Author Topic: Why are 900V semiconductors not common? When 1200 are.  (Read 2143 times)

0 Members and 1 Guest are viewing this topic.

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Why are 900V semiconductors not common? When 1200 are.
« on: September 22, 2021, 11:27:47 am »
Hi. Just little thought.
Why are almost no semiconductors with a 900V rating?
It might be my Eurocentric view where 400V AC mains is industry standard.
But it looks wastefully to me to use 1200V rated ones on 600V DC
With SiC it is not a huge issue, but all Si parts have way worse parameters, IGBTs and Diodes are slow, MOSFETS tend to have variants about that voltage because 1200V is unreachable for them
600V rating makes clear sense, it is nice for 230V AC mains
But why they came with 1200V? Other than it is double the 600V.
Because then 1700V is also relatively common and not directly to 2400V.
 

Offline PartialDischarge

  • Super Contributor
  • ***
  • Posts: 1625
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #1 on: September 22, 2021, 12:47:52 pm »
This paper explains why you need a 1200V rated semiconductor for a 400vac line and so on.

https://library.e.abb.com/public/6f03cdd0f2264ff48f2992e62497dd5a/Voltage%20ratings%20of%20high%20power%20_%205SYA%202051NLay.pdf


To give you another data point, for a 6.6kVrms line, a standard medium voltage distribution voltage, semiconductors, stacked, that can handle up to 20kVpk are needed
« Last Edit: September 22, 2021, 02:34:31 pm by MasterTech »
 

Offline Alti

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #2 on: September 22, 2021, 01:16:08 pm »
600V rating makes clear sense, it is nice for 230V AC mains
For 230VAC and 600V rated semiconductors won't last too long. It is close to impossible to limit mains incoming voltage to 600V during 8us/20us surge test. The least stringent surge test is 1500V via 2R (category I installation) IIRC. You could filter 230VAC low pass RLC and make a ~400VDC, then 600V semiconductors (transistors) should suffice under some conditions, but not directly at AC mains.

I am writing this just to make sure people are not putting 600V mosfets at 230VAC mains.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #3 on: September 22, 2021, 03:20:27 pm »
I've seen plenty of 500 and 550V MOSFETs in 240V products...

Note that most things use a big chunky capacitor before or at the MOSFETs, absorbing the surge.  This does depend on the value, and relative to other things like the CMC's leakage and resistance, and NTC if applicable (which can be helpful in smaller converters, but is too little resistance to help much over say 100W, but capacitance might not be enough until 500W+).

900V Si MOSFETs are common enough, perhaps more common back in the day because conventional VDMOS had specific Rds(on) going as Vds^2 -- so the performance at 900V was already about half that of an equivalent geometry 600V part, and half again at 1200V, making the latter particularly unfavorable.  I spec'd such a part back in 2011 or so, for exactly that reason.  SuperJunction types have quickly taken over since then (they were already rising at that time, alas just not available in the size and voltage needed for that project), and offer linear performance at all ratings -- that is, constant switching area (Id * Vds) per die area, no nasty quadratic penalty.  So higher voltage devices aren't unusually poor, and 600-700V and 1200V parts are very common now.

And we have SiC now too, with an even more aggressive performance curve (the dies are TEENY!).  I forget if they're doing SJ in them now, but even without, they offer great performance at 1200V and up.  At 600 and 900V, you might simply not need the performance edge and Si is fine; or at 600V and below, power GaN is available if you need even higher performance.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Alti

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #4 on: September 22, 2021, 04:06:35 pm »
Typically thyristors and triacs intended for 230VAC applications have 900V rated Vdsm/Vrsm (non-repetitive). For example popular BTA16-800 from ST.
Unless some serious RLC is capable of attenuating surges, but that is limited to low-ish powers to be practical.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #5 on: September 22, 2021, 04:25:19 pm »
600V rating makes clear sense, it is nice for 230V AC mains
For 230VAC and 600V rated semiconductors won't last too long. It is close to impossible to limit mains incoming voltage to 600V during 8us/20us surge test. The least stringent surge test is 1500V via 2R (category I installation) IIRC. You could filter 230VAC low pass RLC and make a ~400VDC, then 600V semiconductors (transistors) should suffice under some conditions, but not directly at AC mains.

I am writing this just to make sure people are not putting 600V mosfets at 230VAC mains.
IME 90% of 230V power supplies use MOSFET of BJT rated for 600V or below.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #6 on: September 22, 2021, 04:37:25 pm »
Typically thyristors and triacs intended for 230VAC applications have 900V rated Vdsm/Vrsm (non-repetitive). For example popular BTA16-800 from ST.
I usually see 600V triacs in 230V products and they work just fine. However what I really don't like and what often fails are 400V optocoupler triac drivers like MOC3021 used in 230V application. They often start to self activate after some time.
« Last Edit: September 22, 2021, 05:24:21 pm by wraper »
 

Offline MiyukiTopic starter

  • Frequent Contributor
  • **
  • Posts: 908
  • Country: cz
    • Me on youtube
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #7 on: September 22, 2021, 04:47:39 pm »
Typically thyristors and triacs intended for 230VAC applications have 900V rated Vdsm/Vrsm (non-repetitive). For example popular BTA16-800 from ST.
I usually see 600V triacs in 230V products and they work just fine. However what I really don't like and what often fails are 400V optocoupler triac drivers like MOC3041 used in 230V application. They often start to self activate after some time.
They have it in the datasheet, they should be also 600V rated (it see the same voltage as the main triac)

From Fairchild datasheet:
The MOC301XM and MOC302XM series
They are designed for interfacing between electronic
controls and power triacs to control resistive and inductive loads for
115 VAC operations.
« Last Edit: September 22, 2021, 04:51:53 pm by Miyuki »
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #8 on: September 22, 2021, 05:24:10 pm »
They have it in the datasheet, they should be also 600V rated (it see the same voltage as the main triac)

From Fairchild datasheet:
The MOC301XM and MOC302XM series
They are designed for interfacing between electronic
controls and power triacs to control resistive and inductive loads for
115 VAC operations.

In some datasheets they are described as such, in some they aren't. But I've seen plenty of them in 230V devices. As if MOC302X is a standard go to part for random phase control @ 230V, like in DMX controllers/dimmers. Zero crossing applications for some reason usually have more appropriate parts.
« Last Edit: September 22, 2021, 05:30:26 pm by wraper »
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #9 on: October 05, 2021, 06:15:45 pm »
Here is an example. PCB from a very expensive Elma ultrasonic cleaner (230V). 400V MOC3043 opto-triac (white 6 leg part) driving a 600V BTA16-600BW triac. Picture taken from another thread but on my desk I have the same PCB with differently populated components from another model.
 
The following users thanked this post: Alti

Offline Alti

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #10 on: October 11, 2021, 06:11:07 pm »
Here is an example.
Lets define what mains rated component is. YMMV. If you take a 20V shottky diode from secondary of mains powered AC/DC 5VDC wallwart adapter then this component is not subjected to mains voltages. Similar situation happens when you RLC filter all mains crap, boost voltage and put some transistor after that. Or if you put a CLC pi filter after bridge rectifier (as in your picture) - this buck converter does not have to withstand mains transients and surges (in the sense of IEC 61000-4).

Same applies to triac. It can be placed in series with load directly across mains but also it might be protected by some RLC, snubber, varistor or other filter. If there were only 400V optotriac on this board then I suspect you did not show us all the picture and there was an input filter that plugs to this board. Or maybe this RC fiter placed in parallel with optotriac/triac/load does the job of limiting surges. RC filter (part of buck filter) is low impedance for high dV/dt so it is going to clamp transients (it has full bridge rectification so it works for both polarities).

Concluding - this 400V optotriac might not be subjected to mains conditions and might not have to be rated to withstand these since it (with triac and load) might be in parallel with chunky RC filter.

Or it might be a poor design - I cannot rule that out.



 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #11 on: October 11, 2021, 06:33:00 pm »
Same applies to triac. It can be placed in series with load directly across mains but also it might be protected by some RLC, snubber, varistor or other filter. If there were only 400V optotriac on this board then I suspect you did not show us all the picture and there was an input filter that plugs to this board. Or maybe this RC fiter placed in parallel with optotriac/triac/load does the job of limiting surges. RC filter (part of buck filter) is low impedance for high dV/dt so it is going to clamp transients (it has full bridge rectification so it works for both polarities).

Concluding - this 400V optotriac might not be subjected to mains conditions and might not have to be rated to withstand these since it (with triac and load) might be in parallel with chunky RC filter.
Datasheet clearly states this optotriac is for 115V mains voltage. Mains filter is  2 Y caps to the earth, 2 ferrite beads, a varistor, X cap, a common mode choke, again 2 Y caps to the earth. This thing works without too much failures only because this is a zero-cross optotriac which is more resilient against overvoltage. If it was a random-phase type, once in the field and years pass, a large number of them would be randomly activating a heater which is directly powered from 230V AC (I have seen plenty of that in other devices). Varistor sort of helps, however it's not likely to clamp everything below 400V if transients arrive. Pictures from a service manual:




« Last Edit: October 11, 2021, 06:46:02 pm by wraper »
 
The following users thanked this post: Alti

Offline Alti

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #12 on: October 11, 2021, 08:14:14 pm »
This is symmetrical CLC (C4+L1+C1/2) low pass and then S10K250. Indeed, one cannot expect S10K250 itself to clamp anything below 700V so I suspect it is this 1uF cap (low ESR?) that does most of the clamping job. Still, this is designed to comply with regulations so I am guessing it is enough and resulting waveform does not exceed Vdsm at any time. Vdsm might be/should be higher than Vdrm (400V) and for triacs is usually Vdsm=Vdrm+100V but for optotriacs Vdsm is rarely mentioned in datasheets. Also, if something is indended for specific application, it does not necessarily mean one cannot use it reliably in a different one (with serious filtering).


This thing works without too much failures only because this is a zero-cross optotriac which is more resilient against overvoltage.
I have not seen any application notes or datasheets that confirm this observation. Any reference?

 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #13 on: October 11, 2021, 08:26:26 pm »
I have not seen any application notes or datasheets that confirm this observation. Any reference?
It's not from datasheets, it's from my experience of repairing stuff. IMHO it's because zero-cross optotriacs by design are made to not activate above about 20V. And the most common failure mode of 400V rapid-phase optotriacs I've seen is sporadically self activating when applied AC voltage reaches its peak.
 

Offline Alti

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: 00
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #14 on: October 12, 2021, 09:36:38 am »
And the most common failure mode of 400V rapid-phase optotriacs I've seen is sporadically self activating when applied AC voltage reaches its peak.

I have never heard about reputable optotriacs and triacs sporadically turn on (=exceed parameters specified in the datasheet). This suggests some kind of gradual wear-out or abuse. I know that these turn on either by pulling a gate/illuminating by LED (typical way), or after exceeding Vdsm, or by exceeding (dV/dt)s but that is specified in the datasheet. Triggering via Vdsm is kind of not characterized but take a look at TOPtriac for which it is. Anyway, BTA16-600BW from ST has dV/dt specified as 1kV/us minimum, at 125oC. Had one tried faster edge, the triac might turn on, this is normal and not destructive.

So if you do not limit the optotriac current adequately (usually Itsm is 1A 100us pulse) hoping to trigger it in SW only at low voltages, then sporadic 500V peak might not destroy it through Vdsm but by killing the triac die (exceeding Itsm), or blowing the limiting resistor.

As for (dV/dt)s, this turn on mode applies to any type of optotriacs and triacs - you can trigger those at any voltage and at any point, even zero-crossing is going to turn on at random angle after exceeding (dV/dt)s.

I suspect that these 400V optos might get strained by turning on by exceeding Vdsm and as a consequence exceeding either Itsm or (dI/dt)c or both (by inadequate limiting impedance of a BTA16 gate pull). IMHO there is nothing wrong in turning a heater element sporadically ON during transients in such ultrasonic cleaner application, 10ms won't make any practical difference.

So 400V opto is ok, if you know what you are doing.
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9331
  • Country: fi
Re: Why are 900V semiconductors not common? When 1200 are.
« Reply #15 on: October 12, 2021, 09:55:35 am »
Depending on country, 230VAC line is expected to go up to 253VAC periodically, this condition is perfectly legal and acceptable. UK has 240VAC nominal and while tolerances are tighter with 254VAC max, 260VAC is not reportedly unusual.

This is already around 367V peak. With 400V parts, there is basically no margin (just 9%).

You can't filter this out using passive filters, because you have to dimension the filter not to activate at 370V, otherwise the filter would overheat and blow up during high (but barely within, or slightly exceeding spec) mains. A lot of blown units will ensue.

Good luck shopping any passive filter component which does not conduct at 370V, but shunts significant currents still below 400V. This would need to be an active crowbar or series pass regulator circuit with voltage reference and comparator driving semiconductors, just to filter. It's less costly to just use said properly dimensioned semiconductors in the actual circuit.

400V rated parts are just simply unsuitable for 230VAC mains, but fine for 115VAC countries. What wraper shows is a clear design mistake, no way around that. It's only matter of place and time when the 400V rating is exceeded, and then all bets are of if the part survives or not, and how it reacts; it doesn't necessarily release magic smoke but can misbehave like described. Operation beyond maximum ratings is never specified in datasheets so that doesn't help.

Regarding original question, 900V semiconductors are fine for 400V rectified three-phase when suitably filtered. The market for such parts is huge, so I guess the answer has to lie within manufacturing processes; possibly 900V rated part is not always much different to manufacture, or they get 1200V rating "for free" even if aiming for 900V. Another reason might be, 400V 3-phase systems are used in rough industrial environments where part cost is secondary consideration, so higher margins are in use. Whereas it's normal to use 550V part for consumer 230VAC device (i.e., 550/(230*1.1*1.41) ~ 54% margin), 900V part for 400VAC application would incur a tad less margin (900/(400*1.1*1.41) ~ 45%). This is undesirable, even if easily filterable, so 1200V part is easier to design with robustly.
« Last Edit: October 12, 2021, 10:02:41 am by Siwastaja »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf