Calibrating Impedence measurement Rig (with Amplifier)

[challenge]

FIXED: SEE POST #8
Hello. I built a dedicated box to measure both frequency response of loudspeakers and also impedance, I also do full dual channel FR measurements with amplifier.

It's plugged like this when measuring speaker impedance with an amp:

I found a nice discussion (cannot post links yet, credits to dcibel at midwestaudio club forum) and followed the schematics there, as it pretty much a simpler ARTA jig without the mic section....

here is the current schematics of the box:

- HOW should I calibrate such a thing in REW? I've read the manual but I think that's for the simpler (just cables without amp).

I see now there are three calibration steps in rew:
1. open circuit
(This measurement compensates for gain differences between the input channels)
2. short circuit
(The measurement compensates for the series impedance of the test leads)
3. reference resistor
(The reference resistor measurement compensates for frequency response differences between the input channels)

or I could input Rsense, Rinput, Rleads.

What is the correct REW calibration procedure for such an amped rig?
I don't think I should short the amp without resistor in place
I have some ideas by reading what the cals do, but would like to know if there is a "best way".

thank you for this beautiful software.

 

[challenge]

ok i got it maybe?
in the manual Rsense is always present in the box and the reference one is another resistor just for the third calibration step...
I thought it was possible to calibrate somehow without another external resistor in this kind of jig, since it has the SPL/IMP switch (SW1 in the diagram above) that allows to bypass the internal Rsense,
---
Tried calibrating.
used the generator, 1khz tone, set the
ref is on scarlett's input 2
imp/DUTis on input 1
set the input gains at same level (within 0,1db, by checking in rew levels).

Rsense 10
Rinput 60000
Rleads 0,01

then start with calibration process:
open cal: no error, almost straight line at 100,3 ohms
short circuit: measurement input should be lower than reference... measurement peak is -23,6 dbFS, reference peak is -21,3 dbFS. Channels may be swapped or test leads not shorted.
Cannot do third step if don't pass second one.

Tried to do a measurement on a 4ohm woofer, and it tells me channels maybe swapped as meas peak is -26,9dBFS and reference peak is -21,3dBFS
If I manually swap the TRS inputs on the focusrite it still shows the same error/offsets, it just swaps the numbers.

double checked settings, they are not swapped. Input 2 is loopback in preferences, and input 1 is input.
same in measure window, input is on input 1, and ref is on input 2.

will double check the trs wirings inside the box to be sure there's nothing wrong there...

 

[John Mulcahy]

Follow the normal REW cal process. Do not bypass the sense resistor, the only thing that changes during the cal is the load - nothing for open circuit, then short the load leads together, then make a measurement with a known, non-inductive reference resistor.

 

[challenge]

Checked the wiring. All is as the schematics above.
Following the cal instructions.
When I do the second step (disconnect the speaker and short the wires), it tells me the ref input is not low enough and cannot move on the next step...
The ref peak is -21,6dBFS, the meas peak is -23,5 dBFS. So just 2 dBFS lower.

is it the 10ohm resistor or the particular cheap amp is am using or what could it be?

 

[sm52]

Try separating the impedance and frequency response measurements. There will be a few more items, but less confusion. My friend also loves versatility and boxes. And for half a year it has not been able to pass the second step of calibration.

 

[challenge]

solved.
now works perfectly, calibrates fully, zero noise, and difference between inputs when shorted is way bigger as it should.

 

[dcibel]

Hey that’s my schematic in your first post, you must have dug that from some real old thread! 1K resistor is just to connect amp common to sound card common, mostly an optional component, but ensures 0V = 0V when measuring.

Here’s the same schematic drawn nicely:

Dual Channel Measurement Jig - HTGuide Forum

Here's a schematic for a simple dual channel measurement jig, which can be used with ARTA, REW, or SoundEasy.

www.htguide.com

 

[challenge]

thank you for the schematic! yeah I've found multiple versions of that schematic via google with the resistor in different position and mixed/mashed a bit of it comparing to arta etc
then saw that version (another guy) with 4 switches and got confused a bit on too many sides, so i posted here.

to be honest the real issue was a ground issue when using my cheesy 12v amp and its transformer i was using...
tested using a commercial hifi amp with resistor and it works perfectly as it should. Resistor stays on... you never know unplugging cables etc

 

[dcibel]

thank you for the schematic! yeah I've found multiple versions of that schematic via google with the resistor in different position and mixed/mashed a bit of it comparing to arta etc
then saw that version (another guy) with 4 switches and got confused a bit on too many sides, so i posted here.

Glad to hear you've got it going. The linked document I wrote specifically because someone was trying to build a measurement jig following a disaster of an instruction and schematic from Audiojudgement, which sounds like it could be the one you are referring to as well. Good news is that whether you use REW, ARTA, or SoundEasy, the same jig can be used as the 2-channel measurement is essentially the same process regardless of software. With REW, latest 5.20.14 early access release should be used.

 

[challenge]

I started to get the hang of it. Also tried measuring caps and inductors and it works fine.

@dcibel i was thinking, while i have it all on desk... and since you all understand electrical circuits better than me...
what if I wanted to use your box to also measure passive crossover outputs (to double check they're still working as they should or to see what they are doing electrically and compare with simulation).

This is the ARTA recommended setup for measuring crossovers, basically just a probe and loopback :

If I replace the switch in your scheme with a three position one (or ON-OFF-ON) so that the "J4/Imp" probe doesn't get the amp signal when measuring crossovers... should be pretty much the same as the ARTA above, right?
I would still get the reference from amp on one probe, and the other probe would just get the crossover output?
and it can still measure SPL and Impedance in the other switch positions.

I would hook the crossover input to the amped signal (at J1), so that the amp goes both to the box and also the crossover, and J2 would be hooked to the filtered output of the crossover (e.g. TW, WF etc), as shown in the image below.
- do you see anything wrong with this?

first time measuring crossovers and just for fun, not my field! I'll have quite a bit to read first, but just want to have the hardware ready for it.

 

[dcibel]

No, don't do any of that, and diagram is incorrect, measurement needs to be from crossover output, as you've drawn you are still measuring amplifier output. No modification to the jig is required. Simply place the jig inbetween your crossover and speaker driver, like this:

You don't need to do anything with the impedance probe, leave it connected or not, doesn't matter as we won't measure that signal. Set REW to single channel mode "no timing reference", and input is the reference channel. Set the jig to SPL, imp probe will measure same signal as reference probe, so you can really use either as the measurement input. Measurement should be un-windowed, so make sure right window is set to 500ms+.

 

[challenge]

I get it now for crossovers. Even easier in single channel mode. Thank you for the settings, very useful.

Ok don't laugh. It all started when I saw the general measurement setup on arta user manual (below) and that's why I got creative thinking of two probes

I thought I could've used both probes like that. By rewiring Switch1 and leaving it open I would get two independent probes? one would measure the amp output as a ref of the signal getting into the crossover, and the other probe a measurement of what would be coming out of it... and the diff would be the filter that's why I thought of it that way...

----------------

I have another question on using this box for dual channel spl measurement with rew, I am following vituixcad's guide for dual channel measurement, but something seems off in my settings. When I use semi-dual with a TRS loopback it works fine. The full dual channel spl is off the chart and messy.
But let me quickly set up a speaker, double check it all first and then will ask something more precise.

 

[dcibel]

Sure, you could do that, but that measurement is just compensating for the frequency response of your amp and cabling, which is quite minor if the purpose of the measurement is mostly verification that you haven't wired the crossover incorrectly. Combination of this transfer function measurement, and impedance measurement of the crossover should provide enough confidence. Addition of the 3 position switch will do the task, but also increases possibility of presenting low impedance to your amp if you happen to put the jig in SPL mode in this configuration. Alternative would be to build the cable probes like in the ARTA document, resistors are cheap so total cost with cable can be around $10.

One alternative solution to provide the amplifier compensation would be to simply run loopback measurement (measure reference channel in normal connection arrangement without crossover), and load that measurement as a soundcard calibration. Then repeat with crossover in place.

I have another question on using this box for dual channel spl measurement with rew, I am following vituixcad's guide for dual channel measurement, but something seems off in my settings. When I use semi-dual with a TRS loopback it works fine. The full dual channel spl is off the chart and messy.
But let me quickly set up a speaker, double check it all first and then will ask something more precise.

I am unsure what the question is, please provide more detail. Make sure latest REW 5.20.14 early access is being used. You can measure each input, reference and mic individually to compare and see where the problem is. Dual channel "loopback as timing and cal" is really a differential between reference and mic input, so each measurement is self-calibrated.

 

[challenge]

Thank you for the explanation. I like the idea of loading the amp as sound card calibration, will write it down and do it that way as it seems safe and easy enough.

Ignore the second question, it all works fine after updating to the latest early access. Forgot to update it on windows as usually I use linux.

 

[The_Maverick]

Hi Folks. First of all I'd like to thank Marius for designing and sharing this schematic. Great work.

I have a simple question about the resistors. Finding a 10 Ohm 5W+ <1%+ Metal film for R1 is impossible, so are there any reasons why I shouldn't use a wirewound instead?

 

[challenge]

Marius (audiojudgement?) created the first image I used in the first post just to explain how i plugged things, as it was quicker to explain than writing a wall of text.

We're not using his 4 switch box schematics here.
the schematics in this thread are much cleaner, simpler and more accurate (the audiojudgement one has the tw cap in the wrong spot too iirc, so it measures wrong!). These schematics are made by @dcibel , different guy.
maybe should I edit all my messages here removing weird diagrams to avoid confusing new users?
---
about resistors...
well I used this 10w one, prob will never max it out but I had it around... https://mouser.com/ProductDetail/Vishay-Dale/NS01010R00FB12?qs=v4SOexxHmwFw9WlYGHGtgw==
the NS prefix means it's non-inductive styles (type NS) with Ayrton-Perry winding.

https://www.vishay.com/doc?30204

I think it's good? when I bought it I didn't read anywhere it had to be metal... I read it just had to be non-inductive. But it's not my field so wait for a wiser opinion.

----
I measured a lot of things with this box in these past days, including crossovers transfer functions (using the soundcard calibration trick is spot on!), RLCs, impedances, SPLs etc... and it just works great.
Zero issues. A must have.
Just have to understand what you're doing and how it has to be done. Calibrating everything etc
Having an amp, is not a toy anymore and requires more attention than the usual unamped cables...

 

[The_Maverick]

Thank you for that.

I thought that it was just the "Calibration Resistor" of a known value should be a non inductive type.

Should all the other resistors in the schematic be the same non inductive types as well?

 

[dcibel]

I use a 5w cement wire wound resistor. Inductive effect of these type of parts is so small, your leads from the jig to speaker will have just as large or greater influence on measurement accuracy. For impedance measurement, it is “calibrated out”.

Sense resistor doesn’t need to be specifically 10 ohm/5w, but something in the range of 10-50ohm, and at least 5w ratings if you want to do any testing above low level T/S stuff. Just use whatever you have lying around, it’s more important that you know the exact value of the resistor, so measure it with a good meter, or buy a resistor of decent precision to begin with. REW requires a second resistor anyway as a double calibration, a step I would prefer to skip but that’s another discussion.