Topics

2N2222A vs 2N3904 sourcing

Kees T
 

I think enough work has been done to say the 3 sets of 2N3904 transistors should be replaced with 3 sets of 2N2222A transistors. The 2N2222A is a great transistor but has been around a while and most retailers have classified them as "obsolete" and dumped the plastic versions to surplus dealers to sell. They still sell the TO-18 metal can versions at quite high prices because the military still needs them to repair "stuff". Many surplus dealers now have the plastic 2N2222A parts available for pennies on places like ebay.

Just FYI

73 Kees K5BCQ

Jerry Gaffke
 

Kees,

Otherwise unobtainable plastic 2n2222A's available for pennies on ebay?
In this particular case, I'm not so sure about the first part in "Trust, but verify". 
But certainly possible, the recipe has been out there for 50 years.

Ebay might be worth a shot if you want a bunch.
For small quantities on prototypes, I'd just give Mouser their $2 for item 610-2N2222A
If you really want a bunch, then Mouser item 583-2N2222A from Rectron is $0.21 each at quantity 10000.
All 2n2222A's on Mouser have metal cans, which should dissipate more than anything plastic.
Especially if you put some sort of heatsink on them.
Or give them a liquid nitrogen bath.

There are parts that are better than the 2n2222A's, especially if interested in operating above 20m.
Search for 2n5109 in this forum for discussion of that and other parts.
Maybe also search for BS170 too, Mouser price of $0.44 for one, $0.12 each for 100. 

Jerry
 

On Thu, Sep 13, 2018 at 01:18 PM, Kees T wrote:
I think enough work has been done to say the 3 sets of 2N3904 transistors should be replaced with 3 sets of 2N2222A transistors. The 2N2222A is a great transistor but has been around a while and most retailers have classified them as "obsolete" and dumped the plastic versions to surplus dealers to sell. They still sell the TO-18 metal can versions at quite high prices because the military still needs them to repair "stuff". Many surplus dealers now have the plastic 2N2222A parts available for pennies on places like ebay.

Arv Evans
 

Jerry

Since the FT is what we need to know for transistor swaps, it might be possible to
make an oscillator circuit that tunes across the predicted FT value.  Point where the
oscillator stops or will not start could be a good indication of FT for a particular device.

Motorola AN-139 is interesting:
Is there any other way to actually measure FT of BJT devices?

Arv
_._


On Thu, Sep 13, 2018 at 3:15 PM, Jerry Gaffke via Groups.Io <jgaffke@...> wrote:
Kees,

Otherwise unobtainable plastic 2n2222A's available for pennies on ebay?
In this particular case, I'm not so sure about the first part in "Trust, but verify". 
But certainly possible, the recipe has been out there for 50 years.

Ebay might be worth a shot if you want a bunch.
For small quantities on prototypes, I'd just give Mouser their $2 for item 610-2N2222A
If you really want a bunch, then Mouser item 583-2N2222A from Rectron is $0.21 each at quantity 10000.
All 2n2222A's on Mouser have metal cans, which should dissipate more than anything plastic.
Especially if you put some sort of heatsink on them.
Or give them a liquid nitrogen bath.

There are parts that are better than the 2n2222A's, especially if interested in operating above 20m.
Search for 2n5109 in this forum for discussion of that and other parts.
Maybe also search for BS170 too, Mouser price of $0.44 for one, $0.12 each for 100. 

Jerry
 
On Thu, Sep 13, 2018 at 01:18 PM, Kees T wrote:
I think enough work has been done to say the 3 sets of 2N3904 transistors should be replaced with 3 sets of 2N2222A transistors. The 2N2222A is a great transistor but has been around a while and most retailers have classified them as "obsolete" and dumped the plastic versions to surplus dealers to sell. They still sell the TO-18 metal can versions at quite high prices because the military still needs them to repair "stuff". Many surplus dealers now have the plastic 2N2222A parts available for pennies on places like ebay.


Nick VK4PP
 

On Fri, Sep 14, 2018 at 07:29 AM, Arv Evans wrote:
Is there any other way to actually measure FT of BJT devices?
 
Art, i found this, but its over my head for now: https://community.cadence.com/cadence_blogs_8/b/rf/posts/measuring-transistor-ft

Kees T
 

Jerry,

I just bought 50 of them for $3.58 including postage from a USA seller.

73 Kees K5BCQ

Dr. Flywheel
 

Most semiconductor FAB shops use ring oscillators spread at strategic location on a wafer to characterize transistor behavior parameters. This is one of the most common techniques to validating the health of a complete wafer. Nowadays this applies mostly to P and N channel CMOS transistors; however the technique is still applicable to all transistors.

--Ron   N7FTZ

On Thu, Sep 13, 2018 at 3:20 PM Kees T <windy10605@...> wrote:
Jerry,

I just bought 50 of them for $3.58 including postage from a USA seller.

73 Kees K5BCQ

Ripley
 

Hello Kees,

I replaced the two pair of driver transistors with metal can 2N2222As early on in my build per Allison’s recommendation to even out the power across the bands. What is the third pair of transistors in which you refer?

Thank you as always

Ripley
KD8UYQ

The words are mine but this iPad does what it will with them.

On Sep 13, 2018, at 4:18 PM, Kees T <windy10605@...> wrote:

I think enough work has been done to say the 3 sets of 2N3904 transistors should be replaced with 3 sets of 2N2222A transistors. The 2N2222A is a great transistor but has been around a while and most retailers have classified them as "obsolete" and dumped the plastic versions to surplus dealers to sell. They still sell the TO-18 metal can versions at quite high prices because the military still needs them to repair "stuff". Many surplus dealers now have the plastic 2N2222A parts available for pennies on places like ebay.

Just FYI

73 Kees K5BCQ

Jerry Gaffke
 

$0.0358 each is not bad.
Let us know how this goes.

Moving to a BS170 would take significant changes to the circuit.
A 2n5109 or 2n3866 is going to cost more than $0.0358

I'd probably build a really simple CE amp with 10dB of gain,
feed it 100mhz from the si5351, see if it gives 10dB.

Jerry


On Thu, Sep 13, 2018 at 03:20 PM, Kees T wrote:
I just bought 50 of them for $3.58 including postage from a USA seller.

Kees T
 

Hello Ripley,

Q92/Q93, Q96/Q97, Q911/Q912

73 Kees K5BCQ

Kees T
 

Nick,

The Spec Ft is the same for both 2N2222A and 2N3904 .....300 MHz, but the Ic max for the 2N2222A is more than twice that of the 2N3904. Someone (Allison ?) had mentioned the 2N3904s were really being "pushed". That's why they are paired.

73 Kees K5BCQ

Michael Hagen
 

For those who worry about the 2N3904.

This is a big brother that was used a long time ago, a step up from the 2N2222A, just bigger.

https://www.newark.com/multicomp/2n2219a/transistor-bipolar-npn-40v-800ma/dp/35C0689?CMP=KNC-GUSA-GEN-KWL&mckv=|pcrid|190524310933|&gclid=EAIaIQobChMI-ajT6Za53QIVAqrsCh0EMgKnEAAYAiAAEgIX7vD_BwE

Mouser has them, but at more $ ?

Mike, WA6ISP


On 9/13/2018 4:28 PM, Kees T wrote:
Nick,

The Spec Ft is the same for both 2N2222A and 2N3904 .....300 MHz, but the Ic max for the 2N2222A is more than twice that of the 2N3904. Someone (Allison ?) had mentioned the 2N3904s were really being "pushed". That's why they are paired.

73 Kees K5BCQ

-- 
Mike Hagen, WA6ISP
10917 Bryant Street
Yucaipa, Ca. 92399
(909) 918-0058
PayPal ID  "MotDog@..."
Mike@...

Timothy Fidler
 

All take care on this transistor in terms of what case size you get it in.. it was manufactured in two sizes T0-39 and the smaller TO-18 metal case size.  It appears from the Newark description  that the larger case size is being offered in this case.  The Toroid king in FL has them the TO-18 size at a good price too (and probably at much lower shipping cost ).  Heat sinking - your mileage may vary.  The  2n3053 part that Newark also list as a semi equivalent is also a good part up to at least 20Mhz but may get a little sad after that.  It has been used in plenty of 7 Mhz and 10 Mhz kits in the past.  (NB it is nominally about 50 percent better on HFE than the first part).  The  2n2222A or 2219 is the better  choice for the predriver as they have lower common emitter capacitance which will be a major driver in that position. Obviously if you put in a more capable device you can drop the emitter resistors  (even by putting  a TH resistor on an emitter leg direct to GND for startup experiments ) , within limits as they are preventing thermal runaway .... and ....it is all fun.

ajparent1/kb1gmx
 

Kees,

Yes they are being pushed and for the 5th time in this conference its more than that
the 3904 does not maintain its gain with increased current unlike the 2n2222A.   
Dig back its here.

Another property is some trasustors exhibit higher gain/bandwidth with increasing
current and some don't.  So comparing them on FT is not as simple as it first appears.

The 2n2219 is the same part as 2222 (internal silicon) in a to5/to39 for better heat transfer
and is higher power.   The only difference in those cases is one has a glass filled base
and the other is metal with glass leads seals.

Allison

Jim Tibbits
 

You guys are just getting silly now ...Pushing 3904's indeed ,,,Just breadboarded the DC circuit in the PA (22 ohm emitter resistor, 100 ohm base bias and 1k collector ..Guess what? 20.1 ma ...EMRFD fig 6.140 Q102 has emitter resistor 100 ohm. base bias 680 ohm and collector 2.2k   20 ma bias also, too....

Just a picture of the "tentative" PA amp chain ...nearly flat across the passband (10 meters is about 5-6 db down ..Still playing with replacements for T10,T11 and maybe T9

Jim
Tentative PA amp chain.jpg

ajparent1/kb1gmx
 

On Thu, Sep 13, 2018 at 06:24 PM, Jim Tibbits wrote:
nearly flat across the passband (10 meters is about 5-6 db down)

Jim,

Is it flat or does it fade by 5-6db at the 10m end?   Seems a contradiction.

BTW what output power are you getting.  For the drivers it needs to be a half watt (27dbm).
for the predrivers it will be the stage gain less but likely 50 or more mW (17dbm).
Even q90 has to be good to maybe 3-5mW (about 7dbm).   I say this as small
signal performance is not always a predictor of high power behavior.  And
since the amp has a serious IMD problems over 2W some stages are
collectively are running short.

The amps pass band for the amp is 3.5mhz to 29.99 Its anything but flat.
As there are many (all) that are not flat as in well over 12W at 80m and maybe 2W
at 10m.  IF it were flat to 3db it would be 6W at 10M not 2w.

This is what I've been saying for months.  The only way to get flat from 3-10mhz with 3904s
is dial the stage gain down to maybe 11DB.  The only problem with with all four stages
weighing in at under 43-45 DB total your well below what is required to  get 10W out
from the last mixer to the antenna.  FYI the mixer output at -17dbm is dirty so its not
a good idea to push for more.  Last I checked from -17 to +40dbm is a mere 57db of gain.

You might get 10W if each stage averages 14.4db across the full band width.
Your SA proves that is not going to happen.  Getting 11db and flat is possible
but you run out of gain and there is no physical place to add a stage.

The part does not have the FT and the Gain/bandwidth product to do that.
It also has a gain(hfe) that decreases with collector current.  At full power the
collector current is NOT 20ma. ( I'd expect about 42 to 45ma per transistor).

The exception for that would be q90 as its true class A and 20ma is more
than enough.  But then even 10ma is enough current for 5mW.

Allison

Ripley
 

Thank you Kees

Is it safe to assume that the emitter resistors for Q911 and Q912 are to be changed to 11 ohms as was done with the driver pairs?

 

Ripley

KD8UYQ

 

Sent from Mail for Windows 10

 

From: Kees T
Sent: Thursday, September 13, 2018 7:08 PM
To: BITX20@groups.io
Subject: Re: [BITX20] 2N2222A vs 2N3904 sourcing

 

Hello Ripley,

Q92/Q93, Q96/Q97, Q911/Q912

73 Kees K5BCQ

 

Ralph Mowery
 

I could be wrong about the device, but read someone well known referred to the 2n2222 as the cockroach that would not die as it seems like that one has been around for 50 or more years.

de ku4pt


On Thu, Sep 13, 2018 at 5:15 PM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Kees,

Otherwise unobtainable plastic 2n2222A's available for pennies on ebay?
In this particular case, I'm not so sure about the first part in "Trust, but verify". 
But certainly possible, the recipe has been out there for 50 years.

Ebay might be worth a shot if you want a bunch.
For small quantities on prototypes, I'd just give Mouser their $2 for item 610-2N2222A
If you really want a bunch, then Mouser item 583-2N2222A from Rectron is $0.21 each at quantity 10000.
All 2n2222A's on Mouser have metal cans, which should dissipate more than anything plastic.
Especially if you put some sort of heatsink on them.
Or give them a liquid nitrogen bath.

There are parts that are better than the 2n2222A's, especially if interested in operating above 20m.
Search for 2n5109 in this forum for discussion of that and other parts.
Maybe also search for BS170 too, Mouser price of $0.44 for one, $0.12 each for 100. 

_._,_._,_

jim
 

LOOK at the picture, please ...Thats WHY I labeled EVERYTHING I could think of ...-27DB input ..O DB output Frequency LABEL across bottom...I agree that you will NEVER get flat gain in the PA "as configured" ...Ever hear of "emitter bypassing:

hint that CAPICITOR in the emitter does the trick ....

Btw I breadboarded Q90 and Q912 ...Put a socket so I could SWAP transistors and was NOT able to reproduce your claims about FT ...The ONLY transistor that reacted was one with an FT on 15 mhz...The circuit (on MY bench at least) is being dominated by the lack of emitter capicitor bypassing and by inductance of the transformers ...NOT by any "ft" cruft ...Nor by Current ..again I refer to the figure in EMRFD mentioned in my previous post ...PS every STAGE in that pa is or should be operated as CLASS A ...

Just to be clear ...-5 @ 0 Mhz ...-1 @ 10 Mhz...-1 @ 15 Mhz ....-4 @ 20 Mhz ...-5 @ 25 Mhz...-6 @ 30 Mhz... -8 @ 35 Mhz...-9 @ 40 Mhz
 
Not FLAT ...But a whole lot better than I've seen anywhere else in this group ...Can YOU do as well (hopefully better)




Jim

 

On Thursday, September 13, 2018, 6:59:26 PM PDT, ajparent1/KB1GMX <kb1gmx@...> wrote:


On Thu, Sep 13, 2018 at 06:24 PM, Jim Tibbits wrote:
nearly flat across the passband (10 meters is about 5-6 db down)

Jim,

Is it flat or does it fade by 5-6db at the 10m end?   Seems a contradiction.

BTW what output power are you getting.  For the drivers it needs to be a half watt (27dbm).
for the predrivers it will be the stage gain less but likely 50 or more mW (17dbm).
Even q90 has to be good to maybe 3-5mW (about 7dbm).   I say this as small
signal performance is not always a predictor of high power behavior.  And
since the amp has a serious IMD problems over 2W some stages are
collectively are running short.

The amps pass band for the amp is 3.5mhz to 29.99 Its anything but flat.
As there are many (all) that are not flat as in well over 12W at 80m and maybe 2W
at 10m.  IF it were flat to 3db it would be 6W at 10M not 2w.

This is what I've been saying for months.  The only way to get flat from 3-10mhz with 3904s
is dial the stage gain down to maybe 11DB.  The only problem with with all four stages
weighing in at under 43-45 DB total your well below what is required to  get 10W out
from the last mixer to the antenna.  FYI the mixer output at -17dbm is dirty so its not
a good idea to push for more.  Last I checked from -17 to +40dbm is a mere 57db of gain.

You might get 10W if each stage averages 14.4db across the full band width.
Your SA proves that is not going to happen.  Getting 11db and flat is possible
but you run out of gain and there is no physical place to add a stage.

The part does not have the FT and the Gain/bandwidth product to do that.
It also has a gain(hfe) that decreases with collector current.  At full power the
collector current is NOT 20ma. ( I'd expect about 42 to 45ma per transistor).

The exception for that would be q90 as its true class A and 20ma is more
than enough.  But then even 10ma is enough current for 5mW.

Allison

ajparent1/kb1gmx
 

Yes!

and over 200 posting about it.

And no forcing 3904s to marginal gain.  Throw the damn things out.  If you want
to use them do it right rather than struggling.  First 5 stages at low gain, its all the 
3904 can do 10-11db as you still need about 60db gain over all.

The goal then was flat 10W, at the point where spurs and  hunting other illnesses
took over I was at 12W at 80 and 6-7W at 10M.

Then I noticed at 28mhz we are barely 30db down for the 17mhz spur!  At that
point a useful and functional 10m capability was a waste of effort.

Then there is the carrier leak... and the lowpass filter failure.

Allison
 

ajparent1/kb1gmx
 

Ralph yes right here about 6 wweeks ago.

We repeat everything here at at least 5 times.  Sometimes more.

Allison