Need help with REW Calibration: Different results with different amplifiers

[EdenCerberus]

Hi Community, this is my first post here. I will keep it short with bulletpoints to not waste your time.

What do I want to do?:
Designing speakers with REW.


What equipment do I have?:
Soundcard: Behringer UMC204HD (audio interface for REW)
ARTA Switch Box (self made)
Microphone: Sonarworks SoundID Reference Measuring Microphone with cal file
Mic-Amp: Behringer UV1 as Mic-Preamp
Amplifiers: Yamaha A-S700, Auna Amp VT and some chinese "AK35"-BT-amp.

unrelated: And some Dayton stuff back from my beginnings (OmniMic V2 and DATS v3)


What I've done so far:
- connected everything together
- calibrated the audio interface (loopback)
- didn't added the Mic-Cal-File yet to make sure it doesn't influence any calibration.



What's my problem?
I can't procede in the impedance calibration.

The 'open circuit cal' with my Yamaha looks like Kermit the Frog’s EKG on ecstasy, while with the Auna AmpVT and the cheap Chinese 'AK35,' it runs nice and flat.
I thought it had something to do with the ground, so I used a wall plug adapter without grounding for testing purposes, but it changed nothing.
However, I can’t get a satisfactory result with the 'short circuit cal' on any of my amps. The lines are totally squiggly and in different frequencies. And when I measure again, they totally different to the previous measurements.
With the Yamaha, it says, the measurement level is too low (the live Input(FR)-preview on the lower left doesn't really move) and with the other amplifiers REW says that the reference input and the measurement input are basically the same (even with 27 Ohm RSense?!?) The difference is 0.1 dB.

I run out of ideas. Do I need an Amp with a specific schematic? Afaik every amp should work. Could this be a problem with the UMC204HD-Interface? (btw: my internal sound card gave me the same results). I wasted more than 2 Weeks worth of evenings and didn't get any closer to a solution.

Every help and every hint/idea will be much appreachiated!

Thank you!

 

[John Mulcahy]

REW's cal process differs from ARTA. If you use an ARTA switch box it should remain in the normal impedance measurement configuration during REW's calibration steps, you only make changes to the connections to the load. You also need to use an amp whose negative output terminal is ground, you can't use an amp with a balanced output.

 

[EdenCerberus]

Hello John ,

thank You for responding .

I just found out that none of my amps are connected to ground. The Yamaha uses the typical PC-Cable, but only has two pins in the inlet at the back side. The other two amps have only 2 pins for the wall outlet.
All amps are connected via RCA to the audio interface. No balanced audio outputs.

My ARTA Switch Box, in the "impedance mode", takes the Speaker Signal, weakens it via the voltage dividers and loops it back to my interfaces inputs. The only difference is, that the measurement-circuit (in contrast to the reference circuit) has the 27 Ohm RSense Resistor inbetween. So I don't think the problem is in the Switch Box.

I've tested my setup on my wifes notebook in battery mode (no power cord attached), but the result was the same.

Am I the only one having graphs like this in the calibration?

Slowly but surely I'm going crazy

John, what would you do to narrow down the problem?


BTW: In case it may help, some additional info and a screenshot from my impedance measurement window:

Resistance between "Speaker ground" and "RCA-Ground":
Yamaha: 0 Ohms (fully connected)
China-Cracker AK35: 12 Megaohms
Auna AmpVT: 0.1 Ohms

Our power grid isn't that stable. If my wife turns on the hair dryer, all LED lamps are flickering and if one of the neighbours is switching on the pool heater, our lights go dimmer. May this be the cause?

Thank You!

 

[John Mulcahy]

By having a ground output on the amp I was meaning a single-ended amplifier, not that it has an Earth connection. Most switching amplifiers have floating outputs with an inverse signal on the negative terminal, those can't be used. To narrow down the issues run calibration without an amp to see if that works.

 

[dcibel]

John is quite right, amplifiers that used a bridged output cannot be used with this type of measurement jig. Easy check would be to run output from your audio interface directly to the amp, turn on the Generator in REW with 60Hz size wave. Check the amp output terminals with a multimeter, from amp output to ground. Red terminal should have AC voltage reading, black terminal should be at or near 0V. If both terminals show the same voltage reading, amp is no-go for this purpose.

For double confirmation before getting too carried away, run a measurement sweep (not impedance measurement) of each left and right input individually (single channel measurement, ie no timing reference) with the jig set for impedance measurement and no load connected. Both input measurements should produce nice flat lines just like a loopback. This is effectively a loopback measurement, with the amp and jig included in the loop.

You're setup is a bit overly complicated to be honest. There's really no need for an external mic preamp, the UMC204HD has a perfectly well functioning preamp, so you can remove the extra preamp, the mic selection switch and R5 from the jig. To switch from impedance to mic input, simply disconnect the impedance probe from the UMC204 and plug in the mic. Leave reference input connected at all times. Cal switch is not needed either to disconnect the load. Simply disconnect the lead wires at the speaker driver, and complete calibration with the lead wires included, ie short at the lead end, not at the jig output.

Here's what I've been using for years, it works very well. No pcb required, the resistors can simply be soldered directly to the TRS connector jacks.

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

From your PCB layout it looks like some labels are incorrect. I would expect, H5 connects to the right input on the interface, and H7 connects to left input.

 

[EdenCerberus]

Hello dcibel, hello John again...

I found the problem... And I have to admid: I'm stupid!

I used quarter inch TS-kables (guitar cables) to connect my "H5" and "H7" to my interface... I soldered the wires on the terminals the false way. So Tip was ground and Sleeve was the signal. I wasn't used to these kind of terminals and I soldered everything up around 3 AM, so I wasn't really fit in the brain anymore and wanted to be finished asap.

Here are the results now (with my non-bridge-Yamaha):

I'm so ashamed right now that I would like to delete this thread but on the other hand it can serve as an information source for others, fiddling around in the middle of the night.

Are the measurements ok?

Thank You for your time and effort!

 

[John Mulcahy]

Yes, looks good, though a bit noisy at high frequencies. There's a noise filter option on the Measure dialog that may improve that, worth trying. Repeat the ref resistor cal with the noise filter on low to see how that looks.