Measurement Adapter [DIY Project]

General Discvssions & Theory/Vnderstanding
1

I don’t think the “I am not sure either, but I am trying”-preamble is needed here.

Building a Measurement Adapter


Startpoint
Since recently-ish, I have an oscilloscope. More an Electronics-Multi-Tool (Scope, Signal Analyzer, Arbitrary Wave Generator, Digital Multi Meter and (basic) Power Supply) that by virtue of being SA and AWG in one package, can be easily used for Frequency Response Analysis (= FR).

This setup right now requires to run a BNC from the AWG, listen to that “In”-signal with a probe and record the “out”-signal with another probe.

FR
FR610×553 11.5 KB

That is a perfectly valid Lab setup where the DUT is the limiting factor and viewed in a vacuum.
Two notes:

  1. My scope is running an 8-bit ADC (at stupid high sampling rate), but still only 8-bit. GWInstek is doing some magical fuckery in Spectrum Analyzer mode (definitely not just normal FFT).
Excerpt: Specs

image
image590×671 61.6 KB

image

  1. “In a Vacuum” is great! It makes your DUT look stunningly perfect when it has perfect controlled conditions where pass/fail can be picked at ones desire. Or even worse “Look at our arbitrary number!” they exclaim, “It is bigger than out competitors arbitrary number!”

The Goal
I would like an adapter box to make setting up a headphone amplifier to test easier.

Requirements: Whishlist:

  1. 3x BNC for Signal in, Signal Sense and Signal out

  2. pair of RCA-Out for DUT

  3. Option to load the DUT via 6.3mm Jack (= Headphone connection)

  4. Option to load DUT with “ideal” resistor (= Internal resistors, either fixed “common” values (32, 120, 300, 600) or pick&choose resistor ladder)

  5. Shielding/Earthing

  6. [Optional] Internal Inductances to approximate some headphones

Point 1) and 2) are just ease of use. I am lazy and since I will use this thing a lot to test DIY projects, I want this adapter to be as Plug&Play as possible.
Point 3) is to prevent the “in a Vacuum”-problem mentioned above. Surely, it is nice to validate output voltage swing into an open circuit to ensure not capacitive issues along the signal path. For me, it would be nice to see how amps fight against whatever weirdness a headphones voice coils present to them.
Point 4) is to explore output power into various loads with more detail than “750mW into 32Ω, 100mW into 300Ω, thanky bye”.

Point 5 :upside_down_face:
Point 5 is entirely because everyone is doing this wireless bullshit at all frequencies creating this wonderful noise floor that limits what I can measure outside of carefully constructed shield arrangements to ±50mV. Tieing my to-be-constructed adapter in a metal enclosure to ground (via binding post and test lead) should keep that BS out of my measurements!

Point 6 is feature-creep. I want this thing to be as compact as it can be reasonably made to be.

Inspiration
This idea is heavily inspired by this here thread on SBAF:

While the “Do Measurements matter?”-discussion is nice and all, I don’t care about that

What I do care about is my own frustration. Because I need five or six cables with various adapters and splitters to set up this test that takes at most a minute to run. The flying spaghetti monster sitting on and near my desk when I set this up, is not pretty and quick to yank my DUT off said Desk.

Expect results some time in Q1/22

So, any questions or ideas?

5 Likes
2

Can you describe the tests/experiments you want to do with this kit? Are you going to test hypothesis or simply explore and learn how different kit interacts?

3

Like hinted at above, I have multiple uses for this test rig.

Number one is to give myself an easy to use option to test my own circuits (mostly proof-of-concepts and the occasional prototype). Especially throwing various loads at my DIY-projects in a manner that does not endanger any of my headphones.

A personal interest of mine (and AFAIK one not explored by anyone else) is how a “subjectively good” synergy looks when measured. Especially dampening, current/voltage correlation, and other time-domain phenomena.

Frequency-Domain (FR, S/N, SINAD, etc.) could be interesting too, especially when looking at a headphone amp running into a real-world load and not a mostly-perfect resistance (or High-Z/open circuit), but that is a question to be answered when the cable mess turns into a repeatable test setup.

One idea to simplify my setup is to use a T-splitter and BNC-RCA adapter to loop AWG-Out and Scope-In-Signal together (close to the DUT). This would then turn this adapter into a glorified load-box, which would be fine in my book.

2 Likes
4

Update

Found a mostly repeatable setup!

  1. Set up this circuit:

    DS0008
    DS0008800×480 6.34 KB


    Which looks like so in practice:
    IMG_7115_2
    IMG_7115_22708×2000 986 KB


    This is not a “real world” test since the PSU is “aftermarket” and there is no load on the output.

  2. Set up the AWG to some frequency (here 1kHz Sine, 300mV P-P) to verify proper signal routing:

    DS0005
    DS0005800×480 11 KB


    Unity-Gain on the DUT results in the blue trace being barely visible:
    DS0006
    DS0006800×480 12.7 KB


    The point of testing at unity-gain is, to my understanding, to avoid the measurement equipment skewing the results due to different voltages making circuits act slightly different.

  3. FRA setup:

    DS0010
    DS0010800×480 7.48 KB


    Yes, the point density is arse not optimal. Still only doing prove of concept here.

  4. Running the test results in this:

    DS0012
    DS0012800×480 5.31 KB


    The blue line is the frequency response. Looks quite flat.
    Ignore the red line for the moment.

    FRA0001
    FRA00011064×470 32.9 KB


    Exporting the data reveals a bit more, also clearly shows how “not flat” the response is. And it shows how bad the 30 data points are resolution wise, especially past 10kHz where there are basically no datapoints anymore.

ToDo:

  • Finalise Adapter box design
  • Order in parts
  • Build the thing

Optional:

  • Look for BNC to TS adapters

Edit:

  • AWG - Arbitrary Wave Generator, a function generator that can synthesize most waveforms
  • FRA - Frequency Response Analyzer/Analysis
  • DUT - Device Under Test
4 Likes
5

Thanks for these, they’re always an interesting read if some of it is way beyond me. I do try to keep up and reference things you mention.

Question for you, on the unity gain front. How would you “know” by ear what’s close to unity gain volume for an amp? Or is it only a mathematical determination that gets one there?

3 Likes
6

Still finding my way here too. Feel free to ask.

I am sure some recording technician or mastering engineer can tell by ear. Best I can do is measure. When input voltage = output voltage, the amplifier is operating at unity (basically: Input Volt / Output Volt = 1). This can change depending on frequency though!

3 Likes