Teensy Laser Control

[TheHermit]

Hi Folks!
I've just started a project to use Teensy 4.1 + audio DACs to control laser projectors. I fell in love with the artform back in the 1980's. With today's inexpensive technology (~$1K per projector vs $30K for 1/2 the performance) I couldn't resist jumping back in.
Unfortunately, I've discovered a huge difference between professional and enthusiasts' control systems. The professional path begins with a $3K price tag for PC software and inline DACs to ILDA standard analog projector inputs, via either USB or Ethernet cabling. The enthusiast's path is lower priced, often voluntary efforts within the 'community', which means less commitment to customer service. So, I'm stuck with $200 DACs per projector, that drop connections, and $300 software that I can't upgrade.
So, Teensy is my intended solution to this dilemma. I can replace the PC software with a DAW and use a MIDI controller for a desk. I can run USB to a (future) Teensy board to generate the imagery and use it's audio DACs to provide all of the analog signals to the projectors' analog ILDA inputs.... except for one.
RGB laser projectors require discrete inputs for X, Y, R, G, B. So, the quad audio board doesn't quite meet the projectors' demand. So, I'm reaching out to the Teensy community for a workaround, of which I'm confident there must be one.
Secondly, once that hurdle is overcome, I would like to open the door to a lucrative opportunity for someone who is far more technical and ambitious than myself. IMO, there is a neglected market demand within the enthusiastic laserist community. All we need is a small PCB with a Teensy 4 and 5 in board audio DACs in order to fill that underserved market gap. Given that, most laserists with modest technical skills can customize their own control solutions, just as I am. How many hundreds of you are capable of designing PCBs and having them produced by PCBway?
JS, for whatever it's worth.
Meanwhile, regarding my task at hand, any guidance towards adding a 5th audio DAC to the existing quad audio shields would be most appreciated.😎

[MarkT]

As I understand it laser controllers need DC-coupled signals, so you need to be aware some audio DACs and CODECs
have in-built high-pass filters designed to remove DC offsets, though typically these are configurable. DC offsets
can damage speakers and it is common to remove these in an audio signal path (digitally or in analog, like the Teensy
Audio adapters which have DC blocking capacitors on linein and lineout).

[TheHermit]

As I understand it laser controllers need DC-coupled signals, so you need to be aware some audio DACs and CODECs
have in-built high-pass filters designed to remove DC offsets, though typically these are configurable. DC offsets
can damage speakers and it is common to remove these in an audio signal path (digitally or in analog, like the Teensy
Audio adapters which have DC blocking capacitors on linein and lineout).

Thank you for mentioning that issue, MarkT. You're correct that DC offsets are necessary in a show environment with sequences of offset 'beam zaps', different projection areas, sweeping flat scans, etc.
DC offsets are also normally used for each of the RGB driver inputs. As a workaround, I'll probably need to use hi-freq PWM outputs with some smoothing capacitors to achieve similar functionality. This is another reason for offering the PCB design to others, who would be able to replace the audio DACs' opto-coupling with DC coupling for a commercially viable product.
However, within my own living room, I'm mostly projecting complex Lissajous imagery above 20 Hz onto a single screen, so I'm only losing ~10% of the total repertoire of imagery. That's sufficient to allow me to develop the code for the waveform generators and their MIDI controls.
Here's a similar approach with Christopher Short's Radiator controller, which is also 'limited' from DC offsets. Yet, he won the 1st place in single projector abstracts at the 2019 ILDA awards with this entry. Last night, I successfully tested 4 quadrature waveform generators, with an LFO for AM, mixed together into X/Y outputs on my o'scope, using my T3.6 with its audio DAC. IOW, the project already has a heartbeat.
Thank you for the heads up.

[TheHermit]

Hi MarkT,
I've searched the forum and haven't found anything regarding 5/6 channel audio for the T4.x. Looks like you're the expert, so here's my question:
I've just ordered a T4.1, quad audio shields, & a PT8211, but would feel much more confident with your reassurance, before soldering it all together. Can the 4 channel audio shield(s) and the PT8211 coexist by simply running a jumper from the PT8211's DIN connection to a T4 pin, other than pin 7, then change that pin definition within the Teensyduino sketch?
Secondly, I've only chosen the PT8211 because the audio shields appear to be limited to a maximum of two I2C addresses, which are consumed in the quad configuration. I prefer to keep the full I2C functionality on X & Y imagery outputs, but need to keep those points synchronized with the RGB outputs, one being from the PT8211. Do you foresee any problems with this configuration?
You mentioned that the in-built high pass filters are configurable. Is that via software or hardware mods? Any pointers to that documentation for the T4 audio shields would be most appreciated.
Thank you for your feedback and guidance.

[Frank B]

https://www.pjrc.com/teensy/SGTL5000.pdf

CS42448 has 6x in and 8x out channels - but not sure if it would work for lasers(?)

[TheHermit]

Thanks Frank!
It's always very helpful to start with RTFM.
The only difference with lasers is that the projectors have VCAs that drive either diodes or galvanometers, instead of speaker cones. The laser diode drivers have 0-1 VDC and the galvos have line levels, just like audio. I don't need any digital sound processing, because that's done as part of the image creation process. IOW, your audio world is far more complex than mine.
After skimming the SGTL5000 docs, I didn't see any software controllable settings to allow DC offsets through the DACs, but only ADC filters, which I see that I can disable (thanks). But, I see two 1uF caps on the Line_Out_R & Line_Out_L, pins 11 & 12, respectively. Is DC coupling simply a matter of replacing those caps with jumpers?
The CS42448 looks like a great candidate for a future Teensy Laser PCB. But, not being an electronics engineer, I have to wonder how much of a rabbit hole I would be jumping into, only to emerge a year later. "Well that's how you learn!" Yeah, haven't we all been down those roads before?
Nevertheless, great info, Frank :-)
Didn't see anything regarding software changeable I2C addresses for the SGTL5000. So, I assume that brings me back to attempting to add the PT8211 to the Quad audio board configuration. Are you able to confirm that it is as simple as running a jumper from the PT8211's DIN to a spare pin on the Teensy 4.1, then altering the sketch?
Am I really only a few no-brainer mods away from (potentially) having a fully functional, 8*Quadrature oscillators, with 5 mono-channel DACs outputting XYRGB line levels?
FYI; I already have the T3.6 communicating with my APC40 MIDI controller and Cakewalk DAW! WooHoo! The parts for the prototype hardware are on the way, then simply convert it to PCB, and flush out the software for the finale'.
Gotta love the world of Teensy, including you guys here, actively monitoring this forum!

[Frank B]

The CS42448 is supported by the audio library - and the digital connections are the same as with any other I2S chip.

But you're right, I'd probably just use the PT8211, too.

[TheHermit]

The CS42448 is supported by the audio library - and the digital connections are the same as with any other I2S chip.

But you're right, I'd probably just use the PT8211, too.

Thanks for the info, Frank.
That makes the CS42448 even more attractive for a PCB. Yes, it was the prospects of a DIY I2S interface that had me the most concerned.
Again, sorry to persist, but:
a) Is DC coupling only a matter of replacing the line output capacitors; and
b) Do I only need to reroute the PT8211's DIN to another T4.1 pin to allow it to coexist with the quad audio shields?

[Frank B]

I don't know.

[TheHermit]

I don't know.

Fair enough. I appreciate your honest reply, rather than providing baseless assumptions. I appreciate the great resources you've already provided.
Perhaps someone else will chime in with the answers I need.
Best regards.

[PaulStoffregen]

Are you able to confirm that it is as simple as running a jumper from the PT8211's DIN to a spare pin on the Teensy 4.1, then altering the sketch?

You can't just choose any spare pin. Only certain pins can transmit digital audio.

The pin functions are documented in 2 places. First is the pinout card, which comes with Teensy 4.1 and can also be found on the Teensy 4.1 product page, and also this page.

https://www.pjrc.com/teensy/pinout.html

The digital audio pins are shows in yellow.



While this serves as reference material, the compact format doesn't really communicate the finer details of exactly how you can really use these pins. For that sort of detail, you need the 2nd main documentation source.

The audio design tool is where the audio library software features are documented.

https://www.pjrc.com/teensy/gui/

If you haven't used the design tool, this 31 page tutorial is a good place to start. There's also a walkthrough video which closely follows the 31 pages of content, if you prefer watching a demo.

https://www.pjrc.com/store/audio_tutorial_kit.html

You might skip to part 2 which begins on page 8, if you just need to learn how to use the design tool.

As you click on each item in the left side, you'll see its documentation appear on the right hand side. All of the input and output features have a "Hardware" section which explains which pins are used.

As a practical matter, if you will connect 2 different audio chips, you will need to create a design using this tool which the software features that communicate with those chips. So before you buy parts, before you start connecting wires, use this tool to make your plan.

When you drag the various features onto the canvas, if you put 2 or more onto your design which can't work together due to a hardware resource conflict, you will see little yellow error triangles appear. A message with a (probably terse) explanation of the conflict might appear too. The point is this can help you to immediately know whether you have chosen outputs which can work together, and you can see the documentation about which pins each uses. That documentation panel offers quite a lot of other info too, which can really help you. Best to read it carefully.

[PaulStoffregen]

One other minor point. I2S is mentioned above. The audio shield with SGTL5000 chip always uses I2S.

CS42448 can use either I2S or TDM. Normally TDM is used, because it requires fewer signals and works just as well. If you will use this board or make a similar design, know that the connections are routed for TDM, not I2S.

Of course you can find TDM fully documented in the design tool, and it's hardware conflict detection is aware of the reuse of the same "SAI" hardware for either I2S or TDM (but not both) on each separate SAI port. So please, do yourself a favor and spend some time experimenting with the design tool and reading the documentation panel for the features you will use. It can save you a lot of trouble by showing you the details and alerting you early on to ideas which would end up causing a hardware resource conflict, and when you do come up with a conflict-free plan, you'll be able to choose the correct pins for connecting the various digital audio signals.

[Frank B]

I2S_quad and PT8211_2 would work. But that needs all pins (ok, without input) from the 2nd I2S (AKA "SAI") port.
However I have no idea if removing the capacitors from the shields is ok.

[PaulStoffregen]

None of these audio DACs are really designed for DC output. They may give some DC offset when they're supposed to be zero.

Most of them also use an internal voltage reference which is "stable" (at least in an audio bandwidth sense) but not "precise", so the exact voltage you get for full scale output is difficult to know and may vary from chip to chip.

Trying to use these parts anyway for a DC coupled control signal will probably require a carefully crafted circuit. You might need trim pots to calibrate / adjust the DC zero and full scale gain.

The fully differential output of CS42448 might be easier, since you'll know the 2 pins are supposed to be the *same* voltage when the output is zero, even if that exact voltage is unpredictable. Maybe a difference amplifier circuit could work? It might need to be followed by an amplifier where you can adjust the gain slightly, so you get the full scale output to the 1.0V level you want.

[TheHermit]

You can't just choose any spare pin. Only certain pins can transmit digital audio.

The pin functions are documented in 2 places. First is the pinout card, which comes with Teensy 4.1 and can also be found on the Teensy 4.1 product page, and also this page.

https://www.pjrc.com/teensy/pinout.html

The digital audio pins are shows in yellow.



While this serves as reference material, the compact format doesn't really communicate the finer details of exactly how you can really use these pins. For that sort of detail, you need the 2nd main documentation source.

The audio design tool is where the audio library software features are documented.

https://www.pjrc.com/teensy/gui/

If you haven't used the design tool, this 31 page tutorial is a good place to start.

https://www.pjrc.com/store/audio_tutorial_kit.html

You might skip to part 2 which begins on page 8, if you just need to learn how to use the design tool.

As you click on each item in the left side, you'll see it's documentation appear on the right hand side. All of the input and output features have a "Hardware" section which explains which pins are used.

As a practical matter, if you will connect 2 different audio chips, you will need to create a design using this tool which the software features that communicate with those chips. So before you buy parts, before you start connecting wires, use this tool to make your plan.

When you drag the various features onto the canvas, if you put 2 or more onto your design which can't work together due to a hardware resource conflict, you will see little yellow error triangles appear. A message with a (probably terse) explanation of the conflict might appear too. The point is this can help you to immediately know whether you have chosen outputs which can work together, and you can see the documentation about which pins each uses. That documentation panel offers quite a lot of other info too, which can really help you. Best to read it carefully.

Thank you for the info, Paul, and I'm honored to receive your personal attention.
Yes, I've already covered most of the info you've pointed me towards, plus searched throughout the forum. Bu, am still unable to find the answer to my 2 seemingly basic questions.
I've already used the design tool to make 4 quadrature waveform generators, with mixers on their X & Y outputs, using T3.6 & the audio shield.
The design tool is a great GUI that makes my dev work not only much easier, but possible from the get-go. I have it bookmarked for my project because it is so useful. The T4.x pt8211_2_2 indicates no conflicts with the quad audio board's SGTL5000 nor the i2s_quad1, which is why I've already ordered those parts.
I will review the documentation you've suggested, before making any pinout alterations.
Gotta commend you for your efforts, Paul. Your work is the most impressive I've seen throughout the world of MCUs.
Thank you.🙏
Kindest Regards.

[TheHermit]

One other minor point. I2S is mentioned above. The audio shield with SGTL5000 chip always uses I2S.

CS42448 can use either I2S or TDM. Normally TDM is used, because it requires fewer signals and works just as well. If you will use this board or make a similar design, know that the connections are routed for TDM, not I2S.

Of course you can find TDM fully documented in the design tool, and it's hardware conflict detection is aware of the reuse of the same "SAI" hardware for either I2S or TDM (but not both) on each separate SAI port. So please, do yourself a favor and spend some time experimenting with the design tool and reading the documentation panel for the features you will use. It can save you a lot of trouble by showing you the details and alerting you early on to ideas which would end up causing a hardware resource conflict, and when you do come up with a conflict-free plan, you'll be able to choose the correct pins for connecting the various digital audio signals.

Thank you again for your time, Paul. Yes, the CS42448 looks very good as a future PCB candidate, but unfortunately, that's above my knowledge level, at the moment.

For your convenience, I've included the Audio Design layout for my project. Hope this helps you to understand my objective.

[TheHermit]

One other minor point. I2S is mentioned above. The audio shield with SGTL5000 chip always uses I2S.

CS42448 can use either I2S or TDM. Normally TDM is used, because it requires fewer signals and works just as well. If you will use this board or make a similar design, know that the connections are routed for TDM, not I2S.

Of course you can find TDM fully documented in the design tool, and it's hardware conflict detection is aware of the reuse of the same "SAI" hardware for either I2S or TDM (but not both) on each separate SAI port. So please, do yourself a favor and spend some time experimenting with the design tool and reading the documentation panel for the features you will use. It can save you a lot of trouble by showing you the details and alerting you early on to ideas which would end up causing a hardware resource conflict, and when you do come up with a conflict-free plan, you'll be able to choose the correct pins for connecting the various digital audio signals.

Thank you again for your time, Paul. Yes, the CS42448 looks very good as a future PCB candidate, but unfortunately, that's above my knowledge level, at the moment.

For your convenience, I've included the Audio Design layout for my project. Hope this helps you to understand my objective.
At the moment, I have the X & Y QOSCs & mixers working on the T3.6 & audio board and controllable via usbMIDI with an Akai MIDI controller and outputting Lissajous imagery on my o'scope.

[TheHermit]

One other minor point. I2S is mentioned above. The audio shield with SGTL5000 chip always uses I2S.

CS42448 can use either I2S or TDM. Normally TDM is used, because it requires fewer signals and works just as well. If you will use this board or make a similar design, know that the connections are routed for TDM, not I2S.

Of course you can find TDM fully documented in the design tool, and it's hardware conflict detection is aware of the reuse of the same "SAI" hardware for either I2S or TDM (but not both) on each separate SAI port. So please, do yourself a favor and spend some time experimenting with the design tool and reading the documentation panel for the features you will use. It can save you a lot of trouble by showing you the details and alerting you early on to ideas which would end up causing a hardware resource conflict, and when you do come up with a conflict-free plan, you'll be able to choose the correct pins for connecting the various digital audio signals.

Thank you again for your time, Paul. Yes, the CS42448 looks very good as a future PCB candidate, but unfortunately, that's above my knowledge level, at the moment.

For your convenience, I've included the Audio Design layout for my project. Hope this helps you to understand my objective.
At the moment, I have the X & Y QOSCs & mixers working on the T3.6 & audio board and controllable via usbMIDI with an Akai MIDI controller and outputting Lissajous imagery on my o'scope.

[TheHermit]

My apologies, Paul. I assumed that your link was only to the IC, instead of a finished PCB. Perfect!
Thank you!

[TheHermit]

None of these audio DACs are really designed for DC output. They may give some DC offset when they're supposed to be zero.

Most of them also use an internal voltage reference which is "stable" (at least in an audio bandwidth sense) but not "precise", so the exact voltage you get for full scale output is difficult to know and may vary from chip to chip.

Trying to use these parts anyway for a DC coupled control signal will probably require a carefully crafted circuit. You might need trim pots to calibrate / adjust the DC zero and full scale gain.

The fully differential output of CS42448 might be easier, since you'll know the 2 pins are supposed to be the *same* voltage when the output is zero, even if that exact voltage is unpredictable. Maybe a difference amplifier circuit could work? It might need to be followed by an amplifier where you can adjust the gain slightly, so you get the full scale output to the 1.0V level you want.

Thank you, Paul
Now that I have 2 T4 audio shields, a PT8211, & 3 CS42448 boards on order, plus my existing T3 audio shield I'll have plenty of opportunities to experiment to determine what works. Meanwhile, I've found a laserist forum that mentions replacing the caps with resisters on commercial audio amps to achieve the same objective. Thank you for the link to the differential amp circuit and suggestions.
Lots to do, lots to learn. Your feedback has been very helpful towards providing me with the tools I need to proceed. Thank you so much. :-)
BTW, OSH Stencils are making a me pair of stencils for the CS42448 board. Hopefully, they will also provide me with the Gerber files, which I would like to share with the community. Would GitHub be the preferred destination for those files?

[TheHermit]

Thanks for the info, Frank. Good to know that's an option. However, I've just discovered Adafruit's down and dirty quad audio DAC that is already DC coupled.
https://www.adafruit.com/product/4470
So, I'm considering to use 3 channels from that to drive the DC RGB color mixing offsets.
Then, I could still use the PT8211 audio DAC for the scanners' X&Y signals.
The Adafruit's I2C address is 0x60.
But, I can't find any specs on the PT8211's I2C address to know whether there's a conflict. Do you have that info?
Otherwise, the spec sheet on the SGTL5000 says:
"Pg23
For the 32 QFN version of the SGTL5000, the I2C device
address is 0n01010(R/W) where n is determined by
CTRL_ADR0_CS and R/W is the read/write bit from the I2C
protocol."

I interpret that to mean, depending upon whether CTRL_ADR0_CS is either high or low, then the I2C address would be either:
0001010
0101010

Does that translate to:
06
26
Or, in Japanese, would it be:
0x60
0x62
Therefore, if the SGTL5000 is configured as with my final assumption, then it should be compatible with the Adafruit's quad DAC board's address of 0x60?
IMHO, either of these options could suit my needs.
What do you think, Frank?

[TheHermit]

Thanks for the info, Frank. Good to know that's an option. However, I've just discovered Adafruit's down and dirty quad audio DAC that is already DC coupled.
https://www.adafruit.com/product/4470
So, I'm considering to use 3 channels from that to drive the DC RGB color mixing offsets.
Then, I could still use the PT8211 audio DAC for the scanners' X&Y signals.
The Adafruit's I2C address is 0x60.
But, I can't find any specs on the PT8211's I2C address to know whether there's a conflict. Do you have that info?
Otherwise, the spec sheet on the SGTL5000 says:
"Pg23
For the 32 QFN version of the SGTL5000, the I2C device
address is 0n01010(R/W) where n is determined by
CTRL_ADR0_CS and R/W is the read/write bit from the I2C
protocol."

I interpret that to mean, depending upon whether CTRL_ADR0_CS is either high or low, then the I2C address would be either:
0001010
0101010

Does that translate to:
06
26
Or, in Japanese, would it be:
0x60
0x62
Therefore, if the SGTL5000 is configured as with my final assumption, then it should be compatible with the Adafruit's quad DAC board's address of 0x60?
IMHO, either of these options could suit my needs.
What do you think, Frank?

On the other hand, would the 4 channels of the Adafruit DAC consume four I2C addresses, including 0x60, 0x61, 0x62, and 0x63, and still conflict with the reconfigured SGTL5000's address of 0x62?
If that's the case, then my final straw to grasp for would be to utilize the T3.6's onboard DAC to provide the 5th audio channel output, along with the 4 channels from either the SGTL5000 or Adafruit's DC coupled DAC.
"Are we there, yet?"🤔

[Frank B]

On the other hand, would the 4 channels of the Adafruit DAC consume four I2C addresses, including 0x60, 0x61, 0x62, and 0x63, and still conflict with the reconfigured SGTL5000's address of 0x62?
If that's the case, then my final straw to grasp for would be to utilize the T3.6's onboard DAC to provide the 5th audio channel output, along with the 4 channels from either the SGTL5000 or Adafruit's DC coupled DAC.
"Are we there, yet?"🤔

I2C is incredible slow compared to I2S. And if the adresses conflict, you have find a way around it...

[Dmax]

HI TheHermit I'm looking at building a controller for my home made laser and would like to you know if you have moved forward with your project ? just in case you didn't know the signal from the controller to follow the ILDA standard it should output differential signals for the Galvo's so most DAC boards made for music/audio will work but you will not get the full range of deflection from the galvo's so I'm looking at designing a 6 channel DAC board with a 12 bit dual channel DAC for the galvo's and two 8bit dual channel DAC's for the RGB signals and obviously one channel spare for now. anyway I just wanted to say I'm looking at getting something together as and when I have some free time so I will let you know how I get on.

[TheHermit]

[QUOTE=Dmax;300443]HI TheHermit I'm looking at building a controller for my home made laser and would like to you know if you have moved forward with your project ?(/QUOTE)

Well, pleased to meet you. Thank you for the reply and info. Yes, I'm plodding ahead with my project. The T3.6 is pushing out 14 waveforms, between 100Hz and 600Hz, with 2 LFOs for AM. It will be piggybacked with the ILDA correction amps, either near to or inside the projector(s), with communications via usbMIDI, which should be fine for short distances at home.

[QUOTE=Dmax;300443] just in case you didn't know the signal from the controller to follow the ILDA standard it should output differential signals for the Galvo's so most DAC boards made for music/audio will work but you will not get the full range of deflection from the galvo's(/QUOTE)

Yes, I'm aware of that and have purchased some uni-polar correction amps from Laserboy (w/ signals converted to differential ILDA standards by the show cards). But, I'm also waiting for3 ILDA standard differential conversion amp boards from HankLloydRight on PLF to arrive. He just posted his new ESP32 ILDA DAC and sold me the ILDA output amps.

so I'm looking at designing a 6 channel DAC board with a 12 bit dual channel DAC for the galvo's and two 8bit dual channel DAC's for the RGB signals and obviously one channel spare for now. anyway I just wanted to say I'm looking at getting something together as and when I have some free time so I will let you know how I get on.

My only obstacle is the T3.x audio shields not being DC coupled. Been searching everywhere for a solution. Tried shorting across the caps, but seem to have fried 3 audio shields,,, so far. I'm guessing that a high impedance buffer needs to be added to limit the output current?
I'd really appreciate any guidance towards a solution that you might be able to provide. Otherwise, your DACs sound like exactly what I need.
Apart from that, the custom MIDI desk with 2 7" Nextion touchscreens, RGB backlit rotary encoders and buttons are breaking the budget. But, wat's the point of skimping on one's own dream system, right?
BR