You could try using multiple transistors in parallel and then heat sinking each one it should be ok if you use 3 of them for 20W a piece, I forgot how the hookup is for mosfats but for NPNs you add small resistors in the emitter to balance out the load between them. As for the heat, I don't think any semiconductor needs to be heated up to work at optimum efficiency, In the datasheet of the mosfat there should be a temperature grafic, I think heat bothers it, the cold is your friend, especially if signal to noise is an issue, as temp ads noise. Water cooling should be fine, although you'd have to design your own water block.
I saw amp schematics using multiple parallel FETs / MOSFETs. I'll take a look a them. Concerning the waterblock I was thinking about mounting the FETs on a copper plate and attaching a MOSFET or CHIPSET WB to it.
So you ended up building buck converters , though if I was designing this, I would rather use discrete mosfets (irlml2030/irlml9301 is a nice small fet with low threshold voltage). For current limiting you can try PTC thermistors. GuilleAcoustic -> the cooler mosfet is - the better.
I'll use the UDN298x part because it is high side (good for buck), and it can be run directly from the PIC. since current limiting is now just on one input line, I was thinking more along the lines of something active , although the PTCs are always a good idea maybe a combo using a ptc as a R sense for a shut off transistor, that way under normal operation the PTC doesent drop to much voltage and if a short happens, it will heat up and activate the shut down transistor
Heating PTC will increase its resistance, there is no need for transistor (acts as resettable fuse) For buck converters for fans - ok, fair enough , although 300mA will be not enough for high rpm fans and pumps are out of question. Also, you can tell if the fan is clogged up from RPM. Edit. UDN298x's switching frequency is anywhere between 83KHz to 430KHz, so keep that in mind.
I know it can be read from the RPM, but you would have to compare the output voltage to the duty cycle of the PWM....but with current and voltage readouts you could know exactly how much power a fan is drawing at a given RPM. And with packs of OP amps cheap as chips, its easy to implement. UDN298x frequencies will be the highest PWM the PIC24FJ256GB can support at 7bits,(128 levels) I think its 125Khz. *I just wanted to increase the performance of the PTC by adding a transistor. thus allowing for a small change in resistance to shut down the system.
I do have a question. I'm looking at an affordable PCIe low profile DAQ card. Do you have any suggestion ? I do not want 30Ms/s (500K will be far enough). A few analog inputs will do the job too (and maybe less than 10 digital I/O)
for such low speed <10MSps any high speed USB DAQ can handle it, and its probably a lot cheaper and practical, then a PCI or PCIe card. For the analog inputs you'd need a DSO, and for the digital a logic analyzer. For the analog stuff I can't recommend any, but at 500ks many microcontrollers can read the inputs and send them through the usb, at full speed only.. Search around the net for well documented projects, although I highly recommend getting a used Analog oscilloscope of ebay, you could get a good one as low as 50USD. now for the digital stuff check out the Open bench logic sniffer by dangerousprototypes.com, it has up to 32 channels and can sample at 100MS/s and its only 50$USD, Also check out the Bus Pirate by the same company, its a universal debugging tool, it allows you to interface with any device that uses SPI, 1wire, I2C, UART.....interface, and all is done via your computer on a terminal window. It can also measure frequencies, be a low speed Oscilloscope, and logic analyzer, and program both AVR and PIC microcontrollers, and the best part is the price its only 30USD
I found some nice stuuf from priggen that could do the job. Since I want them to be internal, I can just make an USB cable that connect to an internal USB header of the motherboard. Pico DrDAQ Pico oscilloscope I had considered this option, but it takes too much place. The WAF wouldn't be on my side . Thanks for this very nice information. I'll read it carefully. I'm used to work with National Instrument PXI chassis ... but they are far too expensive for me. The cheapest NIDAQ card from them cost 400€ ... without the shielded cable (125€) and without the box the connect all wires (>200€) . Maybe the USB stuff would be nice .... and them a DIY project for data acquisition.
commercial DAQ and DSO are grossly overpriced, but at least you get someone to shout at if something goes wrong The Bus Pirate is must have for that price, it makes life so much easier, and you can get free PCB if you wanna build your own, although it would probably cost you more then to get it shipped to your house for 30$
I agree, I found a 4 analog 30Ms/s for ..... $1100 . I've read evrything on the bus pirate site. It's a real bargain at this price .... Definitly worth bying one. I'll search for high speed analog project, I do not need the grossly priced end-user support that comes with NI or Vescow products. My only criterias are : compact and internal, as it is to build a mobile lab computer (like NI PXI chassis but with an embedded LCD) Thanks a lot for your infos. It's priceless for me and help me a lot.
I followed your advise and checked for DAQ project ... I found PCIe card using FPGA, it seams that it is very affordable ... but I'm waiting for a real price : http://www.signatec.com/products/daq/high-speed-programmable-fpga-digitizers.html
That's what I'm afraid of ... but I saw some other that are just "support card" on which you connect daughter card .... thus matching perfectly your needs. Proc board : http://www.gidel.com/PROCBoards.htm Proc daughter : http://www.gidel.com/daughter_boards.htm
You might have a look at what CrystalFontz has done - they have several lines of LCD monitors that have an optional module they call the SCAB that handles a lot of what you're going for. It seems, other than cost, to be close to what you're trying to do. Not that I'm discouraging this - the CF screen I got cost around $100 without the SCAB, and I had to hack together most of the information displays I wanted myself. If nothing else, since i didn't see it mentioned before, might be worth a look at what they've done - drawing from more sources is never a bad thing on a project like this. Also, I'd like to weigh in a vote against Java. I don't think I'm the only person that avoids Java like the plague. I can appreciate it's platform-agnostic nature, but I don't like leaving security holes open if I don't have to. Why not Python or something similar? It can be compiled against different platforms if written right, and can be compiled into stand-alone executables. As for Mac, who cares? Steve's not here to approve anymore, so no Apple owners will want it I'll be following this - I hope like hell this works out!
Thanks for the info, I'll look into it. Btw sorry for not updating this post. I am waiting on some development boards to arive, and I'll start prototyping some extension boards with fan control for them. The project is moving slowly but it is moving.
Better slow than hasty . I received my arduino proto board, it does have pwm outputs. I'll give it a try.