Zeroing out a LM317 reference voltage with a zener - PSU build

[hevans1944]

You already have the "1.225 zener" in place on your original schematic post. Now you just have to provide a small negative voltage through a current-limiting resistor to its anode. I think the "-9V" circuit will provide that, but you don't need the voltage multiplication it provides. So, yes, substitute your device for the 9.1V zener and eliminate the extra parts. Probably you will need to increase the value of R1 and perhaps increase the value of C4 (to get an acceptably small ripple) but the extra diode D3 and capacitor C3 (used for voltage multiplication) are unnecessary IMO.

I know you already have the transformer and full-wave bridge, so wire those up along with a large filter capacitor but don't bother connecting the LM317 yet. From the circuit I posted above, add the capacitor C2, diode D3, current limiting resistor R1, output filter capacitor C4 and two series-connected 1N400x diodes to simulate your LT1004-1.2 (don't want to let the smoke out that by using too small a value of R1). Make sure the cathode of the diode string is connected to the junction of R1 and C4 and the anode of the string is connected to common.

Fire up the transformer and start measuring some voltages. Measure the voltage across the diode string: it should be about -1.4 V with respect to circuit common. Measure the voltage across current-limiting resistor R1: calculate the current through this resistor. Adjust the value of R1 for about 10 mA, which is about one third the maximum reverse bias current the LT1004-1.2 is rated to pass. If everything checks out, replace the two diodes with your LT1004-1.2. Remember to connect its "anode" to the negative output of current-limiting resistor R1 and its "cathode" to common. Go back and check the voltage drop across R1 to make sure the current is approximately 10 mA. Check the voltage drop across your LT1004-1.2 to make sure it is within ±4 mV of 1.235 V. If everything checks out you are ready to wire up the rest of your voltage-regulated power supply. Note that half of that 10 mA current budget flowing through R1 will be used by constant current flowing through your 240 ohm R3 resistor after the LM317 is connected.

The only reason I can think of for using op-amps is for current limiting circuitry. This is a nice feature for any bench supply, especially if it can be set for any current between zero and maximum. However, I think this is unnecessary for your first power supply project. The LM317 protects itself from overload, even extending to a short-circuit from output to common. Plus, you will have digital panel meters to monitor output voltage and load current, so variable current-limiting is just a whistle you don't need to tweet right now.

Please let us know the results of your breadboard experiment.

 

[chopnhack]

I don't have the precision v. reference shunt on hand yet, but I tossed it into spice. Interestingly enough, I left the diode and caps in as they seemed to be effective when I reviewed the circuit in preventing negative voltage during one half of the sine wave - proved to be correct in spice!! When I take D12 out of my schematic, negative voltage appears across the resistor (about 9v) and causes the v. reference to not work. Add the diode back in and it blocks the negative half and all is well. Check it out, I will include the file.

The only reason I can think of for using op-amps is for current limiting circuitry.

Fair enough, I appreciate the input! It helps me refocus and get this thing started and finished!!

 

[hevans1944]

Yikes! You tricked me into launching LTSpice!

Okay, so leave D12 in the circuit if it makes it work. I don't see how it does anything other than allow the voltage on C4 to add to the voltage across D13 when the anode of that diode is negative with respect to common. In fact, if you leave D12 in the circuit, what does removing D13 do? I guess I will have to poke around in LTSpice to find out what's going on, since I still don't have a workbench cleared off in the basement yet.

I did find a brand new, still in its package, LM317 that I bought from Radio Shack last year, so maybe I will breadboard the circuit and use a pair of forward-biased diodes (as described above) in lieu of the LT4004-1.2 that I also don't have and haven't ordered yet. Woo! Woo! Kinda pricey from Digi-Key but Arrow has them for about two bux each. Won't be ordering a whole lot of these to play with, but maybe one or two if I add enough more stuff to the order to make the shipping cost worthwhile. Found some transformers while I was cleaning up the workbench, so maybe I too will build a laboratory bench supply one of these days.

 

[chopnhack]

Yikes! You tricked me into launching LTSpice!

Hahaha, curious eh?!

In fact, if you leave D12 in the circuit, what does removing D13 do?

If we do this, the current across the resistor R7 drops from 2.4mA at inrush to 0.286mA at steady state - and the zener does not get enough current.
As compared to when the D13 is left in, current across R7 starts at zero and ramps up in a saw tooth pattern to 18mA. This is a bit close to the max of the IC so I would go with another value of R7 like 1.5k Ω.

Kinda pricey from Digi-Key but Arrow has them for about two bux each

EEP - yes I saw digikey's price on that and was surprised for a diode, LOL. I think I may have found a 29 cent equivalent - check this one out: LM4041 from our friends at microchip.

I too will build a laboratory bench supply one of these days.

You inspire me, I inspire you works for me boss!

On my desktop(the physical one, not the pc) I have the xfmr, some quick disconnects, old extension cord sans one plug side (previous experiments). I couldn't find anything bigger than a 160mA fuse so I will run this unfused
Will let you know what I find in a bit.

 

[chopnhack]

Here is the setup:

The transformer provided 23.45 VAC unloaded, so the caps I picked where rated at 35vdc, close to max, but I figured under load the xfmr would have some voltage drop so that should keep it safe. Perhaps in the psu it would be more prudent to go with 50vdc for the extra capacity in case of transients and ?wild swings caused by high loads on the output which may be abruptly cut off causing back 'waves'?? (My new thought on current is that its like a wave propagating through the wire)

edit - diodes used were 1n4002, bridge 2A, 800V generic 2W08 high speed rectifier.

Anyway at startup, from R7 to common was -0.756 VDC. I am not sure why yet. Bridge + terminal to common was +39.3 VDC (does that seem correct?)
After 10 minutes the R7 to common junction slowly drifted mV by mV to -1.32.
About 20 minutes later, same junction was at -1.25 - temperature in room consistent.
About 5 minutes later, -1.2.

So, it looks like the circuit works as predicted, the diodes apparently need some kind of 'warm up' time to reach a steady state.
What do you think of the LM4041?

 

[hevans1944]

EEP - yes I saw digikey's price on that and was surprised for a diode, LOL. I think I may have found a 29 cent equivalent - check this one out: LM4041 from our friends at microchip.

It's NOT a diode. It's an analog integrated circuit with two terminals that acts like a zener diode.

The LM4041DYM3-1.2-TR appears to be an acceptable substitute for the LT4004-1.2 with a lower recommended reverse operating current of 12 mA maximum. Absolute maximum reverse current is still 20 mA for all devices in this series. This may be a later part than the LT4004-1.2. With manufacturing experience usually comes decreased cost.

So, it looks like the circuit works as predicted, the diodes apparently need some kind of 'warm up' time to reach a steady state.

Forward-biased silicon diodes have a voltage drop that is a function of current and temperature. So, yes, they do need to reach a stable temperature before their voltage drop settles down. Won't stay there though. The input voltage is unregulated and will vary with power-line voltage. That won't bother your "zener" much as long as the well-filtered current through R7 (get rid of that sawtooth!) stays below 12 mA, but two series-connected forward-biased diodes are not a substitute for your "zener" device. They are just a convenient way to test the negative bias circuit used with your real "zener" device. BTW, datasheet says you need a capacitance of at least 10 nF (0.01 μF) across the LM4041 for stability.

Hey, I thought you said you were not going to use a center-tapped secondary winding? You can use the CT as your circuit common, get half the raw DC out (about 20 V instead of 39-something) and reduce the dissipation on the LM317. And you can now use the negative terminal of your full-bridge rectifier to produce the negative voltage you need for your LM4041. So whatsup with that? Is the transformer in the photo not the one you plan to use?

 

[chopnhack]

It's NOT a diode. It's an analog integrated circuit with two terminals that acts like a zener diode.

Makes more sense and its priced right. I did see recommended operating current and will change R7 when I get my hands on some. (Thanks for reminding me in case!!!!)

get rid of that sawtooth!

Agreed, I think I will play with the values of C5 to see if we can accomplish that.

Hey, I thought you said you were not going to use a center-tapped secondary winding?

These were the two xfmr's that I got for $14, I have to use them, they were a steal!!!
Sorry for the confusion, I probably screwed up in the wiring of the circuit, LOL. I think when I measured from one side to center tap I had less than the expected 20V (not remembering that I had to multiple by rms) and thus hooked it up the way you saw. Sadly I also just updated my spice model and ended up splitting my transformer into two as I was looking at my physical model

No, I know what it is now - this transformer is a bad choice to produce 18vdc out - its 20vct so 10v per winding or 14 vrms... oh, well, moving on

So, if I use one of these xfmr, I will be able to have two separate outputs at 0~13Vdc at up to 2.5A.

I do have c6 across the intended v. reference at 100nF - isn't that sufficient?

 

[chopnhack]

Truest thing ever said of spice regarding modeling and true to life sims:

<<
sceadwianNew Member
Jun 1, 2009
499
37
Those parasitics also don't fully characterize the truly complex nature of an IC. Most IC models for Spice aren't actual TRUE full spice models for the circuit, or they would take about a decade to fully simulate even power up oscillation; they're behavioral models. At best they are nothing more than an approximation if you don't know what you're doing, and the real possible crutches of the real world devices not just the models that simulate them well enough for general purpose use, specifics are a bitch.
>>

Here is an updated schematic to show that the secondaries are yielding 10VAC RMS (14.2VAC). The current ripple that shows across R7 is about 116μA over 16ms - roughly peak to peak.
Is that adequate ripple control? Spice enclosed. The only thing missing from the schematic is the output switch which is a simple toggle switch rated at 125v, 6A which I think should be more than adequate for the 1.5A max output at now ~13v. If we are good here, next need to start calculating max or worst case scenario heat output of the LM317. I will also start to work on the schematic for the psu for the panel meters and then the fixed 'digital' outputs section.

 

[hevans1944]

Okay, it's been a week and you probably have the voltage reference thingy by now. How did that work out?

I think I see where I erred in saying C4 and D13 could be removed. If you connect D12 cathode where the + terminal of C4 is connected (at the common connection of D1 and D3), then C5 will charge on each half-cycle from current passing through D2 and D12. Also, for either circuit (with or without C4 and D13), capacitor C5 will have a negative voltage with respect to ground at the terminal that connects to the anode of D12. You have it incorrectly shown with its positive terminal connected there. Reverse capacitor C5 connections: positive terminal to ground, negative terminal to anode of D12.

As @davenn stated in his post #3, you need to replace the 1N4148 small-signal diodes with 1N4007 rectifier diodes or maybe even something a little beefier. The 1N4007 is capable of handling 1A of DC current, maybe a bit more if you push them hard and keep them cool. The 1N4148s... not so much. How much DC output current are you expecting to draw from the LM317?

 

[chopnhack]

Okay, it's been a week and you probably have the voltage reference thingy by now. How did that work out?

LOL, negatory good sir - I have not purchased anything else other than the xfmr's because they were such a good deal and then some ancillaries, like binding posts, fuse holders, sheet aluminum. I figured I should nail down the design first before placing the order to be economical

You have it incorrectly shown with its positive terminal connected there.

Well spotted Hop!! Thanks, that would have probably popped in my face I should have spotted that in spice...

How much DC output current are you expecting to draw from the LM317?

I am not final on this, but I might go back to using two separate xfmr's so that I can use the 20vac (~28rms per channel). I wanted to have a good range of voltage for experiments. Amperage up to ~2.5A (xfmr states 2.8 (20vct 56va)).

I looked at this schottky diode, but was not sure of my schematic, aside from the bridge, if this was warranted for the other diodes.

I am thinking about using a multiposition selector switch to enable the xfmr to attach to various v. regulator "boards" so that I can tailor the ranges a bit better. For instance, a general 0-20vdc 1A, a 3.3v 1A from the center tap, 5v 2.5A again from center tap, but the two windings in parallel, 12v 1.5A and 24v 2.5A. I would need a second tap selector switch as well in the scheme I am thinking of.

Is it possible to accomplish this with a bridge and a bank of caps then feed this filtered dc to my various "boards" via the selector switch? With switching the vct I see at least three bridges and three banks of caps. The nice thing is that I wouldn't have to buy anymore transformer aside from the tiny ones for the panel meters

Sorry, still analyzing and developing! Maybe I can draw a block sketch of what I am currently thinking of and post it.

 

[chopnhack]

Today's madness brings some design alterations... maybe..

After some thought and some excellent suggestions I have come up with condensing multiple supplies into a compact, usable form. The schematic above is partial - I really just wanted to ask a few questions about the approach. I thought it would be really cool to access the various 'voltage supply circuits' via a multiposition rotary switch. I figured out how to do that for the circuits that use the outer windings. For the two lower voltages that will use the center tap, they each went on their own DPST switches. As is, it looks like I can use either 3.3v or 5v individually or at the same time.

1. Should the bridge and capacitor bank be ahead of the rotary switch to minimize components? i.e. cap/bridge common to all three upper circuits with separate regulators and discretes after?
2. Should the rotary switch be 'make before break' to prevent arc damage on the contacts?
3. As drawn, if the center tap fed circuits are in use, would I still be able to have 20VAC and be able to use the upper circuits (albeit at a reduced current)?

I really like this approach, more flexibility, a few more parts and regulators, but very interesting! If this is a possibility, I would replicate the same group of circuits for the second transformer thus having two identical outputs and the ability to have +/- supplies for audio amps, etc. The xfmr on the bottom is for the panel meters. To save space and number of panel meters, the upper three circuits switched by the rotary switch would share one panel meter - I assume that I would simply need to common their outputs - I haven't fully elaborated on this yet and I am hopefully that this will not be a problem.

Your thoughts?

 

[hevans1944]

My thoughts are: you are over-thinking this.

I can see using a separate switch to enable and disable separate power supplies, all powered from separate transformer windings, but it is penny wise and pound foolish to create an elaborate switching scheme to create series and parallel configurations of a single transformer to allow for different rectifier/regulator power supply outputs. What if you need an output voltage for, say, 3.3 V and 5 V at the same time AND an adjustable 0 to 20 V output and a fixed 24 V output at the same time? Likely to overload the transformer or, in the second case, have only one of the two outputs available when you need both. For example, you have a spiffy CMOS circuit you want to energize with 10 or 15 volts and use it to drive a 24 V DC relay coil you just salvaged from a WWII Mark Z whatchamacallit chassis. Oops! Gotta find another transformer and throw together a 24 V relay supply. Good thing it won't need voltage regulation though.

Maybe I'm just old-fashioned, but I would have appropriate individual and separate secondary windings for each power supply output. Each secondary would have a bridge rectifier, an energy storage (filter) capacitor, and a regulator circuit. Then, if I didn't need a particular output, I would maybe have a switch that disconnected it from the output terminals. Maybe a four-pole switch to disconnect the transformer secondary winding too (saving a watt or two), but that's as far as I would go. In all probability I would leave everything powered on all the time and just select whichever output binding posts I wanted to use on any particular day.

I've done it more times than I care to remember, but to base a project as important as your bench power supply on parts you happen to have because they were purchased cheap is not good engineering. Of course hams do this all the time: they find this nifty boat-anchor of a plate power transformer at a flea market for ten bux and dream of building a kilowatt linear amplifier around it. Maybe it works, maybe it doesn't. But the amplifier design should come first and then the design for the power supply to energize it. And in the case of power supply design, you decide what output voltages and currents you need, assume all the outputs will provide their rated current at their maximum voltage, then you find one or more transformers that satisfy the power requirements.

About your Schottky diode from Cree: I purchased four of these puppies for use with four Cree MOSFET power transistors in a high-frequency switching application (PWM control of the four phases of a stepping motor at 100 kHz or so). Their main claim to fame is very fast switching and high power dissipation. You do not want to use these as your power rectifiers. Their forward voltage drop sucks and their internal resistance is terrible. Scroll down to near the end of the datasheet for a model. Except if you need fast switching (which you do not!) these are not the diodes (or droids) you are looking for.

Datasheet for the LM317A from Linear Technology says this device is good for 1.5 A if you properly heat-sink the K package and pay attention the voltage drop from IN to OUT. You don't need a diode from ADJ to OUT with this device, and the diode from OUT to IN should be a power rectifier diode (same as you would use in the input bridge rectifier) like a NTE576 not a small-signal diode like the 1N4148. You might get away with something from the 1N400x series, but that would be pushing their 1 A forward current rating. Although dirt cheap and readily available, this series is not recommended for new designs.

Hop

 

[chopnhack]

My thoughts are: you are over-thinking this.

Without a doubt - Easy to do though!! To be honest, I have been getting multiple ideas and inputs from various locations. So I have been filtering and processing the various items that pop up and the thoughts that go through my mind, so I apologize for the schizophrenic nature of these posts sometimes.

What if you need an output voltage for, say, 3.3 V and 5 V at the same time AND an adjustable 0 to 20 V output and a fixed 24 V output at the same time?

Indeed, and that is very possible, hence what I asked in question #3. Luckily the separate transformer for the digital outputs is fairly inexpensive, so that is not a deal breaker. Into the plans once more!

these are not the diodes (or droids) you are looking for.

LOL

I appreciate the input! It is hard to know what to plan for not having the experience and knowing what I would need in the future, many thanks

 

[hevans1944]

I must confess I don't own any NTE576 power rectifier power diodes. I have lots of 1N400x types, lots of small-signal 1N914 and 1N4148 types, and even a few germanium diodes I recently spotted at Mendelson's and just had to buy at ten cents each. I will have to eventually purchase the higher current rated NTE576 (or whatever generic equivalent is available) if I do build a serious bench power supply with some serious current capability, but "serious" here would depend on the state of my finances... hence the usual use of 1N400x series diodes to cobble up something on the bench to fart around with.

,,, It is hard to know what to plan for not having the experience and knowing what I would need in the future, ...

Ahh! But experience and knowing are what you are getting now. The future will take care of itself and you will find yourself purchasing "stuff" (like that pair of nifty transformers) with future projects in mind. If you can get "stuff" for bargain prices, and can afford to keep it around until you find your "round tuit" that will allow you to actually build that future project... well, go ahead. I've done that for years and now have enough "stuff" for future projects well into my 90s. But I am sucker for new technology and purchase some of that, too, with no particular project in mind. At some point wife will tell you: enuf is enuf: either build something or throw that junk away. My wife tells me this practically every day, and she isn't just referring to electronics. There is a few hundred board-feet of random widths and lengths of poplar and red oak S2S lumber stacked up in the basement just waiting for me to wood-butcher something out of it. It was purchased sometime in the 1980s IIRC. <sigh>

Hop

 

[dorke]

I have just seen this thread.
Forgive me for not reading it all.
Here are a few thoughts :

1. I wonder why do you need to get down to 0-Volts?
I know all bench power supplies do get to 0 ,but let me tell you ,as far as I remember.
In decades of using various kinds,I have never used one below 3Volts !
So, 1.25V is more than enough for me,should be the same for you ,don't you think?

2. To be practical a bench P.S should be a dual ± one (tracking non-tracking one).

3. Better go for higher current ,3A and add current limit to it as well.

4. For fixed voltages an excellent and very cheap solution is an old PC power supply.
A very simple modification is needed, and voila you have a heavy duty bench P.S (3.3V,5V ,±12v)
Often you just have one laying around somewhere ,so that is free

 

[chopnhack]

It was purchased sometime in the 1980s IIRC.

OMG Hop, I think its dry now Make the Mrs. a cutting board already

I spent some time looking at various designs again, I know.... I am a sucker for eeking out every last drop of <whatever>. There are so many cool features to 'build' into a psu that you feel bad not to at least look at them and then of course - design creep sets in - and the chicken you started designing is a dragon now.... I did realize as I was looking at the various schematics that these features can perhaps be added later, so as per your advice, of several times over, build it already! started to set in, this may not be the last psu I build.

Starting design over with desired project objectives:

1. Digital panel meter for voltage/current for each channel
2. Two 0-20VDC 1-1.5A adjustable supplies
3. Two fixed 12v 3A supplies
4. Two fixed 24v 1.5A supplies

Points three and four would probably be one xfmr per channel, a dual secondary that I can have in parallel for double the current on the 12v circuit
I was told that the digital supplies should not be kept in the same psu as the other supplies. The example given was that you could be powering relays or motors and the inductive spikes could be disruptive of the digital signals. Makes sense and I had not thought of that. So for that, this becomes a separate psu, that I think I can deal with later.

Lets see how complicated I make this

 

[chopnhack]

I have just seen this thread.
Forgive me for not reading it all.

Pleasure to have your input Dorke! No apologies necessary, I tend to go off on tangents so I can not blame you for that. This is but part of the psu build questions that I have asked. There are other posts that pertain to it as well.

1. I wonder why do you need to get down to 0-Volts?

I thought it would be logical to start low and slowly come up when testing circuits. The fact that it was easily attainable with the voltage reference IC and the LM317 made it a cinch to do. You do raise an interesting point though, at 3v or even just the reference voltage of ~1.3 (from the LM317) I doubt there would be much harm to a circuit, well.... , so long as the current was low too!

2. To be practical a bench P.S should be a dual ± one (tracking non-tracking one).

Agreed, I have been with that thought from the beginning, I have not shown in the schematic the duplicate circuit as it will not be a tracking supply and for simplicity in sketching it quickly.

3. Better go for higher current ,3A and add current limit to it as well.

I would love too!! I am first starting with a basic layout and then if I understand the placement of all the components, I will complicate matters with a pass transistor for more current and perhaps we can discuss using op-amps and transistors for current limiting.

4. For fixed voltages an excellent and very cheap solution is an old PC power supply.
A very simple modification is needed, and voila you have a heavy duty bench P.S (3.3V,5V ,±12v)
Often you just have one laying around somewhere ,so that is free

I thought of this too and read a bit about them - many use them and many advocate against them. The one reason that I would probably go against them is that the cap. sections on many of these surplus units are probably toast. From what I recall, so long as you provide a bit of current to a sense wire (indicates the mobo is getting power) the rest of the power sections come online.
Thanks for the input!

 

[hevans1944]

I have just seen this thread.
Forgive me for not reading it all. ...

There is much, much, more... some of it in private conversation. @chopnhack has professional chops in many areas, including home construction and maintenance: drywall installation and repair, plumbing installation and repair, electrical installation and repair, and fine woodworking including cabinet making. However, he is a relative newcomer to the electronics hobby. On the plus side, he is also a fast learner and a critical thinker. Most of all, he is having FUN with his new hobby! His profession, which I cannot disclose here, tends toward perfection. This is an admirable trait for any pursuit, but it can be a hindrance for engineering. In my early years I was very much a perfectionist, but one day an older mentor reminded me: done is better than perfect. I have to keep reminding both myself and @chopnhack of this engineering truism.

... Here are a few thoughts :

1. I wonder why do you need to get down to 0-Volts?
I know all bench power supplies do get to 0 ,but let me tell you ,as far as I remember.
In decades of using various kinds,I have never used one below 3Volts !
So, 1.25V is more than enough for me,should be the same for you ,don't you think?

2. To be practical a bench P.S should be a dual ± one (tracking non-tracking one).

3. Better go for higher current ,3A and add current limit to it as well.

4. For fixed voltages an excellent and very cheap solution is an old PC power supply.
A very simple modification is needed, and voila you have a heavy duty bench P.S (3.3V,5V ,±12v)
Often you just have one laying around somewhere ,so that is free

I agree with all four points. But sometimes it is absolutely necessary to "sneak up" to the rated power supply voltage starting from zero, lest the magic smoke escapes from some critical (and usually expensive or no longer obtainable) component you are playing with.

The point about dual tracking ± outputs is extremely important for op-amp circuits requiring bipolar supplies. I don't think @chopnhack has considered this yet. More current, with an adjustable current limit control, is always a nice feature.

I don't really trust PC power supplies for general-purpose bench use. Although purpose-built to drive computer mother boards and certain peripherals, such as disk drives, it is almost impossible to completely eliminate the switching noise all PC PSUs generate. A conventional linear regulator attached to transformer/full-wave rectifier/large energy-storage capacitor is a very quiet (if inefficient) design that will have little noise on its output. This is essential for prototyping because the last thing you need to do is troubleshoot the power supply along with the prototype.

And finally, @chopnhack is building a bench power supply as much for the learning experience as anything else. It is essential that he understand how it works and how to troubleshoot and repair it. This is very hard to do with an off-the-shelf PC ATX power supply for which there may be no schematic, no parts list, and no source of spare parts. Not that any of that matters: I just buy a new one when one of them fails, often not even bothering to salvage parts from the defective PSU. Keeping the fan clean and the fan motor bearings lubricated seems to be the best way to prolong the life of the PSU, but that applies to anything that uses/requires forced-air cooling.

OMG Hop, I think its dry now Make the Mrs. a cutting board already

I have made cutting boards in the past but no longer do that. Wood retains moisture and bacteria, so we ditched our cutting boards and now use only disposable plastic "cutting" boards, being careful to wash them after use and using only one side. I think my wife gets them at the Dollar Tree store about a block away from our house. They last forever, but we recycle them after a few months of use anyway.

... this may not be the last psu I build.

Gee, I hope not! Please don't try to create an heirloom to display in your old age or pass along to your children. It may turn out that way, but for now its just a valuable tool for learning and to use on your workbench. Add the hand-rubbed walnut cabinet later. I plan to introduce you to high-power pulsed electronics very soon now. You should plan on building many new and different power supplies as your skills and experience improve.

Lets see how complicated I make this

I thought that was my job!? Let's see how soon you can make this... and have it actually work.

Hop

 

[dorke]

I don't really trust PC power supplies for general-purpose bench use. Although purpose-built to drive computer mother boards and certain peripherals, such as disk drives, it is almost impossible to completely eliminate the switching noise all PC PSUs generate. A conventional linear regulator attached to transformer/full-wave rectifier/large energy-storage capacitor is a very quiet (if inefficient) design that will have little noise on its output. This is essential for prototyping because the last thing you need to do is troubleshoot the power supply along with the prototype.

And finally, @chopnhack is building a bench power supply as much for the learning experience as anything else.

Hop

I thought it goes without saying, the PC P.S only for powering digital boards-and they are great for that.

For learning ,
the way to go is building the linear P.S using only transistors,diodes and op. amps ,
no 3 pins regulator ICs at all.

 

[chopnhack]

Thanks for the kind words, Hop, I am blushing over here, LOL

The point about dual tracking ± outputs is extremely important for op-amp circuits requiring bipolar supplies

This I haven't fully understood as I figured it would be easy enough to adjust each supply, no? If you create a tracking supply you no longer have two independent supplies! Wouldn't that be more limiting. I asked about dual tracking supplies last year in my first bench power supply attempt, but never got an answer regarding this, maybe someone can clear it up?
Bench Power Supply

I agree with all four points. But sometimes it is absolutely necessary to "sneak up" to the rated power supply voltage starting from zero, lest the magic smoke escapes from some critical (and usually expensive or no longer obtainable) component you are playing with.

Agreed, had magic smoke fountain out of a cap last year, very unexpected!! I believe that is when I learned about RMS values of ac to dc.

And finally, @chopnhack is building a bench power supply as much for the learning experience as anything else. It is essential that he understand how it works and how to troubleshoot and repair it.

Absolutely, this is critical. I could have placed an order for some of those units we see on e-bay eons ago, but I'd rather learn what these simple circuits do!

Gee, I hope not!

Don't misunderstand, I do intend to do a good job, but this will not be my last p.s.u., lol.

I plan to introduce you to high-power pulsed electronics very soon now.

Negatory, sensei - I must master this, finish the blink circuit and finish the run delay circuits programming as well!!! Only then will I be able to take on a new challenge!

I thought that was my job!? Let's see how soon you can make this... and have it actually work.

Ah!! A challenge, dead bug or professional pcb? LOLOLOL Now to be fair, Hop, any joe can slap a cap, a v. regulator and a resistor together and have a working supply We need a professional, functioning supply that can meet the needs of my tinkering and the future! I will be posting a sketch in the next few days of the basics, essentially little more than a block diagram to flesh out the ideas that I have come to today. Then we can complicate