REW LF bandwidth

[Newmarket]

Anyone knowledgable about low frequency (subsonic) behaviour of REW ?
Summary of scenario:
I'm measuring the frequency response of my computer audio I/O interface. So I measure the 'loopback' response. Looks reasonable = low frequency roll off with some ripple, some action at 50/100/150 Hz (my UK mains is 50Hz) and 'normal' hf ripples before the curve dives down at approx 20 kHz.
So then take that as a "Calibration File".
Then taking another loopback measurement and applying the calibration file - the hf response flattens out , the mains related frequencies persist to varying degrees (I am understanding this as the current reading is phase uncorrelated with the original sweep wrt mains interference). But while the very LF - say below 20Hz looks straightened out in terms of the basic "dB per Octave" response but the "ripples" remain.
Is this an artefact of the REW measurement technique at LF. Am I missing something ? I've spent a fair bit of time wit audio test kit - AP2 / Prism / Neutrik - but quite some time ago and always willing to learn.
Thanks for reading.

 

[ruffrecords]

Is this an external interface or one built in to the computer?

Cheers

Ian

 

[warpie]

Increasing the length should improve the ripple. You won't get it totally flat though. I don't believe it's REW's fault. I get similar behaviour on both REW and fuzzmeasure.

 

[Newmarket]

Is this an external interface or one built in to the computer?

Cheers

Ian

The converters are in external boxes and interface to a PCI DSP board in the PC via TDIF. It's a Soundscape Mixtreme based system if anyone remembers it . It has later Mixtreme card that is compatible with later PC motherboards. But it's "ancient" in PC terms - maybe 2010 or so - and only really viable under Windows XP. The whole Soundscape thing went thru various owners from the Sydec originators to Mackie to SSL who stopped support although using it as the basis of their multi I/O Alphalink etc.
The demise of TDIF is a real shame imo. Very simple and intuitive interface. Can literally split ribbon cables to route etc. Try that with ADAT or USB etc

 

[Newmarket]

Increasing the length should improve the ripple. You won't get it totally flat though. I don't believe it's REW's fault. I get similar behaviour on both REW and fuzzmeasure.

Length of what ?

 

[warpie]

Length of what ?

Sweep time.

5.5sec or more

 

[Newmarket]

Length of what ?

Thanks. Got it. I'll take a look. Only recently got round to running REW so not accustomed to the available options.

 

[trobbins]

You could try using averaging (at least 8x) to prepare a a baseline response to use as the cal file. You could even modify the cal file if you were keen, to remove glitches etc that then apply themselves to every new response made. You can try and start at the lowest frequency selectable - John made me a version that started at 0.1Hz to check some choke impedances once, but I recall it is nominally 1Hz. You can try a lower sample rate setting for your driver preference - my EMU0404 chipset uses a different HPF setting so I get better LF response with 44kHz compared to 192kHz. Do you have signal output options - circuitry may use different filter parts, but you may have to use a scope to measure actual response, as most meters don't go down nicely below say 5Hz. Similarly for signal input options.