Poetcatullus, I cannot answer your exact question as I have no experience with the Sonic Frontiers or Stello units. I have used a Proceed CDD in my system since 2005 with wonderful results. My inspections show it to be a very well constructed transport.
According to Madrigal literature it uses a suspended Phillips CDM 12i industrial transport (all metal) and uses a proprietary anti jitter circuit. The jitter circuit (crystal oscillator) is placed just prior to the digital out, rather than within the transport as in most transports, decreasing jitter relative to those units. The Madrigal proprietary control software for the transport runs on a combination of an Analog Devices DSP chip and XIlinx field programmable device (FDP), the DSP chip is also used to process audio data (timecode, track, index info). Once this data has been read, its presence in the output data potentially increases jitter, with the Madrigal system this data is stripped from the output.
Madrigal also mentions the CDD is optimized for AES/EBU output, my listening sessions with perhaps 15 different digital cables (both RCA and AES/EBU) confirm this.
I am curious as to how the CDD differs from the Mark Levinson No. 37, I can't find anything in the literature that points to differences other than cosmetic, perhaps some functional. I would love to open up a ML No. 37 and examine them side by side, perhaps there is something different in the power supply or such. Anybody out there compare the two side by side? I've seen CDDs go as low as $600 vs. around $1800 for the ML unit, depending on the extent of differences, if any, the Proceed seeems like a damn good deal to me.
In the meantime, I don't think you can go wrong with the Proceed CDD as long as you can be sure about functional issues. I've heard of some transport issues (failing to read discs) with both the Proceed and ML units. My transport was replaced with latest software updates just prior to my purchase back in 05.