sweet, I've been reading some data sheets and basically the limitation is the USB bandwidth that is possible by a MCU only. Has anyone tested Microchips stack for continuous stream of sequential data. The MCU should read 8 channels then send 8x16bits over the usb, and then repeat the process. I get that this is not the best way to utilize the full potential of the USB protocol, but I don't think the MCU has the memory required to fill a full frame and then sent it in one go. I could be wrong since I am new to USB design.
This seems like a easy and cheap project to do, and even at 10KS/s sampling rates it would find many uses, especially for people starting in electronics design. Purpose: To allow a user to simultaneously watch and compare up to 8 analog signals in real time on the computer. Make it as cheap as possible. Build. A singe MCU is all that is needed to both read the channels and pipe the data through the USB. I think pic 24F have a sampling rates of 500KS/s, the PIC32 1MSps.
Minimum amount of external components, probably only a few resistors for the voltage divider and a few switches or mosfats to change the range. 20V 2V would probably be sufficient, although if needed a few channels could have a analog front end to read down to 200mVs. A external reference would be used to allow greater precision
Have the USB connector broken out to pins to allow internal motherboard connection.
Question, how fast could a PIC24F and PIC32F read 8 channels and send the data to the computer, what kind of sample rates could I expect, I am also interested if the sample rates could be increased by lowering the amount of inputs.
It turns out fans for computers have protection in themselves. By specification they are not allowed to draw more then 2Amps of current, and in case of breakdown the chip used in them are made so that when the fan fails it fails in an open circuit. so current limiting won't be needed, although a fuse is always a good Idea. I'll post the update for my schematic if anyone needs to drive anything by current control with a liner system, although because of the dissipation inherent in linear design heat sinks need to be provided on the output transistors.
I've since decided to use Geoff's idea, although I'll increase the pwm frequency 125Khz to lower the C and L values. I'll use 10uHs and 30uF for my circuit, aslo separate shotkys will have to be used, and I'll add a current sense to the mix, with a low side R sense and an op amp.(same as in my circuit) The current sense will be used to detect how the fan works over time, and have the possibility to warn the user that too much dust has accumulated. *I've added NI multisim 11 file of my schematic if anyone would like to play around with it
thanks that looks like a great project, he uses pwm and a large output filter to get different analog output levels.Buthis circuit seems to be missing current sensing or fault protection, maybe it could be implemented by the pic reading the RPMs, but still no real short circuit protection.
*I refined my schematic somewhat, I changed the PWM input section somewhat, making it more stable and component tolerant. the cool thing about it is that its ultra cheap to build and you get current sensing.
I thought some more on this subject and came up with a current controlled design.
one op amp is current sense amplifier, and gives on its output 0-3.3v for 0-500mA current, another Op amp is a error amplifier. one input is from the current sense, the other is driven by a DAC (PWM with RC filter), its output drives the "output transistor".
In this way the MCU could have output voltage and output current over its ADC, and drive the PWM by an algorithm to keep the output transistor within 2W. One good thing about this design is that its inherently current limiting.
Hi I would like to design a fan controller for PC fans that use only 3 pins. GND, VCC, and RPM.
If you PWM drive the VCC you lose the RPM, there is a possibility of stretching a pulse from time to time just to read the RPM, but this brings audible noise into the fan. So I think driving via PWM is impossible. Although if you have any suggestions I would love to hear them.
Liner driving via PWM is covered by this Maxim http://http://www.maxim-ic.com/app-notes/index.mvp/id/3149app note, but the problem with liner driving is power dissipation in the output transistor. I would use either P mos, or PNP in a to220 package. the circuit could be able to provide 500mA of current, which is around 6W if the fan gets stuck and draws 500ma at 1v, then the Transistor will take the full 6W on it self and I don't think it could handle it without a heat-sink.
Would soldering the back of a To-220 to the board provide enough of a heat sink to handle 6W?
Any suggestions how this could be designed without the use of heats sinks?
there is something odd with PK2 and MPLAB X, while pic16f1824 witch I am using right now, and many new chips are suported by pk2 with a device update, it isn't supported in MPLAB x, I dont get this, if the PK2 has no problems twith reading or programing the device, why is there no way to do it from MPLAB 8 or X?
since most of us here are into moding and hacking stuff, you could mod your soldering station so that the cable is housed in a shower metal hose, it should be simple, only prob I see is finding the right holw screw to fit on the station side, on the iron side you could use heatshrink to conect it to the iron.
*sory for offtopic, let me just finish. I bought 2 pics, one to experiment on and the other to actualy use, the second one for some reason just couldn't get the hardware PWM to work, while the first worked perfectly, then the first one died after 12V apered out of *nowere and preaty much made miced meat out of it, and by nowere I mean I placed the wrong jumper on my bredboard PS and insted of 3V I got 12. :D. so I was stuck with this software PWM
arhi, I get them from a local ebay type site, limundo, its kind of a risk, I buy broken, non working phones, for around 2e-4e, and all of the ones I bought so far, around 4 of them had working lcds, from time to time there are even adds for lcds only that also go around 2-3e, you can even get better deals if you buy 2-3 at once, the trick is to buy from the seller that is at the same location as you so you dont have to pay the shipping fee witch is around 2e. also any local mobile phone servis will sell them for around 3e..
p.s. also to repeat my request to route all leftover port pins of a mcu, for future development, like adding more buttons, leds....
I did a solomon control cirtcuit with pic16f1824 using a nokia 3210 for display, it has up, down and enter buttons and a menu, for saveing temperatures, and loading them, it also has the posibilito od usin an analog(pot) for seting temperatures, or digital up/down for fine tuning.
if anyone is interested I could post it here, although, the PID is not coded yet, and the code is not the cleanest of all, for that matter neither is the pcb.
Ian if you have extra pins, add breakouts so I could hookup nokia LCD insted of 2X16, these nokia 3310/3210/3330 lcd are cheap as dirt over here around 3$ a piece ,