@mdp How does it sound? That amount of jitter ought to cause easily audible distortion.

@mansr qualitatively it sounds fine, I was surprised. I will get some oxygen free cables and retest. 😉

@mdp Fine is such a vague term. If you play a 10 kHz tone, you should be able to hear the difference compared to a clean device. Exactly what it sounds like obviously depends on the jitter spectrum.

@mansr yup, pretty early in my investigation. I have some more serious issues with noise on the mic first. of course, also hampered by the mainline mv64xxx i2c controller sometimes stalling. much fun.

@mdp looks fun! I haven't had the chance to prototype anything at work in while (though I've been doing it a lot at home.) For a lot of things it's becoming increasingly impractical. By time the HW group gets me schematics and they pass my review we usually end up able to spin boards that are at least 95% of the way there. Sometimes jumper wires are required or we find minor problems that get changed on the next board rev. I still miss hooking all it up and seeing it work first.

@georgem @mdp How would you plug a QFN device into a breadboard?

@mansr @mdp depends which QFN.


@georgem @mdp For that price you may as well spin a real prototype.

@mansr for $6 with immediate shipping? Could even get it next day if I needed. It's not a cost issue, it's a feedback time issue. In @mdp 's case the prototype hardware is blocked so it's quickest way to proceed. It's often possible to find existing boards with the bits you need to do your initial testing anyway. Believe it or not there are still people that etch their own boards just so they can turn a prototype in a few hours.

@georgem @mansr indeed it's often difficult to find the exact bits you need on a board, right number of xyz interface exposed, much less the right peripherals attached to them. and also, working Linux support for those peripherals. There's a lot of stuff claimed to be working...until you get into the details and find all the corners that have been cut in drivers.

@mansr @georgem I similarly use the (.us-centric) phrase of "it pays my kids' college tuition"... a lot.

@georgem I had to do an entire 3-4 month calendar time project having never received the custom hardware (until a few days go). all my audio processing work done on rpi3, one onsite trip to test some stuff on their board and my colleague testing my stuff on his custom hardware. Still waiting on the hardware/bsp supplier to deliver working ALSA drivers, the stuff they sent has periodic impulse noise that totally breaks AEC algorithms. nmfp :)

@mdp @georgem ALSA drivers for what? Most things should work using the simple-card dt binding.

@mansr @georgem referring to the production hardware which uses A2B. That driver is broken.

@georgem fwiw, all I'm doing here is trying to do a "quick test" of our audio pipeline on some different hardware from the broken custom stuff. not going so quickly atm.

Sign in to participate in the conversation
Society of Trolls

A nice little Mastodon instance. Mild trolling encouraged (keep it local), but not required. Malicious behaviour is not tolerated. Follow Wheaton's law and you'll be fine.