IGBT vs. MOSFET

Den mandag den 23. januar 2017 kl. 01.54.00 UTC+1 skrev Jim Thompson:

as long as you can keep it cool and below the max energy rating

Some forklift trucks use that system for the main traction motors, but they rectify the AC from the switching and use it to power a hold-in contactor for the main supply. If the switcher locks up, the contactor drops out after a few hundred milliseconds.

--
~ Adrian Tuddenham ~ 
(Remove the ".invalid"s and add ".co.uk" to reply) 
www.poppyrecords.co.uk

Ew, IRFP anything!

It's a vacuum tube of a FET: ancient, huge, slow, and 5x worse performance than a proper FET.

That does leave one virtue though: the power dissipation is actually what they say it is...

I don't ever recommend putting a FET under repetitive avalance conditions. MOSFETs don't behave like simple avalanche diodes, and they're always somewhere between iffy and bad under this service.

If you can't swing a full size TVS, then it's suitable to add a smaller one from drain to gate, with a series gate resistor to allow it to do its thing. (Which is more or less what all "protected switch" devices do.)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

The old GE SCR Manual, a classic, is full of commutating circuits. And weird things like cycloconverters.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I said there are better fets. But its avalanche energy is in the common ballpark. I have no idea what the injector solenoid behavior is like, so I can't do more than suggest that an avalanching fet might be OK.

I believed we used that in a linear application. It's interesting to use switching-specified fets in linear mode.

Some mosfets are designed for and rated for repetitive avalanche. It's commonly done. There are also relay driver ICs that are intended to avalanche.

Whether some mosfets have an internal zener structure, or however they dump the avalanche energy, I can't say.

GaN fets are really interesting, but I haven't seen one that will withstand avalanche. They tend to blow out.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

You need additional circuits to turn OFF an SCR, when supplied from a DC bus. Generally, SCRs have a constant on-voltage that can be QUITE a bit higher than a large FET, where the voltage drop can be reduced to tens of mV.

Jon

In a similar vein, I once tested IRF740 to a fair margin beyond the maximum rating.

If you take RthJC(max) as the true value, then TJ was around 200C before failure. More likely, RthJC(typ) was 30-50% lower than the reported maximum, and it failed at a more standard 150-175C.

The die is huge, about quadruple the area of a modern replacement. (That's consistent with the ~5x performance ratio between old and new.)

The equivalent modern transistor failed within 10% of its rating -- a good show of consistency, if not of conservative ratings. Its die was about the size of a SOT23.

The dark ages of bad MOSFETs seem to be behind us; SuperJunction types apparently have full SOA, despite having smaller dies than ever. I'm not sure how or why that works out, but it's nice that it does, I guess.

And you can always opt for "linear duty" FETs, if you don't mind the price hike.

Or... use the old dinosaurs that have more silicon than brains put into 'em!

(On a related subject, I've been meaning to test a SuperJunction type as RF amplifier. The conspicuously low Crss and Coss should prove interesting.)

Possible, but I don't think they ever do, among conventional types. Area's too expensive to waste, I suppose...

Hmm, if they were putting in a proper avalanche diode section, they could probably give it good reverse recovery properties, too.

And then there's "FETky" parts, which obviously have good recovery. I don't know much about them, though. The schottky stuff might have avalanche-capable guard rings. Or they may be designed so the FET part breaks down first, anyway. I also don't know if they are co-pack or monolithic.

They're making proper high voltage power GaN MOSFETs now, which is cool. Still quite expensive, and fragile, though.

I don't know if it's a property of GaN, that avalanche sucks. Perhaps it's a combination of purity, defect and doping problems, perhaps it's more intrinsic.

Blue and white LEDs can't tolerate any reverse breakdown. They're rated for

5V (like all LEDs are), typically handle 30V or more, and fail with an ohmic short (a pinhole failure).

(Red LEDs are modestly capable as avalanche diodes, though you have to sort them: they drop anywhere from 20 to 200V in breakdown!)

Say, did you ever try cracking open a GaN FET and forward-biasing its gate to see it glow blue? If you have any duds laying around, you might still be able to dump enough junction current to see it (or not).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Yup -- while the voltage drop of SCRs and IGBTs does scale with the voltage rating, it's not proportional, as it is with MOSFETs. A 300V IGBT (commonly used for plasma display pulse drivers) drops a hair over 1V at ratings, and an SCR of the same rating might drop a hair less than 1V. A 100V IGBT can't hope to compete, though, which is why they aren't available in such low voltages.

On the other hand, kilovolt IGBTs and SCRs have fantastic efficiency, something MOSFETs can't hope to match (at least until GaN pushes up there, hmm!).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Why not put your 'catcher' (snubber, back-EMF, flyback, etc.) diode in? Wired directly on the solenoid it will clamp the back EMF to around

0.7VDC and the ringing will dampen almost instantly. Considering that these are used in switching supplies with very high frequencies I see no problems. A Shottky diode will give you a faster response if that is needed.

We use those all the time in my industry, just just need a diode with sufficient inrush current handling capability.

John

A standard "catcher" diode slo-o-o-ows the solenoid closure, though it's usually just fine for most relay applications.

My usual approach is to use the switching device also as an active clamp, usually 2X supply which gives toff = ton (provided there isn't a big inductance change between open and closed). ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| STV, Queen Creek, AZ 85142    Skype: skypeanalog |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 

             I'm looking for work... see my website. 

          Thinking outside the box... elegant solutions.

I have thought about this and Mr Doom hasn't provided the spec for the actual turn and turn off times (or I haven't seen them).

We don't know the inductance, if we did we would immediately know the turn on time. Yes I know solenoid inductance is a function of position.

If nominal T = L/R is acceptable the turn on is sorted.

Turn off depends on the highest voltage acceptable at the device, be it a MOSFET or IGBT.

IGBTs allow the highest voltage rating if turn-off time is critical. The simplest circuit is a catch diode in series with a resistor, which at the holding current gives the device say half rated voltage.

--
Mike Perkins 
Video Solutions Ltd 
www.videosolutions.ltd.uk

Shottky diode and a small resistor in series may alleviate the slowness of the release action on solenoids. Now if there was some way to tune the OP's circuit so it behaved more like a flyback/switch-mode supply clamp to reuse the energy...but this is getting beyond my skill set. Still I can see using the turnoff of one solenoid releasing its energy to help provide power to the next one. This is a cycle after all, each solenoid is fired in exactly the same sequence and only the delay between each fire is changed to control the engine speed. Or have I lost it completely?

John

Yeah, specifically the flux required (volts * time taken until current reaches quiescent, i.e. limited by DCR), and whatever equivalent (average) inductance that comes out to.

Yup, which is why IGBTs are great for ignition coils -- fast turn-off and high flyback voltage make for a strong spark. The low cost and relatively low voltage drop (compared to a 400V+ MOSFET) is big win. (The ~2V drop isn't so great relative to cranking voltage conditions -- but there's more than enough current available with the coil DCR already, so it's okay.)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

you could return it to the supply but that would require something like two diodes and two fets one of them on the high side

Den tirsdag den 24. januar 2017 kl. 01.16.58 UTC+1 skrev Tim Williams:

looking at a random Bosch ignition coil it is something like 500mR and a max current of 8A

many cars now use coils with a build igbt

See "Self-Inverting-CD-Ignition.pdf" on the S.E.D/Schematics page of my website.

Though I drew it up for display here in 2001 I actually implemented it in 1971... before there was a PowerMOS to do the switching job.

Works easily at 4V crank.

The concept actually wasn't mine, it was a fellow named Josephson (*) at Philco-Ford when I was there in 1968, that division was promptly closed, so the idea lay fallow until I was at Dickson Electronics in Scottsdale, AZ, and implemented a control system which could cope with the low beta (~3 at 5a) of such high voltage devices of that era.

(*) Not the Josephson junction person ;-) ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| STV, Queen Creek, AZ 85142    Skype: skypeanalog |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 

             I'm looking for work... see my website. 

          Thinking outside the box... elegant solutions.

Well put 100k ohm in series... do you have a part number for blue/white led's that break down at 30V?

George H.

If I sent you some "single photon" red led's would you look at them? AND-113R? there's a 114 too... the one with a "clear" lens is better.

George H.

We are about to test some of the EPC parts that are BGA unpackaged chips. I'll see if they glow blue.

I din't want to use them in production. They are too tiny.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com