advertisement


Distortion Measuring Kit

S-Man

S-Man

StrivingON
As a result of a request in the NCC300 thread I thought it was worth starting a separate thread for the discussion of measurement equipment.

This is what I use to measure the distortion of amps, line stages, active crossovers etc:
1) Ultra low distortion 1KHz sine wave generator;
https://www.eevblog.com/forum/testgear/looking-for-ultra-low-distortion-1khz-sine-generator/

I use Viktor's module which used to be available on Ebay. There is also lots of discussion about this on DIYAudio.
I have found that despite Viktor's instructions that an LM317(HV) reg set to 35V is good enough to supply it, actually the supply is critical. I ended up powering mine from two 9V batteries (which negates the usefulness of the on board shunt regs). If using a mains fed PS then a transformer with an interwinding screen is neccessary, although I can also get good results from a Farnell L30 PSU supplied by an isolation transformer).

1a) Here is another possibility which avoids the supply issues and has a better attenuator setup than Viktor's. It does have higher distortion, but it's good enough!:
https://www.akitika.com/1kHzOscillator.html

2) Some sort of ADC
I use a 1st gen Focusrite 2i2.
The laptop used with this should be powered from its battery. Mains power, at least with my laptop, increases the noise floor by 10 to 20dB.

3) An attenuator - assuming you want to measure things >1V without blowing the 2i2's input stages
- See the "building your own attenuator" on the Akitika page.
I recommend incuding the clipping LEDs... very useful.

4) A lead to connect the DUT to the 2i2:
I can't remember exactly how I did this. It has a stereo 1/4" jack plug on the 2i2 end and stackable 4mm plugs on the amp end. Needs to interface the single ended amp output to the balanced input of the 2i2

5) REW or similar. I will have to check all the settings I use.

Here is the baseline perfomance of the sine wave gen and 2i2. I run all my tests at -12.6dB as I found this gives the lowest distortion. The distortion is dominated by the 2i2, hence my comments about the Akitika generator above.

Siggen-and-2i2-baseline-with-9-V-batteries.jpg


By way of example here is what a JLH 10W (15W in my case) output looks like at my "standard" test condition of ~1W into 10R:

JLH-10-W-Left-at-2-9-V-into-10-R.jpg


Some people get jolly excited by the decaying harmonic profile :D
 
Last edited:
All analogue here. DIY very low distortion oscillator based on design of Heathkit IG18 but with many mods, which does 20Hz - 20KHz. DIY distortion meter built to a design by Ted Rule which was published in Practical Wireless. Also have Marconi Instruments distortion meter and another by Dymar but the DIY one is way better than the commercial ones. I can measure down to about 0.002% or a little lower with this setup.
Also have all the kit to do it digitally as in OP but never use it.
 
Mixture here: M-Audio USB soundcard ADC + Meridian 203 DAC gives me lowest lowest THD (not +N) using ARTA (0.0007%). But, it's also fun to use my valve Heathkit boxes and I have the Marconi wave analyser. These days I'm starting to take more notice of SMPTE IMD as I've observed that with Bluetooth devices the THD can be pretty good, but the IMD less so. In a previous day job I had an Audio Precision sitting on my desk, no longer unfortunately.
 
wd40addict said:
Mixture here: M-Audio USB soundcard ADC + Meridian 203 DAC gives me lowest lowest THD (not +N) using ARTA (0.0007%). But, it's also fun to use my valve Heathkit boxes and I have the Marconi wave analyser. These days I'm starting to take more notice of SMPTE IMD as I've observed that with Bluetooth devices the THD can be pretty good, but the IMD less so. In a previous day job I had an Audio Precision sitting on my desk, no longer unfortunately.
Click to expand...

Same here... but not at any of the hi fi companies I worked for. I'm jealous about the Marconi Instruments Wave Analyser!
 
A little background to the AP purchase: In that company we were suddenly needing to make audio measurements of a telecoms standard for a new product. As we had nothing I brought in my Heathkit IM-58u from home and measured the THD+N, it was -40dB (1%).

A Keithley audio DVM was then ordered to give us some in-house capability. The DVM took a while to be approved and arrive. In the meantime a preliminary datasheet had already gone out, unbeknown to me, quoting -40dB! When I heard this I started to get nervous - how accurate was the Heathkit?

I needn't have worried, the Keithley confirmed the figure and the AP was purchased when we needed to test more demanding standards.
 
Regarding the question of PS for the sine wave generator, this is the best I could achieve with mains power - Farnell L30 PS run off an isolation transformer:

Siggen-and-2i2-baseline-L30-at-40-V-using-iso-tx.jpg
 
I borrowed a couple of 2i2 from work and found the noise floor far too high, are newer ones better?
 
Mine is Mk1.
Did you follow my recommendations re battery power etc?

I think the Mk2 has higher sampling rates and marginally better noise (but not sure TBH).
 
S-Man said:
As a result of a request in the NCC300 thread I thought it was worth starting a separate thread for the discussion of measurement equipment.

This is what I use to measure the distortion of amps, line stages, active crossovers etc:
1) Ultra low distortion 1KHz sine wave generator;
https://www.eevblog.com/forum/testgear/looking-for-ultra-low-distortion-1khz-sine-generator/

I use Viktor's module which used to be available on Ebay. There is also lots of discussion about this on DIYAudio.
I have found that despite Viktor's instructions that an LM317(HV) reg set to 35V is good enough to supply it, actually the supply is critical. I ended up powering mine from two 9V batteries (which negates the usefulness of the on board shunt regs). If using a mains fed PS then a transformer with an interwinding screen is neccessary, although I can also get good results from a Farnell L30 PSU supplied by an isolation transformer).

1a) Here is another possibility which avoids the supply issues and has a better attenuator setup than Viktor's. It does have higher distortion, but it's good enough!:
https://www.akitika.com/1kHzOscillator.html

2) Some sort of ADC
I use a 1st gen Focusrite 2i2.
The laptop used with this should be powered from its battery. Mains power, at least with my laptop, increases the noise floor by 10 to 20dB.

3) An attenuator - assuming you want to measure things >1V without blowing the 2i2's input stages
- See the "building your own attenuator" on the Akitika page.
I recommend incuding the clipping LEDs... very useful.

4) A lead to connect the DUT to the 2i2:
I can't remember exactly how I did this. It has a stereo 1/4" jack plug on the 2i2 end and stackable 4mm plugs on the amp end. Needs to interface the single ended amp output to the balanced input of the 2i2

5) REW or similar. I will have to check all the settings I use.

Here is the baseline perfomance of the sine wave gen and 2i2. I run all my tests at -12.6dB as I found this gives the lowest distortion. The distortion is dominated by the 2i2, hence my comments about the Akitika generator above.

Siggen-and-2i2-baseline-with-9-V-batteries.jpg


By way of example here is what a JLH 10W (15W in my case) output looks like at my "standard" test condition of ~1W into 10R:

JLH-10-W-Left-at-2-9-V-into-10-R.jpg


Some people get jolly excited by the decaying harmonic profile :D
Click to expand...
Thank you for the post :). Time I did some background reading. :)
 
davidsrsb said:
A two tone intermod test is easier to than trying to generate an ultra pure single tone stimulus
Click to expand...

Can you expand on the kit and method for this?

I must admit that so far I have found the actual THD numbers have minimal correlation with sound quality. E.g. the JLH sounds quite different to the NCC, although the NAP200 clone sounds quite similar tothe NCC IMO (see the NCC thread for the spectrums). All 3 are have similar spectrums. Of course this is at only one particular stimulus frequency and one operating point.

One thing that the THD kit is really good for, is finding there is something wrong with the design or implementation i.e. grounding errors are ruthlessly revealed.
 
Really need to get my head round to using this ferrograph RTS2/ ATU1 unit, does any one have any experience with this ?, is it still good enough for a Signal generator / distortion analyser?

IMG_2968 by Alan Towell, on Flickr
 
For SMPTE IMD two tones are generated at 60Hz and 7kHz and a amplitude ratio of 4:1. The analyser detects any intermodulation products that shouldn't be there. Effectively it's an AM radio tuned to 7kHz. I am not familiar with REW, but Arta can do this with your soundcard.

Note: there is another standard that uses tones at 19 & 20kHz so any beat will be at 1 kHz. However I feel the SMPTE test is more in the amplifier's normal range, even though historically it's based on optical film soundtracks.
 
The ATU1 is not of great use but the RTS2 provides a good selection of measurement capabilities in a single instrument. It does rather compromise on the range and accuracy of measurement to get such capability in one instrument though.
It can measure distortion between 400Hz and 1100Hz only and down to around 0.1% ish only. The oscillator itself has 0.025% THD @ 1KHz. It could be modded to give vastly better performance.
 
Arkless Electronics said:
The ATU1 is not of great use but the RTS2 provides a good selection of measurement capabilities in a single instrument. It does rather compromise on the range and accuracy of measurement to get such capability in one instrument though.
It can measure distortion between 400Hz and 1100Hz only and down to around 0.1% ish only. The oscillator itself has 0.025% THD @ 1KHz. It could be modded to give vastly better performance.
Click to expand...

Thanks for the information Jez I would be interested in any mods you know to improve the RTS2 if you get time ?
My battery powered NTI Minirator MR-pro does have good specs (–96 dB (0.0016% @ 18 dBu,Noise floor typ. 15 μVso) So I could just use that and then get a better distortion analyser
https://www.nti-audio.com/en/products/minirator-mr-pro-mr2

Alan
 
337alant said:
Thanks for the information Jez I would be interested in any mods you know to improve the RTS2 if you get time ?
My battery powered NTI Minirator MR-pro does have good specs (–96 dB (0.0016% @ 18 dBu,Noise floor typ. 15 μVso) So I could just use that and then get a better distortion analyser
https://www.nti-audio.com/en/products/minirator-mr-pro-mr2

Alan
Click to expand...

In fact the mods on my mind were to improve the oscillator section to that sort of quality or better...
I believe the distortion analyser section has only the one range and can only be tuned to cover 400Hz to 1100Hz
 
The 60 Hz and 7 KHz test is likely to behave differently in te USA and Europe, due to the mains frequency. It is common in Stereophile amplifier measurements to see 60 or 120Hz power supply artifacts.

Th 19 and 20 KHz test is better as many amplifiers have run out of feedback up there. The intermod is 1 KHz , so a simple low pass filter on the soundcard input can increase the dynamic range beyond the soundcards own performance.
 
I also used a 2i2 (orginal version) for a while, and also found it's performance was affected by its power supply. Best used with a laptop on battery power and an SSD not spinning rust. Laptop being useful for portability as well. If you want real lab bench quality you'd end up using a Benchmark ADC and DAC TBH. That's where I ended up. :)

I'm the other side of the coin to Arkless. :) I now prefer using 24 bit high res ADC and DAC. Means you have full control over whatever test waveforms and analysis process you wish. Indeed, good way to automate almost the entire test rountine and print out the results. 8-]

FWIW For Linux use I wrote a simply sig get app, and a scope/FFT app to use for basic tests. Kept meaning to do a more extensive set of test routines by lacked the round tuits. :) But if anyone wants what I did write, the apps and their source code are on the web.

BTW when doing 'THD' tests also watch out for mains/PSU intermod as that might be more audible, but doesn't show up in THD values unless you do (THD+N)/S where in some cases it might not poke above the noise floor. Also watch for any psu intermod on the 'other channel' as it shows up an amp not rejecting its rail waggles.
 
The other thing people tended not to measure was what I call 'ground bounce' where in a single rail design or asymmetric dual-rail one the near-dc level wobbles as you quickly change the size of the output waveform. In extreme cases like ye ancient Sinclair 2000 the *gain* used to bob about as well! So tone bursts can throw up problems that conventional CW THD don't show.

Then add in a speaker whose linearity may vary with near-dc input along with the audible audio. Indeed, if the crossover components have nonlinarity...
 
You must log in or register to reply here.


advertisement


Back
Top