Enginerding with Weeb Labs

[Weeb Labs]

A thread on which to share my plentiful electronic and electromechanical OSHA violations shenanigans. Engage safety squints.

I have recently been working on quite an elaborate video for the YouTube channel (which will hopefully be complete in a few days) and some of the shots required motion control, for which I hastily cobbled together a rig.

Several steppers and controllers very hastily thrown together but the result was pixel-perfect repeatable motion for the multi-pass shots which required it. No gearing for the steppers but the microstepping in conjunction with loading were sufficiently smooth. For the programming, I took the brute force route rather than attempting to make use of an existing motion control library. The code will probably end up on Github soon.

I am now down to the final few shots but as usual, unexpected complications have arisen. The AVR chosen for the demonstration doesn't seem to be producing any audio via the newly installed post-DSP SPDIF output.

BCK is there, MCK is 24MHz (512*fS), LRCK is fS(48KHz) and the SPDIF output scopes as expected when muted but there is no actual audio. First course of action will be to replace the RCA connector, just in case it happens to be defective.

There is also the possibility that my XDA-1 isn't happy with the WM8804's transformer isolated output, so I will also be trying a TTL output via TOSLINK in the event that the connector isn't the culprit.

More at 11.

 

[Weeb Labs]

As it turns out, there is nothing wrong with the AVR or the SPDIF output. The XDA-1 simply didn't like the 8804's output. Time to move on and obtain those final few shots.

I was fairly certain that this would be the case but it is always a good idea to know for a fact prior to publishing instructional material.

Update:

It's done. It is finally finished. Just a few things to tidy up and it's ready to upload! 40 days of production.

 

[Weeb Labs]

Extracting vectors from a service manual and stylizing them as a little addition to the flowchart sequence. They are all individual strokes, which means that I can selectively highlight or animate them.

 

[Weeb Labs]

After what feels like an eternity, it is finally live!

 

[pozz]

It's really cool, visuals are excellent but I have some complaints about sound:

• The pitched down trailer voice is only funny for a bit. I kept thinking at some point you would drop it and resume in your normal voice.
• Voice is desynced from image. Latter comes late.
• Terrible sibillance (worse because of the pitch down) on f and s syllables.

Warning drags on too. (Kind of like a presenter reading a ppt slide.)

 

[Weeb Labs]

It's really cool, visuals are excellent but I have some complaints about sound:

• The pitched down trailer voice is only funny for a bit. I kept thinking at some point you would drop it and resume in your normal voice.
• Voice is desynced from image. Latter comes late.
• Terrible sibillance (worse because of the pitch down) on f and s syllables.

Warning drags on too. (Kind of like a presenter reading a ppt slide.)

I appreciate the feedback but there is no pitch modification in this video at all; that is simply my voice.

Regarding the sync issues, I suspect that may be a problem on your end, as I have just skimmed through the video for a sanity check and the audio is always perfectly synchronized. Perhaps you could provide a timecode?

I'm not sure about the sibilance; it sounds correct to me, both on my room corrected monitors and headphones.

You're definitely right about the split second of drag which occurs on a couple of sentences. It required quite a number of takes to get this close to the desired result, so I really had to leave some of it as it is.

 

[voodooless]

Very nice video! I do think the lipsync is a bit off as well though, which might be the reason @pozz thought you changed your voice.

One question though: what about volume control? These AVRs tend to use analog volume control chips as you also showed in you video. That means that the AVR does no longer control volume of your main channels now. Seems kind of an issue for. Multichannel content?

 

[Weeb Labs]

Very nice video! I do think the lipsync is a bit off as well though, which might be the reason @pozz thought you changed your voice.

One question though: what about volume control? These AVRs tend to use analog volume control chips as you also showed in you video. That means that the AVR does no longer control volume of your main channels now. Seems kind of an issue for. Multichannel content?

After discussing this with a friend, we arrived at the possibility that the perceived sync issue might actually be an illusion resulting from the apparent mismatch between my appearance and voice, together with the close micing.

Regarding the AVRs, volume might be an issue depending upon one's use case. If the unit in question is an ancient AVR being used solely as a DSP, volume control will be handled by the source and this won't be a problem. For multichannel applications, there are several solutions with one of them being a couple of QuadVol board. I actually use a couple of these already, as they are a convenient means of automating volume control in conjunction with an ESP32.

For situations wherein a PC is the multichannel source, this is also not really a problem.

If you are particularly adventurous, you could even sniff the SPI data destined for the onboard volume controllers and route it to an external QuadVol.

 

[pozz]

The mismatch stands out across two laptops and a phone. I ripped it and after applying 60ms delay to the audio it seems more natural.

There's a definite boost in bass and low mids and the sibilance is pretty broadband between 2kHz—8kHz. There's also some clarity issues in the midrange. Some of this might be a difference in pronunciation, but to me it spells heavy compression (not the knee; rather too low a threshold and too long a release). Looking at the waveform separately seems confirm that. Some quick peaks get through but almost everything else stays in very defined bounds.

If I'm wrong because there's no compressor in the chain, I apologize and hand in my ear/brain combo for repair.

 

[Weeb Labs]

The mismatch stands out across two laptops and a phone. I ripped it and after applying 60ms delay to the audio it seems more natural.

There's a definite boost in bass and low mids and the sibilance is pretty broadband between 2kHz—8kHz. There's also some clarity issues in the midrange. Some of this might be a difference in pronunciation, but to me it spells heavy compression (not the knee; rather too low a threshold and too long a release). Looking at the waveform separately seems confirm that. Some quick peaks get through but almost everything else stays in very defined bounds.

If I'm wrong because there's no compressor in the chain, I apologize and hand in my ear/brain combo for repair.

Having checked the video against the waveform in multiple spots, I can't find anything that looks or sounds out of sync to me. That wouldn't make sense either, as the audio capture was genlocked to the camera and all of the offscreen test references are perfectly synchronized. I do tend to enunciate by leading pronunciation, so perhaps that is what you are seeing. Could you provide a timecode example for me?

You are absolutely right that there is compression and EQ on the track but the bass is actually slightly rolled off via -1dB low shelf at 110Hz. There is a 3.5dB high shelf at 1.8KHz and the 200Hz-1KHz region is reduced by 6dB. I agree that the release is a little bit long but I mastered this to my preference and liked it at the time.

You keep your ears and brain for today.

 

[ernestcarl]

No out of syn issue here on my end. I turned off FIR filtering to my monitors and sent the signal directly to my DAC to get as little delay as possible. The voice still sounds very bass-y after turning off my sub (monitors high-passed at 120Hz), which I do think says more about the characteristics of the microphone being used just not being the best for your voice.

 

[pozz]

200Hz-1KHz region is reduced by 6dB

Ahh. Now I see. A broadband reduction in the mids is the same as boosting the highs and lows. You can actually see this in a real time FFT and watch how much less energy there is in the midrange. There's even significant energy at 70Hz!

I don't have video editing software (really not my area). I used VLC and your intro and outro spiels for reference, playing with the sync tool. 60ms audio delay felt about right, but not exact.

If you like you can record yourself snapping your fingers or clapping, run it through the same chain and see if you're satisfied.

 

[Weeb Labs]

I picked up this little 14" monitor yesterday without knowing very much about it, as there was no trace of it on the internets. It turned out to be a 560 TVL dot mask toob with fewer than four hours of total runtime, RGBS, Y/C and CVBS inputs.

It's a lovely little monitor from a completely unknown manufacturer, so I am currently shooting a short video for it under the new "Wee Blabs" series.

 

[Weeb Labs]

Slightly work-related delays but I am back into production, now with the advantage of realtime grading via an Ultrastudio capture device. Here's a little bit of experimentation with light shaping. Quite a fun monitor to light, with its white cabinet!

The aesthetic really pairs well with an Amiga, so one of those will definitely feature in the final product.

 

[Robin L]

I appreciate the feedback but there is no pitch modification in this video at all; that is simply my voice.

You've got a future in voice-over work.

 

[FrantzM]

You've got a future in voice-over work.

That is my opinion as well. I was kind of surprised by such a professional voice and enunciation coming from a YouTube video for AV DIY.
A bit OT.

Great video.

 

[Robin L]

That is my opinion as well. I was kind of surprised by such a professional voice and enunciation coming from a YouTube video for AV DIY.
A bit OT.

Great video.

Like Ken Nordine, the sound would be more effective as a disembodied voice than attached to a face. Not that there's anything wrong with the face, but there's that ever popular disembodied "voice of authority" going on here.

 

[Weeb Labs]

Glad that you both liked the narration!

Here's a little bit of fun on which I have recently been working. I have taken an EP94Z3E (HDMI/VGA synchronous splitter), passed its 0.3V vertical sync pulse line through a comparator and then into an NE555, when produces a phase and duty cycle adjustable PWM output.

Next, I will be using this synchronized PWM output to drive a pair of transistor bases in order to modulate the LED lamp of a cheap projector. In other words, I am implementing a BFI function for projectors. This will most likely feature in an upcoming video.

Input sync pulses below; output PWM above.

 

[Weeb Labs]

My pseudo-BFI implementation was quite a success. Here is a pursuit shot of this cheap projector without any of my modifications.

And here is one with my pseudo-BFI in effect.

Quite an improvement! Very close to CRT MPRT, here.

 

[maverickronin]

Quite an improvement! Very close to CRT MPRT, here.

Damn...