EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: robse on February 14, 2018, 04:16:50 pm
-
Hello,
I'm designing a simple common emitter amplifier, it will amplify low signals below 100Khz.. nothing complicated
It will be mounted inside a small aluminum chassis connected to ground for improved noise management
Now the simple question: is there any real difference between a 2n2222a TO-18 and TO-92 package ?
Which one should I mount on pcb ? I already have some plastic TO-92 but no metal TO-18
Does the metal cap reduce noise or it is only an heat-sink ?
Thank you,
Robse
-
Well the TO18 package is hermetic, which I suppose might improve its drift characteristics in VERY specific DC applications but or anything normal, the TO92 is fine.
-
It's been a long time since I've seen anything modern with TO-18 transistors, are they even still made outside of perhaps a few specialty components?
-
Unlike the larger medium power TO-39 package which was never really replaced, the TO-18 package has an insulating base so it does not even offer better power dissipation than a plastic TO-92 package. There might be some advantage to using the metal can TO-18 package in UHF applications where the metal can is soldered into the printed circuit board and some old designs did exactly this.
In higher power applications there was never really a replacement for the TO-39 package except perhaps the TO-126/TO-225 but transistors selection there is poor. A better option if you need a higher power TO-92 part is to use several in parallel with emitter ballasting if necessary.
-
How can you solder the metal cap to ground since the metal cap is connected to the collector ?
I don't have a TO-18 in my hands now, I've just red this, I hope it's correct
-
Now the simple question: is there any real difference between a 2n2222a TO-18 and TO-92 package ?
Some transparent-case steampunk/audiophile design can benefit from TO-18. Otherwise it's just waste of money unless you are repairing some old industrial/precision stuff where TO-18 was chosen for some reason. TO-92 PN2222 is 10x cheaper.
-
How can you solder the metal cap to ground since the metal cap is connected to the collector ?
I don't have a TO-18 in my hands now, I've just red this, I hope it's correct
For TO-18 JFETs, I think the gate lead is always connected to the case but for at least some bipolar transistors, no lead is connected and the case is floating so it may be connected directly to the ground plane. The Tektronix S-2 sampling head (http://w140.com/tekwiki/wiki/S-2) and 7T11 sampling timebase (http://w140.com/tekwiki/wiki/7T11) are examples where this was done as shown below.
-
It's been a long time since I've seen anything modern with TO-18 transistors, are they even still made outside of perhaps a few specialty components?
Sure. Plenty of people see "2N2222(A)" and just buy that. Regardless of whether it's 10 times the cost, or whether there are better alternatives.
Suppliers dutifully deliver what has been ordered: a TO-18 metal case 2N2222(A). If the customer wanted TO-92, they surely would have ordered PN2222(A).
Or a more suitable transistor like 2N3904, or any number of SMTs.
Tim
-
The Tektronix S-2 sampling head (http://w140.com/tekwiki/wiki/S-2) and 7T11 sampling timebase (http://w140.com/tekwiki/wiki/7T11) are examples where this was done as shown below.
Literally Art Of Electronics (https://www.amazon.com/Art-Electronics-Paul-Horowitz/dp/0521809266) :) Modern boards does not look so sincere & soulful.
-
How can you solder the metal cap to ground since the metal cap is connected to the collector ?
I don't have a TO-18 in my hands now, I've just red this, I hope it's correct
For TO-18 JFETs, I think the gate lead is always connected to the case but for at least some bipolar transistors, no lead is connected and the case is floating so it may be connected directly to the ground plane. The Tektronix S-2 sampling head (http://w140.com/tekwiki/wiki/S-2) and 7T11 sampling timebase (http://w140.com/tekwiki/wiki/7T11) are examples where this was done as shown below.
That board is a piece of art - worthy of an I-spy thread for component choices, and techniques! The more I gaze at it, the more I see!
(What on earth is going on to the right of the K31 silk screen? :-\)
-
That board makes me think that a lot of old metal can packages are not hermetically sealed, just potted with epoxy resin seal on the bottom.
-
No, the epoxy is there to mechanically protect the tiny fragile glass to metal seals.
-
No, the epoxy is there to mechanically protect the tiny fragile glass to metal seals.
OK. I see it now. The bottom is cup shaped, and gold plated. So fancy.
-
The Tektronix S-2 sampling head (http://w140.com/tekwiki/wiki/S-2) and 7T11 sampling timebase (http://w140.com/tekwiki/wiki/7T11) are examples where this was done as shown below.
Literally Art Of Electronics (https://www.amazon.com/Art-Electronics-Paul-Horowitz/dp/0521809266) :) Modern boards does not look so sincere & soulful.
That is how you get 10+ GHz and 10 picosecond performance out of through hole parts. Similar or better performance could be achieved with a 2 layer microstrip surface mount implementation except for the lack of modern tunnel diodes. I am not sure how one could make up for them without using custom integrated circuits.
(What on earth is going on to the right of the K31 silk screen? :-\)
It is difficult to see in that photograph and I could take a better one if someone is interested but that is a beam lead mounted tunnel diode with a very fine wire used to couple the trigger signal in for its inductance.
-
It is difficult to see in that photograph and I could take a better one if someone is interested but that is a beam lead mounted tunnel diode with a very fine wire used to couple the trigger signal in for its inductance.
Oh, that's neat! .... and fun to assemble in production. You've got to hand it to them for the lengths they were prepared to go to, to achieve the performance in those days.
-
Hello,
I'm designing a simple common emitter amplifier, it will amplify low signals below 100Khz.. nothing complicated
It will be mounted inside a small aluminum chassis connected to ground for improved noise management
Now the simple question: is there any real difference between a 2n2222a TO-18 and TO-92 package ?
Which one should I mount on pcb ? I already have some plastic TO-92 but no metal TO-18
Does the metal cap reduce noise or it is only an heat-sink ?
Thank you,
Robse
Many months ago I purchased a bag of pn2222a(to92) and 2n2222a (to18) and built a jig and put them head to head.Now before we get into anything deep here it helps if we have a basic understanding of transistor manufacture.Transistors are made on circular silicon disks(called dies) each disk can contains many hundreds or even thousands of transistors per disk.What manufacturers can't do is make all the transistors have exact same parameters, now the low spec. ones end up in as to92 whilst the good ones are housed in a to18 package.
This is why the spec. sheet for a particular transistor is so broad so as to fit both .Sometime back when I was testing and comparing to92/to18 I found that the bandwidthproduct,in the to18 to be superior than the plastic to92.The to92 pn2222a (Central semi's)(7pence/pc) were very good for a to92 part, worst was ksp2222a(2pence/pc)
-
Many months ago I purchased a bag of pn2222a(to92) and 2n2222a (to18) and built a jig and put them head to head.Now before we get into anything deep here it helps if we have a basic understanding of transistor manufacture.Transistors are made on circular silicon disks(called dies) each disk can contains many hundreds or even thousands of transistors per disk.What manufacturers can't do is make all the transistors have exact same parameters, now the low spec. ones end up in as to92 whilst the good ones are housed in a to18 package.
This is why the spec. sheet for a particular transistor is so broad so as to fit both .Sometime back when I was testing and comparing to92/to18 I found that the bandwidthproduct,in the to18 to be superior than the plastic to92.The to92 pn2222a (Central semi's)(7pence/pc) were very good for a to92 part, worst was ksp2222a(2pence/pc)
First of all, the silicon disk you are talking about is called a "wafer". A "die" is a single chip cut out of this wafer, it might be an IC, it might be a transistor, it might be a diode or whatever.
On one wafer (or rather, one stack of wafers), the dies are pretty much identical. The parameter spread you mention is between wafer stacks. Eg., a wafer produced in May might not have the same parameters as one produced in June.
This is because production conditions vary, such as mask alignment, etching depth, doping percentage, ion implantation intensity etc. and is normal in an environment where different devices are produced on the same production line.
Your observation about differences between epoxy and metal case devices is IMO purely coincidental. No manufacturer will select dies, that would be far too expensive. A simple "go/no-go" test is performed for the device parameters, that's it.
-
Please excuse me you are quite right regarding the dies and wafers, wafers typically have hundreds of transistors formed on them. Sorry but I must disagree regarding your go no go test scenario, rather when the test is run it goes toss to "to92 bin" or "to18" bin.Very few of the transistors are thrown away and you pay for what you get.
Again, very sorry for getting die and wafer mixed up, have you read the other replies as they reveal nothing about the original question?.
Cheers
Commie