The ESP32-Cam appears to have very severe operational voltage sensitivity…
I’m beginning to doubt the practicality of this device.
Here is my experience with trying to even mount this device on a simple prototype board.
For background, I’m just running the sample CameraWebServer example from the Arduino ESP32-Cam library.
Running from Adafruit FTDI directly connected with short leads. Supply Voltage 5.23V Current Draw 210mA → Camera running smoothly OK!
Running from Adafruit FTDI with short ProtoBoard pin jump & same short leads as before. Supply Voltage 5.24V Current Draw 170mA → Camera NOT running smoothly, running a bit but freezing often.
The voltage drop across the male / female pins on the ProtoBoard is essentially zero - With 5.3V on the FTDI being delivered to the female socket on the ProtoBoard as 5.3V
I also measured the resistance between the male/female VCC/GND pins at around 0.1 Ohm [including my measurement wires]
So I’m at a loss to know what is going on here - except to note that the ESP32-Cam [or the Camera] appears to be running at the outer limits of its voltage / current draw ability when running off 5V.
I would note that even running off my lab power supply @ 5V it again only manages to run smoothly with very short direct connections.
Is this everyone’s experience with the ESP32-Cam? If so I’m not sure this is a viable product.
I bought this to drive a robot car with vision, but I it can’t run - even in a simple ProtoBoard socket, then I can’t see this project continuing.
Any input, ideas, assistance here welcome.
Hello John. While I am a rank amateur to esp32, I managed to program 6 of the ai-thinker cams via a daughterboard that came with them rather than ftdi and they were running for most of the day delivering video (A little ‘noisy’) but okay.
Then I hung 2 off a usb 3 cable each straight from my computer, the usb cables were around 2 metres in length.
The others were plugged into various phone chargers and they all worked.
All of them were connected to the daughterboard (Also a programmer for them) as I didn’t have any ftdi boards and none of them missed a beat.
Is there a possibility that you might have a marginal one, possibly even bad?
Thanks for the quick reply!
Mystery solved! - But very annoying for my project
If I re-jump the ESP32-Cam to lift it back off the ProtoBoard, while still passing current through the protoboard board - problem fixed.
It would appear that the problem is associated with WiFi reflections off the metal in the Makerverse ProtoBoard - but I really want mount this ESP32-Cam in a protoboard to simplify the other connections associated with my robot car!!
I also have some Adafruit protoboards - Does anyone know if these also have the same type of WiFi reflection problems?
Well, I’ll be shaved like a sheep.
Reflections are possible with anything and given the fact that the cam boards usually use just the built in antenna, my logical (Amateur radio) brain would consider that a likely prospect.
I had not considered that problem with my projects mine as they dangled at the end of sheathed cables but it’s logical.
I also bought 4 external antenna connectors to put on some of my cam units to get them well away from the unit itself or any metal but forgot to consider my eyesight in relocating the 0 ohm resistor so i can use them. And everything else is packed away so can’t do it right now.
And would you believe how short the connecting wire is? I might have the same problem you do with that. There are much longer connecting cables that I had not considered when I set mine up. Found a link somewhere.
More to the mystery - Just tried an Adafruit Protoboard - No WiFi reflection problems , but narrower, only one accessible row of pin holes either side of the ESP32-Cam - Why I chose the Makerverse Protoboard. I’m not surprised by this as I have some other WiFi projects working OK on Adafruit boards.
Maybe I could try an external antenna with the Makerverse board.
Note to Core: You may want to cross check my finding here and there is possibly nothing that can be done, but this may well be a limiting issue with WiFi projects being built on a Makerverse Protoboard.
Also, if I am correct, this post may better be re-titled as a warning regarding the use of Makerverse protoboards with WiFi projects
Very interesting finds, how did you have the ESP32 mounted on the Adafruit protoboard? I’m no RF expert but if you have the WiFi antenna dangling off the end it should have a much higher chance of working.
If I remember my amateur radio theory correctly, an RF field around an antenna is shaped like a torus and near the middle of the antenna. The magnetic field is 90 degrees to that. Might help a bit planning where/how to position it in relation to anything.
External antenna would give more separation from any metal sources with a longer cable, not the short ones the dealers seem o provide.
My teacher might smack me in shame:)
I just poked the pins of the ESP32-Cam through the Adafruit board and connected VCC/GND from the other side.
The antenna was in the middle of the board and no issues were encountered. As I say, I have other WiFi projects mounted on Adafruit boards without any issues. I just wanted to cross check that there were no special issues with the ESP32-Cam, I had used Adafruit Huzza8266’s in the other projects.
There does appear to be something fundamentally different in the structure of the Makerverse & Adafruit protoboards in relation to their interaction with WiFi when the uP is mounted close to the board.
Hi Miki and all.
You have to factor in the cable loss. This might more than negate any improvement you get with antenna positioning.
The longer tracks on that Makerverse photo board are 65mm and 75mm which is a bit over 1/2 wavelength at 2.4GHz which may have an influence. Could be a help if the bluetooth antenna could be kept at 90º to them.
The shorter (outside) ones are closest to a resonant length so if you are not using them or can arrange not to use them try cutting them off, that is cut off that section of board.
Good point Robert
The fact that the Makerverse board has two 75 mm tracks straight down the middle of the board, directly under the antenna, may also be an issue with WiFi interference. The Adafruit board has similar 75 mm rails, but not down the center of the board.
I think this is probably the main difference, explaining why the Adafruit boards don’t exhibit similar WiFi interference.
That is probably correct. I am not familiar with the Adafruit boards so was not aware of the difference.
I highlighted the outer tracks as these are closer to a resonant length.
John and Miki
If you are not using these 2 centre tracks or can do without them and want to do an experiment drill out all of the holes. This will break the tracks up into small bits.
Be aware all !!! The Core Makerverse board is double sided. If you want to cut a track you have to cut both sides of the board.
If I choose to go with an external antenna, I have an old 2.4 GHz antenna from a decommissioned router
As far as I can tell, this looks like female [both inner & outer] RP-SMA connection???
I’m guessing the correct lead between my antenna and the IPEX connector on the ESP32-Cam is
Interface Cable RP-SMA to U.FL
I’m hoping the radio experts can let me know if I’ve got this wrong
Thanks for all your help.
Another, simpler, solution to the Makerverse - WiFi 2.8 GHz resonance problem
Lift the ESPCam-32 further from the Makerverse socket using a second set of Feather Headers.
This geometry appears to fix the WiFi resonance with the central Makerverse rails that was leading to unreliable / intermittent video transmission.
It is yet to be seen if this still works when I’m not sitting underneath my router - but I’m hopeful.
Very interesting discussion! I had a chat to @Brenton and if you are able poking the antenna off the edge of the protoboard the signal quality will increase a ton.
Is is analogous to designing a PCB for an ESP32 module and having a keepout area surrounding the antenna.
RF Microwave/EMF engineering is definitely a deep deep rabbit hole and a specialty!
Especially antennas, now that really is a black art.
A fair part of my working life involved working with and assisting the engineer (I think he was actually employed as a Scientist) who designed a fair percentage of the TV transmitting antennas in Australis. Now that is something that is difficult to experiment with. When it is erected it has to work. For instance Ch 9 in Sydney had 192 half wave elements on it and when installed measured a return loss of better than 40db (VSWR of approx 1.02:1). Anything better than this is arbitrary as the reference 50Ω load used was 1% (40db RL).
Aviation Nav Aid antennas also in the mix. All up a very interesting period in my working life and somehow I feel privileged in having a working acquaintance with some extremely smart people who did not mind one little bit sitting down and explaining, in terms someone with my pay scale could understand, anything I wanted to know. At some point we dabbled in everything pretty much from broadcast-HF-VHF right through to about 16GHz.
PS. There was also something we used which was an “Artificial free space”. Yes I too thought “you’re kidding” but this sort of thing exists and has a use. I have actually used such a thing.
Hi Liam & Robert
Thanks Liam for the pointer to the ESP32 WiFi datasheet.
I did test the ESP32-Cam module with the antenna hanging out from the end of the Makerverse board, and yes this is a working solution as you suggest.
I also suspect this is the better solution, but the geometry does not suit my install position, given the angle of dangle of the built in camera and some other design issues - so I’ll probably stick with my elevated Feather Solution for the moment.
But I’ll certainly keep your wise antenna advice in mind for any future projects.
You will find the Antenna connector information on page 22 here https://cdn.sparkfun.com/assets/0/9/7/b/f/DS-18583-ESP32-S2_WROOM_Module_-4MB__PCB_Antenna.pdf
EDIT: Sorry that link did not work. You will have to navigate to the data sheet from Liam’s reply above.
Sorry to bring up an older topic but I came across this in the ESP32 design guide:
It suggests the even the side of the boards that it sits on makes a difference, maybe shifting it a bit to the side that the antenna is fed on to the PCB might help.
Link to said document (page 17)