Is this a Bug? Weird trace directly via DAC on OS X

[Gaffer]

Before I bought my DAC I was measuring my HiFi response via REW on OS X (both updated) and a minDSP UMIK calibrated mic via the headphone out on the MBP. So signal path was:
REW (digital) -> MBP on board DAC -> Analogue out via headphone port -> 3.5mm to phono lead -> analogue signal into amp -> analogue to speakers

Here is the in-room response:

So when I bought my Arcam irDAC (upsampling 24bit/192kHz) I thought it would be great as most info would be kept in the digital domain till the end for a better trace:

REW (digital) -> MBP DAC not used, info still digital -> Digital out via USB -> digital USB lead to Arcam -> analogue signal into amp -> analogue to speakers

but this is the trace I get:

Where did all these bumps and spikes come from? If I subtract this from the top image, the resultant trace "added" by the Arcam looks horrendous wether I've set it to upsample to 192kHz or use 44.1kHz (in midi preferences):

Now is this a problem with the OS X software, the REW software or the Arcam DAC?

Is there some setting I have not enabled somewhere?

The DAC itself sounds great and doing sweep tones by ear, it sounds smooth all the way and definitely not lumpy/spikey as that trace would suggest.

Can anyone help as I'd like to continue using the REW via the DAC as it's easier to connect and doesn't rely on the less capable Apple on board DAC.

Thanks

 

[John Mulcahy]

Can you post an mdat file for a measurement? Are you using single sweeps?

 

[Gaffer]

I think on that i used multiple sweeps. the mdat includes tests with my BT receiver too.
I have attached the 192kHz DAC in 192kHz and mdat with DAC in 44.1kHz modes and one with various upsampling modes as well.
My BT receiver is a philips AEA2700 (MBP connected via BT) and my asyncronos DAC is an Arcam irDAC 24bit/192kHz (with pcm1796 chipset - MBP connected via USB)

 

[John Mulcahy]

Use single sweeps. The replay and record sample clock synchronisation required for multiple sweeps to provide correct results is usually only possible when using the same device for both.

 

[Gaffer]

Aah, thank you. That's it. Using single sweeps, now I'm getting proper traces using the Arcam:

Thanks for your help