Pages: [1] 2 3   Go Down

Author Topic: 2N2222A TO-18 vs TO-92  (Read 877 times)

AC9QC

  • Posts: 57
    • HomeURL
2N2222A TO-18 vs TO-92
« on: May 31, 2021, 06:39:51 AM »

What's the difference between the two of these besides the price? From what I understand the TO-92 actually has 50% better dissipation than the TO-18s do. I'm looking at doing the driver replacement in my UbitX V3 so I'm just trying to figure out what to order. Is it just that TO-18s look cool and are old that keeps them around or is it a technical merit of the design?
Logged

AC2EU

  • Member
  • Posts: 2793
    • McVey Electronics
Re: 2N2222A TO-18 vs TO-92
« Reply #1 on: May 31, 2021, 07:02:22 AM »

What's the difference between the two of these besides the price? From what I understand the TO-92 actually has 50% better dissipation than the TO-18s do. I'm looking at doing the driver replacement in my UbitX V3 so I'm just trying to figure out what to order. Is it just that TO-18s look cool and are old that keeps them around or is it a technical merit of the design?

There used to be "crown" heat sinks that would fit over the TO-5 and TO-18 styles. Maybe they sill make them to accommodate those designs which used the sinks?

Otherwise figure out the dissipation, but Id go with the TO-92 for better dissipation.

KB1NO

  • Member
  • Posts: 95
Re: 2N2222A TO-18 vs TO-92
« Reply #2 on: May 31, 2021, 08:53:22 AM »

The TO-18 is a metal package, vs TO-92 which is plastic.    Generally the metal package (TO-18) can dissipate more power than the To-92.  This may vary with manufacturer so best to look up the actual datasheets.  The key specs to look at are the "Theta J-A" which is the thermal resistance from the junction (chip) to ambient at room temp.     Lower value of thermal resistance is better and keeps the chip cooler at a given power.
 The maximum power dissipation spec would also show the difference.   The pinout will also vary.    TO-18 is a triangle and TO-92 is a straight line.    You can usually bend the leads to fit.      Of course, the collector is bonded to the metal TO-18 case, so if it can short if there are wires near it.   
Personally, I would replace it with what is recommended by the design. 
Good luck
 
Logged

WB6BYU

  • Member
  • Posts: 20896
    • Practical Antennas
Re: 2N2222A TO-18 vs TO-92
« Reply #3 on: May 31, 2021, 09:07:38 AM »

I agree that, in practice, the TO-18 case can dissipate
more power, either with or without a heatsink.

The thermal ratings aren’t always obvious:  both
packages may contain exactly the same silicon chip,
but the thermal resistance will be much higher
for the plastic package.  (The max power rating
often assumes a perfect heat sink.)

If you application uses a heatsink, then the
TO-18 is certainly a better choice, or if it gets warm
to the touch during normal operation.

K6BSU

  • Member
  • Posts: 313
Re: 2N2222A TO-18 vs TO-92
« Reply #4 on: June 01, 2021, 05:07:30 PM »

If your 2N2222 is getting hot, it's time to think of a different transistor with higher rating.
Logged

W9IQ

  • Member
  • Posts: 8866
Re: 2N2222A TO-18 vs TO-92
« Reply #5 on: June 01, 2021, 05:33:20 PM »

The UbitX V3 makes extensive use of the 2N3904. Where is this 2N2222?

- Glenn W9IQ
Logged
- Glenn W9IQ

God runs electromagnetics on Monday, Wednesday and Friday by the wave theory and the devil runs it on Tuesday, Thursday and Saturday by the Quantum theory.

N7EKU

  • Member
  • Posts: 1471
Re: 2N2222A TO-18 vs TO-92
« Reply #6 on: June 02, 2021, 07:21:16 AM »

Hi,

The v3 doesn't use any 2n2222a transistors.  However the 2n3904 used in the transmit chain run out of steam at higher frequencies and at the gain that is being attempted in the stages that use them.  As a result, the power drops off significantly in the upper bands.  My stock v3 tests out to make around 10W on the lower bands and tapers down to less than 1W on 10m.  The 2n2222a transistors hold up better at high frequencies under higher currents, so it's been recommended to use them to help correct this power drop.  One could also use single 2n5109 transistors instead of pairs of 2n2222a's.

Another improvement to be made is in the first TX amp stage Q90 transistor that needs to have its gain compensation improved.  This also helps balance the power across the bands.  A transistor change can also be made here and this was this was done on the newest board versions.

Details of these changes are on the group.io wiki page.

73.
Logged
Mark -- N7EKU/VE3

W9IQ

  • Member
  • Posts: 8866
Re: 2N2222A TO-18 vs TO-92
« Reply #7 on: June 02, 2021, 07:47:32 AM »

Thanks for the background. That helps to bring pertinence to your question.

Within the same manufacturer, the 2N2222 die is identical. So all non-thermal parameters would be the same regardless of the package style. You may find differences between manufacturers using the same package but it probably won't be significant and they may in fact just be marketeering.

The thermal resistance of the die to the case, RθJC, is nearly identical for the TO-18 vs the TO-92 version. So the thermal performance with a heatsink will be the same.

What might surprise you is that the thermal resistance of the die to air, RθJA, is often higher for the metal TO-18 case. I suspect the reason for this is that die is attached to the internal metal base but the upper portion of the die and the bond wires are open to internal air compared to the TO-92 case where the entire die and all bonds are encapulated in the epoxy. No matter the cause, the result is that the TO-18 case will often dissipate slightly less power than the TO-92 package. But check the specific spec sheets to compare RθJC if that fits your application.

Make sure to buy your 2N2222 from a reputable, US distributor unless you are prepared to characterize a batch of them yourself. The eBay, et al crowd tends to sell "bargain" versions that you may have to pick through.

- Glenn W9IQ
 
« Last Edit: June 02, 2021, 08:00:27 AM by W9IQ »
Logged
- Glenn W9IQ

God runs electromagnetics on Monday, Wednesday and Friday by the wave theory and the devil runs it on Tuesday, Thursday and Saturday by the Quantum theory.

W1VT

  • Member
  • Posts: 6071
Re: 2N2222A TO-18 vs TO-92
« Reply #8 on: June 02, 2021, 08:05:40 AM »

A TO-18 case is designed to be hermetic to better protect the transistor die.
Logged

W9IQ

  • Member
  • Posts: 8866
Re: 2N2222A TO-18 vs TO-92
« Reply #9 on: June 02, 2021, 08:27:26 AM »

A TO-18 case is designed to be hermetic to better protect the transistor die.

It would be better to say that a TO-18 case can be made to comply with hermetic standards, such as MIL-STD-883C, but unless so specified, they generally don't have hermetic properties any different than an epoxy TO-92 package. Put another way, you will know you when you are paying for hermetically graded parts!

- Glenn W9IQ
Logged
- Glenn W9IQ

God runs electromagnetics on Monday, Wednesday and Friday by the wave theory and the devil runs it on Tuesday, Thursday and Saturday by the Quantum theory.

AC9QC

  • Posts: 57
    • HomeURL
Re: 2N2222A TO-18 vs TO-92
« Reply #10 on: June 02, 2021, 09:30:29 AM »

Hi,

The v3 doesn't use any 2n2222a transistors.  However the 2n3904 used in the transmit chain run out of steam at higher frequencies and at the gain that is being attempted in the stages that use them.  As a result, the power drops off significantly in the upper bands.  My stock v3 tests out to make around 10W on the lower bands and tapers down to less than 1W on 10m.  The 2n2222a transistors hold up better at high frequencies under higher currents, so it's been recommended to use them to help correct this power drop.  One could also use single 2n5109 transistors instead of pairs of 2n2222a's.

Another improvement to be made is in the first TX amp stage Q90 transistor that needs to have its gain compensation improved.  This also helps balance the power across the bands.  A transistor change can also be made here and this was this was done on the newest board versions.

Details of these changes are on the group.io wiki page.

73.

I found stuff for the 2N2222As but not really any info on the 2N5109s. Do you have to mod the bias in the same way for them? I'd rather go that route in some ways, the big one being that the board real estate isn't great if I wanted to put some heat sinks on the TO-18s they're just a bit too close. I bought some 4.5mm tubing to try and make something but not really sure about that. May have to get the lathe going and produce something that way.




Make sure to buy your 2N2222 from a reputable, US distributor unless you are prepared to characterize a batch of them yourself. The eBay, et al crowd tends to sell "bargain" versions that you may have to pick through.

- Glenn W9IQ

Yeah, even going through the ones on Digi Key there seem to be a few different dissipations listed even with the same package. I figured for a few bucks I'll just buy a pack and then I'll be good to go instead of trying to do the ebay/NOS route.
« Last Edit: June 02, 2021, 09:51:24 AM by AC9QC »
Logged

KM1H

  • Member
  • Posts: 11155
Re: 2N2222A TO-18 vs TO-92
« Reply #11 on: June 02, 2021, 01:59:57 PM »

The 2N5109 was developed for CATV line amps in the late 70's and large systems had coverage from ~50-200 MHz.

Extremely rugged, low noise figure, and very low distortion products. One of my all time favorites for all sorts of RF use.

Carl

Logged

N7EKU

  • Member
  • Posts: 1471
Re: 2N2222A TO-18 vs TO-92
« Reply #12 on: June 02, 2021, 05:30:29 PM »

Hi,

Did you look at the bitx20 groups.io wiki page?  That has all the info on changes.

I don't think there's a lot to worry about with heatsinks.  The output to the PA finals is only about 1W I think, so that is not a lot when spread out among the transistors.  The board is just using SMD transistors right up to the finals and their ratings are much lower than leaded parts.

73.
Logged
Mark -- N7EKU/VE3

AC9QC

  • Posts: 57
    • HomeURL
Re: 2N2222A TO-18 vs TO-92
« Reply #13 on: June 03, 2021, 10:00:28 AM »

Hi,

Did you look at the bitx20 groups.io wiki page?  That has all the info on changes.

I don't think there's a lot to worry about with heatsinks.  The output to the PA finals is only about 1W I think, so that is not a lot when spread out among the transistors.  The board is just using SMD transistors right up to the finals and their ratings are much lower than leaded parts.

73.

I've been reading through the 330+ page mega thread on doing mods to the output. I need to get a bunch of parts together, it seems as if a ton of effort were put into trying stuff with this. It looks  as if you do the mods as if you were using 2n2222As with the 2N5109s just using less transistors in all.

In terms of the heat sinks I'll probably wait and see. If I have a couple around I may use them anyways, I'd like to do digital with this and it would probably be good to have a hair of margin on the transistors under used or not.
Logged

N7EKU

  • Member
  • Posts: 1471
Re: 2N2222A TO-18 vs TO-92
« Reply #14 on: June 03, 2021, 12:18:46 PM »

Hi,

Yes, with the 5109's, you just need one to replace two of the 2222a's.

The group is kind of a mess with many people trying things randomly and getting random results.  To me it's not really worth digging into all the threads.  Just go to the wiki page and down to the very last link: 

https://groups.io/g/BITX20/wiki/RF-power-and-gain

The changes here were done by Allison KB1GMX who is an expert EE and put a lot of effort into researching and testing the changes (I've already done the LPF relay re-arrangement mod she designed and it worked perfect at getting the harmonics down to legal levels).

There are basically two areas that need fixing:  Q90 biasing change to help tamp down the gain on the lower bands, then replacing all the 3904's which can't give the gain needed on higher bands.  The page also suggests changing Q90 to something with a better Ft because it is running kind of hot, and suggests good replacements.

73,  Mark
Logged
Mark -- N7EKU/VE3
Pages: [1] 2 3   Go Up