Home
JAQForum Ver 24.01
Log In or Join  
Active Topics
Local Time 22:36 27 Nov 2024 Privacy Policy
Jump to

Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.

Forum Index : Electronics : Guidelines for derating MOSFET datasheets

Author Message
LadyN

Guru

Joined: 26/01/2019
Location: United States
Posts: 408
Posted: 10:45pm 02 May 2019
Copy link to clipboard 
Print this post

As I start to design circuits using MOSFETs, I need to understand guidelines for derating MOSFET datasheets and I look at you all to get this valuable input.

Let's consider one of the favorites:

HY4008P: At 80в, it can support upto 200A.

The datasheet claims the 200A rating is valid at 25C, but at 200A, it's likely to be dissipating (40000 * 0.003) = 120W of power so unless all of it is dissipated, the die temperature will increase

The TO220 packaging itself restricts Ids to 90A (as per datasheet) and following the curve we see at 100C, it still should carry 60A.

So what margins and derating should a designer follow?
 
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 4406
Posted: 11:08pm 02 May 2019
Copy link to clipboard 
Print this post

Its purely a judgement call.

Work it out for the worst expected ambient temperature using whatever heatsink you plan to use, and call that the absolute maximum recommended figure.

How far below that you wish to run, depends on your appetite for risk.

Nobody can say for sure what that risk is.

The idea that you can run a 100 watt rated device at 99.9 watts and have total reliability, but 100.1 watts will destroy it instantly is just not realistic.Edited by Warpspeed 2019-05-04
Cheers,  Tony.
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 12:12am 03 May 2019
Copy link to clipboard 
Print this post

These guys are very careful ladyN with very good reason

that is one of the reasons what they tell you is as good as gold

I have four of the 247 hy4008 on My Little inverter

I have ran the heck out of it now but how long will it last who knows

I run a 5000 btu ac for hours at a time and run power tools at same time

That's 4 mosfets so I now think 24 is overkill and been told by experts that it increases the likelihood of failure

So I'm in the same boat as you how many is enough but not too many to increase likelihood of failure

I know for a fact I will not go over 12 no matter what and that's for my whole house

Another thing that the experts have said that makes really good sense is how long are you going to be able to run these giant wattages anyways

Are you going to have a battery Bank the size of Texas I know my pockets ain't that deep

If you have big loads it's best to run in during the day when the sun's out and you can backfeed a grid-tied inverter 2 up your wattage and charge batteries

These are all things that make really good sense to me that I've read

What I would like to know is how many of the four mosfets are actually working at one particular Moment In Time

I know it is not all four but is it two of them or just one

If I knew that I can figure out what wattage and amperage I am pushing through how many mosfets

and be able to get a better understanding on how many amps these things will actually do at normal temperatures

be warned i am good parrot but Dumber than a box of rocks
 
LadyN

Guru

Joined: 26/01/2019
Location: United States
Posts: 408
Posted: 12:26am 03 May 2019
Copy link to clipboard 
Print this post

My concerns are many about paralleling MOSFETs.

I barely understand a single MOSFET, so paralleling MOSFETs increases the time I need to make something.

Like all engineering decisions, there are trades to be made with the devil here.

Paralleling MOSFETs connect their Rds in parallel that reduces their net Rds, if they loadshare equally.

That also parallels their capacitances, which in parallel, increases them.

Also, each additional lead and PCB trace is additional capacitance and inductance to worry about.

Furthermore, we would like to keep TVS diodes across the gate and DS to try and increase their reliability, which again, adds capacitance and inductance to the mix.

This, on the whole, severely complicates the drive circuitry as well as debugging issues when things go wrong, as they always do.

I would like to figure out how to minimize complications as much as possible.
 
Solar Mike
Guru

Joined: 08/02/2015
Location: New Zealand
Posts: 1138
Posted: 12:42am 03 May 2019
Copy link to clipboard 
Print this post


The device spec under consideration has to be treated as just a part of the complete system your are engineering, assuming the data sheets are true then it is the system that you are putting it in that will determine how many devices to use based on the individual spec.

eg, Currently I am building a small 24 volt DC to 230 vac portable inverter, using a LF H-Bridge, I have size constraints on the packaged unit, these determine the physical size of the heat sinks I can use. Looking in the RS data sheets there are some good heat sinks with known spec and would seem to be ok to use as the rear panel of my case. So my heat sink is 0.5 C\watt in still air (I don't want to use fans)

Lets assume the max average current required is 200 amps from the 24v battery, and we are using the HY4008W TO-247 device because they are inexpensive.
Ignoring device switching transition losses.
200 amps rms per full full sine wave = 200 amps per diagonal series pair.
2 devices in series @ 0.003R = 0.006R * 200 * 200 = 240 watts for 1/2 the period, so the total power for 4 devices in the bridge = 240w or 60w per device.

The spec "Thermal Resistance-Junction to Case" = 0.38 so the internal temp would increase to 60*0.38 = 23C + whatever temp the heat sink + ambient, the heat sink temp with 240w = 0.5*240 = 120C, ambient = 35C so total temp on the die now increases to 23 + 120 + 35 = 178C, BANG!

If 6 devices are paralleled their combined RDson = 0.0005 so now the total H-Bridge power loss is 200*200*(0.0005+0.0005) = 40 watts averaged between 24 devices, clearly this will work, our heat sink will only warm up 20C above ambient and the mosfets are happy.

But..., there is always a but, if the parallel devices don't switch on\off at the same time, then the first device to so could destroy itself, leading to cascade failure and explosion. More devices we have in parallel generally means bigger separation between each device and the driver chip, separation = time delay and other issues, so it becomes increasingly critical to engineer a circuit layout to work. Our big driver chip wont work in this situation, we need to split out the drivers so each mosfets or perhaps pair of mosfets have a driver mounted with little separation, these drivers can now happily be driven from a master device as the drive currents to them are low and like logic signals.

There is another But.. now that our 6 mosfets are all switching within nano seconds of each other they can then start current sharing that 200 amp load, as they heat up their RDson also increases and assuming they are all at the same temperature things will be rosy. BUT it is difficult to get larger numbers of mosfets to be mounted on a heat sink such that their temperatures are even, ones on the ends may run cooler than in the middle etc, differences in device temperature will mean lower temperature ones will hog more current and if they are running too close to max spec may also go bang.

Clearly building high power H-Bridges is not easy.


Cheers
Mike




 
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 4406
Posted: 01:20am 03 May 2019
Copy link to clipboard 
Print this post

For a low frequency square wave inverter its only really the dc conditions and conduction losses that require consideration for thermal design and de-rating.

A high frequency PWM inverter or switching power supply will also have switching losses which might add considerably to the thermal design problem. Its often a case of building the thing, and then doing some very careful testing and measurement.

It should then be possible to compare measured results with theoretical expectations, and if the numbers all come out fairly close, to be reasonably confident that all is well.

Quite often something does not work out as expected, and then the cause needs to be identified. It may be a sub standard component, or the circuit is behaving in an unexpected way.
Cheers,  Tony.
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 01:24am 03 May 2019
Copy link to clipboard 
Print this post

Goodness gracious I wish I could wrap my head around that

I did wrap my head around that last phrase and I'd have to agree 100%

I really thought that Electronics was going to be the same as everything else has been in my life jump in there and eventually I'll get it

Now I understand it takes a lifetime Too Perfect

That's why I have decided I am not try to design Anything

I am perfectly happy riding the coattails of others that don't mind me doing so

My biggest problem right now is a power jack inverter trying to come up with a new control board for it

then I'm onto a warp speed and inverter I hope I am up to it

I have no problem making and winding Transformers

Other than that it has to be monkey see monkey doEdited by BenandAmber 2019-05-04
be warned i am good parrot but Dumber than a box of rocks
 
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 4406
Posted: 01:30am 03 May 2019
Copy link to clipboard 
Print this post

Ben, some of us here have been doing this for fifty years for our day job, and its a continual steady learning process that goes on and on and never really stops.
Cheers,  Tony.
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 01:35am 03 May 2019
Copy link to clipboard 
Print this post

I would like to also copy warpspeeds little inverter one day also

If you haven't seen it it's well worth looking into really cool

All the respect in the world go to you guys your job is definitely not easy

Warpspeed and all the other great minds on hereEdited by BenandAmber 2019-05-04
be warned i am good parrot but Dumber than a box of rocks
 
LadyN

Guru

Joined: 26/01/2019
Location: United States
Posts: 408
Posted: 01:38am 03 May 2019
Copy link to clipboard 
Print this post

IF you don't get to build it before I do, I will help you in any way I can.

It might very well be I'll have to learn how to make toroids from scratch, from you, but I have not progressed to the AC section yet.

This is all still in the DC section, infact at the very OR section of the Warpverter that I'm working on.Edited by LadyN 2019-05-04
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 01:56am 03 May 2019
Copy link to clipboard 
Print this post

Thanks ladyN me and Amber are finishing our RV remodel and getting all the solar battery stuff in

I have been working on it and using that awesome little inverter that poida set me up with

I know everybody's tired of hearing about it but that little thing is just awesome

I will be using it even after I get my big one done I'll probably use this one at night for the very low idle current

It will run the fridge lights and TVs computers

that's what I was saying earlier it don't take that big of an inverter

until you go heating water cooking heating your house drying clothes

if someone didn't have a problem using natural gas or propane they could really get by cheap on the solar side of things

That's my opinion anyway
be warned i am good parrot but Dumber than a box of rocks
 
Revlac

Guru

Joined: 31/12/2016
Location: Australia
Posts: 1026
Posted: 03:33am 03 May 2019
Copy link to clipboard 
Print this post

Think this might help a bit. AN-1140 Has an example as well.

Cheers Aaron
Off The Grid
 
Chopperp

Guru

Joined: 03/01/2018
Location: Australia
Posts: 1057
Posted: 07:23am 03 May 2019
Copy link to clipboard 
Print this post

@Aaron,

Good bit of bedtime reading there. Interesting.

Thanks for the post.
ChopperP
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 01:27am 04 May 2019
Copy link to clipboard 
Print this post

I think it goes back to warpspeed said

Get in there build something and do lots of testing

Then you'll have a good idea on what you can get by with

Or you can go with a design that someone with a lifetime of experience made why re-invent the wheel



And if you want to change it up just a little bit here and there might be a lot easier than coming up with one yourself

I bought a board that was already made and had loads of fun with it

And not only that it made me realize that there are still good people out there in the world

I was starting to lose my hope


Edited by BenandAmber 2019-05-05
be warned i am good parrot but Dumber than a box of rocks
 
LadyN

Guru

Joined: 26/01/2019
Location: United States
Posts: 408
Posted: 05:09pm 06 May 2019
Copy link to clipboard 
Print this post

While we are having a discussion about derating the current rating

Is there a similar process for voltage derating?
 
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 4406
Posted: 10:01pm 06 May 2019
Copy link to clipboard 
Print this post

  LadyN said   While we are having a discussion about derating the current rating

Is there a similar process for voltage derating?

Not usually.
There is a maximum voltage rating in the data sheet, and anything above that the device usually violently breaks down and is permanently destroyed.

Counterfeit chips sometimes die at a lower voltage than expected due to poor quality control. Good devices will fail over a fairly narrow voltage range amazingly close to their maximum rating.

Some mosfets are avalanche rated, but most are not. The avalanche rated mosfets will behave like a high voltage zener and go into non destructive conduction above the maximum rated voltage.

This does have some serious limitations, because the power dissipation will be extremely high under avalanche conditions. So its only usable to swallow very brief occasional voltage spikes, and only if the avalanche current can be limited so something reasonable by having sufficient series impedance.
Cheers,  Tony.
 
BenandAmber
Guru

Joined: 16/02/2019
Location: United States
Posts: 961
Posted: 05:20pm 07 May 2019
Copy link to clipboard 
Print this post

Reviac that is a very good link.
(AN-1140)that you gave for newbies like me to help understand how mosfets work in their limitations I recommend anybody new to read it

Thanks to all of you guys for helping us newbies learn and understand
be warned i am good parrot but Dumber than a box of rocks
 
Revlac

Guru

Joined: 31/12/2016
Location: Australia
Posts: 1026
Posted: 01:59pm 08 May 2019
Copy link to clipboard 
Print this post

There are App notes around for many electronic components, It can sometimes be difficult understanding them, some of them get a bit technical.

Just keeping with the topic.

This one AN11599 Using power MOSFETs in parallel

Section 3. MOSFET mounting.
This had me thinking, In some inverters I have seen all the mosfets crammed together as close as possible on the heatsink, and the heatsink was small and heats up easily, I thought that was just to save space, looks like there is a bit more to it.

And Section 6. Gate drive considerations

Anyway its a good idea to collect a few relevant app notes for reference.... I usually look for them after thing's go wrong.
Cheers Aaron
Off The Grid
 
Print this page


To reply to this topic, you need to log in.

© JAQ Software 2024