9 Jun
2009
9 Jun
'09
9:38 p.m.
Subject: Re: Transistor Substitution
From: ard at p850ug1.demon.co.uk (Tony Duell)
Date: Tue, 09 Jun 2009 18:53:39 +0100 (BST)
To: cctalk at classiccmp.org
Way too much information.. What he needs to know is what can he buy now that
should work to replace a 2n3009?
Answer: the 2N2222A (metal can not the plastic PN2222) has the same or close
enough FT, Ic and breakdown voltages. It would be my first choice if I could
not purchase/salvage the exact part. Also the 2n2222A is available and usually
cheap. I buy them usually in for groups of 25 for about $0.08US each.
DEC used a lot of 2n2907, 2n2905. 2n2219, 2n3553 and 2n2222A equivilents under
different marking schemes. These parts are still very common.
To assure the substitution it would help to know how that 2n3009 was used
but from working on a lot of DEC gear the 2n2222 is a good bet.
The only case where they type transistor is a bit fussy is some of the faster
flip chip cards (logic) and SMPS.
FYI: Older DEC machines like straight-8 and friends used a lot fo the ceramic
epoxy parts similar to the 2n3638 and 2n3563. Those are ceramic headers with
a block of epoxy covering the die and are prone to popping the epoxy and the
symptom works when cold or mechanically intermittent.
Allison
Okay: I admit it, I am sometimes pig-ignorant about basic hardware
questions. :/
I have a couple of DEC machines which I need to replace a few components
on, and also stock up spares of others. With the transistors and diodes,
however, I often can't find a direct replacement - and don't know how to
figure out what a modern substitute is.
For a 2N3009, for example, I can find basic information and a datasheet
online easily enough - but as for choosing a functional, available
substitute for it, I'm honestly not even sure where to begin.
Is there a basic resource for determining modern equivalents for older
transistors and diodes? Can someone helpfully provide me with a clue
here? :)
It depends on what the transistor is being used for.. In general
transsitors used in the logic (particulalry low-speed ones, for things
like lamp drivers) and transsitors used in linear PSUs are fairly easy to
substitute. Transsitors used for high energy pulse work (SMPSU choppers,
horizontal output transsitors in monitors) are a lot more critical, and
the data sheet may bot help.
Basically, you need to get the following right :
Tpye/polarity (NPN or PNP etc)
VOltage ratings high enough to stand voltages in the circuit (if the
ratings of the substitute exceed those of the original, you should be OK)
Current ratings (particularly Ic, collector current). Again, if your
substitute is better than the original it should be OK
Gain (hfe and all its varients) is not too critical in most classic
computer applications, particularly for switich transiustors which are
driven hard into saturation. But try not to use a transistor of too low
a gain.
Max frequency (Ft, etc) is something you should look at in logic
transistors, particularly in higher speed circuits, clock oscillators, etc.
But my experience is if you pick something of the same polarity and
similar characteristics, it'll work in most classic computer circuits
(except for SMPSUs and horizontal output stages). Probaly 99% of
small-signal transistors can be replaced y 2N3904 (NPN) and 2N3906 (PNP) :-)
Diodes are evem more generic. Most switching diodes can be replaced with
1N4148 :-). For PSU rectiifers in linear PSUs (again, SMPSUs are another
story), choose something with sufficient votlage rating (PIV -- peak
inverse votlage) (2*sqrt(2)*output voltage should be safe, say 3 times the
output voltage or more) and sufficient current rating (If, forward current).
When there are similar diodes of different PIV ratings, I normally buy
the highest. It's not much more expensive, and I'm on the side of safety.
-tony 
10 Jun
10 Jun
2:21 p.m.
On Tue, Jun 9, 2009 at 5:38 PM, Allison<ajp166 at bellatlantic.net> wrote:
Way too much information.. ?What he needs to know is
what can he buy now that
should work to replace a 2n3009?
I could use that info, too, since I also have older DEC equipment that
uses the 2N3009.
Answer: the 2N2222A (metal can not the plastic PN2222)
has the same or close
enough FT, Ic and breakdown voltages. ?It would be my first choice if I could
not purchase/salvage the exact part. Also the 2n2222A is available and usually
cheap. ?I buy them usually in for groups of 25 for about $0.08US each.
What is different about the metal can and the plastic case part? I
have a few of each, and have so-far interchanged them in circuits
where performance isn't an issue (i.e., where a 2N3904 would work
fine, too). Obviously 2N2222s aren't expensive, but I'd like to know
why I should set aside the metal-cased ones for repairing specific
items.
The only case where they type transistor is a bit
fussy is some of the faster
flip chip cards (logic) and SMPS.
What categories of FLIP-CHIPs are you thinking of when you say
"faster"? I have three R-series CPUs and a few DF-32s. I have yet to
delve into component-level repair, but I know I'll need to soon.
FYI: Older DEC machines like straight-8 and friends
used a lot fo the ceramic
epoxy parts similar to the 2n3638 and 2n3563.
Good to know (since I have more than one item from that era).
Those are ceramic headers with
a block of epoxy covering the die and are prone to popping the epoxy and the
symptom works when cold or mechanically intermittent.
Is that visually apparent, or is that an internal fault? If I
identified a bad transistor and removed it from a FLIP-CHIP, would it
be difficult to set up something with an oscilloscope to watch some
output then tap on the transistor to watch the trace "change" to
verify mechanical fragility?
Thanks,
-ethan
1:57 p.m.
Ethan Dicks wrote:
some of the older boards that used transistors for logic. one place
you would see those is
in an 8l or straight 8 and part of the core memory subsystem of the
omnibus machines.
Its rare to see the top epoxy pop off but it happens. usually the part
either fails
or is most often thermally intermittent.
Lot's of years working with transistors from that era and noo small
amount of DEC scrap
boards during my time with DEC for spares and as resource.
Allison
On Tue, Jun 9, 2009 at 5:38 PM, Allison<ajp166 at
bellatlantic.net> wrote:
They are essentially the same die but due to mounting and the absence of
the metal the
pn2222 does not stand he same Ic and power dissapation. if you really
need the plastic
part you might as well use the 2n3904.
Way too much information.. What he needs to know
is what can he buy now that
should work to replace a 2n3009?
I could use that info, too, since I also have older DEC equipment that
uses the 2N3009.
Answer: the 2N2222A (metal can not the plastic
PN2222) has the same or close
enough FT, Ic and breakdown voltages. It would be my first choice if I could
not purchase/salvage the exact part. Also the 2n2222A is available and usually
cheap. I buy them usually in for groups of 25 for about $0.08US each.
What is different about the metal can and the plastic case part? I
have a few of each, and have so-far interchanged them in circuits
where performance isn't an issue (i.e., where a 2N3904 would work
fine, too). Obviously 2N2222s aren't expensive, but I'd like to know
why I should set aside the metal-cased ones for repairing specific
items.
The only case
where they type transistor is a bit fussy is some of the faster
flip chip cards (logic) and SMPS.
What categories of FLIP-CHIPs are you thinking of when you say
"faster"? I have three R-series CPUs and a few DF-32s. I have yet to
delve into component-level repair, but I know I'll need to soon.
FYI: Older DEC
machines like straight-8 and friends used a lot fo the ceramic
epoxy parts similar to the 2n3638 and 2n3563.
Good to know (since I have more than one item from that era).
Those are ceramic headers with
a block of epoxy covering the die and are prone to popping the epoxy and the
symptom works when cold or mechanically intermittent.
Is that visually apparent, or is that an internal fault? If I
identified a bad transistor and removed it from a FLIP-CHIP, would it
be difficult to set up something with an oscilloscope to watch some
output then tap on the transistor to watch the trace "change" to
verify mechanical fragility?
Thanks,
-ethan
5:43 p.m.
Thanks for all the replies so far. While what I've seen so far is
definitely helpful, particularly regarding the 2N3009, I'd like to take a
step back and generalize the question a bit.
Let's say you have a system where something dies, and you trace it down to
a particular transistor. You do a search for the part number, and find out
that particular part is no longer manufactured and/or available. You find
a datasheet, though,, and between that and the circuit you get a pretty
reasonable idea what the transistor's specifications ought to be like. So
far, so good.
The problem I run into at this point is how to go about finding a modern
substitute. Some of you are up on your specs for modern transisitors to
the point where you can just casually say, "Oh, well, I know a part that
ought to be able to handle that," and you're home free. It's admirable,
and I'm glad many of you are in that position. Unfortunately I'm not, and
I suspect I'm not the only one. :/ I'm _hoping_ there's an alternative
to just blind-referencing thousands of transistor specs in hopes of
running across a suitable one. (I suppose "repeatedly pestering those of
you who _have_ kept up" is an alternative, but somehow it doesn't seem
like the most acceptable one. <g>) *
Soooooo...: I guess my real question is, once you know what properties
you want in a transistor, how do you go about finding a transistor that's
currently available which matches or exceeds those specs? Is it simply a
question of plowing through datasheets, or are there resources available
to help narrow down the search to a handful of possibly-suitable parts?
...It's not just the 2N3009 I'm after, obviously. :)
-O.-
* - If somebody who _has_ kept up were to offer a "Transistor
Cross-Reference Service" for a slight fee, I suspect the hobbyist market
could keep them in fairly adequate beer-money. :)
11 Jun
11 Jun
5:42 p.m.
Thanks for all the replies so far. While what I've seen so far is
definitely helpful, particularly regarding the 2N3009, I'd like to take a
step back and generalize the question a bit.
Let's say you have a system where something dies, and you trace it down to
a particular transistor. You do a search for the part number, and find out
that particular part is no longer manufactured and/or available. You find
a datasheet, though,, and between that and the circuit you get a pretty
reasonable idea what the transistor's specifications ought to be like. So
far, so good.
The problem I run into at this point is how to go about finding a modern
substitute. Some of you are up on your specs for modern transisitors to
the point where you can just casually say, "Oh, well, I know a part that
ought to be able to handle that," and you're home free. It's admirable,
That's the easy case :-). You tend to remember the common transistors
(2N2222, 2N3904, 2N3906, 2N2219, 2N2907) and look up the specs of those
first (if you can't remember them).
Waht I used to do (I'll explain the 'used' in a minute) was based on the
obvious fact that the substitute was only useful if I could get it. So
I'd gra the catalogues for 2 or 3 suppliers and look down the list of
transistors (normally only 2 or 3 pages) which included brief specs to
see which would be suitable. If I was in doubt, I'd look up more detailed
data on the couple of 'possible' devices.
Why 'used to'? Well, you can't get paper catalouges from some suppliers
any more. They expect you to use their web page. And I find it impossible
to find transistors that way. The web pages were not set up by engineers,
or even people remotely clueful in electronics. And you get some sillies
(like some transistors being listed witha maximum collector current of
1A others 1000mA. Of course a search will find one and not the other...)
I've kept paper catalouges from about 10 years ago, I look in those for
suitable candidates and hope they're still available...
-tony
12 Jun
12 Jun
9:59 a.m.
New subject: [personal] Re: Transistor Substitution
----- Original Message -----
From: "Tony Duell" <ard at p850ug1.demon.co.uk>
To: <cctalk at classiccmp.org>
Sent: Thursday, June 11, 2009 6:42 PM
That's the easy case :-). You tend to remember the common transistors
(2N2222, 2N3904, 2N3906, 2N2219, 2N2907) and look up the specs of those
first (if you can't remember them).
If it helps anyone (probably in UK the most) there is a stock of old parts
at - http://www.mclennan.co.uk/stockdisposalmenu.html - eg qty 3500 x
2N3906, also with a few TTL and CMOS Ic's, all brand new.
I run that site, I'm not trying to advertise, but if there is anything there
for list members let me know, I'll bypass the requirements at the top of the
page, it's better you use the bits than they sit on the shelves.
Mike
Web - www.soemtron.org
PDP-7 - www.soemtron.org/pdp7.html
Email - mike at soemtron.org
Looking for a PDP-7 (some hope!) and an ASR33
What I used to do (I'll explain the 'used' in a minute) was based on the
obvious fact that the substitute was only useful if I could get it. So
I'd grab the catalogues for 2 or 3 suppliers and look down the list of
transistors (normally only 2 or 3 pages) which included brief specs to
see which would be suitable. If I was in doubt, I'd look up more detailed
data on the couple of 'possible' devices.
Why 'used to'? Well, you can't get paper catalouges from some suppliers
any more. They expect you to use their web page. And I find it impossible
to find transistors that way. The web pages were not set up by engineers,
or even people remotely clueful in electronics. And you get some sillies
(like some transistors being listed witha maximum collector current of
1A others 1000mA. Of course a search will find one and not the other...)
I've kept paper catalouges from about 10 years ago, I look in those for
suitable candidates and hope they're still available...
-tony
10 Jun
10 Jun
6 p.m.
Allison wrote:
Its rare to see the top epoxy pop off but it happens. usually the part
either fails
or is most often thermally intermittent.
TMK, only Fairchild made those packages, although they had JEDEC numbers
(TO-105 and TO-106).
I kind of liked them somehow, perhaps a nostalgia thing. (Never seen them fail
mechanically, myself.) In all sorts of commercial equipment around the mid-60's
to 70's. 2N3638 was very popular.
FYI: Older DEC machines like straight-8 and friends
used a lot fo the ceramic
epoxy parts similar to the 2n3638 and 2n3563.
Those are ceramic headers with
a block of epoxy covering the die and are prone to popping the epoxy and the
symptom works when cold or mechanically intermittent.
Is that visually apparent, or is that an internal fault? If I
identified a bad transistor and removed it from a FLIP-CHIP, would it
be difficult to set up something with an oscilloscope to watch some
output then tap on the transistor to watch the trace "change" to
verify mechanical fragility?
6:09 p.m.
On Jun 10, 2009, at 2:00 PM, Brent Hilpert wrote:
TMK, only Fairchild made those packages, although they
had JEDEC
numbers
(TO-105 and TO-106).
I kind of liked them somehow, perhaps a nostalgia thing. (Never
seen them fail
mechanically, myself.) In all sorts of commercial equipment around
the mid-60's
to 70's. 2N3638 was very popular.
I got a bag of 2N3638s at a hamfest when I was a kid. I think
they're the first transistors I ever did anything fun with, with the
exception of the 2SA/2SB-somethingorothers in the Radio Shack 65-in-1
kit.
-Dave
--
Dave McGuire
Port Charlotte, FL
7:17 p.m.
Ethan Dicks schrieb:
On Tue, Jun 9, 2009 at 5:38 PM, Allison<ajp166 at
bellatlantic.net> wrote:
What is different about the metal can and the plastic case part? I
have a few of each, and have so-far interchanged them in circuits
where performance isn't an issue (i.e., where a 2N3904 would work
fine, too). Obviously 2N2222s aren't expensive, but I'd like to know
why I should set aside the metal-cased ones for repairing specific
items.
Older transistors were generally put into a metal, and sometimes into a
ceramic package, case
for manufacturing reasons and better thermal distribution. Even earlier
transistor cases were
made of glass, painted black (you can make good photo transistors from
old OCxxx devices
when scratching the black paint away).
Plastic packages came up around late 60s when industry found materials
with both good insulation
and good thermal conduction. Nowadays, putting a semiconductor chip into
a casted plastic case
for low power and low frequencies is mainly a factor of production
costs, so if power dissipation
permits there is no longer a reason to prefer a metal case over a
plastic one. Same holds for
old eight-legged opamp ICs which came in a round metal case. For a few
high frequency applications,
though, the metal case makes a Faraday cage which is connected to ground
then. And still, if a lot of
power has to be carried away, the thermal resistance is lower for metal
obivously. For the work horses
2N2222 and 2N3904, this is no real argument though, the 2N3055 with its
15A maximum Ic there are
pretty many Watts to be distributed.
--
Holger
7:11 p.m.
>> I have a couple of DEC machines which I need
to replace a few components
>> on, and also stock up spares of others. With the transistors and diodes,
>> however, I often can't find a direct replacement - and don't know how to
>> figure out what a modern substitute is.
>>
>> For a 2N3009, for example, I can find basic information and a datasheet
>> online easily enough - but as for choosing a functional, available
>> substitute for it, I'm honestly not even sure where to begin.
[...]
Way too much information.. What he needs to know is
what can he buy now that
should work to replace a 2n3009?
There is a saying in England :
'Give a man a fish, you feed him for a day. Teach a man to fish, you feed
him for life'.
I try to use that principle when posting here. I try to show how solve
similar problems in the future. None of us will be around for ever [1]
and I think it's important the information, methods, 'tricks', etc get
passed on
[1] It would only take one careless mistake when repairing an SMPSU...
As the bit of the original message I've left above semems to suggest, the
2N3309 is an example, and the OP possibly needs to substitute other
transistors as well.
The only case where they type transistor is a bit
fussy is some of the faster
flip chip cards (logic) and SMPS.
Maybe in DEC equipment... But I can assure you that the HP9100 is
'touchy'. It's not particularly high-speed, it's certainly not an SMPSU,
but you will have 'fun' working out what transistors to use.
-tony
11 Jun
11 Jun
10:38 p.m.
Tony Duell wrote:
There is a saying in England :
'Give a man a fish, you feed him for a day. Teach a man to fish, you feed
him for life'.
It has made it across the pond, but with some modifications:
'Give a man a fish, you feed him for a day.
Teach a man to fish, and he'll sit in a boat all day drinking beer.
Teach a fisherman to create an artificial shortage of fish, and he'll
eat steak.'
Warren
5689
days inactive
5692
days old
10 comments
10 participants
participants (10)
-
ajp166＠bellatlantic.net -
ajp166＠verizon.net -
ard＠p850ug1.demon.co.uk -
ethan.dicks＠gmail.com -
hilpert＠cs.ubc.ca -
holger.veit＠iais.fraunhofer.de -
lists＠databasics.us -
mcguire＠neurotica.com -
mike＠brickfieldspark.org -
ohh＠panix.com