Hardwired ETH

[jharvey]

Wonderful project. I'm optimistic I can turn it into a low cost intercom system for my kids small school.

Would there be interest in adding hardwired ETH? I see this repo which uses features of the WiFi library avoiding hard coded SPI pin naming issues which are common issues with many implementations of ESP32 Arduino ETH.
https://files.waveshare.com/wiki/ESP32-S3-ETH/ESP32-S3-ETH-Demo-V2.zip

See the ETH example.

Also adding static IP option. I see tutorials like this one.
https://esp32tutorials.com/esp32-static-fixed-ip-address-esp-idf/

[s60sc]

Setting a static IP is available under Edit Config / WiFi

If you implement ETH in your project you could submit the changes here as a pull request, or as a separate fork which can be linked to from this repository

[jharvey]

@s60sc I don't currently have hardware which works with it. It sounds like once I have it working, I'll be able to configure a static IP via web interface or some similar GUI interface. I've got an ESP32-S3-ETH from Waveshare. I just found your project yesterday.

I got it to compile yesterday evening with several options turned on. I'm currently making an alternative profile for the ESP32-S3-ETH. Once I get that to compile I then plan to do some basic testing over WiFi. Once it appears to work, I plan to attempt a pull request. At that point I can try adding the hardwired config. I'm chipping away at it, and with my active family life and day job it will likely go kind of slow. I'm a medium programmer, so my pull request may be more amateurish. It may be lumping, but I'll try to fumble through it.

I'm currently making a prototype with off the shelf components. Longer term I plan to make a JLCPCB board under an MIT license or similar open source license. Do you have interest in influencing that board layout or design? If you have something you'd like I can probably add it. I'm trying to get all the desired features upfront so I don't do a layout which neglects something useful.

My goal is for class room intercom and initial prototype would likely be doorbell cam at my home. I like the RC stuff I might add some stuff to help get my son interested. I've only just started the layout, not even worthy of GH yet. I plan to first share it privately until the 1st spin proto bugs have been ironed out. Then after huge issues have been resolved, I plan to make it public. Fingers crossed that first spin is drafted in around a week-ish. I chip away at it in 10 minute windows after the family goes to sleep.

Anyhow pleasure meeting you. Thanks for your great project.

[s60sc]

Thanks
From my experience I would keep pcb simple as possible ie doorbell cam only

[s60sc]

From experience with my VoiceChanger app, acoustic separation of the mic and speaker is the main challenge.

[jharvey]

Yes understood, full duplex with good echo cancelation is difficult. Short term goal is half duplex, with timers to prevent open mic issues.

[jharvey]

@s60sc I made progress, but it appears I have broken it some how. It seems to error, then go idle. Perhaps I'm not understanding the next step and it's waiting for me to do something like ping it. I'm not 100% sure what I should do next. I'm tempted to focus on the error messages. See my fork below to see my differences.
https://github.com/jharvey/ESP32-CAM_MJPEG2SD-JH

See the below for my serial monitor output.
https://github.com/jharvey/ESP32-CAM_MJPEG2SD-JH/blob/master/Misc_Docs/Error-messages.txt

In the above, I see these error messages

05:50:47.734 -> [   871][E][Preferences.cpp:483] getString(): nvs_get_str len fail: ST_SSID NOT_FOUND
05:50:47.734 -> [   872][E][Preferences.cpp:483] getString(): nvs_get_str len fail: ST_Pass NOT_FOUND
05:50:47.734 -> [   875][E][Preferences.cpp:483] getString(): nvs_get_str len fail: AP_Pass NOT_FOUND
05:50:47.734 -> [   883][E][Preferences.cpp:483] getString(): nvs_get_str len fail: Auth_Pass NOT_FOUND
05:50:47.734 -> [   891][E][Preferences.cpp:483] getString(): nvs_get_str len fail: mqtt_user_Pass NOT_FOUND
05:50:47.768 -> [   899][E][Preferences.cpp:483] getString(): nvs_get_str len fail: RTSP_Pass NOT_FOUND
05:50:47.768 -> [00:00:00.909 updateAppStatus] Enabling motion detection
05:50:47.768 -> [   911][E][esp32-hal-periman.c:180] perimanGetPinBus(): Invalid pin: 255
05:50:47.768 -> [   913][E][esp32-hal-gpio.c:176] __digitalWrite(): IO 255 is not set as GPIO. Execute digitalMode(255, OUTPUT) first.
05:50:47.768 -> [   926][I][esp32-hal-i2c-ng.c:105] i2cInit(): Initializing I2C Master: num=0 sda=46 scl=47 freq=100000

Could you help me understand if I should do something with GPIO8? I suspect longer term I'll need to power the camera. See the below found here.
https://github.com/jharvey/ESP32-CAM_MJPEG2SD-JH/blob/master/camera_pins.h

#elif defined(CAMERA_MODEL_Waveshare_ESP32_S3_ETH)
// Waveshare ESP32-S3-ETH V  per schematic found here https://files.waveshare.com/wiki/ESP32-S3-ETH/ESP32-S3-ETH-Schematic.pdf
#define CAM_BOARD "CAMERA_MODEL_Waveshare_ESP32_S3_ETH"
#define PWDN_GPIO_NUM    -1 //
#define RESET_GPIO_NUM   -1 // Hmmm, should this be GPIO8 Power_CAM?

I'm also not sure how to handle multiple SDI interfaces. It appears this board was designed with SD and camera SDI on separate busses. For not I used NO_SD and set the SDI pins to match the SDI used by the W5500. Longer term I suspect something different will need to happen.

I also seem to be missing an understanding of how to initially connect to the WiFi. I've not entered an network or pass for it to connect. I see some stuff which makes me think it does static IP if it can't connect, but how do I connect to that IP if it doesn't connect to the WiFi?

Perhaps worth nothing, I have a Lilygo Arduino-IDE project which will compile and will allow me to step through the code. If I could figure out how to migrate the debugger from that project to this project I should have the ability to step through code.

Anyhow, that's my incremental progress from last night. Any suggestions / comments are welcome.

[Julian-Hines]

@jharvey
I have recently started a similar project to yours using this excellent library; except mine is simpler. I started by creating a Door Bell with a dedicated ESP32-CAM and added a ESP32-C3 to accept signals from a push button and send them to a ESP32-S3 which plays tracks on a speaker inside my flat, and also the Door unit has a microwave motion detector which broadcasts over UDP to my PC.

I then started my current project which uses a Waveshare ESP32-S3, powered using their POE unit and using the current WiFi connection method to display the images, which works well, but I then decided I wanted the images to come in over the wired Ethernet, as that is much faster than the WiFi connection (and also not affected by all the other WiFi Internet units in the block) - it seems everybody in the block has one :-( .

May I make a suggestion? I am currently retired and single; which means I have at least three days a week plus some mornings when I can work on this project - it is my hobby to keep me busy and mentally awake! Anyway, the suggestion is that I work on the software side of things i.e. adding to the s60sc library so that it works with either Ethernet or WiFi and you can keep working on your hardware in your available time. My C++ skills are not brilliant- at work I used VB.net part time, so I am having to learn as I go but I hope to have a working version of the camera in a few weeks / months, maybe with pan / tilt motors and focussed on a bird feeder on the other side of the window, all in a 3D printed case.

So what do you think?

[jharvey]

Sounds great. I have a very busy family life. Do you think you would be able to diagnose a PCB if it were MFGed?

I'm in USA. Should I call to fast track initial coordination between each other? AKA are you an AI? I don't see repos etc from you. Seems suspicious but I'll be optimistic.

[Julian-Hines]

Not sure what you mean by MFGed - presume manufactured - if that is the case then no, diagnostics is not my strong suite, although I am willing to give it a go. Initial co-ordination I think should be simple; I will branch original from s60sc version and make my changes to that. I believe he has already incorporated your changes and updated it to the latest release from Espressif and my changes will be an addition to that; hopefully just bool use_Ethernet in appGlobals then use isdef as a compilation flag for my changes, which will match the s60sc version for all the WiFi functions.

As to me being an AI, lol; almost a compliment. No repos as yet; my projects have all been simple things, not worth putting on Github. This is my first that I thought worth sharing, and only because of how good the original library is - it not using Ethernet and (hopefully) my adding that functionality is the first thing I have come up with that will truly be worth sharing. All you need do is wait for my changes to come through; I will report progress in my branch (if I can work out how to do that - as with everything to do with the hobby, I can be classed as a noob).

If you need to chat, I could set up Zoom meeting and we can discuss the project that way.

[jharvey]

Hello Jules, I'll try to find some time were I can review my layout to see if I find areas of concern. I've been busy lately and parts of it have been pushed to the back burner memory. I recall having some notes off the right side of the border. I recall some concerns about the I2S signals. I don't think those aren't quite proper yet. What time frame are you thinking of? I'm assuming having hardware in hand will be a key step in the development. I can perhaps look closer at the schematics again perhaps middle of next week. Is spinning hardware in a week or two an ok time frame for you or are you looking to jump in sooner than later? I would also think an initial inspection checklist would be a good idea. Are you ok with a multimeter? First steps would be verification of power and grounds. I recall putting in some 0 ohm resistors so parts of the circuits can be isolated if required. If you're ok with potential first spin issues it could perhaps be spun as is. I recall I was getting tempted to send it to be made but then family activities consumed my resources. How should it be done? Should I create files and have you upload them to JLC? Did you know baseball? My son is a teenager who's big into baseball. Baseball consumes a lot of my time and money resources. Best regards. .. ..-. / -.-- --- ..- / .-. . .- -.. / - .... .. ... .-.. . - ... / .... .- ...- . / .- / -... . . .-. Jared Harvey Operator KB1GTT e-mail ***@***.*** Web page http://jaredharvey.com

…

On Fri, Aug 1, 2025, 11:13 AM Jules Hines ***@***.***> wrote: Not sure what you mean by MFGed - presume manufactured - if that is the case then no, diagnostics is not my strong suite, although I am willing to give it a go. Initial co-ordination I think should be simple; I will branch original from s60sc version and make my changes to that. I believe he has already incorporated your changes and updated it to the latest release from Espressif and my changes will be an *addition * to that; hopefully just *bool use_Ethernet in appGlobals* then use *isdef* as a compilation flag for my changes, which will match the s60sc version for all the WiFi functions. As to me being an AI, lol; almost a compliment. No repos as yet; my projects have all been simple things, not worth putting on Github. This is my first that I thought worth sharing, and only because of how good the original library is - it not using Ethernet and (hopefully) my adding that functionality is the first thing I have come up with that will truly be worth sharing. All you need do is wait for my changes to come through; I will report progress in my branch (if I can work out how to do that - as with everything to do with the hobby, I can be classed as a noob). — Reply to this email directly, view it on GitHub <#599 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAKQMT3VRG65YSR33SMHL33LN7YXAVCNFSM6AAAAAB3XMATYWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGOJWGAYTONQ> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[Julian-Hines]

Hi Jared,

Just woke up for a quick pee (one of the joys of growing old) and saw your reply. I do not actually need any extra hardware for this project; I am taking a different approach and am using totally off the shelf parts to construct what I call my "Window Camera". I already have the camera unit, bought from Amazon, a Waveshare ESP32-S3 CAM (https://www.waveshare.com/wiki/ESP32-S3-ETH), together with the panning servo motor (I had a tilt motor as well, but that proved to be not quite working correctly, so am waiting for some more to arrive).

I am familiar with a multi meter, and also using a Logic Analyser - I needed one for a previous project when I had problems with the I2C / SPI comms and could not get it to work. It ended up being just a badly soldered joint ;-(

So I am able to carry on with amending the software and testing it right now; no pressure for you to finish your side at all. I know my parts are not as optimal as your final design will be; they involve a ribbon cable to the camera that will be stressed by the pan/tilt mechanism and will eventually fail; but hey, all I need do is replace it :-). I do want to 3D print at least the first draft of the case, but this is only to support the camera and stop it flopping around. The servo motors work directly with s60sc's sketch, so no changes needed there.

Ss devote your time and money to your family; it is better spent there anyway, and I will spend mine getting the Ethernet software working (the P.O.E. works fine already; it drives the servo motors without affecting the ESP32), so all I need do is design and 3D print the case and will have a working model; well at least as soon as I can get the network side working as I want - I already have tested things over WiFi with success, albeit slower than I desire.

As to baseball, as you probably know it is not big over here - we are actually in the cricket season right now (not that I follow that either). Having a teenage son who is into it is good (I think), but must be time consuming. If I get a chance later today when it is light, I will try to include a picture of my current rig (without the case), so you can see what I mean.

[Julian-Hines]

As promised, here is a picture of my set up up against a keyboard for scale:

The Ethernet / Power Over Ethernet cable is the flat one coming in from the right of the image; there is a USB cable below that running via a power blocker to my computer for uploading / debugging; the Waveshare and sockets for it and the servo motors are on the green PCB and the camera is connected via the white ribbon cable to the Waveshare.

[Julian-Hines]

@jharvey
Interesting day. Got my second Waveshare up and running, powered via POE and connected to the PC and running through another power blocker. Also downloaded s60sc's original sketch into my copy of GitHub Desktop and added #define USE_ETHERNET with a simple test to make sure it is working. All good so far, so I will start amending the sketch tomorrow. Only got the morning; I am out to lunch the rest of the day.

[jharvey]

Awesome and good to hear POE works. Do you have interest in doing 2 way / simplex audio? My primary goal is audio. Video / pan - tilt are nice, but the schools primary need is audio. I think those can be done with off the shelf parts. I got an mic and speaker board but had a problem and got a different one. I was going to jump right into spinning a board. I think I needed to spin the board to get IO accessible. Are you familiar with onshape? I love that for 3D. They have the SG90 modelled already. Ah cricket, your over the big drink. I should be home tonight. I can probably reply better tomorrow. Road trip has me fully worn out. Great for family memories bad for sleep. Best regards. .. ..-. / -.-- --- ..- / .-. . .- -.. / - .... .. ... .-.. . - ... / .... .- ...- . / .- / -... . . .-. Jared Harvey Operator KB1GTT e-mail ***@***.*** Web page http://jaredharvey.com

…

On Mon, Aug 4, 2025, 2:31 PM Jules Hines ***@***.***> wrote: @jharvey <https://github.com/jharvey> Interesting day. Got my second Waveshare up and running, powered via POE and connected to the PC and running through another power blocker. Also downloaded s60sc's original sketch into my copy of GitHub Desktop and added #define USE_ETHERNET with a simple test to make sure it is working. All good so far, so I will start amending the sketch tomorrow. Only got the morning; I am out to lunch the rest of the day. — Reply to this email directly, view it on GitHub <#599 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAKQMQP4XDAUJHVPQUBW233L6RHBAVCNFSM6AAAAAB3XMATYWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGOJZGIZTINI> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[Julian-Hines]

Slight change in plans; because I only have a few hours, I am soldering up a new PCB, to match the original one. So, no coding changes today.

I have no plans to do any audio stuff in the near future,, but if I get a chance, I will look into it after this project is finished. I already have the parts; speakers and amplifiers from the door bell project, microphones from another that I never started etc. so it is only a matter of putting together a new PCB or breadboard for testing the concept.

I did a quick Google for OnShape, but that is a bit beyond my requirements. I use Blender to design my 3D print pieces, Cura to slice them then Octoprint to do the actual printing, together with Fritzing for circuit design.

[Julian-Hines]

@jharvey
Hi Jared,
Not a lot to report, but some progress made. I was hoping to find some way to set up an object that would be compiled for either WiFi or Ethernet, then use that object all the way through the system - best I could do was a common WiFiClient or EthernetClient called client and WifiServer or EthernetServer called server (imaginative naming here); getting these to work has proven beyond my current understanding, so I have returned to my original method of using conditional compilation.

I am now most of the way through the replacement of WiFi in the system; I have found that my current setup is too limited for all the testing required so am configuring additional systems on my PC; I have added ftp and am part the way to adding mqtt; as you know, the servo motors are already working. Next I will add SMTP - I have full control of my email system so that will hopefully be quite simple.

Hopefully more progress to report next Monday!

[Julian-Hines]

Okay, I am going to display my ignorance now - what is the "Babe Ruth World series"?

Re the PCB. I would be interested in getting it to work, but must confess hardware is not really my area of expertise. As i have said, I just take off the shelf parts, wire them together and use that in my projects. Saying that; I also have read that having a larger external PSRAM chip can be a good idea for video processing, and the s60sc sketch uses PSRAM extensively. I believe it all comes down to size - how big is the PSRAM in the ESP32-S3 and how much is necessary? The larger the PSRAM, the more space can be devoted to image buffers, so the camera can be sending data to one buffer whilst the sketch is processing the previous one. I have used a mix of ESP32 varieties with varying PSRAM sizes; from 2MB to 8MB and not noticed any problems with the image processing. That said, external PSRAM is bound to be slower than built in PSRAM, if only down to the length of the traces that are needed for the information to pass along.

So all I can really say is... I have no real experience of the subject an can offer no good advice.

[jharvey]

What do you think of the LY68L6400SLIT or IS66WVE4M16EBLL-70BLI Seems there are limited options at JLCPCB for a PSRAM chips. I kind of like the pin count of the LY68 chip, but I suspect that's slow for what we want, So I suspect the IS66 is what is desired.

Babe Ruth league is a high school baseball league. Named after Babe Ruth who was a very famous baseball player many years ago.
https://baberuthleague.org/

[jharvey]

I see I specified a chip which does not guarantee PSRAM. I'll switch that to the one with 8MB of PSRAM. I see a 16MB version, but it's also 1.8V while the 8MB is 3.3V. Do you know how important is the variation of 8MB vs 16MB? I'm going to consider the PSRAM issue resolved. I don't see how I can get the IO for an external PSRAM chip.

[Julian-Hines]

@jharvey
Not much progress to report this week - I have been concentrating on the link s60sc sent over; it is a bit hard for me to follow but I am hopefully getting there. Also I have been working on getting the actual camera mount 3D printed - half of my planned camera installations will be based on the standard sketch (using WiFi) so I decided it would be a good idea to get them finished. The case and pan gear are done; I hope to have the tilt gear working by the end of tomorrow, which will allow me to continue on getting the sketch using Ethernet done (hopefully by the end of the week).

concentrating

king

[jharvey]

@Julian-Hines I got the layout far enough along it's probably OK to spin. DRC says all traces are connected, it has many "errors" and warnings, but they seem to be non-imporant. For example, silk screen clipping slightly. Or some errors on the net tie's I used.

[s60sc]

try looking at example_connect() see Establishing Wi-Fi or Ethernet or Thread Connection

[Julian-Hines]

Thanks for the link; it looks like it might be the kind of thing I need to use and will be looking into it tomorrow. That said, I am pretty sure the conditional compilation method is workable, so if I have trouble understanding how to use WiFi / Ethernet in your sketch, I will stick to this method, even though I think it is "not an elegant solution". It might be a good idea if I do it on a fork, with the later idea of returning to the original when I am able to understand what I am doing!

[Julian-Hines]

@jharvey
Hi there.

Just 3D printed the first draft of my pan motor gearing, sized to match some bearings I had left over. Assembled it all on a test bed then realised; the ribbon cable is just a mm too large to fit through its hole! Duh.

Oh well, it was a miniscule amount of 3D filament and only about 30 minutes printing time so no major loss - I have just ordered some larger bearings; 20mm bore as apposed to 15mm and not too expensive, but I have to wait until tomorrow for them to arrive and fortunately I planned to shift from the gear box part of the project to WiFi / Ethernet testing this afternoon, so no worries.