Author Topic: Convert transistor switching time to frequency  (Read 1256 times)

0 Members and 1 Guest are viewing this topic.

Offline gkmaiaTopic starter

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: nz
  • Electronics Hobbyist
Convert transistor switching time to frequency
« on: May 13, 2019, 02:42:53 am »
I've got a vintage transistor datasheet that shows on and off times for the switching in microseconds. 

But I would like to convert that to MHZ so I can compare to my other transistors.

If I were to sum the time it takes to turn on, store and off it would be around 2uS and that is pretty much 0.5Mhz which is a bit weird.

If there a way to define frequency base on on/off switching time?
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Convert transistor switching time to frequency
« Reply #1 on: May 13, 2019, 03:47:23 am »
They are not really comparable.  Switching times assume saturated operation and current-gain-bandwidth product assumes linear operation.  Add to this that anti-saturation will remove storage time and there are actually three different sets of specifications.

Other characteristics like output capacitance sometimes take on more importance in switching applications.

Incidentally, you might think that gold-doped RF transistors would perform well as gold-doped fast saturated switches but my experience is that they do not.  There must be some fundamental difference in construction.

Now if you have a switching application where you know frequency and duty cycle, the switching characteristics tell you everything you need to know.
 
The following users thanked this post: gkmaia

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Convert transistor switching time to frequency
« Reply #2 on: May 13, 2019, 01:43:04 pm »
Broadly speaking, I would guess low MHz, maybe fractional MHz.

But that's just a guess.  David's spot on, and trying to make any better of a guess really isn't meaningful.

Why, what do you need fT for?  Or Fsw, or whatever frequency it is you're after?

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: gkmaia

Offline gkmaiaTopic starter

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: nz
  • Electronics Hobbyist
Re: Convert transistor switching time to frequency
« Reply #3 on: May 13, 2019, 07:59:22 pm »
Basically I harvested heaps of N1F4M, N1L3M, AA1F4M, AA1F4N transistors from an old stereo system.

https://pdf1.alldatasheet.com/datasheet-pdf/view/112804/NEC/AN1F4M.html

But their datasheets do not mention frequency.

And usually when I need to mach a transistor for replacement one of the things I check is the frequency.

Without knowing what frequency these harvested transistors are I can't really use them.

Unless I learn how to define their frequency based on the datasheet... I just do not have the knowledge to do that. But I could learn if there is a way to.
 

Offline nwaelder

  • Newbie
  • Posts: 1
Re: Convert transistor switching time to frequency
« Reply #4 on: May 13, 2019, 08:47:03 pm »
The relation for one dominant lag is:
f(-3db) = 0.35/Tr where Tr is the rise time measured between 10% and 90% of the step input.

Lookup risetime-bandwidth relation.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Convert transistor switching time to frequency
« Reply #5 on: May 13, 2019, 11:17:09 pm »
Ah...

They'll probably replace TIP31/32 or 2N3055 sorts of things.  Within voltage, current and power ratings of course.

I'd just throw 'em out, they're pretty weak spec.  Or, perhaps only useful because they're so bad?  (You can make the mistake of putting in a too-good transistor in a weak circuit, and it oscillates like mad!)

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

Offline bson

  • Supporter
  • ****
  • Posts: 2497
  • Country: us
Re: Convert transistor switching time to frequency
« Reply #6 on: May 14, 2019, 01:59:44 am »
One thing worth paying attention to is the actual test circuit used for measurement.  Does it look like how you plan to use the part?  Can you infer something for your particular uses from it?  If not, or very different, the specs it's used to measure will be meaningless...

 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Convert transistor switching time to frequency
« Reply #7 on: May 14, 2019, 02:32:01 am »
The relation for one dominant lag is:
f(-3db) = 0.35/Tr where Tr is the rise time measured between 10% and 90% of the step input.

Lookup risetime-bandwidth relation.

That works for a linear system and Ft (1) assumes that as it is a small signal parameter but even without other factors, switching is a large signal behavior and the switching specifications reflect this.  There is a relationship but it includes too many other factors to reliably use hfe alone.

Basically I harvested heaps of N1F4M, N1L3M, AA1F4M, AA1F4N transistors from an old stereo system.

https://pdf1.alldatasheet.com/datasheet-pdf/view/112804/NEC/AN1F4M.html

But their datasheets do not mention frequency.

That is true even of some modern datasheets like the MJE371/MJE521.  The BD135 to BD140 series have no specified Ft from ST or Fairchild but the Savant (?) datasheets do specify it.  Can the Savant specifications be trusted?

Quote
And usually when I need to mach a transistor for replacement one of the things I check is the frequency.

Another good thing to check is capacitance but that is not much help here.  Looking at the datasheet you linked, I can see why there is no gain-bandwidth product specification.  The AN1F4M is a biased transistor with internal series and shunt resistors.  It is not really suited for linear applications so switching times are all that matter.

Quote
Without knowing what frequency these harvested transistors are I can't really use them.

Unless I learn how to define their frequency based on the datasheet... I just do not have the knowledge to do that. But I could learn if there is a way to.

For a lot of applications the current-gain-bandwidth product is largely irrelevant.  In principle it is not difficult to measure with slow transistors with just a function generator and oscilloscope.

(1) Actually fT but I am lazy.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf