I have been designing a PCB board to drive a Mercedes Benz PWM radiator cooling fan. The board will take inputs from up to 3 themisters. Water Temp, AC condenser temp and optionally under hood temp. The schematic and board layout is presented below. An Arduino Nano 3.3BLE will be used to control the fan
Automotive power comes in from connector JP4
The three thermisters are connected to the connector JP3 at pins 11, 12, 13, 14, 15, 16
Pins 4, 6, 8 and 10 will be used to power a try colored LED for trouble shooting.
The fan output at pin 2 of connector JP3.
This is my first board so I would appreciate the critique of veterans.
Have you breadboarded the circuit for testing ?
a 2N3904 seems a bit light duty for the start stop current of the fan, or is there an external snubber & relay / driver?
I would to add filtering to the inputs and in the power supply - using a regulator and filter ( the input voltage could easily rise to 15v and that is “ pushing” the nano regulators
Ability ). Car electrical systems are very electrically noisy and you need to allow for that in the hardware and software .
You mention pwm control of the fan - you don’t seem to have a driver stage for that - it will need a large power transistor for that . On /off control would be fine. ( unless of course the pwm control is already part of the car )
Any particular reason for the 1k resistor going to 3v3 instead of the 5 v line ?
Hi,
Keep your tracks as wide as possible, especially the power and gnd tracks.
I fact try and leave as much copper on your PCB as possible, as a gnd plane.
The thinner the tracks, the less support they have when soldering to pads.
Can I suggest you use a DC-DC converter to get 5V, rather than put your vehicles noise ridden DC onto the controller.
Some LEDs as indicators would also help any debugging and indicating that the PCB is powered up.
Some 0.1uF bypass capacitors and possibly a 10uF or 100uF across the 5V supply, and 0.1uF caps across each of the thermistor inputs.
You need to remember that even modern cars are noisy electrically.
Good to see you haven't crowded the PCB, nice job.
Tom...
PS Some mounting holes in each corner to make fixing points for the board.
I have breadboard tested it with pots in place of the thermisters. Still waiting for my thermisters to come in. The current to the fan is not the fan current. It is a controller embedded internally in the fan motor. I did measure it and found the current to be ~20ma.
The Nano33 BLE is rated at 21V input. I have seen several different opinions on whether it needs filtering or not. Some say yes some say no. My first though was to put an LM1710 regulator before the input. But I'm going to try it without anything at first, especially since this will unlikely be the last board I make for this. I have not considered any filtering on the input. These input are only thermisters for the moment. Is there something you propose?
I believe what you refer to as the driver stage is part of the motor. It is a 600W motor with 8ga wires going to it. But I only control ~20ma to something inside the motor.
I'm using 3.3V because 5V is not available on the Nano 3.3 unless you power it from the USB input. I will be applying power to the Vin pin.
This cooling fan does not run with the car off. It maybe a feature I add later but for the moment I prefer to leave it out. There is nothing to attach a heat sink to in this design except maybe some of the chips on the Nano. For the moment the plan is to just measure temperature inside the enclosure and evaluate.
Is it the engine cooling fan? On every car I've ever worked on that had a fan, the fan could come on to cool the engine at any point, even if the engine was off.
I'm not sure about the circuit, but the board itself looks fine. Your traces are a bit thin, maybe, and the lack of mounting holes is concerning, but overall I see no issue.
Don't quote me saying it's okay though. I've never made a working PCB (hopefully that will change in the next month or two).
I have a question on the tracks. The default from the Eagle software is 0.006. It doesn't give me the units but if its inches of mm that's too small. In the picture of post 1 the width is set to 0.006 and 0.01 for the main power and ground. I also need to check with the manufacture what they will accept.
DC-DC converter. I assume you mean like an LM7805. The input voltage for this board is 5 to 18V. Would operating it at 10V be better with an LM7810. Not sure if I want to do any further filtering as there is plenty of debate as to whether its needed or not, as well as others have done this without it. I will probably pass on it for the first go around. As you can see, I opted out of the opticoupler because the 817 opticoupler has a MTBF of 5000 hr which is limited in an automotive environment. So I am interested in durability > 5 years or so.
I do have output pins on connector 4, 6, 8 and 10 available LED's. I could put one on D9 which runs at 10HZ and is easy to see that the Arduino is giving output to the transistors.
OK I understand 0.1uF caps on each thermister. Not sure what that does.
Not sure I understand this."Some 0.1uF bypass capacitors and possibly a 10uF or 100uF across the 5V supply"
Sensors are limited to sensors that will physically mount onto the required components. Therefore I will be using a GM sensor with a 1/2" NTP thread mounted in my radiator. I needs to work up to 150° C. A little fuse clip on pipe sensor will be used for the AC condenser. Linearisation is not really required because the fan is driven in set points of % duty cycles of 10. 20, 30, 40, 50, 60, 70, 80, 90%. I only need the resistance at each desired point. However I may liniarise the sensors anyway to make adjustments simpler.
I do have these but they appear to be difficult to mount onto a PCB. I wish they put the 4 holes on each corner instead of on one edge. I would hate to cantilever the mounting of it on pins where it could vibrate and fatigue.
The way I understand this, the manufactures have finite sizes that they make. For example I might not be able to get a 0.055" trace but I would have to select from one of there stock sizes maybe 0.06" or 0.05". In any case are there minimum sizes you would recommend.
I ASSUMED filtering would be part of the Nano, I guess No
Difficult for me to not over think. I'm an engineer. A mechanical engineer but an engineer. Yes I am not sure what my electrical counterparts do and I'm sure the rules are different. In the Aerospace mechanical world we qualify a part for 1/5 the lifetime. Therefore we would consider a 5000hr MTBF component to have a 1000 hr lifespan. This particular part will in operation 100% of the time the car is running and will run at an average of ~50% duty cycle. That would qualify this component for 2000hr of time in the car running. That would equate to about 3 to 4 years of service for the average driver. And considering the rest of the components in this project are not MIL spec qualified, I agree I might be over thinking it.
So if I understand this correctly, I need to add 4 capacitors 1 for each thermister and 1 across the 5V side of the buck converter. Nothing is required on the 13.8V side?
Also since the Arduino Nano 33 runs between 5V to 18V input. Is it best to run it at 5V or would something more in the middle be more appropriate. That particular Buck converter I have can be set up with a fixed output voltage of 5V, 9V or 12V.
I made a separate thread for that because those features will not be on this first board. They are not required for the vehicle this board is going on.
The vehicle is a modified Mercedes 1986 560SL with a 1996 SL600 M120 V12 engine. The ECU is from the 1996 SL600. Neither vehicle was equipped with an electric fan so this controller has to stand on its own. The fan being installed is from a later model C230.
They are using a board-level module with its own bypass caps.
It shouldn't need more, unless they go for the pre-regulator idea (which would be a good thing.) (still, putting in footprints for extra bypass caps isn't a bad idea.)
0.006 inch is a typical default ("6 mils"), which fits most board house "minimum track width" requirements even for the cheap boards. But "minimum width traces" are generally to be avoided. I like 0.3mm (abut 11mils, .011 inch) as a practical minimum (though it may be difficult to fit those between the 0.025inch pin holes needed for the module.) 0.4mm if the traces don't need to go between pins. Power wiring would be at least 0.8mm (with GND expanded to a plane.)
See https://www.instructables.com/Make-hobbyist-PCBs-with-professional-CAD-tools-by-/ (not entirely apprioriate, since it accounts for homemade boards as well as "amateur" soldering. But probably still useful.)
I have a preference for routing through-hole component traces on the bottom layer; I think it provides slightly more mechanical integrity (tug on a component, and it pulls the trace toward the board instead of off.)
