Announcement

Collapse

Midwest Audio Fest

It’s that time audio enthusiasts! Registration for the 2019 Speaker Design Competition is now open! Visit midwestaudiofest.com for details and to list your speaker project. We are excited to see all returning participants, and look forward to meeting some new designers this year, as well! Be sure your plans include a visit to the Parts Express Tent Sale for the lowest prices of the year, and the Audio Swap Meet where you can buy and trade with other audio fans. We hope to see you this summer! Vivian and Jill
See more
See less

Basic DIY amplifier measurements?

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • jhollander
    started a topic Basic DIY amplifier measurements?

    Basic DIY amplifier measurements?

    I’m getting ready to build a few small chip amps, one of which I’d like to use for measuring speaker frequency response with my OmniMic.

    It occurred to me how do I know if the little amps are not distorting the signal? Meaning are there some basic amplifier tests I can do? May be something using OmniMic? I do not have an oscilloscope.

    My fall back plan is to do some A-B tests using different amps, same speaker, same close mic position and fixed voltage measurement at the speaker then compare FR graphs.

  • Ron_E
    replied
    There is a multi-part series of articles on using sound cards for audio measurements at AudioXpress.

    Ron
    A great test and measurement article series by Stuart Yaniger, which started to be originally published in audioXpress, June 2015.

    Leave a comment:


  • Millstonemike
    replied
    Originally posted by 1100xxben View Post

    Are you sure it was test equipment failure? I had a Sure class-D board that had its output filter resonance at 17 kHz. When testing with no load, it showed a response just like that as the output filter was un-damped.
    Maybe not.

    I measured the two line level signals with my scope. Since I don't have a 10X probe, I couldn't use the scope for the full power out measurement - the sine wave's peak voltage exceeded the scope's range. So the power out was done with a DVM; a $5 Harbor Freight unit. I had done some comparisons between the scope and the cheapo HF unit and it was good down low. So I assumed it was the DVM for the anomaly up high. It may be. Or perhaps the old saying: "*** U and Me" applies here - except it's all on me this time.

    So you make a good point. In hind site, I should have measured at a lower power out - something within the scope's range. Alas, retesting is not on my radar, but if I ever do, I let you know.

    But the fact remains that many of these low cost Class D amp modules have some degree of loss in the low end. It can be corrected on modules whose "on/off" switch is really the chip's "On / Stand by" input (the board remains powered). On those modules you can increase the value of the interstage DC isolation caps to get the inherent LP filter down in the single Hz range. But the modules that actually use a switch to control DC power to the board - increasing the same capacitors results in on / off speaker pop.

    Leave a comment:


  • 1100xxben
    replied
    Originally posted by Millstonemike View Post

    ...(ignore the green response above 5K Hz - test equipment failure).
    Are you sure it was test equipment failure? I had a Sure class-D board that had its output filter resonance at 17 kHz. When testing with no load, it showed a response just like that as the output filter was un-damped.

    Leave a comment:


  • Millstonemike
    replied
    Originally posted by 4thtry View Post
    I use your fall back procedure of A/B testing with different amps. This is how I found out that the little Lepai chip amp that I was planning to use for testing was messing up my low frequency measurements. .... They all give the same results, but the little Lepai amp rolls off the deep bass.

    Bill
    I've analyzed a bunch of chip amps. All exhibit low frequency roll off as mentioned by 4thtry. It's inherent in the low cost, simplistically designed amp modules - the chips' target market .

    Here's a $20 BT 2.0 module based on the very respectable TI TPA3116 chip. Red is BT chip's FR to the received test signal - into an op-amp stage. Blue is after the op-amp stage at the chip amp's input. And green is the chip's power output response to the speakers (ignore the green response above 5K Hz - test equipment failure).


    Click image for larger version

Name:	2 x 120 W BT AMP FR Graph.png
Views:	1
Size:	36.8 KB
ID:	1385900

    Leave a comment:


  • JRT
    replied
    TI's PCM4222EVM might be useful in this, is an AD converter with differential input for balanced impedance interconnection, exhibits high signal to noise ratio, low THD thru 20kHz, and provides AES3 output. You would need a computer digital audio interface that accepts AES/EBU input. A proper AES/EBU input is transformer coupled, providing galvanic isolation between the digital audio interface attached to the computer and the AD converter board.

    Leave a comment:


  • jhollander
    replied
    Here's a drawing of the measurement set up. I wasn't sure if a series load bank was correct or if I needed a voltage divider (L-pad) network. I can always check if I'm clipping in ARTA and the sound card.



    Leave a comment:


  • 1100xxben
    replied
    Originally posted by jhollander View Post
    Ben, thanks I get it. The cost of the 2-channel balanced sound card might kill this. Can I assume that all microphone audio inputs are balanced? So, something like the Behringer U-Phoria would work? https://www.sweetwater.com/store/det...horia-umc202hd
    Yes, that has a combo TRS/XLR input, so it should be a differential input. You'll likely want to use the 1/4" TRS input as that should be a "line" input, not a microphone input.

    Leave a comment:


  • wogg
    replied
    Originally posted by TN Allen View Post
    Just out of curiosity, given the variety of test equipment different people posting test results on the forum are using, how consistent are the results likely to be? I can see where different amplifiers could significantly affect Omnimic and Holm results, but how much variation is likely to result from different sound cards, and CD or DVD players, rather than initiating the test signal via computer memory? Is there a practical way to quantify or allow for the range of variations when looking at response graphs, a percentage or plus or minus a given number of decibels?
    In reality the variations from most sound cards and amplifiers are minuscule compared to the speaker being measured. There will be perhaps +- a fraction of a DB in response and a fraction of a percent of distortion, but I consider that way within tolerance. You can get bigger variation simply re-measuring in the same spot with the same settings consecutively. Loop back mode solves that entirely. Sent from my SM-G960U using Tapatalk

    Leave a comment:


  • jhollander
    replied
    Ben, thanks I get it. The cost of the 2-channel balanced sound card might kill this. Can I assume that all microphone audio inputs are balanced? So, something like the Behringer U-Phoria would work? https://www.sweetwater.com/store/det...horia-umc202hd

    Leave a comment:


  • 1100xxben
    replied
    Originally posted by TN Allen View Post
    Just out of curiosity, given the variety of test equipment different people posting test results on the forum are using, how consistent are the results likely to be? I can see where different amplifiers could significantly affect Omnimic and Holm results, but how much variation is likely to result from different sound cards, and CD or DVD players, rather than initiating the test signal via computer memory? Is there a practical way to quantify or allow for the range of variations when looking at response graphs, a percentage or plus or minus a given number of decibels?
    This is why I prefer ARTA for my acoustic measurements in loop-back mode. When using loop-back mode in ARTA, the soundcard uses one channel to measure the output of the amplifier and one channel to measure the acoustic response. It then does the computation in the background using the MEASURED output of the amplifier instead of assuming that the digital source signal is completely unaltered by any of the analog electronics. Some sources of error can be sound card frequency or phase response, amplifier frequency or phase response, and amplifier output filter/speaker impedance interaction. Using the loop-back mode should virtually eliminate any of those errors in frequency response. The reality is that most systems will only have slight alterations at the ends of the frequency spectrum, and for crossover design, those differences are way outside the bands of interest.

    Leave a comment:


  • TN Allen
    replied
    Just out of curiosity, given the variety of test equipment different people posting test results on the forum are using, how consistent are the results likely to be? I can see where different amplifiers could significantly affect Omnimic and Holm results, but how much variation is likely to result from different sound cards, and CD or DVD players, rather than initiating the test signal via computer memory? Is there a practical way to quantify or allow for the range of variations when looking at response graphs, a percentage or plus or minus a given number of decibels?

    Leave a comment:


  • wogg
    replied
    Originally posted by 1100xxben View Post

    Jon, I don't have time right this second to put out a full write-up, but I'll give you some key information and I'm sure others will chime in. I might have some time this weekend to do more of a full write-up with detailed wiring diagrams and instructions.... <SNIP>
    Well said and spot on! I'd be super nervous hitting the input of a fairly expensive Focusrite with a speaker level signal myself

    Leave a comment:


  • 1100xxben
    replied
    Originally posted by jhollander View Post
    I’m getting ready to build a few small chip amps, one of which I’d like to use for measuring speaker frequency response with my OmniMic.

    It occurred to me how do I know if the little amps are not distorting the signal? Meaning are there some basic amplifier tests I can do? May be something using OmniMic? I do not have an oscilloscope.

    My fall back plan is to do some A-B tests using different amps, same speaker, same close mic position and fixed voltage measurement at the speaker then compare FR graphs.
    Jon, I don't have time right this second to put out a full write-up, but I'll give you some key information and I'm sure others will chime in. I might have some time this weekend to do more of a full write-up with detailed wiring diagrams and instructions.

    The basic setup is just sound card output into your test amplifier and output of the amplifier into sound card input (possibly using an attenuator). You'll also want to terminate the output of the amplifier with a resistive dummy load (typically 4 or 8 ohms).

    Ideally, you're going to want a soundcard with balanced, differential inputs for a couple reasons. The first reason is that most pro-audio USB audio interfaces will be able to naturally accept higher input levels, and you may not even need any kind of attenuator. I use a Focusrite Scarlett 6i6 and it can take something like 20 V-peak before the input of the soundcard clips. This would allow you to test your amplifiers up to an output level of about 24 watts (into 8 ohms). If you want to go higher than that, you can make a really simple resistor divider. I've done this before so I can measure the output of amplifiers with 200 V-peak output levels. The Scarlett 2i2 would also probably be a good option, and is quite a bit cheaper.

    The second reason for using a differential input, and this is a BIG one, is so you can measure the output of bridged amplifiers. When you hook up the amplifier output to the soundcard input, you will have 2 connection points. If you have a soundcard with sindle-ended inputs, then you'll have to connect one side of your amplifier output to the ground of your soundcard. If your amplifier is single-ended, then is not a big deal. However, if the output of your amplifier is bridged, then you're going to be shorting out one side of your amplifier, through the input cable of your soundcard, which is very likely to cause problems, of which the worst case is destroying your amplifier output and/or the input to your soundcard.

    For software, ARTA is a great choice using STEPS. You don't actually need a scope to "see" the output of the amplifier, assuming it was built correctly. It is actually easier to see the onset of clipping using a spectrum analyser, which ARTA has. You will essentially run a frequency sweep and look at the THD. When you get to the point of clipping, your odd-order distortion will rise VERY rapidly. If you take a look at any of the datasheets for amplifier chips, they should have graphs that show THD+N vs output power. Take a look at figure 19 on this LM3886 datasheet: http://www.ti.com/lit/ds/symlink/lm3886.pdf . The sharp knee in the curve is where the amplifier begins to clip. ARTA also will tell you the signal level coming into the sound card after running each measurement. When you see distortion rise in your measurement, make sure that you're not clipping the input to the sound card first, before you assume it is amplifier clipping.

    Ideally, you should also have a low-pass RC filter on your sound card input to help reduce some of the switching noise from the amplifier, but a lot of sound cards input anti-aliasing filters will also take care of this for you.

    I hope that's not too much info too fast. Again, I'll see if I can't find some time to make a little write-up with more specific wiring.

    Leave a comment:


  • jhollander
    replied
    Thanks Don. Maybe if I knew how to use it I could tell if it was decent...

    Leave a comment:

Working...
X