AVR-Chat

Hello again :) Thank you for the kind words. 1.)  The low value gate (10 Ohm) resistor helps to control ringing - here's a link regarding the why-for's in a switching regulator design:    www.fairchildsemi.com/an/AB/AB-9.pdf2.)  You probably won't need the ferrite bead on a SM board layout.  It helps to control parasitic oscillations by absorbing short, fast transients.  
3.  RE: a MOSFET driver; see this link:    http://homepages.which.net/~paul.hills/SpeedControl/MosfetBody.html Some other comments (if you don't mind): 1.)  Presuming that you are connected directly to the AVR uP for driving the FET, then, if you haven't already done so, review the AVR pin that you are using for its sink & source specs and (if listed) how much capacitance it can drive.  Some devices can sink (or source), better than the other way around.  If it can't handle the 470pF load, then you need to buffer the output somehow.  I haven't yet worked with the AVR device (pending, tho) and I have too much on my plate right now to review it for you - sorry.  You know your design, so you'll have to review it a bit more.2.)  I suspect that you might have some issues driving the electromagnet and the above ref'd links might give you some insights.  Depending on what you are trying to design, you might want to consider some type of circuit for the magnet that adjusts the 
current (there might be an addressable constant current or 'volume control' device out there somewhere that might work), rather than PWMing the coil outright.  What might play a issue here, is the 'loading' that the magnet will 'see'.  I would suspect that if you intend to lift or suspend an item using the PWM technique, then there will be some side effects that reflect back into the driver, so great care and consideration must be given to the design.  Also, is that load dynamic or steady?  Some programming routine might have to have an 'initialize' the magnet routine (give it a big shot of current) and a sub-routine for its fine or steady-state control.3.)  You might want to use a separate power supply, if you aren't already, for the heavy current loads - that solenoid is drawing quite a bit of current and its repetition rate is fairly high.  This will give a great deal of isolation from the drive and uP circuits - no more under and over shoots on the supply line!  
Obviously, the grounds would be common - careful on that node!4.)  In my opinion, the construction of the solenoid should be closely reviewed, even if you have to take one apart.  You need to know it's construction in every detail to retain your product performance.  If the internal wire has too thin an insulation, it might breakdown over time.  If the working environment is hot / humid, the insulation might degrade, too.  I have seen many failures in designs where the design engineer did not look at the long-term picture.  (Hey!  It works!  Ship it!)5.)  If you have good drive to the FET, then it shouldn't get too warm, even at 1Amp, as the 'on' resistance is in the milliohms (so says the specs).  6.)  ALWAYS take the approach that every voltage and every current is at the design maximum (include the spec'd tolerances of the parts, too) when calculating power dissipations and choosing resistor wattages and heatsinking.  The effort will pay off in a quality design that gives 
great performance every time.  I have seen many resistors and devices fail because the designer chose the average current (based on rep rates or duty cycles), rather than the peak current that was going through the resistor / device.  Built-in failure - I guarantee it!  That peak heating accumulates over time and parts fail.  7.)  If you have separate (individual) fuse clips mounted on the PCB, MAKE SURE that they are in axial alignment and that there is no displacement laterally, but don't wave or reflow solder with the fuse in the clips - it'll get damaged!  You'll have to use a wooden dowel or such to fixture them in place.  If the fuse doesn't sit right in the clips, it's another failure is in the making.  (I saw this in a high current power supply that caused permanent heat damage to the PCB - many PCB's were charred to carbon around the fuse clips.)8.)  'Partition' your circuits on your PCB layout in a rational manner - the 'node' concept helps here, too.  Analog - 
digital - high current - overall cuurent flow; where is their common node? That's all for now. I hope I didn't over-do or appear insultive - certainly not intended. Keep at it. Frank *****************************First, thank you for the incredible reply! It will take me a day tochew it over. I will use a shottky diode as you suggest. Division ofthe high and low current paths on the PCB is a good thing to consider.Very important especially considering how an analog line next to orparallel to a digital high current line can easily get swamped,possibly damaging components even! The "node" power supply issomething I always do. That's been engrained. The solenoid draws approximately 1 amp continuous. I have notmeasured peak-on, but it is probably several times that, which is whyI picked this MOSFET - to have some safety zone. Repetition rate isslow. Maybe a cycle time of 1/4-1/2 second minimum. However, I also used this same circuit to drive an electromagnet, thatis current 
controlled through PWM of 500Hz. (and plan to continue todo the same for the "next generation" circuit)I have 3 questions for you or anyone else...1) What is the purpose of the 10 ohm gate resistor?2) What is the purpose of the ferrite bead? I have seen this done inother applications as well... One problem I have is the FET I use issurface mount (D2PAK). I don't know if one would fit.3) Are there any benefits towards using a "mosfet driver"? (Whenwould one be required?)Thanks again,Ryan--- In AVR-Chat@yahoogroups.com, "" <fxguthri@...> wrote:>> Hello: I hope I can help. I have 40 years of electronicexperience. The IRL520 is a hefty device, rated at 100V @ 9.2A. BTW -what type of solenoid requirements are involved - high current draw,high mechanical load, repetition rate, etc.? Do you really need touse this type of FET? In my opinion, I think that the design canimproved by controlling the current flows involved, so as to minimizespikes, etc. The reversed biased 
diode to 12V is OK. If you want,use a 'fast' (Shottky) diode to help quench the circuit. Altho it isalways best to quench at the source (the solenoid, in this case), onecan also place components in a way that have 'preferrential' currentpaths. The inherent diode that is part of the FET (Drain to Source)should be adequate for that current path, as it is at least as good asthe FET itself. The way to do this is to (mentally) divide thecircuit into two parts; a low current (the gate circuit) and a highcurrent (the solenoid). 'Pick' a > point (constructionally) for the power and the ground. On one'side' of this 'point' is the low current and on the other side is thehigh current. Between the two points, add an electrolytic (~100uf),bypassed with a .1uF. The best place to 'divide' the circuits isright at the electrolytic leads. (I speak from experience, as I havedealt with this type of problem before.) This approach can beconsidered as being 'nodal', where a node (point) is 
chosen and not'moved' for the current path involved. As there are two circuit'loops' (low and high currents), then they must share the common nodesinvolved. Uncontrolled circuit paths (no nodes) create complexcurrent paths, leading to problematic designs. Think of theuncontrolled paths as a ladder array of resistors, with differentconnections placed anywhere - the currents get complex very rapidly. The other advantage of the 100uF, is that it provides a lowimpeadance source for supplying pull-in current and acts as a > low impeadance 'load' for the quench aspects. Cool! If you keep allof the solenoid currents on that side and all of the currents for thegate drive on the other side, then there should be no real issue withdriving the circuit. This FET has a gate capacitance of about 490pF -not bad for the device, actually. Otherwise, your use of thepull-down resistor on the gate is commendable, as most engineersoverlook that subtlety entirely. However, you should have a 
smallresistor (~10 Ohms) and maybe a ferrite bead in series with the gate.Rise and fall times are important for driving the FET gate - makesure that the uP, or other circuits not shown in your schematic, arecapable of driving the gate properly, especially if you need fastpull-in and drop-out times for the solenoid. If the gate signal istoo slow, then the solenoid performance will suffer and the FET willdissipate excessive power while ramping up and down during the on andoff states. A circuit I designed > many years ago resolved all of these problems and was very, veryquiet from a digital and analog point-of-view. If you decide that theproblems are unresolvable, then you might consider using an optoisolator driving the FET, which might help, but certainly at highercost and more parts, too.Divide and conquer! Good luck!

_______________________________________________
Join Excite! - http://www.excite.com
The most personalized portal on the Web!


[Non-text portions of this message have been removed]