On Fri, Jul 1, 2022 at 11:43 AM, <brent@...> wrote:

a wiring diagram would help

 
--
Mike Miller
PA, USA

The specs for the EMU-0404 USB soundcard are pretty good

Input THD+N: -101dB (1kHz @ -1dBFS)
Input S/N: 113dB (A-weighted)
Output THD+N: -100dB (1kHz @ -1dBFS)
Output S/N: 117dB A-weighted

--
Mike Miller
PA, USA

On Fri, Jul 1, 2022 at 09:18 AM, newaag wrote:

It is not perfect, or even nearly as good as the 339A source or the 24 bit sound card

For the 1kHz audio source, I use the output of the soundcard to the Ext In of the Panasonic VP-8175, not the Panasonic's internal audio generator.

Selecting [Lock frequency to RTA FFT (64k)] on REW's audio generator allows use of a rectangular window for maximal spectral resolution of the FFT plot by restricting the generator output frequency to the nearest FFT bin for the current RTA FFT length. {RTA = real time analyzer}

Mike Miller
PA, USA

On Fri, Jul 1, 2022 at 10:12 AM, M.K. wrote:

I know and I use 24 bit cards as well butthey are not as good as you think .

I agree the stuff you attached supports your point above. My actual experience is with a M Audio Audiophile 192 card.  I spent lots of time playing with this in the old days, maybe 10-12 years ago.
Windows XP days. It has performance similar to what this guy did with an EM-U 0204
https://www.diyaudio.com/archive/blogs/alexcp/1017-measurements-sound-card.html

All I can say is don't be misled by poor sound card "source" (D/A) performance, which is very commonly worse by 10-20 dB than the A/D using an external high quality analog source. I actually found that the software driving the source (generating the sine wave) could make a 10 dB difference in the source distortion. Some programs were really bad on that.
Which led me to pursue and buy a couple analog sources (SG505, 339A). Use these, at optimized levels into a 24 bit sound card and see what the A/D can really do with a better source.

If the ASR tests are based on loop back tests using the internal source, the card may be able to measure a lot better.    

The dynamic range limit you mention for sound cards may be true for typical ones built into a laptop or motherboard, but 24 bit ones are way better than 90-98 dB.

I know and I use 24 bit cards as well butthey are not as good as you think . A decent card will have its own  distortion products that will be in the -100dB  range but almost no cards have distortion below 120dB and some are in the 70dB range.  That is why digitizing the output of the THD meter  with a bit of gain will allow much better measurements of the THD.  Assuming the THD meter circuit is not adding THD itself to the signal.

here are many ADC SINAD measurements.  There are good cards in the $150 and above range.




https://www.audiosciencereview.com/


here is the one of the most affordable good cards with SINAD of 102dB

 https://www.lrultrasonics.com/solutions/clock-lube-136   

Look at the SDS for Clock Lube. Interesting stuff!  I do keep an ultrasonic - might be the ticket for some radio items. 

Peter Wieck
Melrose Park, PA 

Relevant to this:  My wife and I spent a few years in Saudi Arabia, she as a school teacher, me as a property manager. While shopping in the Souks of Al Khobar, I found Cramolin Red. Now, in Saudi, "proprietary ingredients" are not permitted, and all compounds are required to have their contents called out. 

On the 350 ml spray can of Cramolin Red was a paper label, in English and Arabic:

Contents:    Oleic Acid - 5%
                    Hydrocarbon solvents and propellants - 95%

I took that knowledge back to the US with me - and communicated at-length with Mr. Turner, who acted upon it, doing a favor for the hobby. Notes:

• Why Coleman Fuel?  Coleman Fuel has anti-corrosive and anti-gumming additives. Pure naptha does not. 
• If one is using naptha - purchase *LAB GRADE" material, not VM & P stuff. That is the poorest grade of that material. 
• Why lighter fluid? Although far more expensive by-the-ounce (ml) than Coleman Fuel, it also contains anti-gumming additives. Lighter-fluid is also considerably more refined than Coleman Fuel (meaning it consists of fewer fractions). Lighter Fluid is very nearly lab-grade + additives. 
• Oleic acid may go by mail, and is available from Amazon:  https://www.amazon.com/Oleic-C18H34O2-Liquid-Space-Saver-Bottles/dp/B01M0EJZB4/ref=sr_1_4?keywords=oleic+acid&qid=1656694433&sr=8-4   That would be a life-time supply for the lot of us, by the way - unless we are busy restoring brass clock movements. 

NOTE:  If you are getting paid to do work on equipment owned 'by-others', I counsel against using home-brews of any nature. Should there be any inadvertent consequence, you will 'own' it whether related or not. But for yourself - and with the understanding of the materials involved - it is the way to go (again, if done safely). 

Peter Wieck
Melrose Park, PA 

very simple but the test gear makes all the difference
The RF generator needs to have internal or external low distortion source and low distortion FM modulator.
The THD measurement device will notch out the fundamental tone with a high q notch filter and measure what is remaining which is THD and noise.
You can also use a PC card on the output of the THD device to get even better results.

On Thu, Jun 30, 2022 at 07:55 PM, Mike Miller wrote:

Equipment: External soundcard (EMU-0404 USB), laptop running REW software, Panasonic VP-8175A generator, Linear Audio Autoranger II (soundcard attenuation/amplifier)

Still trying to understand how this is connected...a wiring diagram would help.  Sorry for the noob question - I understand how to measure THD in an amplifier with my 339A because it has an input for the oscillator, but just not understanding the connections when testing a tuner.  I have been searching and can't seem to find any information on how to do THD tests on tuners - lots of theory about it...but not a lot of information on how to connect something like an HP 331A/334A/339A to do THD measurements.  What you are doing should be similar so a wiring diagram would really help me understand this.

Oleic acid - Both a lubricant and contact cleaner -
"Oleic acid is a fatty acid that occurs naturally in various animal and vegetable fats and oils. It is an odorless, colorless oil, although commercial samples may be yellowish. In chemical terms, oleic acid is classified as a monounsaturated omega-9 fatty acid, abbreviated with a lipid number of 18:1 cis-9. It has the formula CH₃(CH₂)₇CH=CH(CH₂)₇COOH.^[2] The name derives from the Latin word oleum, which means oil.^[3]"
https://en.wikipedia.org/wiki/Oleic_acid

Alan Douglas made an interesting post here about oleic acid - https://antiqueradios.com/forums/viewtopic.php?t=204421
"I once ran across a circa-1930 European trademark registration for "Gramolin," a treatment for DC motor commutators. I presume it was from "Gramme Ring" and "olein," the German for oleic acid. "Gramolin" became "Cramolin" which more recently became DeOxit when the US importer broke off from the German parent (it's still Cramolin there, or was several years ago)."

Bill Turner at one time sold Oleic Acid in small bottles and recommended you mix it with naptha - https://web.archive.org/web/20121114091837/https://dialcover.com/components.html
page down to "Cease and Desist" contact cleaner
"My latest find is a super contact cleaner for copper and brass contacts that you can mix at home.  It consists of OLEIC ACID mixed with standard Coleman Fuel.  Oleic acid is an organic acid found in things like cocoa and olive oil and not considered dangerous.  It is used for many things in industry ranging from lubricants to cosmetics to food flavorings.
My recommended mix with the Coleman fuel (naptha) or regular lighter fluid is 5%. "

I was also thinking "clock  lube". It does dry up on clock movements eventually as time goes by.

On Thu, Jun 30, 2022 at 11:16 PM, M.K. wrote:

if you use a twin t notch filter and some gain stage after that

This is a good way to look at the actual distortion waveform too. The dynamic range limit you mention for sound cards may be true for typical ones built into a laptop or motherboard, but 24 bit ones are way better than 90-98 dB. The method works fine for THD, but I've never had the need to use it for tuners - I don't really trust any tuner distortion reading below ~0.04%, or -68 db.

With sound cards, you still have the basic problem of matching the signal to be analyzed to the sound card input range. Which is usually not as flexible as the calibrated inputs on a box type distortion analyzer. Where you have options for both gain and attenuation.  

The best window to use for this sort of thing is "flattop". It has a very low error for amplitude accuracy. The rectangular window can have gross amplitude errors when used incorrectly. If the time domain signal is not at a zero crossing on both ends of the time domain plot (corresponding to what is being FFT'ed), you have FFT amplitude errors. The only time I use rectangular window (or boxcar) is for impulsive type single shot signals that decay fully within the sample window.     

The other point - the VP-8175 stereo generator is way better than the 8640B for THD on FM output, but...
It is not perfect, or even nearly as good as the 339A source or the 24 bit sound card. So the generator will still be the limiting factor, and you can still get 2nd harmonic cancellation between tuner and generator. 

Olson Electronics was, indeed, "something" like Allied but was more similar to the original Boston-based Radio Shack or Lafayette. Olson sold electronic components and such but was very big into Japanese-made stereo components - everything from amps and tuners to receivers and speakers. Their gear was usually bargain priced but not of the highest quality. They were headquartered in Akron, OH. And sometime in the late 60's or early 70's they were purchased by the Teledyne operation which led to the franchising of stores across the country - again, rather like Radio Shack. When the stereo boom died off, Olson was very hard hit and ceased operations in the 1980's.

On Thu, Jun 30, 2022 at 09:07 PM, Mike Miller wrote:

mploying averaging did not, as I observed, lower the fundamental levels (2nd, 3rd, etc) and resultant THD% calculated value, but did average out the fluctuations that they experienced sample to sample while down around the -80dB level. The noise floor was perhaps 10dB below that, and of course, was flattened out as averaging was switched on. Sort of like cutting the grass.

In other words the THD numbers with averaging did not drop, but stop jumping around like a toddler so I could get a reading.
Tomorrow I will take a screen shot of the FFT of the Pioneer F-99X with and without averaging. What will not be visible though, is the fluctuations of the 2nd, 3rd, etc without averaging. I will try to get a visual estimate of the fluctuating numbers. REW provides a numeric, as well as spectrum, readout of THD, THD+N, N and the harmonic levels.
--
Mike Miller
PA, USA

Mike

if you use a twin t notch filter and some gain stage after that you will get much better results.
lets say you get a -60dB twin t active notch and add a gain of 20db after that notch filter. you will get a 110....120dB resolution of dynamic range for your FTT. if you already have a THD meter with output  you just need to amplify the output of the meter and feed it to the sound card.  most cards have a dynamic range of 90....98dB  and this will allow much greater resolution.

Maciej

Maciej

Employing averaging did not, as I observed, lower the fundamental levels (2nd, 3rd, etc) and resultant THD% calculated value, but did average out the fluctuations that they experienced sample to sample while down around the -80dB level. The noise floor was perhaps 10dB below that, and of course, was flattened out as averaging was switched on. Sort of like cutting the grass.

In other words the THD numbers with averaging did not drop, but stop jumping around like a toddler so I could get a reading.
Tomorrow I will take a screen shot of the FFT of the Pioneer F-99X with and without averaging. What will not be visible though, is the fluctuations of the 2nd, 3rd, etc without averaging. I will try to get a visual estimate of the fluctuating numbers. REW provides a numeric, as well as spectrum, readout of THD, THD+N, N and the harmonic levels.
--
Mike Miller
PA, USA

On Thu, Jun 30, 2022 at 07:55 PM, Mike Miller wrote:

averaging of 32 samples

Averaging makes a big difference, versus instantaneous readings.
Using it with high resolution (64k samples) FFT analysis (i.e. power spectrum) can make THD readings better by 6 dB or more...  
By lowering the noise floor under the harmonics. 

On Wed, Jun 29, 2022 at 03:14 PM, AmadeusMozart wrote:

80 dB is about at the limit of what the HP can do

When you get to -80 dB levels, the chasm between THD and THD + noise becomes fairly wide.
An FFT measures THD, analog meters like the HP 339A measure THD + noise. So the -80 dB number above is for THD + noise, using the 339A analyzer section.

I have seen and measured a lot of audio generators here, using a very good 24 bit sound card and FFT THD programs.
The HP 339A audio source is clean to well over -110 dB via 24 bit FFT THD. It is one of the best commonly available low distortion sources, and that short list includes the old Tek audio analyzer and ST sources as well. My Tek SG505 matches the 339A, both are reference quality analog sources.
One can build single frequency generators that are as clean, DIY, but these 2 old dogs cover the broad audio frequencies and also have adjustable output levels.