Hi guys,
I am repairing an old HP 6253A powersupply. I bought two of these half a year ago, with one of them listed as broken.
When investigating I found that in the powersupply There were some clear indicators of damage, with a few of the bias diodes and resistors in the series regulator being turned to ash. On of the resistors literally turned to dust after removing. After removing the components I could also see some serious damage to the board.
Now, when digging in further I found that one of the main power transistors, Q7, measured a short between collector and emitter. This is where the problem lies: All other components I can find the specs of in the service manual. In the service manual, they are just labeled as:
Q6,7 Power NPN Si. HP 1854-0225.
(Quick note: the actual device markings say motorola and 1854-1017)
So my question now is, how do I find a replacement? I tried using some of the transistor I had laying around in one of the other channels, but none really worked very well. I would usually look at the datasheet and try finding a similar part based on that but I can't seem to find any datasheets on this part. And yes, I can buy NOS transistors of this type but at often 10-15 euros a piece (if I include shipping) I want to try and look at other options first.
So how would I go about trying to replace this transistor? The worst case I can imagine is use the fact that I have at least 7 more of these transistors in working order, and I could use one of those to get a rough idea in terms of expected values...
I've run into those before when looking and unfortunatly no luck. Neither the 1854-1017 or the 1854-0225 are in those lists (unless I looked over them but I checked twice)... It's like HP pretends this device never existed.
Such low output power supply of yesteryear, the first transistor is the 2N3055 for NPN or MJ2955 or PNP that you should consider trying.
Those were yesteryear popular power transistors.
Watch it with replacing power transistors in old HP kit. Sometimes the fT and hFe is pretty high on new 2n3055’s compared to older parts. They oscillate badly. If this happens, stick a few tens of pF between base and collector (add miller capacitance). Just solder across the pins underneath the board.
Also and this is bloody annoying, if it’s one of the older supplies with the heat sink bolted to the board then the legs on some of the new ones aren’t long enough to go through the board.
Watch it with replacing power transistors in old HP kit. Sometimes the fT and hFe is pretty high on new 2n3055’s compared to older parts. They oscillate badly. If this happens, stick a few tens of pF between base and collector (add miller capacitance). Just solder across the pins underneath the board.
Also and this is bloody annoying, if it’s one of the older supplies with the heat sink bolted to the board then the legs on some of the new ones aren’t long enough to go through the board.
Will look into doing that. I'll order replacement components for the rest of the supply later today, or perhaps try and replace one of the transistors in the working supplies, and see if it works before ordering replacements for everything else.
Thanks for the help already. Will try and keep you guys updated
Watch it with replacing power transistors in old HP kit. Sometimes the fT and hFe is pretty high on new 2n3055’s compared to older parts. They oscillate badly. If this happens, stick a few tens of pF between base and collector (add miller capacitance). Just solder across the pins underneath the board.
Also and this is bloody annoying, if it’s one of the older supplies with the heat sink bolted to the board then the legs on some of the new ones aren’t long enough to go through the board.
Will look into doing that. I'll order replacement components for the rest of the supply later today, or perhaps try and replace one of the transistors in the working supplies, and see if it works before ordering replacements for everything else.
Thanks for the help already. Will try and keep you guys updated
Pull one and shove it into a component identifier if for nothing more than a Hfe measurement.
Supposedly the 1854-0225 is a ' selected 2n3055 ' but selected for what I don't know.
Supposedly the 1854-0225 is a ' selected 2n3055 ' but selected for what I don't know.
More than likely Hfe 'range'.
If there's more than one it's always wise to in some way attempt to match them. One of the reasons I still have my shitty old
Victor Vichy VC99 with its Hfe tester.
HP part # 1854-0225 (Transistor NPN VCE-60V IC-15A PD-115W FT-0.5MHz Silicon TO-3) was superceded by part # 1854-1017, which I think was an epitaxial 2N3055 selected for lower gain (~15) & higher SOAR to match the original selected homotaxial RCA 2N3055.
Something like a 2N3771/2 is the usually recommended substitute though, as bd139 mentioned, you'll probably have to make minor mods to ensure stability with the higher Ft of modern parts.
stick a few tens of pF between base and collector (add miller capacitance). Just solder across the pins underneath the board.
"Watch it" on street side modification, if it is unnecessary, don't add sluggishness in response to your power supply. If this is a modern power supply, I would consider it as "failed" because of output voltage dip during power demand.
The series pass transistor is regulated by the feedback biasing network so it would not be necessary. If it does, the better approach is to lower the Ib drives.
Edit: Of all you know with modern parts, you may be lucky to end up with a better power supply with faster response like more of the modern power supply with faster response and higher speed.
That doesn’t always work as intended though. Part of the feedback loop is the transient response time. Stick a few watts of resistor on the end of a couple of meters of wire and see what happens. With the output cap and too fast a response you get a nice LCR circuit with an amplifier to drive it.
You can reduce base drive but that requires slightly more drastic modifications and experimentation.
Only saying this because I fecked up an HP supply doing exactly just that. Learned a lot about loop stability then
Don't need to remind you of the abundance of "Fake" 2N3055 around.
With the output cap and too fast a response you get a nice LCR circuit with an amplifier to drive it.
With moderns and as instrument progresses to higher frequencies, gigahertz, terahertz and beyond, it is worthy of the reminder of the miller oscillator of the yesteryears and the practice of good proper wirings, things like miller cap and miller inductance etc.. and the caution of blind adding of capacitor to base of transistor. For oscillation to sustain, then 180 degrees phase shift has been unknowingly created.
I am not saying is wrong, but "DO" add it for a known good purpose though.
I am just reminding of unknowing created oscillator.
Something like a 2N3771/2 is the usually recommended substitute though, as bd139 mentioned, you'll probably have to make minor mods to ensure stability with the higher Ft of modern parts.
For new parts to replace an old 2N3055, I usually recommend the On Semiconductor 2N3771G, 2N3772G, 2N3055AG, or MJ15015 but even they are not a sure thing. I would test and if necessary trim the transient probably with a series RC network between the collector and base.
Don't need to remind you of the abundance of "Fake" 2N3055 around.
I always have to wonder why anyone bothers to fake such low tech and widely available parts. Surely the TO3 package costs more than the silicon it contains.
Don't need to remind you of the abundance of "Fake" 2N3055 around.
I always have to wonder why anyone bothers to fake such low tech and widely available parts. Surely the TO3 package costs more than the silicon it contains.
I stopped wondering when someone pointed out that someone is faking 2N3904s. I have a pile of likely fake BC550 and BC560 transistors.
One of the problems with the 2N3055 is that a lot of other power transistors get marked as 2N3055 when they do not meet their original specifications.
Don't need to remind you of the abundance of "Fake" 2N3055 around.
I always have to wonder why anyone bothers to fake such low tech and widely available parts. Surely the TO3 package costs more than the silicon it contains.
I suppose it has to do with market value. Let's say market value is $1.50 and your "all in" copied and short changed manufacturing cost is $0.10 cents a piece. You distribute to shops at $0.50 cents a piece. You produced 100,000 pieces. You would make $40,000.00. But you are not limited to just fake 2N3055. Your small factory need to let the machines running and your workers fed, so you decided to make 1,000 fake devices per year. So you would make $40 million per year, hypothetically speaking. Ok, let's reduce the maths a bit. Say you make overall "ALL IN" net profit at 1 million a year. For a small boss factory say in china, I think the logic make sense.
I still have some old NOS 2n3055 laying around that I got from some older engineers that were cleaning up their labs, so I might have a look at trying one of those.
A few people have suggested stickign the transistors into a component tester, but I only have one of those cheap ebay 20 dollar testers, and was under the impression their measurements were innacurate at best. Could I just use some powersupplies and measure it that way? (put current in base measure how much is passed through the emiter?)
The $20 tester in fact will be Good
enough, I would say.
Not a very complex measurement.
Just don't rely it to measure capacitance for example.
There is also a guy on this forum who has got hold of a real big lot of original HP and Tek spare parts and who seems to be willing to part with some. Maybe run the numbers by him? Post was from this year!
Don't forget KD502, KD503, KD3055 from Tesla. Soviet Russia Czechoslovakia used to build beefy devices. (Machines and transistors). Might be much cheaper/better from a surplus seller in eastern europe than buy new crap from china