Fixing old 723 variable power supply. Should I replace the guts?

[David Hess]

I have an old 723 based variable power supply that I built when I was about 13.  Finding myself in need of another bench supply, I dug it up and fixed the enclosure.  I may change the big input capacitors, but what about designing and building one of my super precision regulators, adapted to variable operation, to replace the insides, with noise, load, and line regulation under a microvolt?

The transformer has 1 amp 14.5 volt center tap and 0.5 amp 43.5 volt center tap outputs.

The printed circuit board came from the long gone R&D Electronics on Orangethorpe avenue in Fullerton, California, not far from Orvac Electronics.

[KE5FX]

Heh, reminds me of the 'classic' I just recapped:



Two LM723s and two 2N3055s in that one.   

In other news, through-hole electrolytics cost more than eggs these days.

[David Hess]

In other news, through-hole electrolytics cost more than eggs these days.

And axial ones are more expensive than that.

The two 1100 microfarad 50 volt capacitors were trashed with 25 ohms of series resistance.  I replaced each one with a pair of NOS (New Old Stock) 2200 microfarad 35 volt capacitors in series which are "only" 30 years old.

[TomKatt]

The two 1100 microfarad 50 volt capacitors were trashed with 25 ohms of series resistance.  I replaced each one with a pair of NOS (New Old Stock) 2200 microfarad 35 volt capacitors in series which are "only" 30 years old.

A few years ago I was working on my HP 180 analog scope that was built in the late 1960's...  I figured that while I was in there I should probably replace the main power supply caps, which were big screw top 'computer grade' cans made by Sprague.  I was shocked when they tested good and showed virtually no ripple.  So I left them in the scope, which continues working well to this day.  Just as well, because even a few years ago they were listed around $20 each and there are 6 in the scope - that's more than I paid for the scope to begin with lol.

I've heard stories about some old caps built to last - but didn't really expect to run across any that did.

[floobydust]

Everyone needs a linear PSU. My favorite was a LM317 one I made when I was 14.
1.5A is enough to not make too much smoke.

I recently did a full redesign of it using discrete op-amps and it ended up being a lot of work. Still need to add UVLO so the output does not spike on power up/down. I wanted adjustable current-limiting.
I would say it has not been worth it, aside from the fun designing and building it. I just don't need that much performance when powering things.
I also have a few other linear PSU's, Topward I just bought and recapped and it's meh ok.

That '723 board pretty vintage but it's missing a couple protection diodes so I'm surprised the 1967 2N3055 is alive.

[TomKatt]

Everyone needs a linear PSU. My favorite was a LM317 one I made when I was 14.
1.5A is enough to not make too much smoke.

I recently did a full redesign of it using discrete op-amps and it ended up being a lot of work. Still need to add UVLO so the output does not spike on power up/down. I wanted adjustable current-limiting.
I would say it has not been worth it, aside from the fun designing and building it. I just don't need that much performance when powering things.
I also have a few other linear PSU's, Topward I just bought and recapped and it's meh ok.

That '723 board pretty vintage but it's missing a couple protection diodes so I'm surprised the 1967 2N3055 is alive.

Back in the day it seems like virtually every electronics hobbyist built a linear power supply, decade box and things of that nature - they were things you needed and building them provided valuable experience.  Now it seems like everyone starts with module boards like Lego blocks...  Seems like things are generally more digital these days.  I don't think that's good or bad, just different.

I built a power supply similar to the OP, but my Siglent SPD1305X is like Buck Rogers in comparison lol.  Though I still use the decade box I built in the 70's - it's hard to find a decent one today at an affordable price.

[David Hess]

A few years ago I was working on my HP 180 analog scope that was built in the late 1960's...  I figured that while I was in there I should probably replace the main power supply caps, which were big screw top 'computer grade' cans made by Sprague.  I was shocked when they tested good and showed virtually no ripple.  So I left them in the scope, which continues working well to this day.  Just as well, because even a few years ago they were listed around $20 each and there are 6 in the scope - that's more than I paid for the scope to begin with lol.

Big can capacitors have so much excess electrolyte that they rarely dry out, but what does happen is when the seal becomes leaky, oxygen gets pumped in with pressure changes and corrodes the aluminum strip which connects to the leads.  My PS503 capacitors, shown below, failed this way.  The replacement capacitors were selected based on diameter so that they fit in the original mounting clamps.

Everyone needs a linear PSU. My favorite was a LM317 one I made when I was 14.
1.5A is enough to not make too much smoke.

I recently did a full redesign of it using discrete op-amps and it ended up being a lot of work. Still need to add UVLO so the output does not spike on power up/down. I wanted adjustable current-limiting.
I would say it has not been worth it, aside from the fun designing and building it. I just don't need that much performance when powering things.
I also have a few other linear PSU's, Topward I just bought and recapped and it's meh ok.

I have a pair of Tektronix PS503A power supplies, but it would be handy to have another simple supply.

I am leaning toward just designing and and building another super precision power supply with its own enclosure.  There is no need to sacrifice an old working supply.  I might add a simple variable 317/337 dual tracking supply to this one with the extra transformer winding.

That '723 board pretty vintage but it's missing a couple protection diodes so I'm surprised the 1967 2N3055 is aliveV.

That is not the original 2N3055.  I replaced it at least once.

[floobydust]

If I'm doing analog circuit design, then I need quiet power.  How many people even notice the difference between 5mV or 200mV ripple?

People conflate a PSU with fancy display/keypad with high performance. Modern PSU's are frequently crappy.
Have you put the Siglent in CC mode at 10mA with say 25VDC out and shorted the output? That's when you find the big fat 470uF output cap...
Charge a gel-cell with it and see what happens if you switch it off/mains outage occurs. Or if the setpoint is low a few volts and you connect it to a 12V battery.

Homebrew linear PSU's done right can take the abuse and shit. A faithful horse- dependable, repairable. Not something you have no schematic for. With a hurricane cooling fan lol.

[CaptDon]

I find the 723 series to be easily damaged and often short lived especially in those 13.8vdc 20 amp ham radio supplies. I have experienced brand new old stock that simply would not work at all in more modern power supplies. I guess there was an increase in current and possibly voltage capacity between 723 and 723A and so forth. So many prefixes also LM, uA, CA, SA and more. We had probably 100 from mixed manufacturers all at least 20 years old. I gave about 20 of them 'a ride home' to fix many makes and models of friends 'dead' power supplies. In almost every case the 723 was kaPut and the ones I put in generally died within a year. O.E.M. supplies probably pushed them to their limits. Many blew out after short interval mains 'power flickers'. I had better luck with the Philips ECG and NuTone NTE replacements. They seemed a bit more robust than the OEM parts.

[floobydust]

Many 723 power supply designs are minimalist and have weaknesses.
When you look at the circuits, they are relying on a bit of luck.
Example: missing resistor to limit 723 output current for base-drive to the pass transistor(s).
So the IC gets damaged when a PSU is under heavy load etc. OP's seems to have no driver transistor(s)... A naked Darlington is not so tough.
Another missing resistor for the current-limit transistor base. I could go on but the subtleties need to be looked after for a really good PSU.

I had one project that could not work with 20mV ripple (at load) and when I dug in, it was just poor star-grounding moved a few nodes and got it down to 2mV.

[KE5FX]

These are all true, and they are among the reasons why I'm rather attached to this old Radio Shack job.  They did their homework:



Nothing flashy, but nothing is missing that needs to be there.  It's one of the supplies I trust to run expensive OCXOs for months at a time.

[coppercone2]

I think it should be replaced unless you fix that ugly burned PCB up

[ME]

These are all true, and they are among the reasons why I'm rather attached to this old Radio Shack job.  They did their homework:



Nothing flashy, but nothing is missing that needs to be there.  It's one of the supplies I trust to run expensive OCXOs for months at a time.

Can you upload the full schematic diag ? , Looks like a good project.

[floobydust]

Micronta 22-121, I couldn't see it around 1980's Radio Shack catalogs. It's way under-rated 0-15V 1A x 2 channels.
https://elektrotanya.com/micronta_no.22-121_dual-dc-power-supply_sm.pdf/download.html
I have the Project Kit dual PSU dumpster rescue uses a few 741's.

[KE5FX]

Can you upload the full schematic diag ? , Looks like a good project.

I snagged the service manual here.  I'll upload it to ko4bb.com in case that site stops working.

[David Hess]

So the IC gets damaged when a PSU is under heavy load etc. OP's seems to have no driver transistor(s)... A naked Darlington is not so tough.
Another missing resistor for the current-limit transistor base. I could go on but the subtleties need to be looked after for a really good PSU.

I checked the board and there is no current limiting except for the 723's Vbe current limiting.

Have you put the Siglent in CC mode at 10mA with say 25VDC out and shorted the output? That's when you find the big fat 470uF output cap...

I just noticed that there is no output capacitance on this design.  Somehow it does not oscillate.

[David Hess]

The design is based on the positive voltage regulator with NPN pass transistor in the 723 datasheet, but with the reference divided down to 2.5 volts to lower the minimum output voltage.

[DimitriP]

I have an old 723 based variable power supply that I built when I was about 13.  Finding myself in need of another bench supply, I dug it up and fixed the enclosure.  I may change the big input capacitors, but what about designing and building one of my super precision regulators, adapted to variable operation, to replace the insides, with noise, load, and line regulation under a microvolt?

The transformer has 1 amp 14.5 volt center tap and 0.5 amp 43.5 volt center tap outputs.

The printed circuit board came from the long gone R&D Electronics on Orangethorpe avenue in Fullerton, California, not far from Orvac Electronics.

I miss the old neighborhood:)

Which "super precision regulator" did you have in mind ?
I figured if it was good enough to run the HP10811 ..it's  good 'nough for me

[RoGeorge]

Nice power supply! 
It is not clear from the pics if the entire enclosure is made of aluminium, if so, might worth changing the 723 with a better design.

Though, I like 723 power supplies, too, particularly because they were usually put in boxes with passive cooling, so they were quiet.  Same for the first HCs (Home Computers) from about the same time with 723 ICs.  The computers were quiet, running hot but quietly.

Annoying fans started (for me at least) with the apparition of the IBM-PC and with the invasion of desktops at home.  That was somewhere around the 90's I think, and since then, I've never manage to get rid of all the fans and their noise.  All the desktops and laptops and lab instruments nowadays are more-or-less noisy.

[David Hess]

I have an old 723 based variable power supply that I built when I was about 13.  Finding myself in need of another bench supply, I dug it up and fixed the enclosure.  I may change the big input capacitors, but what about designing and building one of my super precision regulators, adapted to variable operation, to replace the insides, with noise, load, and line regulation under a microvolt?

The transformer has 1 amp 14.5 volt center tap and 0.5 amp 43.5 volt center tap outputs.

The printed circuit board came from the long gone R&D Electronics on Orangethorpe avenue in Fullerton, California, not far from Orvac Electronics.

I miss the old neighborhood:)

Which "super precision regulator" did you have in mind ?
I figured if it was good enough to run the HP10811 ..it's  good 'nough for me

Something like that shown below, but with a separate bias source for the reference and a negative bias supply to allow operation down to zero.  The reference is divided by a poteniometer to make it variable so the feedback network impedance is fixed making frequency compensation better.  Using a fixed regulator for the power pass element adds free thermal and short circuit protection, but I would likely add adjustable current limiting with a second operational amplifier.

I have oodles of TO-3 fixed and variable regulators, as well as big power MOSFETs, bipolar, and Darlington transistors available for any power pass element configuration, but the transformer can only supply an amp at most unless I dig up another transformer.  Using a fixed 7805 or 7808 has some advantage because it lowers the output range requirement of the operational amplifier.

I could also design a fast response low output capacitance power supply using TO-220 ring emitter transistors.

[David Hess]

Nice power supply! 
It is not clear from the pics if the entire enclosure is made of aluminium, if so, might worth changing the 723 with a better design.

Only the front panel is aluminum.  The top and bottom are thin plywood, and the sides are particle board which is where the front panel screws pulled out and needed repairs.  Keep in mind that my father and I made this when I was like 13.

So far I have repaired the enclosure and replaced the big input capacitors.  I will add a small output capacitor, but it works as intended now.

Though, I like 723 power supplies, too, particularly because they were usually put in boxes with passive cooling, so they were quiet.  Same for the first HCs (Home Computers) from about the same time with 723 ICs.  The computers were quiet, running hot but quietly.

Linear power supplies become impractical at high power.  I remember CP/M systems with linear supplies where the power supply was a separate cabinet and required two people to move safely.  I think all of them relied on the 723.

Annoying fans started (for me at least) with the apparition of the IBM-PC and with the invasion of desktops at home.  That was somewhere around the 90's I think, and since then, I've never manage to get rid of all the fans and their noise.  All the desktops and laptops and lab instruments nowadays are more-or-less noisy.

Where forced air cooling is required, I like to build a small linear fan control circuit to adjust speed based on temperature.  Of course me being me, I use only the integrator part of a PID controller and tune the step response, so the fan varies speed while the temperature remains constant until overload.

[floobydust]

Do you know why the resistors burned up? It could maybe be tap-switched.

When I made my first bench PSU at around age 13 the shop teacher freaked out because I was on the drill press for quite a while, surrounded by electronic parts. It was a class where you made a sheet metal funnel or changed a bike tire lol and it was as if I was from another planet. I didn't have a dad or a drill at home. It worked out OK, mostly dumpster dive and Radio Shack parts, an LM317K and oil-can selenium rectifier from the army surplus store. It powered hundreds of breadboard projects for many years.

You'll have to be clear on what features you want. Last iteration I started out by wanting adjustable current-limiting which sent me down a rabbit hole more complicated than I wanted. OP-07, LM4040, 2SC5200 kind of thing with an added transformer to make +/-12V for the IC's.

Magnetics cost a small fortune nowadays. I must have a few $grand worth stockpiled over the years.
After the pandemic, my rule is "use what you've got" instead of buying new parts.

[David Hess]

Do you know why the resistors burned up? It could maybe be tap-switched.

Two resistors burned up at different times.

The obvious one is in series with the voltage control potentiometer to limit the voltage range.  My guess is that it failed when the output transistor shorted and the potentiometer was turned one way.

The other burnt resistor is part of the Vbe current limit and just overheated.

You'll have to be clear on what features you want. Last iteration I started out by wanting adjustable current-limiting which sent me down a rabbit hole more complicated than I wanted. OP-07, LM4040, 2SC5200 kind of thing with an added transformer to make +/-12V for the IC's.

I was looking for suggestions about this particular power supply, but you are right; I should consider what would be most useful.

If I was building a supply from scratch, I would want a low noise precision voltage and current tracking supply with low output capacitance and high isolation, but this is beyond the scope of this project.  Achieving high isolation and low noise might require a high frequency sine inverter, and this is a future research project for me to test feasibility.

Magnetics cost a small fortune nowadays. I must have a few $grand worth stockpiled over the years.
After the pandemic, my rule is "use what you've got" instead of buying new parts.

I have quite a collection of big and small power transformers, including a pair of plate transformers which are too heavy to easily move.

[coppercone2]

the low Cout is actually a good reason to keep the supply and just fix the burned thing that's irritating me

scratch or nibble it out and coat with epoxy or sandwich in teflon and use epoxy with glass beads or fiber to patch it


adding anything to the supply that effect its Cout is a detriment to its capability , build another supply for other capabilities

for real hacking that c is a big problem


it could be improved though by putting the power resistors on big 'turret' standoffs, fixing the wires to PCB issue, leaving the SLCC capacitor alone, repasting the transistor with a modern thermal grease, checking to see if that through hole can can use a heat sink, washing the pcb, adding fuses to the chassis if it does not have any, replacing the banana holes with binding posts, adding a chassis binding post if its compatible with the design (along with a shorting jumper), etching a scale for the potentiometer and adding a bright mark to show the position (unless its a multi turn, then it could benefit from a vernier), putting a better panel meter and cleaning the chassis

[KE5FX]

At which point he will have the Power Supply of Theseus.