ADC section of EMU has excelent performance. I bet that is comparable with any Audio Precision etc. The problem is found in output - either in the DAC or in the analog output interface, probably the second - and it is the unacceptable noise floor. Then, for accurate measurements we lost the very good and flexible signal generators included in the meas. software. Then the only remains, it is the use of an external benchtop signal generator. But yet in this case, as Bob mentioned very correctly above, we have the problem of synchronization i.e. the trigger level. I think that can be resolved from the settings on meas. software. There are enough parameters for callibration, like edge, level, delay etc. With some experiments i believe we can find the correct settings for synchro. I think that MI does it by alone if the trigger placed in auto mode. Acording, i checked again (but this time with correct settings) my Hameg HM8130 function generator. See the results in the picture bellow

An externally hosted image should be here but it was not working when we last tested it.


Of course, i have violate my previous statement and i placed the sampling rate at 192K instead 48K :D Also i placed the level in -1dBfs (90%) this time to get the most possible bigger S/N ratio. Also i used A-weighting this time. The rest are the same as my previous measurement.
I think that, although it is a function generator and not clear audio generator, the HM8130 has enough good performance after all the above tricks.
THD A- weighted = 0,061%
NL = -81,7dBV
THD+Noise A-weighted = 0,063%
SNR = 77,5dB
For some measurements, a conjuction of EMU as ADC interface and the HM8130 as generator can gives good results.
Moreover who is trusting the accuracy of results given from an Audio Precision? From a respect to its brand? SURE! :D :D :D Everything it is relative and not absolute in our world. ;)
Greetings and no waste of money. In these days it exists a real economic chaos.
Fotios
 
syn08 said:


Nope. EMU 1212M uses the CS4398. Second best today after the PCM1704, not counting (yet) ESS Sabre.

Correctly. In all EMU models used ADCs of AKM and DACs of Cirrus Logic, except of EMU0404 where is used DAC of AKM.
Also in EMU 1212M PCI included the new AK5394A of AKM instead the previous AK5385A used in EMU 0404. If you are interested, you can download the datasheets of ADC and DAC used in your device directly from the sites of Asahi-Kasei (AKM) and Cirrus Logic, to make your comparissons.
Comparing the EMU0404 (which i bought before one year) with EMU 1212M, the second includes the AK5394A which is the dirrect improved model of AK5385A. Instead the DACs are of different constructor. By a rough estimation, the differences are found mainly in THD+N and S/N. These are improved between 7 to 9 dB in 1212M.
To be sincere, i don't know how much improvement can offer those, but i have the sense that will be a waste of money an upgrade without signifficant improvement e.g. i bought the EMU0404 for 190 euros and to sell it i can't expect more than 100 euros. The same time the 1212M cost is 150 euros. Thus in the best case, my money loss will be: 90 euro + 60 euro = 150 euro (the whole cost of EMU1212M), thus, NO THANKS. EMU products as mentioned by Bob, are addressed for music processing and not as lab measurement interface. For measurements, the USB interfaces of e.g. NI DAQ, Pico etc. although have depth from 12 to 16 bits as much, are most accurate. But their cost it is 3 to 5 times bigger than sound cards. There is also a drawback in most sound card based FFT softwares, which is that can only recognize audio codec devices due to Windows OS. A significant reason why i bought the MI was the promise of company that very soon they will added drivers for lab equipements. Indeed, after one year in the last version of MI included drivers for two NI DAQs as well ASIO compatibility for our sound cards. Maybe in few months, our tortures will end-up? I don't know.

Fotios
 
syn08 said:


- Linear Technology AN43 application note.
- Linear Technology AN67 application note.
- Tektronix SG505 schematic http://www.diyaudio.com/forums/showthread.php?postid=502295#post502295
- State variable oscillator http://www.uwm.edu/~msw/AudioOsc/index.html

If you don't decide to buy a second hand SG505, I would build the last one in the list.


Those are great links, than you. In the link thread above referring to the SG505, one person describes building a band pass filter. Here is a link to a site that makes this design very easy.
http://sound.westhost.com/project63.htm

You could build one of these around an AD-797, and then switch the caps to cover, say, 5 common test frequencies. I'm thinking about doing a little board for this. The bandpass filter could be used to further filter the existing sound card output, or greatly improve a more common standalone source with, say, .01% distortion. Depending on where the worst harmonics are, (not 2nd) you could easily improve .01% to maybe in the range of .001% or even .0007 %.
And it would certainly clean up the noise on the soundcard DAC, to not make it the limiting function in the noise floor.
Building that would be a lot easier than building a super low distortion oscillator, and easier to switch to different frequencies.
 
BFNY said:
Yes, you can get tricky and make the input frequency an exactly perfect ratio of the sample frequency so that no window is required (uniform window). But I should point out this also requires reliable triggering on the waveform, and trigger delay such that the sine wave starts precisely at a 0 crossing for the first sample, and ends at a zero crossing for the last sample. This may be accomplished in sound cards automatically when the source DAC and input ADC are both running from the same sample clock and data acquisition start and stop are syncronized. Or not. But will not be the case for an external signal from more pure sources like your Cordell generator. Any slow drift of that external signal will also muck it up.

After reading again all posts, there is a very correct remark of Bob depending on internal clock and external clock's precision:

"Or not. But will not be the case for an external signal from more pure sources like your Cordell generator. Any slow drift of that external signal will also muck it up."

Be carefull Stefanoo before you make anything regarding external generators. Bob's statement it is absolutely correct and WARNING!

Regs
Fotios
 
I've been playing around with software sources, and it is very interesting. I'm using an older version of FFT software that does not have a 24 bit source built in, but does allow for 24 bit FFT's. And using it with an M-Audio Audiphile 192 sound card.

I found a few free programs that work good. The first was Dr. Jordan 24 bit freeware test version here

http://www.dr-jordan-design.de/Downloads.htm

I noticed it puts out only 777.1 Hz. The mystery starts. The version of this one must purchase to get 24 bit support is $79. - kinda high if you only need a nice pure sine generator.

It worked very well when set to an output level -13dB, so became the benchmark. With some finagling, it produced much better results for the M-Audio card than shown previously.

I ended up using a block size of 131,072, with 100 averages. In this and most cases, using "triggered" measurements gave better results, but the trigger level needed to be set pretty low (2-3% of input). I also found some low level 60 Hz noise initially, as I was using an 8 foot non-shielded twisted pair cable initially for analog loopback. By replacing the cable with a 12 inch BNC and adapters, the 60Hz noise went away.

Next I found this one,

http://www.tropicalcoder.com/ASIOTestSigGen.htm

which works very well, same as Dr. Jordan with one exception - it only runs for 60 sec. max. But it does allow you to program any frequency you like. Not bad for free.

Next I tried the Virtins "free trial" generator, and had a hard time getting it to work, until I turned off the Spectrum analyzer and Scope. It worked the same the above two packages when set up the same way.
http://www.virtins.com/page3.html

I tried 3 or 4 more, and found them lacking, almost all worked in 16 bit mode only, even though some had settings for 24 bit. Try before you buy! Found them here
http://cricket-sound-generator.qarchive.org/

Last I tried this one, the DSSF3 Basic from Yoshimasa Electronic Inc..
This being a complete packe, tt was tricky to get setup for generator only, but when all set to ASIO was fine. It worked OK, but oddly had an average noise floor about 5 dB higher than the other packages. It must be a a math thing, like a software rounding error, and tells you all generators are not created equal.

http://dssf3-basic.yoshimasa-electronic-inc.qarchive.org/

Now, the mystery of the frequency used correlating to THD results. It seems that most packages loop the generator signal, which means you may have the same problem with discontinuities at end of record creating distortion products. So it appears you really want an integral number of cycles to exactly fit in the time record for the generator as well. I can't say for sure why, but using frequencies like 777.1 (48kHz sample rate, 131072 block size) and multiples show lower THD and fewer harmonics than nice even numbers like 1,000 Hz. Graphs of results later.
Bob
 
here's an FFT plot for 1000 Hz

Settings were as posted above - block size 131,072 100 averages, triggered acquisition, source at about -20dB

Notice all the upper harmonics. There is still a lingering LF noise problem, which was less late at night...
 

Attachments

  • 1000hz_image.jpg
    1000hz_image.jpg
    90 KB · Views: 438
Hi Bob- that's really interesting. Given the number of things one has to worry about, and seriously doubting any two people can get the same results, it makes me appreciate my old analog equipment all the more ;-) Speaking of which, those spectral plots would make me think the waveform on a scope should be absolutely pristine. Do you have a scope and have you looked at it?
 
Conrad Hoffman said:
... those spectral plots would make me think the waveform on a scope should be absolutely pristine. Do you have a scope and have you looked at it?

Yes, there is about 20mV pk-pk of HF noise on the waveform with a 500MHz scope. I see similar stuff on my ono-sokki FFT when I turn off the anti-aliasing filters - the noise floor goes up. I think is typical for any digital based measurement.

Any digital source like a DAC or CD player likely puts out as much or more...especially the people that use zero oversampling DACS that shun output filters (way more).

You probably want to excite any tendency to oscillate while testing, rather than get surprised later. But yes, digital stuff has low level RF, and likely you could get rid of it, with a nice passive output filter. I guess the real test would be to hook it up to typical audio products and see how much really comes through.
Bob
 
Hi guys,

i’ve red your discussion about EMU… I’ve a EMU0404 and I would like to learn to measure T&S speakers with LIMP. I’m very beginner in DIY and I don’t speak well English so … it’s quite difficult for me to explain :)
How should I set the volume pot IN?
I set the output volume about a ¾ . When I measure the impedance of speaker, the impedance change if I change the input volume. The setup of the generator of LIMP seems to be non influential.
If I setup the generator with input volume at +3db and then I measures the speaker, I got a graph; but then I setup the generator with input volume at … +6db or 0db, the graph goes up… or down

Where am I wrong?!
Thanks !

marco
 
There is a PDF that describes what you want to do. Look around on the ARTA/STEPS/LIMP site under Support.

LIMP is for impedance measurements. Demo is free.




Sigurd

Kreisky said:
Hi guys,

i’ve red your discussion about EMU… I’ve a EMU0404 and I would like to learn to measure T&S speakers with LIMP. I’m very beginner in DIY and I don’t speak well English so … it’s quite difficult for me to explain :)
How should I set the volume pot IN?
I set the output volume about a ¾ . When I measure the impedance of speaker, the impedance change if I change the input volume. The setup of the generator of LIMP seems to be non influential.
If I setup the generator with input volume at +3db and then I measures the speaker, I got a graph; but then I setup the generator with input volume at … +6db or 0db, the graph goes up… or down

Where am I wrong?!
Thanks !

marco
 
If you don't know already, there is another one wonderfull software (totally free) the "Speaker Workshop" which can do the same things like ARTA. I reffer this software, because there is a nice tutorial about it, writen from an Italian in your language except English.

Regs
Fotios
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.