Author Topic: Power bipolar transistor die -> parallel???  (Read 3707 times)

0 Members and 1 Guest are viewing this topic.

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1989
  • Country: de
    • Richis-Lab
Power bipolar transistor die -> parallel???
« on: November 10, 2019, 11:02:40 pm »

Hi all,


I did some pictures of a Sanken-Transistor (2SC2922):

https://www.richis-lab.de/2SC2922.htm


It seems that the transistor consists of a lot of small transistors connected in parallel:




But how does this work? Paralleling bipolar transistors (without resistors at the emitters) leads to more current for the hottest transistor and so to failure of this transistor. You know what I mean.


Can anyone explain why the engineers were able to connect these transistors in parallel? Or is there a misunderstanding on my side?


Greeting,


Richard

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Power bipolar transistor die -> parallel???
« Reply #1 on: November 10, 2019, 11:20:00 pm »
Since the transistors are on the same die they are largely the same. However hotspots will develop and there is a distinct difference between transistors made for switching and audio applications. The former have a very small DC SOA.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Power bipolar transistor die -> parallel???
« Reply #2 on: November 10, 2019, 11:27:07 pm »
My guess is they rely on 1) good match between smaller bjts because they are on the same die 2) thermal coupling between bjts.
 

Offline m3vuv

  • Super Contributor
  • ***
  • !
  • Posts: 1738
  • Country: gb
Re: Power bipolar transistor die -> parallel???
« Reply #3 on: November 11, 2019, 02:29:45 am »
rf power transistors have been made this way for years,a lot of times they dont totaly fail just several elements,ie transistors on the die,leading to reduced output.
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5569
  • Country: us
Re: Power bipolar transistor die -> parallel???
« Reply #4 on: November 11, 2019, 03:08:28 am »
I don't know about this one, but some have had small emitter resistors included on the die.  But as far as I know most depend primarily on the similarity of all of these transistors that are very close on the same die.
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1989
  • Country: de
    • Richis-Lab
Re: Power bipolar transistor die -> parallel???
« Reply #5 on: November 11, 2019, 08:08:05 am »
Thanks for all you input!  :-+


Since the transistors are on the same die they are largely the same. However hotspots will develop and there is a distinct difference between transistors made for switching and audio applications. The former have a very small DC SOA.

I know this relation but the 2SC2922 is a typical transistor for an audio amplifier, so it always has to deal with linear operation. That particularly puzzled me.  :-//


My guess is they rely on 1) good match between smaller bjts because they are on the same die 2) thermal coupling between bjts.

I also had this guess.
The "thermal instability" of bipolar Transistors seems to be not as critical as I thought.


rf power transistors have been made this way for years,a lot of times they dont totaly fail just several elements,ie transistors on the die,leading to reduced output.

Interesting! Never heard of this!
But I don´t think these audio power transistors were designed this way.


I don't know about this one, but some have had small emitter resistors included on the die.  But as far as I know most depend primarily on the similarity of all of these transistors that are very close on the same die.

I also thought of little emitter resistor designed into the die. Technically that should be no problem but it will add a lot of power loss and that is absolutely not something you want to have in such a transistor.  :-//
It seems that the similarity is the important point.


The good old 2N3055 in contrast doesn´t use this technique:


https://www.richis-lab.de/2N3055.htm

Perhaps the 2SC2922-type of integration is more complicated in production and control of parameters.
Actually the 2SC2922 has a marking explaining the hfe of the device: It can reach from 30 to 180!
https://www.semicon.sanken-ele.co.jp/sk_content/2sc2922_ds_en.pdf
« Last Edit: November 11, 2019, 08:13:47 am by Noopy »
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Power bipolar transistor die -> parallel???
« Reply #6 on: November 11, 2019, 09:15:13 am »
The "thermal instability" of bipolar Transistors seems to be not as critical as I thought.

Idk, but I observed that Hfe at first rises with temperature, but then goes down when it becomes hot. How hot? Hot-enough to burn my fingers, didn't really do comprehensive testing. I observed this behavior on several different bjts. I think one of them was 2n3904 from onsemi. Could someone else confirm this behavior?
 

Offline rfeecs

  • Frequent Contributor
  • **
  • Posts: 807
  • Country: us
Re: Power bipolar transistor die -> parallel???
« Reply #7 on: November 11, 2019, 06:55:36 pm »
This transistor has been described as a Ring Emitter Transistor:
https://www.diyaudio.com/forums/solid-state/53508-skinny-sanken-2sa1216-epitaxials.html

Quote
Sanken's 2SA1216, and its counterpart , the 2SC2922, are ring-emitter transistors.
They also used to be called multi-emitter devices.

Sanken also produces the 2SA1215 and 2SC2921, 150 watt versions.
Nowadays the planar epitaxial technique is solely connected to Sanken.
Sanken produces a great number of other planar epitaxial devices in MT200, e.g. the 2SA1494/2SC3858.
Toshiba produced ring-emitter devices for a relative short period: the 2SA1095 and 2SC2565.

Simply put, a ring-emitter device consists of a great number of transistors on 1 die.
Which translates in a large SOA with an unusual curving, and a very high bandwidth.
The 1216/2922 have a 40 MHz bandwidth, the 1215/2921 do 50 MHZ.
Toshiba's 150 watt 2SA1095/2SC2565 devices even higher at 60 MHz.
High bandwidth translates in high speed, in return this requires less feedback correction.
Less feedback is favorable for low TIM distortion.
Ring-emitter's have a high Hfe, which reduces the current capability of the driver devices, and consequently less current gain from the front end of the amplifier.

Fujitsu described their Ring Emitter Transistors as "constructed with mulitple emitters connected through diffused ballast resistors which provide uniform current density".

There may be diffused ballast resistors there that are not visible.
 
The following users thanked this post: Noopy

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1989
  • Country: de
    • Richis-Lab
Re: Power bipolar transistor die -> parallel???
« Reply #8 on: November 11, 2019, 08:05:44 pm »
Very interesting!
Thanks a lot!
 :-+ :-+ :-+

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3362
  • Country: au
Re: Power bipolar transistor die -> parallel???
« Reply #9 on: November 11, 2019, 09:05:31 pm »
I don't know about this one, but some have had small emitter resistors included on the die.  But as far as I know most depend primarily on the similarity of all of these transistors that are very close on the same die.
Even if they were all identical I imagine there would still be some risk. For example, with just one large die it’s characteristics ought to be uniform across its area but they can develop hot spots.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Power bipolar transistor die -> parallel???
« Reply #10 on: November 11, 2019, 10:05:36 pm »
Consider a finite die on an ideal heatsink.  Consider the sub-transistors to be perfectly matched, evenly distributed over the surface, and operating at some load power.

The key insight is this: the transistors in the center are surrounded by other transistors, while those at the edge are open to one side.

Silicon has finite thermal conductivity, so there will be a difference in temperature rise over the die, with a hot spot in the middle.

This is still missing one more bit of info, actually I think -- the transistors cannot run right up to the edge of the die, there must be some buffer space there for dicing and guard rings.  This makes a ring of silicon around the edge which is not dissipating power, but is thermally conductive, thus lowering the temperature around the edge.

This could be mitigated by placing cells slightly more densely around the periphery, or with extra emitter resistance of course.  Preferably with PTC, though I doubt there's anything readily available (i.e., doped or poly-Si, or aluminum metallization, and that's about it..) that has quite a strong enough tempco to compensate Vbe.

The initial or low-power hot spotting in this scenario is very modest, but it is exponentiated by the tempco of the device, so there will always be some point where the exponent goes critical and one transistor (preferentially, one near the center) will hog it all and fail.

What matters, of course, is whether that critical point is even on the SOA.  Ideally it's not.  Most audio power BJTs have it towards the edge, not quite off the plot but leaving enough area that it's still plenty useful.  Old MOSFETs did (thanks to a poor power density), previous generation MOSFETs largely didn't (high power density, relatively high exponent -- hence the distinction between switching and linear FETs), and current generation (SJ) FETs again seem to (which is pretty amazing given their yet higher power density; I'm not sure what it is that drives this, but in any case, I guess it gives them a low exponent).

And since the driving force is temp drop across the die -- simply running at low power helps.  Fullpak types are largely good for rated power without 2nd breakdown effects, simply because they're limited to a paltry 30W or so; whereas their metal-tab versions may suffer from it as usual.  A SOT-23 part is very unlikely to have such problems, its die is tiny (under 1mm?) and can only dissipate a fraction of a watt.

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

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3362
  • Country: au
Re: Power bipolar transistor die -> parallel???
« Reply #11 on: November 11, 2019, 11:32:17 pm »
Old MOSFETs did (thanks to a poor power density), previous generation MOSFETs largely didn't (high power density, relatively high exponent -- hence the distinction between switching and linear FETs), and current generation (SJ) FETs
Tim
Not trying to derail the thread, but an important and often misunderstood point about mosfets - many people will tell you that they current share across the die area because if one part of the die gets hotter the on resistance increases and causes the hot area to conduct less. All well and good if the mosfet is fully switched on, but in linear operation (which of course is the situation where they are more likely to get hot) the increase in on resistance when hot counts for practically nothing, whereas the gate threshold voltage reduces, causing that hot part of the die to turn on even harder....  :bullshit:  :scared:
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Power bipolar transistor die -> parallel???
« Reply #12 on: November 12, 2019, 02:12:56 am »
Yeah, easy to misread the various appnotes and think it's one way or the other, when actually it's both, conditionally.  The two operating regimes of the MOSFET have different (more or less opposite) characteristics. :)

A lot of people are also stuck thinking that 2nd breakdown is some immutable foundation of BJTs, or MOSFETs, or both.  Nah; as I noted above -- there are numerous designs that perform well, and that do not.  Check the SOA, it contains all the info you need*!

*Although it can vary (nonlinearly) with Tc, and you almost only ever see the SOA at 25°C.  So it's not clear if it'll be alright under other conditions.  Budgeting for a prototype testing or component qualification phase is always a good idea in such a case.

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

Online magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
Re: Power bipolar transistor die -> parallel???
« Reply #13 on: November 12, 2019, 08:09:29 am »
I think some early MOSFETs (lateral perhaps?) were truly free of second breakdown and that's why people keep repeating this mantra till this day. And it's not just switching people who talk about it, but also audio.
I suppose it is a matter of RDS(on) being high enough that its tempco makes more difference than Vth tempco.
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5569
  • Country: us
Re: Power bipolar transistor die -> parallel???
« Reply #14 on: November 12, 2019, 04:42:29 pm »
Consider a finite die on an ideal heatsink.  Consider the sub-transistors to be perfectly matched, evenly distributed over the surface, and operating at some load power.

The key insight is this: the transistors in the center are surrounded by other transistors, while those at the edge are open to one side.

Silicon has finite thermal conductivity, so there will be a difference in temperature rise over the die, with a hot spot in the middle.

This is still missing one more bit of info, actually I think -- the transistors cannot run right up to the edge of the die, there must be some buffer space there for dicing and guard rings.  This makes a ring of silicon around the edge which is not dissipating power, but is thermally conductive, thus lowering the temperature around the edge.

This could be mitigated by placing cells slightly more densely around the periphery, or with extra emitter resistance of course.  Preferably with PTC, though I doubt there's anything readily available (i.e., doped or poly-Si, or aluminum metallization, and that's about it..) that has quite a strong enough tempco to compensate Vbe.

The initial or low-power hot spotting in this scenario is very modest, but it is exponentiated by the tempco of the device, so there will always be some point where the exponent goes critical and one transistor (preferentially, one near the center) will hog it all and fail.

What matters, of course, is whether that critical point is even on the SOA.  Ideally it's not.  Most audio power BJTs have it towards the edge, not quite off the plot but leaving enough area that it's still plenty useful.  Old MOSFETs did (thanks to a poor power density), previous generation MOSFETs largely didn't (high power density, relatively high exponent -- hence the distinction between switching and linear FETs), and current generation (SJ) FETs again seem to (which is pretty amazing given their yet higher power density; I'm not sure what it is that drives this, but in any case, I guess it gives them a low exponent).

And since the driving force is temp drop across the die -- simply running at low power helps.  Fullpak types are largely good for rated power without 2nd breakdown effects, simply because they're limited to a paltry 30W or so; whereas their metal-tab versions may suffer from it as usual.  A SOT-23 part is very unlikely to have such problems, its die is tiny (under 1mm?) and can only dissipate a fraction of a watt.

Tim

Though I am not aware of anyone actually doing this, with today's silicon it wouldn't be too hard to implement a resistive heater ring around the perimeter, closed loop controlled to match the center transistors temperature.  Providing a very close approximation to an isothermal environment.    Perhaps the best implementation of this would be to use the boundary transistors as heaters.  And it is totally not clear that such a part would have any market advantage.

Patent trolls note.  This constitutes public release of this concept. 
 

Online Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: Power bipolar transistor die -> parallel???
« Reply #15 on: November 12, 2019, 04:50:31 pm »
Yesterday I have dug through a lot of different small high voltage depletion mode mosfets. Some of the datasheets claimed they are of DMOS type and also free from 2nd breakdown.  :-//

For example this one: http://ww1.microchip.com/downloads/en/DeviceDoc/LND150%20C041114.pdf

Is that really true or not?  Those Supertex/Microchip datasheets for these depletion mode trannies are often pisspure junk with incomplete information, sometimes even lacking characteristic plots completely (I am talking you, DN3145!) or having the characteristics incomplete, measured in areas the part will never be likely used at.   |O
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Power bipolar transistor die -> parallel???
« Reply #16 on: November 12, 2019, 06:56:23 pm »
For example this one: http://ww1.microchip.com/downloads/en/DeviceDoc/LND150%20C041114.pdf

Ha! Just recently I was looking for a depletion-mode fet. Found none back then. Here you are!
Looking at characteristics, it feels it's just a poor jfet. I can imagine it's free from secondary breakdown because maximum current is just 13mA continius, 30mA pulsed, maximum voltages is +-20V. Try kill something with that! Gate leakage current is also not stellar.

Afaik, I've seen jfet for large currents, like for 5 or 15 amps. But they are rare.
 

Online Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: Power bipolar transistor die -> parallel???
« Reply #17 on: November 12, 2019, 07:44:24 pm »
No, it is not a JFET. It is a MOS-FET, it has insulated gate. But sure, both types are depletion mode.

There are surprisingly quite many high voltage small signal depletion fets out there, just have a search.  (BSS126, LND150, DN3145, BSP135, ...)

Sure, LDN150 is very low current one, but its specifically good for small high voltage current sources (as well as a very good analog input protection circuitry). These are not switching high current types.

But this is a bit offtopic here.
« Last Edit: November 12, 2019, 07:46:12 pm by Yansi »
 

Online magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
Re: Power bipolar transistor die -> parallel???
« Reply #18 on: November 12, 2019, 08:55:42 pm »
Afaik, I've seen jfet for large currents, like for 5 or 15 amps. But they are rare.
You have probably seen silicon carbide power JFETs. They are relatively new and pricey and AFAIK used mainly for switching high voltage with low RDS(on), although of course some exotic manufacturers are building audio amplifiers with them too ::)
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Power bipolar transistor die -> parallel???
« Reply #19 on: November 13, 2019, 10:07:06 am »
Ah, I confused maximum gate voltage with Vds, sorry :(.

I wonder what are the advantages of depletion mode mosfet vs jfet. Seems jfet supposed to have higher gate leakage, better works as a current source, and gate will conduct if reverse biased. Anything else?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Power bipolar transistor die -> parallel???
« Reply #20 on: November 13, 2019, 10:47:02 am »
gate will conduct if reverse biased. Anything else?

Well, forward biased.  Well, both I suppose.  Well... so will MOSFETs, in fact you get increased leakage near Vgs(max), it's just that it's not enough current to prevent it from going pop (it's not an avalanche breakdown phenomenon), and then you have a MOSFET that's very leaky and doesn't turn off all the way. :P

On that note, never tried JFETs (of any sort) in avalanche breakdown.  Wonder if it's okay for small amounts, or large amounts too, if it has a breakover effect like BJTs, or if's just destructive.

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

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Power bipolar transistor die -> parallel???
« Reply #21 on: November 13, 2019, 12:28:21 pm »
The 2SC2922 is a a ring or perforated emitter transistor.  This is not quite the same thing as multiple smaller ballasted transistors in parallel which is common for RF transistors and integrated power transistors.

The special emitter structure prevents current crowding so bandwidth and current gain are improved at high currents.  The safe operating area is also improved and they have a reduced or sometimes no secondary breakdown region.  These characteristics can be used to identify them if the manufacturer does not otherwise say.  Apparently they also have very low base and emitter resistances making them low noise and low saturation.  What is not to like?

They are very popular for audio output stages but also work well in switching applications and high performance linear regulators.  On Semiconductor makes a lot of them.  I think the old Zetex Super E-Line transistors were constructed this way which is why TO-92 sized parts can be found which can switch several amps.
 
The following users thanked this post: exe, Noopy

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1989
  • Country: de
    • Richis-Lab
Re: Power bipolar transistor die -> parallel???
« Reply #22 on: November 13, 2019, 10:58:49 pm »
The 2SC2922 is a a ring or perforated emitter transistor...

Thanks for the explanation!  :-+


On that note, never tried JFETs (of any sort) in avalanche breakdown.  Wonder if it's okay for small amounts, or large amounts too, if it has a breakover effect like BJTs, or if's just destructive.

I did some research regarding the LF355 in the Input circuit of the Gould 4074 (https://www.richis-lab.de/Gould407X.htm).

The LF355 has J-FET-Inputs. The Application Note 447 from National Semiconductors (http://www.ti.com/cn/lit/an/snoa736/snoa736.pdf) says:

"If either input is pulled above V+, nothing happens until the
difference between the input and V+ gets near the breakdown
voltage, typically 50V. At this point, the FET’s
gate-source junction avalanches and will draw all the current
it can. Limiting this input current to something less than 3 mA
helps prevent damage."


and

"Failure to clamp the voltage or limit the current adequately
may not destroy the part, but the offset voltage and bias current
will be permanently degraded."
« Last Edit: November 13, 2019, 11:01:09 pm by Noopy »
 
The following users thanked this post: T3sl4co1l

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Power bipolar transistor die -> parallel???
« Reply #23 on: November 14, 2019, 02:19:15 am »
"If either input is pulled above V+, nothing happens until the
difference between the input and V+ gets near the breakdown
voltage, typically 50V. At this point, the FET’s
gate-source junction avalanches and will draw all the current
it can. Limiting this input current to something less than 3 mA
helps prevent damage."


and

"Failure to clamp the voltage or limit the current adequately
may not destroy the part, but the offset voltage and bias current
will be permanently degraded."


Fascinating, an unusually candid explanation.  I would guess IC FETs to be perhaps a bit more consistent than discrete ones; so, not that they'll necessarily handle any more current (discrete ones are typically rated for ~mA as well, under forward bias that is), but that they give a hand-waved chance of failure (rather than "outside of warranty / you're on your own", say) beyond that is unusual.

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


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf