EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: nour on May 30, 2017, 11:41:22 pm
-
I was looking for a Mosfet which is capable of working in the linear region to build a Dc load and all that I can find for this job is Mosfets from IXYS!
I am not quite sure but is there any other manufacturer that supply such a Mosfet?
The reason I am asking about this because I want to see all the available options on the market and also to compare cost\performance and limiting this to one manufacturer is very irritating.
I couldn't find on the parametric search on Digikey something to sort and find with such a Mosfet type!
Maybe someone can tell me something I am missing over here, does all the big names Electronic DC Load manufacturers uses this kind of Mosfet, I mean the one which is capable of dissipating power safely in the linear region? or there are other design approaches that I don't know about(certainly there is ;D)
-
you should look for older, higher V, higher on resistance parts - ideally with SOA graphs showing power limited SOA only
diyaudio.com should have some parts suggestion threads
-
There's been at least a couple of informative threads on this subject, and I've provided a lot of detail on extensive
testing that I've done as well. Short answer is - there are a few but mainly DO NOT believe the curves that are
supplied. Make your own. I'm not at my office, so can't provide anything more for now.
-
It's much easier to use bipolar in parallel, with just really small value resistors on the emitters and a copper plate on the heatsink between the components. Now, this is if you want to dissipate overs 300 watts of power. When it comes to mosfets, the 2 strategies is for low voltage apps, find the lowest VGS devices in TO-3 or TO-220, purchase many extras, build a high current VGS measuring test fixture and bin them and match them for each of your hardware devices you are building. For higher voltages, go for older high VGS mosfets and perform the same VGS matching.
I don't know the speed of your electronic load, but, I would use an MCU PWM driven N channel mosfet fed through a kick-ass inductor and avoid the linear and matching problem all together, where all of the heat will be dissipated in the inductor.
-
Voltage, current, package, quantity?
The ISOTOP package is pretty easy to work with and can have huge power dissipation capability (500W+)!
-
I am still doing my own investigation on this topic, I am not sure yet about the specs. I just have a general spec on my mind, but the more I do research on the topic the more my understand get better and maybe at the end I will have a totally new specs.
But generally, I want to have two main Electronic DC Loads(if possible) one for higher voltage and higher power up between 0.5KW and 1KW and other below 250W
As I said I am still doing my research and not yet have a lot of knowledge about it, in the meanwhile I was watching this and came here to ask about the
https://www.youtube.com/watch?v=WUPrj03UbTM&t=710s (https://www.youtube.com/watch?v=WUPrj03UbTM&t=710s)
https://www.youtube.com/watch?v=Vr6YsT403DM (https://www.youtube.com/watch?v=Vr6YsT403DM)
In those two clips, it seems to me that it was very easy to build a controllable 500W DC Load with just a few components and very big heat sink :D
He seems to me that he was advertising for them but even though if those specs are reliable and trusty, that would make those devices very good, right?
And maybe I should just buy some of them and use them to avoid the headache of the whole design process!
I will also see Scullcom Electronic DC Load series and see what can I get from it!
-
I think you guys are using the wrong type of MOSFET packages by using TO-264s or smaller packages. I just bought 5 of IXYS IXFN520N075T2 (Datasheet (http://ixapps.ixys.com/Datasheet/DS100193A(IXFN520N075T2).pdf)) which come in SOT-227 miniblocks and have a much better case to heat sink heat transfer characteristic. They cost more, but if you're going for a 1 kW+ dummy load, forget about piddly TO-247s/TO-264s (or the even more ridiculous TO-220s) and go for miniblocks (unless you want to mount dozens of the little guys in parallel on your heatsink(s) and good luck matching them up without a crap-ton of additional circuitry and tuning).
I'd rather mount a handful of SOT-227 packaged devices and put 250 W+ (if not 500 W+) through each of them with ease, given a large enough heatsink and convection/liquid cooling, than have to deal with the poor case/heatsink heat transfer characteristics of the various puny plastic 3-pin devices. Also, consider exactly how you are going to wire their pins. With a SOT-227 I can screw in a connector off a thick copper low gage cable. With TO-264, it's going into a circuit board and there is going to be plenty of voltage drop at high currents (which may not be desirable if you are trying to minimize the resistance of the circuit for high current loads at low voltages).
-
Agree. One of the problems is that manufacturers grossly overstate the thermal capabilities of tab-style packages. The claims made for TO220 in particular, of 90W dissipation or whatever, are simply impossible under normal working conditions. Maybe sandwiched between two anvils with an iceberg balanced on top.. Claims made for the larger TO247 etc are more realistic but even then they're on the limit of credibility.
The good ol' TO3 was hard to beat, although also harder to mount. The better ones had a slightly dished underside to give a really firm central contact.
Though, the price of SOT227 devices is rather outside the range of the experimenter. A cheaper and equally effective option would be a bank of 2N3773's. These are TO3 NPN devices designed for guitar amps and the like. Quite robust and relatively cheap, main issue is the relatively low gain which usually calls for a TO220 or similar driver stage.
-
The linked MOSFET IXFN520N075T2 is a switching type and thus not a good choice for high power loss operation. The FBSOA is not that great - at least they show a credible curve that include the break due to thermal instabilities. Even though rated at 900 W, its only 300 W at 30 V (if the case is hold at non realistic 25 C). At 60 V it would be more like 150 W theoretical power limit and maybe 100 W usable at 40 C case temperature. For the 60 V rating, it would be about 2 of the large srew terminal FETs to replace 3 of the IRFP250 in TO247. This does not sound like a good choice for me.
Not sure if these are typical limits or minimum guaranteed values - there might be scattering between individual samples and thus prepare for possible FET failure during the mandatory burn in test.
It depends very much on the voltage range for linear operation. For very high currents and low voltage the mini-block and screw on cables might be a good choice. For higher voltages a few mOhms of wire resistance does not matter but the FBSOA gets important.
-
It depends very much on the voltage range for linear operation. For very high currents and low voltage the mini-block and screw on cables might be a good choice. For higher voltages a few mOhms of wire resistance does not matter but the FBSOA gets important.
I am shooting for <=24 V max in general for my load and in particular very high currents at much more common voltages <=12 V. I figure if I can (theoretically) get 100 A at 12 V and maybe close to 200 A at lower voltages, that's good enough for me and this is per device, mind you. Of course I will most likely hit realistic thermal limits way before that, even with convection cooling, so 500-750 W per device is the final real world goal with a properly cooled nice heat sink. I picked this sucker up for $45 USD on eBay - (9-1/4"x9"x5-1/2" = 235mm x 230mm x 140mm, 25 lbs = 11.4 kg). It was the best (and coolest looking ::) extruded aluminum configuration I could find for mounting multiple MOSFET modules on for "reasonably cheap":
(http://i.ebayimg.com/images/g/MUYAAOSwWKtUrvSn/s-l1600.jpg)
(http://i.ebayimg.com/images/g/LsIAAOSw-W5UrvSt/s-l1600.jpg)
I can match the MOSFET modules, by picking the closest two (or four?) out of the five I bought:
(https://media.digikey.com/photos/IXYS%20Photos/TO-227B,MINIBLOC.jpg)
..., and put them in parallel for a nice 1-2 kW electronic load for under $150 (well maybe).
I can also mount two 120mm fans on the heatsink for the temperature controlled convection cooling aspect (e.g., Noctua NF-F12 PWM):
(http://noctua.at/media/catalog/product/cache/2/image/9df78eab33525d08d6e5fb8d27136e95/n/o/noctua_nf_f12_pwm_3.jpg)
And this folks is how you build a proper high wattage electronic dummy load that puts Kerry Wong's concoction (https://www.youtube.com/watch?time_continue=1&v=Vr6YsT403DM) to shame...
-
Even with good cooling, I would not go above about half the rated power. Even with just 24 V operation the FBSOA limit is already lower for that MOSFETs - the full Ptot limit is valid only for voltages below about 12 V. Higher temperatures tend to shift that voltage limit even lower. So the practical limits would be something like 200 W at 24 V and maybe 300-400 W at 12 V.
Even with matching, it is not really practical to have the MOSFETs directly in parallel, as current sharing would still need large source resistors (e.g. 0.05 - 0.1 Ohms range for 24 V operation). So the sensible way would be a separate current controlling loop for every MOSFET. With such high capacitance gates, it might need an extra driver stage (e.g. transistors as emitter follower / driver). Even a cascode configuration of 2 MOSFETs might be possible.
-
Well known problem, and the advice for individual matching seems to be the only viable approach.
What I'm curious about: has anyone ever looked if the VGS and RDSon match of selected MOSFETs is maintained over temperature? Or do they spread apart?
-
It's much easier to use bipolar in parallel, with just really small value resistors on the emitters and a copper plate on the heatsink between the components.
He's talking MOSFETs, because -- presumably -- he's talking >200V load conditions.
Bipolars don't exist which can do that. Most MOSFETs can't either, hence the purpose of the thread. :)
Tim
-
Having MOSFETs in parallel in the linear range is tricky: compared to BJTs the temperature coefficient of the control voltage is something like 2 - 3 times larger. Thus they need a ballast resistor (source or emitter) that is larger by about that factor. The second point is that variations are usually larger (Gate threshold vs. U_BE) - so it takes more effort in matching than with BJTs.
High power at higher voltage (e.g. > 100 V) is difficult with BJTs and MOSFETs. Up to about 100 V there are audio grade BJTs, that sometimes even come FBSAO tested with a reasonable price tag (e.g. MJ15022 and similar). FBSAO tested MOSFETs are rare and expensive, like some of the initially mentioned IXYS ones.