I was thinking of something like that. Maybe you can send via the USB some formated informations resulting from, for example, a TOP or PS shell command. Maybe csv like formated data, then your setup tool can just translate those data into displayed infos. I see this like this is you setup : - The command to be executed - The data format - The result on the screen - The refresh interval Very good question : I think that 4 fans sensor and 4 analog (share for temps, voltage, etc.) could be nice. I think that it could be a nice addition to be able to create a sensor lib. This will allow advanced user to create their own sensor and set the "attributes" (voltage range, etc.) depending on, for example, the termistor used to sense a temperature. Outputs ... I really have no idea. For sure a LCD/VFD output.
If you are designing a whole new uController from scratch, then slap in as many as you want. Possibly adding/taking away to create different tiers of products?
For a PCB I would suggest black. Black is always badass. You could have: 4 digital inputs (switches), 4 analogue inputs (temp, fan/pump speed); 4 digital outputs (LEDS, motors, relais) 4 analogue outputs (fans, pump). Separate connector for the LCD screen. You may want to include a power/status LED and a Piezo buzzer for warnings. Your biggest challenge will be to make sure each port can manage some hefty wattage. You may have to mount the PCB on a heatsink which could function as its housing --a bit like with the mCubed T-Balancer MiniNG.
3.5" is a good size, that way If you lack a floppy bay you can always bung it in with your hard drives and just run it off software.
Since we'd be using one of thoese MCU with a huge load of pins I think a 100. so many more inputs/outpus could be posible, the only limitation is that Analog outputs that requra more current like Fans/pumps, and such require external circutry, witch adds to the cost and increase board size. Although I think 8 outputs should be posible. PWMs are simple and easy to directly drive via the MCU so digital only outputs are posible in even greater numbers. As for the sensor library, there are many of them around even today so that wouldn't be an issue. there could be a drop down menu, from which you chose the sensor you have. Also a good idea would be to have a Thermistor calibration within the setup utility, as well as some standard values for comonly used ones.
I dont think the wattage/heat will be an issue since even 3 pin fans could be run of a PWM modulated power signal, keeping the RPM signal intact, although further research and testing will be done to see if this doesent in any way shorten the life span of a fan.
You'll want linear voltage regulation for 3-pin fans, otherwise they'll start growling at <50-60% speed, and if you use low-side switching - no RPM monitoring. Also don't use low-side switches for 4-pin fans, then ground is disconnected the circuit completes over PWM pin (will fry the MCU). Suggestion for peripherals: IR remote receiver, 1-wire support for precision temperature monitoring. If you need any help writing firmware or developing pcb/schematic - pm me. p.s. whats the point of making inputs 12v tolerant ( other than making it idiot proof)?
you mean the LVDS connector ., the little flat cable used in mobile phones? Posibble. Linear voltage regulation is difficult to do with a MCU, but not imposible. The article I linked explains how 3pin fans could be driven by PWM and still be able to read the RPM. Although highside control will be implemented(PMOS) along with a 300mA current limit for short circuit protection. Or if someone deciedes its a cool idea to hold the fan while power is applied to it. all the output power pins will have current monitoring for break down/dust notifications. Any 4pin PWM peripheral will have a constant 12v powersupply, and be driven like intended over the pwm pin. basicly all 4 pin outputs will require at least 4 posibly 5 pins of the MCU, 1 for PWM 4pin fans, one for Analog/PWM 3pin control, 1 for reading the RPM, and another to read the current. *like stated earlier the 3pin option will be tested extensivly, and no design tha shortens the lifespan on increases sound, or rejects rpm monitoring wont be accepted. A header for IR, and 1 wire interface will be provided, since the board will be inside the PC case no ir sensor will be placed on it, although if the demand is there it will be done. Unfortunentaly thermocouple support probably wont be added because it requires lots of external support circutry and would drive the cost up, althogh generic serial to thermo-couple boards will be supported as addons. *this project is still in the Idea stage so it won't be finished any time soon, once and if it given a go ahead, links will be provided to folow the design process as well as updates on this thread. *the Idea is to make it cheap but still very high quality, and safe.
All those pins will need connectors, which take up space. Keep in mind the physical dimensions you are working within.
You are quite right, now that I think about it will like a forest of headers although I think that it will still be manageable BTW not all MCU pins have to be used, and I think many will be used internally as well. so far here is what is intended 6x4pin outputs that could drive 4pin and 3pin devices. (32mmX9mm) 8x2pin digital i/o (drives leds/switches,rpm,etc....) (25mmX3mm) 6x2pin analog inputs (20mmx3m) 16x1pin LCD (possibly the lvds) (40mmx2mm) 5x2idc USB (17mmx3mm) 5x1pin UART/SPI/I2C/1wire...(x2)posibly. (17mmx2mm) 5x1 programing header(witch mat or may not be installed) as the device will be sold preprogrammed, and all firmware updates will be done by usb. with a 3x4 inch board or 75mmX100mm there should be plenty of space on the boards ed so far 83 pins, of which many will be connected to the ground, 25 to be exact. all of this should be possible to do with high quality and still keep the price low and accessible *feature I forgot to mention*the LCD will be contrast and backlight controllable with the computer. *the highest design challenge will be the 4pin outputs to make them work on both 3pin and 4 pin devices.
I looked trough some 3 pin driving options and it seems PWM driving is out, the only way to do it reliably is pulse stretching which could make weird noises. so it seems linear is the way to go. I found this note by Maxim that explains how PWM could be converted to high-side Linear PMOS driver. By adding another pnp and a sense resistor current could be limited easily *also after cheeking out some Fans and pumps its obvious that pumps require more juice so 2 separate PWM 4pins with higher currents will be provided.
You need some sort of goals, spec. draft, budget and end price. It's easy to say "I can do that and that, oh and that." But at the end of day you'll have limited resources and wont be able to do everything at once. You say you are planning to have 6 channels for fans, you could have all of them fully support 4-pin fans, but only 3-4 that support 3-pin fans. for 3 pin fans you could build full buck converters or something like in maxim's white-note - your choice will, most likely, be influenced by end cost. What wattage per connector are you aiming at? You want to build current-limiters for all outputs, but that will take pcb space and cost money, are they really needed for all the outputs? 75x100 board, are you planning to place components on one side or both? How many layers? Where are you planing to fabricate them? What mcu are you planning to use? Maybe I missed your answer, but for what purpose do you need 12v tolerant inputs? I doubt you'll fit in 40$ with everything you listed (maybe for components alone + PCB if you order them from PRC). Sorry for nitpicking, but I've seen to many projects fail only because developers were overenthusiastic.
Very good points. I'll try to answer some of them. Full buck converters take up too much space and external components and are very hard to control digitally. And since 3pin fans don't require perfectly stable voltages, the maxim white note is perfect. the smaller transistor can be bought in IC arrays dirt cheap, all of the smaller transistors for the 6 outputs will come out to <1USD , only the output power ones would add to the cost, but that could still be relatively low. wattage per connector is something I haven't figured out yet I think most fans dont draw more then 300mA of current..this is something I would need your input as well, how much current do your varius fan setups draw. Current limiting only requires one small transistor and a sense resistor, once the voltage over the resistor is above base voltage of the transistor, it shuts down the larger output transistor. its cheap and reliable. Doing it with the MCU would add to much unreliability. And yes it is needed for all outputs that will draw power from the board. I don't want to be responsible for killing someones PS or worse. (I know any PS worth a dam will have CS protection, but still there are too many computers with cheap PS's. Pumps would have to be driven of a separate 4pin connectors as they require a lot more current. Boards will be 2 layered with the components on one side, and manufactured and El tested in PRC. Not low quality PRC. Most of the cost comes form the parts and assembly, the boards them selves are very cheap, you can order boards of this size for 25 bucks for 10 boards, in higher quantity is even cheaper. most lcd can be driven directly off the MCU, only the backlight and contrast would require additional components. The MCU used will be Microchip PIC, 16bit most likely. As for the 12V tolerant inputs, mainly for protection against miss use, but there my be some applications to it as well, like PS monitoring..... PIC24s have HW USB so no external chip will be needed for that, resistors and caps are negligible cost wise, although it does go into the manufacturing cost trough increased pads that need to be soldered. So far its still possible to do <40USD and make some profit, of course if the demand is high and larger quantities go into production the cost will go down. *Of course I could be all wrong since it is still just an idea.
*I am about to take attempt to take the foot out of my mouth After some calculations it became clear that any kind of liner control would require heat sinks, and generally be inefficient. I found this circuit by Geoff, and I will be basing the output design of of it. current sense will be added to provide fan consumption history/detect too much dust.. also the frequency of the PWM will be increased to minimize the inducer and cap required. *It came to my attention that fans have built in short circuit protection, so independent current limiting will not be necessary, although, there will be a current limit on overall output and fuses as well. http://geoffg.net/fancontroller.html
What kind of regulator are you going to use ? LM517 ? Heat and space are always an issue, I'm encountering this problem with my amplifier design (especially since I want to use class A). Good luck
For powering the MCU and digital low amp stuff a generic 3.3v linear regulator will be used on the USB line, as it is On even when the computer is shut down. The analog outputs will be powered from the computer PS 12V so no voltage regulation will be used only one clamp transistor for current limiting the line... *Class A is more of a heater then an amplifier, just kidding *if you are using TO-220 cases, just get the thickest piece of Alu you can bend and make a u shaped heat sink, and screw the to220 to the bottom of the U. If you wanna incorporate PC into the design, get a generic box cooler, drill a hole in it and screw the To220 to the bottom of it , and stick a fan on top pic of a lm317 with a u shaped heat sink http://it.wikipedia.org/wiki/File:Regulador_LM317_en_disipador.JPG
Class A are quite the heaters but are not class A power efficiency certified . Nice tip about the U shaped aluminium, but it won't work with class A power mosfet (around 60W per mosfet to dissipate) and fans will induce noise in the audio signal (unless they have a separated PSU circuitery). I naively thought I could make compact class A monoblock, but the PSU part (power transformer + LC pi filtering) is already quite big. The amplifying circuit is very small by itselfs, but the heatsing is a monster ! I tought about watercooling the mosfet, but I do not know if they have to reach a certain temp to operate at their full potential.