Module ESP32-2432S integrated with Annex

[botman]

If you have changed R7 to 47K and left R8 at 0 Ohms and left R9 at 68K, you have lowered the overall gain of the amplifier by a factor of 11. That alone is a significant improvement, and it could be enough.

If you also changed R8 to 22K, you have lowered the overall gain of the amplifier by another 25% for a small improvement. Now the overall gain of the amplifier would be less than 1 (about 0.9).

It is my opinion that there is nothing to be gained by lowering the overall gain by another factor of 3 by changing R9 to 22K Ohms. It won't reduce any distortion, and it will limit the maximum output volume to about one third of the amplifier's potential distortion-free output capability.

I guess my old analog roots are showing.....dating myself.

[Fernando Perez]

My knowledge of electronics applied to audio is very rusty, but I think I can deduce, from the attached application note, that the gain of the amplifier will be twice as much as dividing Rf by Ri, in our case, R9 divided by R8. If these resistances are equal, the gain will be 2.
Looking at the diagram provided by Cicciocb, it is seen that to the initial design of the 8002B note, the board manufacturer has added two low-pass filters in cascade, consisting of R4/C11 and R7/C12, which should suppress frequencies higher than 33kHz.
A curious thing is that each filter cuts at the same frequency, since the component values are multiplied/divided by 10 in each case.
https://www.digikey.es/en/resources/con ... ass-filter
What I don't know how to calculate is how much resistors R4 + R7 attenuate the signal.
In any case, I am disappointed with the module in terms of audio, due to its design and conception it will never be possible to achieve a quality sound similar to that of an authentic external I2S converter.
And no matter how much you play with the gain, an audible background noise remains.
I think I'll dedicate it only to graphics, hoping that one day the touch pad fails. In which case, it would surely go to the memory drawer.

[cicciocb]

My knowledge of electronics applied to audio is very rusty, but I think I can deduce, from the attached application note, that the gain of the amplifier will be twice as much as dividing Rf by Ri, in our case, R9 divided by R8. If these resistances are equal, the gain will be 2.
Looking at the diagram provided by Cicciocb, it is seen that to the initial design of the 8002B note, the board manufacturer has added two low-pass filters in cascade, consisting of R4/C11 and R7/C12, which should suppress frequencies higher than 33kHz.
A curious thing is that each filter cuts at the same frequency, since the component values are multiplied/divided by 10 in each case.
https://www.digikey.es/en/resources/con ... ass-filter
What I don't know how to calculate is how much resistors R4 + R7 attenuate the signal.
In any case, I am disappointed with the module in terms of audio, due to its design and conception it will never be possible to achieve a quality sound similar to that of an authentic external I2S converter.
And no matter how much you play with the gain, an audible background noise remains.
I think I'll dedicate it only to graphics, hoping that one day the touch pad fails. In which case, it would surely go to the memory drawer.

I think that putting the resistor values as per the original schematic (as I did), the quality is not bad, sufficient for most of the applications. It is clear that the direct DAC output (noisy and only 8 bits) is very far from the quality with an external DAC but is cheap and very simple.
The more priced model 7" has an ESP32-S3 and the I2S dac/amplifier (only mono ) and the quality is really HI-FI.
After all, the module is very cheap, probably because is technically a bad designed board with, in addition, wrong resistors on the amplifier

[cicciocb]

I have a little riddle for you.
How do you make 3 devices work on the SPI bus with different connections when the chip has only 2 available SPI buses?

[Helmut_number_one]

Yes, you shouldn't expect too much.
But the 7" also have a larger pixel range and 8 Giga, is it supported by Annex?
Which display is it exactly?

[cicciocb]

Yes, you shouldn't expect too much.
But the 7" also have a larger pixel range and 8 Giga, is it supported by Annex?
Which display is it exactly?

I'm working on actually. The display is 800x480 LCD RGB parallel 16 bits

[botman]

I don't know the answer to your riddle about 3 devices being served by only two SPI busses, but since you are discussing display boards other than the 2.8 inch CYD, I would like to see if you would consider modifying Annex32 to support a display board that I bought recently. It is an ESP32-3248S035R CYD relative that has a 3.5 inch TFT. It solves your riddle by having the touch controller connected to the TFT SPI bus, so the other SPI would be available for the SD card which is connected on separate IO pins. I believe that the SD is on the same IO pin numbers that are used for the SD on the 2.8 inch CYD board. With the proper custom IO pin choices I have demonstrated to myself that the touch function works properly with the TFT screen without any wiring modifications. The display on this new board is 480 by 320 pixel resolution, which is twice the number of pixels on the 2.8 CYD. Another improvement on the 3.5 board is that the multicolor LED device is on the front, display side of the circuit board instead of the back. On AliExpress, if I buy 3 or more of the 3.5 boards at a time, they cost me less than $14 apiece including shipping. It should be easier to add support for this new inexpensive board with a larger TFT screen and more resolution than it was for the 2.8 inch CYD. Do you think that you would want to do that?

[cicciocb]

I don't know the answer to your riddle about 3 devices being served by only two SPI busses, but since you are discussing display boards other than the 2.8 inch CYD, I would like to see if you would consider modifying Annex32 to support a display board that I bought recently. It is an ESP32-3248035R CYD relative that has a 3.5 inch TFT. It solves your riddle by having the touch controller connected to the TFT SPI bus, so the other SPI would be available for the SD card which is connected on separate IO pins. I believe that the SD is on the same IO pin numbers that are used for the SD on the 2.8 inch CYD board. With the proper custom IO pin choices I have demonstrated to myself that the touch function works properly with the TFT screen without any wiring modifications. The display on this new board is 480 by 320 pixel resolution, which is twice the number of pixels on the 2.8 CYD. Another improvement on the 3.5 board is that the multicolor LED device is on the front, display side of the circuit board instead of the back. On AliExpress, if I buy 3 or more of the 3.5 boards at a time, they cost me less than $14 apiece including shipping. It should be easier to add support for this new inexpensive board with a larger TFT screen and more resolution than it was for the 2.8 inch CYD. Do you think that you would want to do that?

I'm curious to know what you plan to do with all these modules!
No, I don't plan to implement the support for all the modules, this can rapidly become a work for me with the unique benefit for the sellers.
After all, the value is in the software but the money is in the hardware because people assume that the software is free.
The problem comes after when people discover that without the software the product is useless (or very hard to use).
The demonstration is that you are ready to buy because it works with Annex32.

Anyway, as you said, that module already work as is (with the resistive touch controller) except for the SDCARD so I think that it worth to buy 3 of these units as the software is valueless (pardon free ).

[botman]

It's true that I have probably bought more of these CYD modules than I will use, but I plan to share some of them with friends, kids, and grandkids. My biggest interest is robotics. I have built two SpotMicro 3D printed robot dogs, two of my sons have each built one, and my granddaughter built one for a senior high school project last year. My latest one is controlled by an M5stack running Annex32.

20221002_111253.jpg

20221002_111246 (1).jpg

It can walk, and it uses I2C interfaced TOF distance sensor and accelerometer/magnetometer boards to walk around the house without bumping into things and to build a map of its travels. These CYD boards are significantly less expensive than the M5stack, and I like the larger display and the touch instead of separate buttons. The RTrobot 16 servo controller that I use can connect to the serial port like it does on the M5stack. With all of the options for I2C add-on modules, there is little need for additional directly connected IO. The audio is nice to make it bark and talk.

[Electroguard]

That is revolutionary... breeding robotic dogs with annex brains !