OK to replace this 2N3054 with a 3055 ?

[precaud]

I'm bringing an old HP 3570A back to life. Gotta love the construction quality on this generation of HP. Anyway, the+5V rail is down. Problem was traced to a blown fuse and a bad 2N3054 pass transistor wrapped around an LM723 regulator. See attached scan, the defective part is pointed to in green.

I have 2N3054 spares but they're TO-66 and this one is a TO-3 package. I don't see any 3054's in TO-3 being offered these days, even on Fleabay.

Is it OK to replace it with a 2N3055? Or any other ideas?

[David Hess]

I have never seen a 2N3054 in a TO-3 package.  Are you sure the original was not a 2N3055?

I assume the SCR crowbar triggered blowing the fuse so you might want to verify that there was not a failure to cause that other than a shorted pass transistor.

[Richard Crowley]

Do you have a heat-sink solution for an extra TO3 transistor? 
TO-3 clearly will not fit into the TO-66 footprint of the 2N3054
How much heat sink does it need?  Can use a TO-220 version and provide an extra heat sink?

[floobydust]

A TO-220 package fits on a TO-66 footprint. The middle (C) lead is not used, as the tab is screwed down. The TO-220 leads are bent 90 degrees.
So a MJE3055T would work or others.

[precaud]

I have never seen a 2N3054 in a TO-3 package.  Are you sure the original was not a 2N3055?

Nor had I. Perhaps it's a collector's item, then.. or, it was 
Yes, it's on the parts list as a 2N3054. And its definitely a TO-3. HP part 1854-0072

I assume the SCR crowbar triggered blowing the fuse so you might want to verify that there was not a failure to cause that other than a shorted pass transistor.

Have checked all the diodes and they're good. Haven't checked the SCR yet, will do manana. There are two 15000uF lytics in parallel filtering the recitified AC. The ground wire between them had snapped off, leaving one of them ungrounded. I doubt that would have caused the transistor to fail. The supplies easily disconnect from the circuits they power for troubleshooting.

Do you have a heat-sink solution for an extra TO3 transistor?

Or visa versa, as would be the case here. Why an extra TO-3?
It obviously runs warm, they placed it right in front of the cooling fan. There isn't much clearance between it and the top of the case, nowhere to add thermal mass if that's what you're thinking. I doubt a TO-220 would cut it; they fused it for 10A fast blow.

This unit has all options installed, so it is loading the 5V supply to the max...

EDIT: The plot thickens... maybe it isn't a 3054 after all. The part # on the unit is 1854-0245, not 1854-0072 as the manual says. I have two of these units and they both have 0245's there. Hmmm...

[rsjsouza]

Quite interesting... Keysight part number search shows it as a TO-66 packaging, but certainly a 2N3054 judging by its specifications.

The part search engine:
https://www.keysight.com/my/faces/fapHomePage.jspx

[precaud]

Since both units have the 1854-0245 part, I have to assume the parts list is in error. Keysight shows it as "Transistor NPN Silicon TO-3 PD-150W", protected supply. What does that latter mean?

I found one listing that shows nearest commercial equivalent as 2N3771.

[David Hess]

I have never seen a 2N3054 in a TO-3 package.  Are you sure the original was not a 2N3055?

Nor had I. Perhaps it's a collector's item, then.. or, it was 
Yes, it's on the parts list as a 2N3054. And its definitely a TO-3. HP part 1854-0072

I would believe that the parts list is wrong before believing that there was a 2N3055 in a TO-66 package.

I assume the SCR crowbar triggered blowing the fuse so you might want to verify that there was not a failure to cause that other than a shorted pass transistor.

Have checked all the diodes and they're good. Haven't checked the SCR yet, will do manana. There are two 15000uF lytics in parallel filtering the recitified AC. The ground wire between than had snapped off, leaving one of the ungrounded. I doubt that would have caused the transistor to fail. The supplies easily disconnect from the circuits they power for troubleshooting.

The regulator's control loop could have failed triggering the SCR crowbar blowing out the pass transistor and fuse.  Or the pass transistor could have failed triggering the SCR crowbar blowing out the fuse.

Since both units have the 1854-0245 part, I have to assume the parts list is in error. Keysight shows it as "Transistor NPN Silicon TO-3 PD-150W", protected supply. What does that latter mean?

I found one listing that shows nearest commercial equivalent as 2N3771.

The 2N3771 is a likely choice in the form of the ON Semiconductor 2N3771G and would be my first choice.

[precaud]

I found one listing that shows nearest commercial equivalent as 2N3771.

The 2N3771 is a likely choice in the form of the ON Semiconductor 2N3771G and would be my first choice.

I'll go that route. Mouser has them, I'm about ready to order from them anyway.

I saw a list of HP power supplies from that era that use the -0245, will check if there's a carcass in the storage shed first.

[David Hess]

I found one listing that shows nearest commercial equivalent as 2N3771.

The 2N3771 is a likely choice in the form of the ON Semiconductor 2N3771G and would be my first choice.

I'll go that route. Mouser has them, I'm about ready to order from them anyway.

I saw a list of HP power supplies from that era that use the -0245, will check if there's a carcass in the storage shed first.

They are one of my choices for replacing old 2N3055s when a "slow" transistor is needed and a modern 2+ MHz 2N3055 is too fast.

2N3055 0.3MHz   Old
2N3055 0.8MHz   Newer but still old
2N3055 2.0MHz and higher   Current

2N3771G 0.2MHz   Current
2N3772G 0.2MHz   Current
2N3055AG 0.8MHz   Current
MJ15015 0.8MHz   Current

[amspire]

That circuit seems to have two Q3's that has to be wrong. Definitely an error. The 2N3054 is traditionally a lower power driver transistor for a 2N3055, so I would expect to see the transistor attached to the regulator IC output as the 2N3054 in a TO66 package, and the regulator power transistor as the 2N3055. I would have thought 2N3055 transistors were still readily available, but the 2N3771 would probably be an excellent replacement.

You would not normally want to replace the 2N3054 driver transistor with a 2N3055.

[amspire]

That circuit seems to have two Q3's that has to be wrong. Definitely an error. The 2N3054 is traditionally a lower power driver transistor for a 2N3055, so I would expect to see the transistor attached to the regulator IC output as the 2N3054 in a TO66 package, and the regulator power transistor as the 2N3055. I would have thought 2N3055 transistors were still readily available, but the 2N3771 would probably be an excellent replacement.

You would not normally want to replace the 2N3054 driver transistor with a 2N3055.

Ah! the reason there are two Q3's is that one is on the controller board and one is on the chassis. The Q3 on the power supply controller module A37 has an actual part reference designation of A37Q3 and it is a 25W 2N3054 in the parts list.



Here is the layout of the controller board and the 2N3054 will be the TO66 power transistor on the left of the board. The HP 1854-0245 Q3 power transistor will be the TO3 transistor on the chassis.

The chassis TO3 transistor is an 150W transistor so a 2N3055 is inadequate. Need a 2N3771.

[precaud]

Ah! the reason there are two Q3's is that one is on the controller board and one is on the chassis. The Q3 on the power supply controller module A37 has an actual part reference designation of A37Q3 and it is a 25W 2N3054 in the parts list.

Here is the layout of the controller board and the 2N3054 will be the TO66 power transistor on the left of the board. The HP 1854-0245 Q3 power transistor will be the TO3 transistor on the chassis.

The chassis TO3 transistor is an 150W transistor so a 2N3055 is inadequate. Need a 2N3771.

You are exactly right! It was my mistake, thanks for catching it. And the chassis-mounted Q3 is listed in the replacement parts as an 1854-0245.

I think I will also remove the group-delay option pcb's from the machine, to relax the load on the regulator. "Group delay" was a fashionable measurement back in the 70's-80's, not so much any more. No need to power all those logic circuits - it can be done in software if needed...

[precaud]

Well I'm lucky. An old HP 6266B carcass in the shed had a -0245 and it tests good!
These are the moments where it pays to be a hoarder 

[precaud]

I replaced the -0245 and the SCR and it's alive again!
Now to hook it up to a 3330B and see if it behaves.
Thanks for your help, guys.

EDIT : Well unfortunately it has other issues as well. It responds to all controls but the level and phase displays don't change. So more work to do.
I bought this unit from a Honeywell surplus sale 8 or 9 years ago. Last cal date was June of 1989 !

[precaud]

And here it is in all its glory, a 99.9% functional 3570A, sitting atop the world's fugliest 3330B. I used this combo (not these exact units) for a number of years back in the 80's and 90's and this brings back some good memories. Performance-wise, they really are quite good.

As I mentioned last post, after getting the power supply up, the displays were frozen. A week or so later, after moving the inputs from the rear to the front panel, moments after powering it up again, smoke started pouring out of a board in the test channel analog card cage. I guess such things are to be expected when a unit has sat untouched in the garage/warehouse for over a decade. Surprise, surprise... a shorted tantalum cap. Replaced it, and the choke that precedes it (which is what burned), and it powered up again, display still frozen. Went through the troubleshooting tree, found an out-of tolerance frequency on the clock board (A32). Pulled the board, examined it, didn't see anything obvious, plugged it back in, powered it up, and now the display is working, though some digits drop in and out at random. I like them kind of fixes. Checked the levels on each channel and they were not even close. Signal paths checked out OK, so I troubleshooted the A/D boards, didn't find anything wrong, plugged them back in, and the levels are now in the ballpark but phase is off.

At this point it was clear that there were intermittent contacts between the boards and their edge connectors, so I pulled each one and cleaned the gold fingers and the sockets too. This thing is in meticulous condition inside. With a last cal date of 1989, I wouldn't be surprised if it had sat unused for a decade or two before I got it.

After the great cleaning, powered it up, and there it is, working fine. The only remaining problem to solve is an intermittent in the 2nd digit from the right in the phase display. It will stay stuck on either 4 or 6 for a while, and then work fine. I suspect the DIP sockets for the ribbon connectors, they're the old tall ones that make side contact with the pins. Will replace them with better ones.

After warmup, tweaking the mag and phase adjustments (I must say, I LIKE having level and phase trims on the front panel... who needs thru sweeps?), and measuring some familiar things, this rig is as good or better than any other VNA I've used in this freq range. Noise floor -108dbV with 100Hz RBW. Phase stays within 1/2 a degree even with signals 80dB down. Impressive.

The only thing that would stop me from using this on a regular basis is the 3330B. Not the unit itself, it's actually quite good. It's the fan noise The 3330B fan is ridiculously loud. It's even louder than my 3577A's fan, which I find intolerable for bench use.

Has anyone found a quieter fan with the same airflow?