"Jeff Liebermann" wrote in message
...
On Mon, 18 Apr 2011 23:43:35 -0400, "Barry" wrote:
To begin, it is really a conventional NPN, not an IGBT.
I'm not sure. Getting a Vce(sat) of 2V at 30A is a difficult feat
with most conventional NPN xsistors. Low saturation voltage is the
major benefit of IGBT. However, the original data sheet specifies the
device beta, which is not normally specified with an IGBT, which acts
more like a MOSFET.
From the original data sheet at:
http://www.datasheets.org.uk/T30G40*-datasheet.html
With a saturation voltage of 2V at 30A, it's certainly an IGBT.
Look at the very high base current. This is how the low Vce(sat) is
obtained. Also note the Vbe(sat).
Ib = 5Amps and Vbe(sat) = 2.5V. Now, I'm sure it's an IGBT. Such a
high base voltage and current is a characteristic of IGBT devices. See
typical curves at:
http://en.wikipedia.org/wiki/File:IvsV_IGBT.png
If it were an NPN transistor, the curves would be compressed around
Vbe(sat) = 0.7v to 1.0v (or slightly more if it has emitter current
balancing resistors inside) and have an upward slope.
Much beyond a volt or so, the curves should be linear, i.e. the Shockley
relationship becomes dominated by a pure resistance. I would expect the
device has multiple emitter balancing resistors to keep a hot spot from
forming at high currents.
I'm still undecided if it's an IGBT or a common xsistor, but I'm
begining to agree with you.
I'll guess the package is a TO-264 or TO-3PL
It's a TO-247 case.
I beg to differ (again). The original T30G40 data sheet shows the
package as 20.5 x 26.5 mm with a 3.3mm hole. The TO-247 package is
16.0 x 21 mm with a 3.0mm hole. The TO-264 is 20 x 20 mm with a 3.3mm
hole. Neither is a perfect match, but methinks the TO-264 is closer,
especially due to having the same larger hole size.
http://www.fairchildsemi.com/products/discrete/pdf/to247pdd_dim.pdf
http://www.fairchildsemi.com/products/discrete/pdf/to264.pdf
My bad. I only looked at the lead spacing. With an obsolete part with
unusual specs, matching the case is just icing on the cake. Matching the
electrical specs is usually difficult enough.
Actually starting the search with the package style,
in my opinion, is a poor place to start.
I beg to differ (yet again). Finding a suitable replacement that
doesn't fit in whatever is being repaired is rather useless. By
reducing the search to only those items that can actually be used as a
physical substitute, the search scope is drastically reduced.
You need to match the critical
specs first. In this case, high breakdown voltage, high current, and
low
saturation voltage were the critical specs to match first.
I went for package, Vceo(max), Vce(sat), Ic(max), and Pt(max) in that
order.
I chose Vceo, Ic, Pt, and Vce(sat) and left the case as "icing."
Mounting arrangements can be modified, but you cannot change the
electrical specs.
To avoid all the suspense, look at an OnSemi MJW18020. OnSemi even
has Spice2, PSpice, and Spice3 models as well as a Saber model for it.
While Digi-Key does not stock them, Mouser has them at less than $6 a
pop.
http://www.onsemi.com/pub_link/Collateral/MJW18020-D.PDF
Nope. Other then my objection to the package (hole too small), the
beta is too low. The original has a minimum beta of 80, while this
xsistor shows a minimum of 5.5 at 20A.
Unfortunately the Shindengen device does not list the conditions at which
hFE is measured. The application is an induction heater so hfe (ac) is
more important than hFE (DC). The higher fT of the OnSemi device
suggests that the hfe falls off less rapidly than in the Shindengen
device. With most induction heaters relying on self-excited power
oscillators, the actual hFE should not be that important. Do you
remember in the early days of transistors the exceptionally wide Beta
spreads of a given device? We, as engineers, learned to design circuits
that were not sensitive to Beta spread. Of course my experience is with
BIG vacuum tube dielectric heaters. [Need any NOS 3CX10000A3's?]
Unfortunately, the device I excavated doesn't specify the beta.
http://ixdev.ixys.com/DataSheet/98573.pdf
Argh!
The IGBT is a voltage driven device and the gate only draws current due
to its capacitance, so Beta is a meaningless term with an IGBT. Instead
you use gfs measured in amps/volt (much like gm in a pentode tube).
I know we are dating ourselves here, Jeff, but bipolar transistors are
much like vacuum tubes in that triodes are triodes and with some
changes in circuit values (and rewiring, of course) just about any
triode
can be made to work in a pinch. Actually bipolar transistors generally
are even easier to substitute than vacuum tubes.
I like to think of FET's as a solid state analogue to vacuum tubes. I
was in college when the transition from tube to semiconductors was
taught. Over the 6 years of various skools, I started with tubes,
delved into bipolar, and graduated with FET's. Keeping them untangles
was not a trivial exercise, but I'm comfortable working in all three
areas.
In my case both vacuum tubes and bipolar transistors were taught as an
undergraduate. By graduate school, tubes were dropped and field effect
devices were added. The nice thing about tubes was the fact that they
were usually forgiving of short term mistakes! :-) And then there is
the built-in pilot light!
Do you have an NTE replacement book handy?
Books? They tend to obsolete as soon as they're printed. I prefer
online lookups.
http://www.nteinc.com
Notice how many
transistors can be replaced by the NTE123AP.
http://www.nteinc.com/specs/100to199/pdf/nte123ap.pdf
The online lookup doesn't have a reverse NTE - JEDEC lookup so I
can't check how many the devices the NTE123AP replaces.
I quit counting after I found several hundred devices in the first few
pages. My point is that a small signal NPN will generally substitute for
most other small signal NPN's.
Look up the specs on
a 2N3904. The 2N3904 is far cheaper ($0.07) than the NTE part and can
probably be used anywhere the NTE123AP ($1.25) can be used.
Sure. The NTE123AP is probably a relabeled 2N3904 or 2N2222, both
very common parts. You're also paying for handling, bubble packaging,
cataloging, distribution, and retail floor space for the NTE part. The
2N3904 part is sold by the thousands, while NTE parts tend to be sold
one at a time. Of course the NTE parts are more expensive.
But if you have a 2N3904, or a 2N706, or even a PN2222, why bother to
order the NTE123AP? My point is that unless the circuit is unusual, it
will probably work fine with what you might have on hand.
One other thing when using the Digi-Key or Mouser search engines...
If you need a part with, say, a BVceo of 100 volts or more, you can
highlight 100, 150, 200, and 250 volts at the same time.
Oops. I never noticed that you can select more than one. Thanks.
--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Tomorrow I am off to visit WA5KSC, who I knew in high school, to
troubleshoot an old Heath Marauder (HX-10). After high school, we went
off to different colleges and lost touch. We reconnected after 45 years.
We were best friends then and we are best friends again now. That is one
of the nice things about ham radio. And working on old vacuum tube gear
is still fun!
73, Barry WA4VZQ