Warning: Creating default object from empty value in /homepages/11/d132647312/htdocs/Amfone/mkportal/include/SMF/smf_out.php on line 47
measuring distorsion




 
The AM Forum
November 23, 2017, 05:11:22 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: [1]   Go Down
  Print  
Author Topic: measuring distorsion  (Read 1942 times)
0 Members and 1 Guest are viewing this topic.
PA0NVD
Member

Offline Offline

Posts: 55


« on: November 16, 2016, 12:48:05 PM »

Hello all. Because an other branch of my hobby is designing and building tube amps, I was in need of a system to measure IMD+N. these systems and meters are quite difficult to get here in Europe and quite expensive. So I decided to build me a simple automatic tuned notch filter that could measure IMD from 10% down to 0,01%
It may be a little off-topic in this forum, but when someone is interested I will post the information to build it. I did design and build in 2 days, so quite simple and no special materials needed, all E12 resistors and capacitors. The measuring frequency is 700 Hz in order to have most harmonis within tha audio band and not go too low that transformes make the difference. But other frequencies can be selected without a problem.
Attached some photos and a result of measuring a small LM386 amp


* THD meter 001.jpg (62.73 KB, 780x585 - viewed 91 times.)

* THD meter 002.jpg (134.27 KB, 780x585 - viewed 110 times.)

* THD meter LM386 amp 2,2%.jpg (76.5 KB, 780x585 - viewed 68 times.)
Logged
w3jn
Johnny Novice
Administrator
Member

Offline Offline

Posts: 4554



« Reply #1 on: November 16, 2016, 08:12:50 PM »

Nice project - thanks for sharing!

Schematic???
Logged

FCC:  "The record is devoid of a demonstrated nexus between Morse code proficiency and on-the-air conduct."
W1TAG
Member

Offline Offline

Posts: 13


« Reply #2 on: November 16, 2016, 08:18:28 PM »

I'm guessing that the picture captions are correct, and this setup measures THD+N with a single tone. IMD measurements would require two tones.
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 6728



WWW
« Reply #3 on: November 16, 2016, 10:23:49 PM »

I'd like all the info too. Low distortion is a decent Grail to seek. It does not have to be 0.1% but a reduction to 3-5% would be a large improvement on many stations.

One question - that is probably designed to check sine waves. If the sine wave is good, can we assume the strange compressed and clipped audio of hams passed through the modulation system will be as free of further-added[/b] distortion as the sine wave is?
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
WBear2GCR
Member

Offline Offline

Posts: 3418


Brrrr- it's cold in the shack! Fire up the BIG RIG


WWW
« Reply #4 on: November 16, 2016, 10:43:08 PM »

Realistically today you can do all this with either a computer & free software or a computer + soundcard and pay for software, or just buy a QA400 or QA401 distortion analyzer (goes to the USB port).

Even a laptop built in sound chips will usually do down to ~0.005% or better.
Most good soundcards, even "last generation" will do ~0.001% and newer ones will do lower.
The Quant Asylum ("QA") devices do <0.001% or ~-120dB or better. The now older QA400 new was $200USD.

There is a benefit to a notch filter, and that it that you can extend the effective performance of a sound card or other analyzer
by ~20dB or so, sometimes more.

Attached is a screen shot of the QA400 looking at a McIntosh 50w-2 amplifier (tubes)

      _-_-

PS. on http://www.diyaudio.com you can find a ton of information on this topic, and various analyzers, and homebrew
set ups like this one, as well as ultra low distortion oscillators...


* McIntosh-50-W-Amp-Hi-Pwr.jpg (364.23 KB, 1480x1050 - viewed 74 times.)
Logged

_-_- bear WB2GCR                   http://www.bearlabs.com
MikeKE0ZUinkcmo
Member

Offline Offline

Posts: 258



« Reply #5 on: November 16, 2016, 11:15:58 PM »

An excellent signal source is presented in Linear Tech's AN-67, published in 1996.   The signal generator is presented starting on page 62.   Distortion is claimed to be in the parts per billion region.

Mike
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #6 on: November 17, 2016, 03:47:19 AM »

I'm guessing that the picture captions are correct, and this setup measures THD+N with a single tone. IMD measurements would require two tones.

You are right, sorry typo. I meant THD+N
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #7 on: November 17, 2016, 06:19:59 AM »

First of all the comments. I use the THD meter for tube circuits, so I consider anything better than 0,01% absolutely useless.  The meter has 3 sensitivities, 0,1% FS, 1% FS and 10% FS. You can select the sensitivity just by 1 resistor, so if you like to have other sensitivities, no problem at all. The meter has an output where you can compare the original signal wuth the distorsion to see where the distorsion originates. I will include a few pic's of a positive clipped signal wich makes that very clear.
It can be discussed as well that I should measure true RMS instead of what I do. But I use the meter just to find the approx amount of distorsion and to find the reason (clipping, compression, cross-over etc.)
In addition, you can build at least 5 of these circuits for the price of a sound card. And I don't have a computer in my hobby corner
Ok, here is the information and the schematic diagram
I used cheap chinese LM5532 amps. Using originals or better opamps may increase the temperature stability or/and reduce noise.
The input goes via a 500K pot to an opamp in order to have a high input impedance. The input level should be adjusted to 200 mV eff, a meter circuit is included wich should be calibrated . The input to the filter is also led to a comparator in order to have a reference signal for the nulling loops.
The filter is a variable state filter connected such that there is little dependence of component value's, normal E12 resistors and caps can be used. The filter has a notch output (via an amplifier), a band pass output (BP) and a 90degrees output, used in the nulling loops.
When build as such, the filter will give you approx a notch with a factor 50 - 100 attenuation, wich is not sufficient for most THD measurements.
But the Notch output is in counter phase with the BP output, so when a small amount of BP signal is added to the NOTCH output, one can get a very deep null. But the frequency becomes extremely sharp ( a part of a Hz) and tuning becomes almost impossible.
Therefore, I added a nulling loop for the frequency by adjusting the drive resistor of the second integrator with a LDR driven by a red LED. Both are mounted together in a piece of shrink sleeve. The tuning information comes from the phase of the REF signal and the 90 degrees signal. These are 90 degrees out of phase when tuned correctly.
In order to test, first test the filter without the LDR for the frequency loop and the nulling loop connected. One should get a Notch at approx 650 Hz. Set the DC NULL so that there is no DC at the NOTCH output.
Now adjust the loop gain for the freq. loop to midrange and connect the LDR for the freq. loop. The tuning should go to apprix 700 Hz and the loop will have much less frequency dependency. Now increase the loop gain until the loop starts oscillating and back-off a little.
There should be a remainder of the 700 Hz visible at the NOTCH output.
This remainder can be nulled by a variable resistor of 100 K between the NOTCH and the BP. You should adjust between both the FREQ adjustment and this resistor, but again it is a very sharp null. So the NULLING loop is connected also using a LDR wich is connected between the BP and the NOTCH. (instead of the proposed trim potmeter used just to see if it works.)
The Null should inmedeately become very deep. Now adjust the FREQ ADJ trim pot for minimum NOTCH signal and only the harmonics of the 700 Hz frequency source are now visible. Adjust the nulling loop gain at the max gain possible without oscillatory effects.
Calibration
use a low distorsion generator at 700 Hz and adjust the level.
Let the meter warm up for 10 minutes and fine adjust the freq. adjust to have the deepest null.
The output meter should be calibrated for 0,1% FS, 1%FS and 10% FS.
The null should be less than 0,05% THD. Mine is 0,025% for my old LEVELL generator.
Now define the generator output impedance (mine is 600 Ohms ) and connect a second generator via a resistor of 100X the generator impedance (so e.g. 56 kOhm) to the generator output. Adjust the second generator for the same output level as the 700 Hz generator but at a frequency between 1,4 and 10 kHz. This should add 1% distorsion to the 700 Hz signal and adjust the NOTCH GAIN until the meter shows 1%
BUILDING
I did build at non-etched PCB using islands glued to the surface with cyano glue. The IC's are glued upside down. Please note that the pincount at the bottom view is CW. mark where the pin 1 is with a felt pen. Building this way gives a very good ground plane and one can build very fast just following the diagram.
Attached also a pic of positive clipping resulting in 0,05% distorsion. When I remove the clipping, it is absolutely impossible to see the distorsion at the original sine wave. A clipping of 0,5% can be seen as only half a linewidth . A 5% clipping is clearly visible
I will attached these pics in an other post because I can't add more attachements



* THD meter schema 1.jpg (47.48 KB, 780x547 - viewed 82 times.)

* THD meter schema 2.jpg (40.91 KB, 780x547 - viewed 77 times.)

* THD meter, pos clipping, 0,05%.jpg (64.99 KB, 780x585 - viewed 50 times.)
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #8 on: November 17, 2016, 06:24:15 AM »

positive clipping
Attached some pics of 0,05% distorsion due to positive clipping and  5%.
Also a pic of the distorsion of my LEVELL generator of 0,025%


* THD meter, 0,025% LEVELL generator.jpg (65.99 KB, 780x585 - viewed 47 times.)

* THD meter, pos clipping, 0,05%.jpg (64.99 KB, 780x585 - viewed 43 times.)

* THD meter, pos clipping, 5%.jpg (70.89 KB, 780x585 - viewed 50 times.)
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #9 on: November 17, 2016, 06:33:37 AM »

some more info
I used the CD4046 phase comparator 1 for the tuning information. Just for simplicity. The powersupply should be minimum +/- 6V to max +/- 7,5 Volts. I used the LM78L05 and the LM79L05 both with a red LED in the common to increase the voltage to +/- 6,6 Volts. The supply voltages should be well decoupled with a few caps.
The input level meter circuit will contain distorted signals that can capacitively couple to the high impedance input, so screen the meter from the input.
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #10 on: November 17, 2016, 06:42:43 AM »

I'd like all the info too. Low distortion is a decent Grail to seek. It does not have to be 0.1% but a reduction to 3-5% would be a large improvement on many stations.

One question - that is probably designed to check sine waves. If the sine wave is good, can we assume the strange compressed and clipped audio of hams passed through the modulation system will be as free of further-added[/b] distortion as the sine wave is?

You are right in my opinion. If you measure a distorsion less than e.g. 1% single tone, evrything you add later (positive enhancement etc) should be free of other distorsions. I think that the main benefit will be finding the reason of a distorsion in your modulation or your amp, clipping compression, cross-over etc. It makes it very simple. Just hookup a generator better than you are looking for and you can see the result at the scope. But when measuring a modulated signal, it will be difficult though to have a detector of the RF signal that doesn't give distorsions. A syncronous detector may be required for the measurement.
Logged
W1RKW
Contributing
Member
*
Offline Offline

Posts: 4038



« Reply #11 on: November 17, 2016, 04:53:04 PM »

for many years I relied on the Sound Technology 1700B distortion measurement system. A very reliable system. I've seen them on Ebay in supposedly excellent operating condition for as little as $200.  They're hard to come by though.
Logged

Bob
W1RKW
Home of GORT. A buddy of mine named the 813 rig GORT.
His fear was when I turned it on for the first time life on earth would come to a stand still.
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #12 on: November 18, 2016, 03:42:42 AM »

for many years I relied on the Sound Technology 1700B distortion measurement system. A very reliable system. I've seen them on Ebay in supposedly excellent operating condition for as little as $200.  They're hard to come by though.

And when imported to Spain, you can add at least an other 200$ for customs and transport
Logged
WBear2GCR
Member

Offline Offline

Posts: 3418


Brrrr- it's cold in the shack! Fire up the BIG RIG


WWW
« Reply #13 on: November 18, 2016, 11:09:36 PM »

Nothing wrong with building your own...

but FYI:

https://www.quantasylum.com/content/Products/QA400.aspx

It's good to see your harmonic ratios, and also to see the effects of biasing...



* QA400.jpg (102.25 KB, 1024x501 - viewed 54 times.)
Logged

_-_- bear WB2GCR                   http://www.bearlabs.com
WD5JKO
Member

Offline Offline

Posts: 1659


WD5JKO


« Reply #14 on: November 19, 2016, 07:57:42 AM »


Nice discussion. Lots of ways to measure distortion. If the audio amplifier being tested is part of a modulator circuit, then we can measure the modulation distortion with an SDR receiver, and use either a single tone, or two tone modulation. I include images of my QRO modified Central Electronics 20A which uses 7591A finals. One plot is on LSB, single tone at 40W pep, and the other is 2 tone running AM, at 10W carrier. I include a third plot breaking down the intermodulation products.

Those with a good SDR receiver have a very useful piece of test equipment.

Jim
Wd5JKO


* 2nd_Day_1Khz_40 watts_LSB.jpg (242.59 KB, 1557x662 - viewed 68 times.)

* 2_Tone_10W_Carrier_700hz_1900hz_100Percent.jpg (210.78 KB, 1559x662 - viewed 79 times.)

* 20A_10W_AM_2_Tone_100_Percent_Mod.jpg (60.34 KB, 906x649 - viewed 73 times.)
Logged
PA0NVD
Member

Offline Offline

Posts: 55


« Reply #15 on: November 19, 2016, 09:11:43 AM »

That are impressive pieces of equipment. Unfortunately I am a moron with computers. Embarrassed So for me they are out of reach and I don't have an application for a measurement system that can see less than 0.01%
What I like is that you can seen how far the harmonics reach. But on the otherhand, for a modulator it is good engineering practice to have a low-pass filter anyway to avoid splatter with over modulation or due to distorsion.
What I do not like, is that you can't see where the distorsion originates, what kind of distorsion it is, so what to do about it. In the set-up with a notch filter, that is often very clear, see the pics of the clipping. You can inmedeately see that there is clipping and at which polarity. With a PP amp and one soft half, you can see which half is bad. The spectrum just shows you even harmonics that also can be generated in the driver. (see the pic of the LM386, the second harmonics are out of phase with the sine, so they most probably are originated in the driver stage). When there is cross-over, you see the distorsion at the zero crossing of the sine. Clear as water and you can do something about it.
And again, for my modulators and tube amps, I am not interested in distorsion less than 0,1%.
Logged
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 1619

Making Amplitude Modulation GREAT Again!


« Reply #16 on: November 20, 2016, 07:36:08 PM »

Jim,

Which rx sdr are you using in those pics above?

--Shane
KD6VXI
Logged
WD5JKO
Member

Offline Offline

Posts: 1659


WD5JKO


« Reply #17 on: November 20, 2016, 09:18:23 PM »


N VAN,

   You raise several interesting points for using an analog based distortion device. Some years ago I purchased a student version of the DR. Jordan distortion program. There is a free signal download that is quite capable. Open up two of them for 2 tone tests. Knowing the FFT response to symmetrical overload versus asymetrical overload is a useful indicator of what is happening. In either case, more a whatever your used to doing works better.

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

Shane,

   I am using a Flex 3000 for those screenshots. I had a fixed 40 db tap off my dummy load, and a 0-90db rotary attenuator to get the Flex signal level up around S9+20. The software I am using is PowerSDR 2.7.2 with the Ke9NS version T11 enhancements.

Jim
Wd5JKO
Logged
Pages: [1]   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.048 seconds with 18 queries.