Why does this power supply use low-side regulation?

Common parts bin. Both power transistors, Q1, Q6 are npn, which would have been the most expensive device in the circuit at the time. Pnp power transistors were quite a bit more expensive than npn...

Regards,

Chris

Well, you've got plenty of voltage overhead to drive the bases of the pass transistors -- you don't have to give up overhead for that, or make an auxiliary supply, or anything. There's not a whole lot of active devices in there, it may be that all of the alternatives needed more.

Then why do the little arrow thingies point in different directions?

Qn and Q(n+5) are complimentary for all n in [1, 5]. They designed the circuit once, then replaced all the transistors with their complimentary equivalent for the other side.

Good point; if one actually had to sit down and design the thing over again with high-side regulation, it might be a bit more obvious.

The capacitors in that thing are absolutely huge. The replacements I used are better rated and about 1/3 the size of the originals!

Thanks,

---Joel

Responding to myself here...

Joerg has been kind enough to point out that those 100uF caps are actually between the output rails and ground. Given that on my own marked-up version of the schematic I have an annotation of "+15v" less than an inch away from one of them, you'd think I'd have understood that, but apparently half of my brain just couldn't get over the bottom part of the circuit being the output rather than ground.

---Joel

ion,

the

lar

nyway

n the

es me

That particular design is a clever way of keeping the two transformer secondaries electronically isolated, completely eliminating any possibility of circulating DC currents and other unintended interactions.

Absolutely right, mea culpa. Obviously need some better glasses :-(...

Regards,

Chris

Laziness. If the designer had wanted to be really lazy and smart, he'd have used the same regulation circuit for both outputs and simply wired one as negative.

C5's negative can is not grounded.... regardless of the schematic symbol's 'polarity'. I wonder if your replacement will also last 40 years.

Just because it's HP, doesn't mean it's a good example of engineering design, regardless of the vintage. We don't know who this guy was trying to fool, or for what reason.

RL

That requires a bit more effort in that the output of one supply is used for the main feedback reference for the other, doesn't it? (the -12.6V rail goes through R8 to provide the feedback signal for the +15V supply, and the +15V rail goes through R7 similarly)

Yeah, apparently to the draftsman the straight/curved line representation of the cap just means "electrolytic" and isn't meant to imply the curved end is the cathode as one typically assumes today (it is wired correctly, polarity-wise, in the real circuit). They also appear to have drawn S2 incorrectly -- as shown, in the 230V position it would just create a shorted winding rather than placing the two windings in series as (presumably) really happens.

Does seem kinda iffy, doesn't it? These days most I expect even companies like HP only design for an expected life of, e.g., 5 years.

---Joel

Have restored a lot of old hp kit for my own lab and some of the older circuit design looks quite laboured for no good reason, while other areas were about as tight and spoton as one could expect. I put it down to the fact that engineers of varying experience and skill worked in the product team. The more experienced guys working on the more complex areas, while fresh grads did psu's and other simple stuff, while building experience. To some people, a mirror image looks elegant, while using a

• reg circuit for the - rail could offend the aesthetic sensibilities. I'm sure hp had more than a tad of that kind of attitude back in 1968. Their whole reputation was built around pursuit of excellence and no compromise engineering...

Regards,

Chris

...although one wonders why they chose +15V/-12.6V rather than just +/-15V!

And only one side has overvoltage protection on it.

But overall it seems like a reasonable enough design.

---Joel

I have an hp 1120a fet 500Mhz x1 probe (uses the 1122a psu) schematic in front of me. There's no regulation in the probe other than an 8.2v zener and the schematic is very much discreet parts, fet, dual bipolar transistor etc. There's little psu decoupling in some areas, which probably accounts for the 100uF caps on the psu, to keep ripple and noise to a minimum. No inbuilt psu rejection, unlike ic's...

Regards,

Chris

The curvy side *used* to mean the outer foil. That meaning has been lost. I'm converting all of our drawings to straight plates for non-polarized caps and straight-curvy (/w polarity sign) for polarized.

Our design life was 10 years.

There's no denying, it's an elaborate little box.

I'm always pissed off when I see this kind of bollox in the power section. Sticking fresh grads on 'simple' stuff has always crippled power electronics development.

I wonder where they stick fresh grads in companies that do nothing but power electronics? Probably programming PICS.

RL

That's not beedback, but an ~ attempt at start-up coordination and fault limiting. Both outputs are independantly regulated.

RL

Viewed through the lens of 1968, it doesn't look particularly over the top. More or less standard issue at the time. You could economise by replacing each differential pair with a single transistor, with zener reference in the emitter, but you then get uncompensated vbe drift summed into the o/p voltage. Bearing in mind that the unit was designed to drive probes working at the mV level, it would need a well stabilised, low noise psu with good long term stability.

I do find the older manuals interesting, from any of the quality vendors. Just look at the schematics and try to work out why it was done that way. Fluke, Tek, Hp and others were all masters of the discreet component analog art and well worthy of study.

Nothing wrong with fresh grads so long as they are properly mentored, not just chucked in the deep end because everyone is so busy now. I would think hp would have been a very good place to work back then, with good mentoring that emphasised the hp way and no excuses, do it right attitude. I can just see junior engineers being sent back to do it again if it wasn't up to scratch :-).

I think we could do with a bit more of that kind of attitude now, but perhaps that's just my age...

Regards,

Chris

I'm puzzled by Legg's almost vituperative criticism of this hp design.

Perhaps he could elucidate by writing down a detailed description of how it works?

I suspect none of the complainers here actually know how (and why) it works :-) ...Jim Thompson

These days we have software updates issued on a weekly basis. :-)

The hp design is just too subtle for the critics to fathom ;-)

I suspect it is because there are very few people with device-level design skills around anymore.

Personally I think it's rather clever how they got around the lack of later-day chip functions. ...Jim Thompson