056 :: mini research of MOSFETS

(UPDATED 5th RUN - added some low voltage MOSFETs, found some LFPAK-D2pak package errors in my chart so when browsing, do read the original PDF for final confirmation)

thanks to the EEVBLOG community for teaching me some tricks about MOSFETS (http://www.eevblog.com/forum/projects/switch-mode-inverter-totem-pole-mosfet-driver-need-help/)

that would then translate into this very useful chart and formulas. (updated for 3D folks reference on logic driven MOSFETS. column 3 gate voltage 2.45 and below. 1.7v preferred)

List of MOSFETs (it is better to save this pic and open with zooom, 4th update, the list is getting quite troublesome to view completely)

the thermal envelope calculations are just very fast numbers done as guides to help identify if the power dissipation will fit onto SMD PCB as heatsink scenario. for more robust fitting, you need to know your loads well and calculate accordingly. (alternatively, drop a post at EEVBLOG and seek some help like me from the helpful peeps there)

for 3D-ers, switching just DC. focus on low RDS, working voltage and gate drive of 1.7v (see example circuit and calculation below)

For example : basic DC switching below 1kHz, do try to add in the schottky protection (D1 and D2) if on very heavy loads.

in this case, to know the power dissipation :
P = I x ( I x R) = 39.5A x 39.5A x 0.0036ohms
= 5.62 watt (make sure you use a heatsink, typical bare packages do not do well without heatsink ABOVE 2 watt) ** we could use P = IV, it would be the same results **
for cheap shottky diodes, try using MUR4xx series or PMEG series (eg MUR415G, about 40cents. MUR 4 series are 4 amps, PMEGS are good for very high speed but at a lower 2A)

Warning : at 39.5ampere, please do make sure ALL your other current carrying wires are up to the capacity. if this MOSFET is replacing the ramps, your ramps will burn. choose a mosfet below 20A for PCB app unless, the PCB copper tracks have been rated for over 20A. at 39.3A, a lousy wire with 0.1 ohm resistance will burn with the ferocity of 154 watt power ! it will be fiery and bright !

The above is a fancier and highly exaggerated version of the clamping with zener+schottky. the zener values should be custom to what ever safe voltage you wish to impose usually a few volts above DC supply but below MOSFET voltage limit. the very small inductive load of 1uH is a parasitic example giving me a kickback that can soar over 15v. the zener is suppose to be a high speed version, but i doubt ANYthing you can find in farnell will tell you it can do that, so having a larger safety margin is the way to go. the smaller schottky PMEGs are to make sure the zener only conduct in the reverse only. R4/3/6 are simulated parasitic resistances. C1/2/3 are spike arrestors, but they can only do so much. as you can see, where there is a inductor along side a schottky (U3), the inductive kickback is huge (136A, again this is an exaggerated example). the simulated osciloscope shows the orange trace of the MOSFET source, it is sitting on Y= -1 as null, the swing goes over 20volts and slightly below zero volts also, and thats just 12VDC. so if you use 24VCD, expect something over 30v in spikes easily (again exaggerated simulation but it does happen). the schottky in use may seem odd, but its one of the fastest (17ns @ S$0.82cents, 8A/100A surge) i can find in element14 and it doesnt cost a bomb. the next candidate would be Vishay V40107C (10ns @ S$3.60, 40A ! 200A surge).

for safe use of MOSFET without a heatsink, the thermal output should be below 1 watt (at 1 watt, the package will be burning at probably 80-100degree celcius without heatsink). always remember, where is all the heat going to go, make sure sure they go somewhere safe.

when in doubt. ask someone. im sure EEVblog/3D printing google groups have many experts. (NOTE : im not an expert, i just happen to have some time)

*edit : added, 13.5Mhz simulation for METCAL PSU. looks like this setup will give the RF output around approx 110w raw power based on the RMS 9.7A of current going around L1. the original DIY project was started by Mr Chris in EEVblog, i was very intrigue as its the first time i see a soldering machine uniquely using RF power dissipation as a heating method for soldering iron, its like an induction cooker squeezed into a small tip. the DIY circuit originally uses a IRF510 MOSFET with a RDS of approx 0.5ohms (see chart for full comparison). so i started to dig around on the MOSFET, something which i think will fit on the PCB w/o any additional heatsink (yes tall order, in this case i suppose it could well be possible by way of parallel connection of 2 to 4 of these MOSFETs and additional brass heat spreaders)

the key in digging out all these MOSFETs is to look at ways if waste heat can be reduced. if by way of ratio comparison. the IRF510 dissipates/wastes at least 4-5 times more heat. but then again, i have never tried to use MOSFETs in RF applications before.

problem with simulations is that the SPICE data provided are only estimates of what could happen. it is definately not going to happen like what is going on in the diagram below. if it does, i want to kiss the engineer who wrote the SPICE simulation data file :D !

EDIT : on 4th chart update. added few more logic capable MOSFETs like BUK6215.


Popular Posts