Setting the LM317's MAX voltage output.

[CharlesGibbsNam]

Good day everyone.....

Hope you are all well, my first post here.

I have been on the know it all google and YT trail for a few days now, have worked through some examples and think I have a relatively ok understanding of how to slap together an ok variable supply using the LM317.
I will be using the LM337 to generate a negative 1.25V rail to get down to 0V.

My question is this: 

Other than having an exactly unregulated 33Vdc which of the 2 resistors on the voltage divider is actually setting your max output voltage.... ? Is it R1?
There are so many conflicting views it's hard to be able to make sense of it all?
I do understand that the resistor wants to maintain a 1.25V difference between output and the adjust pin......... That though still does not answer my question.... 

Using figure 23 on page 17 of the TI datasheet as a starting point for a variable regulator, is R1 going to determine the max voltage output of that circuit?
My transformer is a 12-0-12 ac transformer so roughly 34 volts out when smoothed.

I am going to set the circuit up so that the taps are switched when going over a certain voltage to keep heat dissipation down.

I would be grateful if either someone could point me to ONE of the 20 pages worth of posts where this has specifically been answered or if possibly not, then please help me attain a suitable answer.

Thanks,
Regards
Charles.

[Ian.M]

The LM317 needs a worst-case minimum load current of 10 mA to maintain regulation.  (A specific LM317 may or may not regulate at a lower load current at a significantly lower current, but even if it does, the output voltage may be out of spec.)

Therefore a starting point for calculating the upper resistor (R1) in the feedback divider is almost invariably 120 ohms which guarantees a min. 10 mA load current.  Many are misled by the LM117 in the same datasheet which has a min. load current of 5mA so can use a 240 ohm upper resistor.

Neglecting the tiny ADG pin bias current, that same current flows through the lower resistor (R2), setting the ADG pin voltage, and thus the output voltage 1.25V higher.  The max. value of the lower resistor is thus what sets the max. output voltage.  The exact equation is given at the beginning of the 'Application Hints' section of the National Semiconductor datasheet, but as a rough rule of thumb, you need 100 ohms per volt.  Unfortunately there is a limited selection of available potentiometer track resistances so you generally have to settle for a larger pot resistance and accept that the max output voltage (as permitted by the input voltage) will not be at the pot's endstop position which means that past the 'max.' voltage there will be a further increase in voltage, followed by a 'dead zone' where the voltage doesn't change (much) as the regulator is in dropout.  An alternative is to use a smaller pot resistance and decrease R1 to increase the current through the pot to bring the voltage up to your desired max, when the pot *is* at its endstop.

Also, be aware that the no load rectified output voltage of your 12-0-12 transformer *will* be higher than the 34V you have calculated, and if it has particularly poor regulation or if the supply voltage is above nominal (so called 'high line' conditions), it may exceed the max. permissible input to output voltage  differential for the LM317, with a risk of damage to it if the output voltage is set to minimum with no load.

[CharlesGibbsNam]

@Ian.M

Hi Ian. Thank you. I am incorporating an LM334 to make sure I have a good minimum load. I have been made aware of that need.

So, in essence, A smaller R2 pot (5K?) and a smaller R1 (Possibly 100 or 90 ohm?) resistor to see what the max Vout then is?
Hope I understand this correctly?

Also, be aware that the no load rectified output voltage of your 12-0-12 transformer *will* be higher than the 34V you have calculated, and if it has particularly poor regulation or if the supply voltage is above nominal (so called 'high line' conditions), it may exceed the max. permissible input to output voltage  differential for the LM317, with a risk of damage to it if the output voltage is set to minimum with no load.

I used the 24Vac x 1.414 equation this gives me 33.96Vdc I understand this is not precise but seeng as the regulator can take up to 40Vdc it should be ok?

Regards.
Charles

[Ian.M]

You want max. 30V out, and are subtracting 1.25V at the bottom end of the pot to let the output go right down to zero.  The output voltage is thus I(R1)*R2.   A 5K pot wont pass 10mA with 30V across its track, so is no good.  Try a 2K pot, which will need 15mA through it to develop 30V out, so calculate R1 accordingly.   82 ohms should be fairly close.   Probably easiest to take the next higher value and trim down (i.e. voltage up) with a select on test resistor or preset in parallel.

Whether or not Vunreg will go over 40V depends on your transformer's regulation (percentage change in output voltage between full load and no load) and the max. permitted 'high line' voltage in your country.  (+10% of nominal in many countries.)   A small transformer may have 15% or even worse regulation, but a large transformer can go as low as 3%, or even lower for very large transformers.   Measure the unloaded voltage than multiply by 1.1 (assuming +10% max. 'high line' voltage), to see if you may have a problem.  If it goes over 40V you'll need a LM317HV which can handle up to 60V.

N.B a LM317 makes a lousy bench power supply if you need much output current or adjustable current limiting.  Maybe look at LM723 based designs?

[CharlesGibbsNam]

HI Ian.....

Hmmm.

Yeah I see 5K will only do 6mA. Problem is where I am situated getting a 2K multi turn pot is not a viable option...

I am in Africa, to import a 2k Bourns is just not feasible. Not the price tag of the pot...... The import cost of it.... It's just outrageous.
Do you think I could get away with putting a 2K resistor maybe even a 3k across the pot? (The secret life of pots) kind of trick?

I do understand that this changes the linearity of the pot.

I would like to do a multi turn.....

Again thank you.

Regards
Charles

[Ian.M]

As the pot is being used as a variable resistor not a potentiometer, no you cant 'help' it with external resistance across its track.  You've got two options here - go down to a 1K pot, passing 30mA, or add a constant current sink circuit to meet the min load current requirement.  As you've got a negative rail this could be as simple as a NPN transistor, collector to output, base to 0V, with a resistor between its emitter and the negative rail to set how much current it sinks.

N.B. If you follow the simple LM317 circuit, momentary loss of wiper contact while turning the pot can result in the output voltage spiking up to its max. limit, or even a couple of volts below the input voltage.

See: https://www.eevblog.com/forum/projects/question-about-potentiometer-wiper-failure-in-bench-power-supplies/ for how to avoid that.

[CharlesGibbsNam]

Will go and have a look the link you've posted there.

Thank you.

[CharlesGibbsNam]

Hi Ian

As you've got a negative rail this could be as simple as a NPN transistor, collector to output, base to 0V, with a resistor between its emitter and the negative rail to set how much current it sinks.

Is the attached schematic correct?

[langwadt]

I will be using the LM337 to generate a negative 1.25V rail to get down to 0V.

An LM337 will not generate a negative supply, it'll regulate an already negative supply

[Ian.M]

Yes.  I assume there's either a capacitively coupled bridge rectifier,  or a low voltage secondary feeding a rectifier, to generate a raw negative rail so the LM337 only has to regulate it.

@Charles,
The current sink schematic layout is correct but R6 is the wrong value.  The emitter resistor R6 has Vneg-Vbe across it.  You need to sink 4mA (as 6mA flows through your 5K pot).  Taking Vbe as 0.65V, gives 0.6V across R6, which must therefore be 150 ohms or a little lower.

[Doctorandus_P]

I do understand that the resistor wants to maintain a 1.25V difference between output and the adjust pin....

This is not true. The resistor does not "want" anything, it is just a passive component, and combined with:

Other than having an exactly unregulated 33Vdc which of the 2 resistors on the voltage divider is actually setting your max output voltage.... ? Is it R1?

Just look at it this way:
The LM317 measures the voltage between it's output pin and the adjust pin. If that voltage is lower then 1.25V, then it opens up the (high current) output transistor more, if the voltage is smaller then 1.25V, then it reduces the drive strength of the output transistor.

Usually the output voltage is set with a second resistor (or potentiometer), and in that case the voltage ratio between those two resistors is equal to the resistance ratio of those two resistors, because the current through both resistors is the same (except for the small constant current through the adjust pin). And that is the basic way the LM317 set's it total output voltage. But it does not have to be with resistors. You can also connect some voltage reference or an opamp output to the adjust pin. If you look around you will find some creative ways in which to (ab)use an LM317.

My transformer is a 12-0-12 ac transformer so roughly 34 volts out when smoothed.

That is getting quite close to the maximum voltage an LM317 can tolerate, combined with the negative voltage on the adjust pin, it's getting uncomfortably close.

Also note that the LM317 only sees the voltage differences between it's own pins. It is for example possible to use an LM317 with a 200Vdc input, and then regulate that voltage to a 190Vdc output (just 10V difference) but you have to carefully design the circuit so the voltage differential over the LM317 itself never exceeds it's limit.

[CharlesGibbsNam]

Good day @Ian.M

Sorry, on the schematic I just wanted to know if the layout was correctly understood as worded, I just used an arbitrary valued resistor.

That is getting quite close to the maximum voltage an LM317 can tolerate, combined with the negative voltage on the adjust pin, it's getting uncomfortably close.

This is most certainly very true.
I have factored in a +10% increase in voltage and reach a value of 39.20Vdc again, these are all perfect world values as there are still losses involved in converting to dc.
Transformer secondary unregulated AC output at the upper tap is 24.4Vac x 1.1 = 26.84Vac then taking that x 1.414 = 37.96Vdc. + 1.25 = 39.20Vdc. This is not including any loss in wiring voltage drop across diodes etc..

The LM317's data sheet says it can regulate up to 37V? I understand that these are maximum voltages, but, as I "theoretically" do not want to go that high I should remain with in this margin, again, "theoretically".
But, if I am to look at all that has been said it would seem that I would face exactly the same challenges with the LM723, not only that, but the cost?
The LM723 also has only a 40V limit, if all this top end voltage is of a danger to the LM317 is it not the same then for the LM723? 
I also do understand that there is a big difference in regulation (LM723 being a lot better) but, with that comes the price tag.

The LM723 again is not locally available, I will either have to wait for it to be imported which could only be next year, or it has to be specially ordered at a minimum of 10pcs, and in my money at 50N$ a piece
that means 500N$ out of pocket (not including shipping, which as already stated is just ridiculous), the LM317 is here, available, and cost me 15N$ a piece, again, I am in Africa, which makes getting hold of certain items not viable The LM 317HV is also not locally available.

As I am not using the LM317 to directly supply the main current to the output but am rather shunting the current via a set of power transistors, would it help if I could "limit" the the input voltage "AFTER" the dropper resistor to the LM317? Say for example by using a TL431 or possibly zener diodes to regulate to 33V, do you think this could be a viable solution? Obviously using the correct transistors to supply the needed current to the LM317.

I urge all who read this to try and understand that we (in certain parts of Africa) do not have a lot of the luxuries that many of you take for granted, eg. quick door to door deliveries, LES's that stock almost everything you want, or, that will order in anything you need at a decent and affordable price, I design with what is available to me, here in Africa, often times import, depending on the product can cost up to 2x what the product itself costs, please bear this in mind and do not take what you have for granted and think that i am afforded the same amenities as you have available to you.

Yes, there is a capacitively coupled negative rail, from this the LM337 will only have to "regulate" to the required 1.25v.

On that note, addressing some of the other comments.

I am not natively English speaking, I work in and live in a natively speaking Afrikaans population, This means: THAT I DO NOT ALWAYS WORD MY SENTENCES CORRECTLY!!

If all you want to do is attack my poor command of the English language then please keep your comments to yourselves.

For the constructive and meaningful parts of your explanations for which there were answers to the asked questions, I thank you.

Regards.
Charles

[Zero999]

The person who was critical of your English is also a non-native speaker. It's better to learn from it, rather than get upset.

I've done this before. Here's my circuit, which is a bit more complicated and uses an op-amp to read the 1.25V LM317's reference, invert it and give -1.25V.



A capacitive negative voltage doubler and zener diode are used to generate a negative rail to power the op-amp.


In your case, you can use the above circuit to power the LM337, if you don't want to use an op-amp.

[Ian.M]

The LM723 + an external pass transistor allows you to power it separately, at a lower voltage than the raw supply to the pass transistor(s) it is controlling.  e.g. See my LTspice sim of a Philips PE 1535/00 40V 0.5A PSU, which can be found here: https://www.eevblog.com/forum/projects/how-do-make-a-current-limiting-knob-using-lm723-for-linear-reg-power-supplies/msg1212923/#msg1212923
In that circuit, the IC is powered at 16V with Zeners limiting the control supply, but the raw unreg supply to the pass transistor is well above the IC's 40V maximum.

I agree the LM723 (and any other uA723 equivalent) is getting a little hard to find in stock from major distributors, and the TO5-10 metal can package has always been a bit expensive. Jameco still stock the LM723 in its PDIP-14 package at an affordable price, as do a number of reputable independent suppliers, so its worth checking if any S.A. based independent supplier has it in stock at a reasonable price.

[CharlesGibbsNam]

The person who was critical of your English is also a non-native speaker. It's better to learn from it, rather than get upset.

HI @Zero999,

Sorry, you misunderstand, I was not getting upset, I was merrily stating that both those comments do not address the question I am posing, ergo,
all it does is clutter up the post with unnecessary talk, no?

Address the questions posed rather than critique how I said what I did?

As for your recommendations, thank you. Yes, I am using the second method "charge pump style" .

As I am Unable to get hold of either of the other more suitable components I need to rather now look at pre regulating the input to the 317....

Thank you for your suggestions.
Have a good day. 

[CharlesGibbsNam]

Hi @Ian.M

I agree the LM723 (and any other uA723 equivalent) is getting a little hard to find in stock from major distributors, and the TO5-10 metal can package has always been a bit expensive. Jameco still stock the LM723 in its PDIP-14 package at an affordable price, as do a number of reputable independent suppliers, so its worth checking if any S.A. based independent supplier has it in stock at a reasonable price.

I have already asked around to our LES's they source from SA only due to the aforementioned import problems, the only other option is to wait till mid Feb 2024 when the 317HV is "supposed" to land..... No one has stock of the 723IC this is my current problem.....

As for your schematic, I will go and look at that as an example, thank you kindly.

[Ian.M]

I feel your supply-chain pain!

Staying with the ordinary LM317, as the unregulated voltage can only go slightly over the maximum permitted input to output voltage, an alternative would be to add a LM317 based adjustable current limiter in the feed to the voltage regulating LM317.  The additional voltage drop should keep both safe during unloaded high line condition.

See: https://www.eevblog.com/forum/projects/lm317-_labratory_-power-supply/
and my comments on how to replace the 'unobtanium' JFETs in that circuit with ordinary NPN BJTs here: https://www.eevblog.com/forum/beginners/lm317-variable-power-supply-question/msg687356/#msg687356

[CharlesGibbsNam]

Staying with the ordinary LM317, as the unregulated voltage can only go slightly over the maximum permitted input to output voltage, an alternative would be to add a LM317 based adjustable current limiter in the feed to the voltage regulating LM317.  The additional voltage drop should keep both safe during unloaded high line condition.

Yes, Not easy, Since the dreaded "human malware" that hit SA is having a hard time getting some much needed silicon variety. And, hence where I currently find myself....

With your above statement would this be similar to the attached captured schematic?

Thanks again.

[Ian.M]

Yes, that's the one, but the JFETs are more or less unobtainable unless you can find them N.O.S.  hence my suggestion to replace them with NPN BJT current sinks.

If you've got a LM337 based -1.25V negative rail for the bottom end of the voltage pot, you don't need Q2, D3,D4, and you can simply replace Q1 with the grounded base NPN BJT current sink I showed you earlier.  A 120 ohm emitter resistor should give you an approx. 5mA sink.

[CharlesGibbsNam]

Yes, that's the one, but the JFETs are more or less unobtainable unless you can find them N.O.S.  hence my suggestion to replace them with NPN BJT current sinks.

If you've got a LM337 based -1.25V negative rail for the bottom end of the voltage pot, you don't need Q2, D3,D4, and you can simply replace Q1 with the grounded base NPN BJT current sink I showed you earlier.  A 120 ohm emitter resistor should give you an approx. 5mA sink.

Hi Ian,

Ok, Thank you, this gives me enough to work on for now, Would you suggest me possibly still going down the TL431 route to try and manage the Vin for both the LM317's? Or, just use use the current regulating LM317 ahead of the Voltage regulator?

Thank you.

This helps tremendously.

Regards
Charles

 

[Zero999]

There's no need for a current sink, just set R1 so it draws over 5mA. 

An addional current limiter is a reasonable enough suggestion and antoher LM317 will work, but it requires a large voltage overhead of around 4V. If I were to add current limiting, I would add a sense resistor with an op-amp to pull down the LM317's adjust pin, when the limit is exceeded. It could also easilly be made adjustable, but it's a lot of comlexity, for a beginner.

The transformer voltage being slightly over 40V is no big deal. It will drop, if the LM317 is loaded anyway.

A better solution to avoiding excess power dissipation is to switch to the transformer centre tap, when the LM317 output voltage drops below a certain level. This can easilly be done with the TL431 a few ressistors, a transistor a small relay.

[Ian.M]

The problem with over 40V no load unregulated input is that if the output is suddenly shorted the LM317 gets transiently overvoltaged till the reservoir capacitor has discharged some.  Will it survive this transient?  Probably, at least once, but any time you exceed a datasheet's published abs. max. limits you are taking a gamble. Here be dragons!

I doubt the availability of HV OPAMPs in S.A./Namibia is any better than that of HV regulators.  :'(  I'd bet its difficult to find anything for more than +/-18V supplies (36V total).

I believe you only need a volt or two more safety margin?  If so then yes the additional voltage drops in the 2x LM317 current limited circuit  will keep both safe.
If not, I'd pre-regulate and smooth with a capacitance multiplier, with its base clamped with a 39V 5% Zener (anode to ground) to keep its output under 40V.
See: https://www.eevblog.com/forum/blog/eevblog-1116-the-capacitance-multiplier/

Be careful of the input voltage to the LM337 negative bias rail regulator.  You should probably limit the available current by choice of 'flying' capacitor(s) value in the charge pump, and Zener clamp its input e.g to -12V

[CharlesGibbsNam]

Thanks chaps,

@Zero999
I am using a tapped transformer, ac out is 12.2Vac x2. so the issue is not with the lower tap obviously, what everyone is worried about is the higher tap, and after punching the numbers given by Ian.M that is relatively close to the input max.

@Ian.M over 40+ volt opamps are nowhere to be found here, most when asking for something OPA type they all just go........ "Huh?"

So I have abandoned that ship long ago....

Thank you both gents, very kind of you to offer up your time and knowledge.
I have enough to keep myself busy with for tomorrow before getting back to the grindstone on Sunday.....
First work day of the week for me is Sunday and the last is Friday so I only have Saturdays in which to "tinker".

Thank you both for the links and suggestions.
Regards
Charles

[Zero999]

1) It's unlikely to be 40V, except under extreme circumstances.

No load voltage from a 50VA 12-0-12V transformer.
https://docs.rs-online.com/2c38/A700000008857540.pdf

13.1V, so that's 26.2V, when used as a 24VAC transformer.

The data sheet specifies that with 230V on the primary, yet the mains can be up to 253V, which is 10% higher, so muliply by 1.1.

26.1*1.1 = 28.71V

Giving a peak voltage of 28.71√2 = 40.6V

Minus the voltage drop in the diodes, say 1.2V minimum worst case, at low currents 40.6 - 1.2 = 39.4V

See, still below 40V and it's unlikely to happen often.

2) The voltage on the smoothing capacitor will fall rapidly, under overload conditions.

I say, don't worry about it. It's certainly not worth the complexity for a simply hobby project. Perhaps if it were to be used on an industrial machine, I'd make it more robust, but not for beginner project like this.

[Ian.M]

The 2x LM317 PSU isn't a *GOOD* way to build a bench supply, but for a novice still gathering experience its worth a try if a low max. current (e.g. 500mA) is sufficient.    The additional voltage drop is of course problematic at the high current, high voltage 'corner' of its output characteristic but that + the high dissipation may be acceptable in exchange for the design's simplicity (compared to other >25V out designs with variable current limiting).

An alternative to cope with the slight overvoltage for the simple LM317 circuit would be to add one or more diodes between the bridge rectifier + and the reservoir capacitor for additional voltage drop.

[CharlesGibbsNam]

@Ian.M

This is what I have come up with, circuit attached, the 470 Ohm will give about 15mA across the TL431,
the TL431 has 0.1% resistors so they are not bang on a set resistance I have a multiturn above them to
be able to fine adjust the setpoint out to the LM317.

I have left out the transformer and negative charge pump for simplicity.

Vunreg will go directly to the power stage, there will be constant current I have also left this out for simplicities sake.
I have followed Lindsay Wilsons directions for the pot connections.

There is also an LM334 to ensure a constant current on the output, this will be behind the sens resistor for the current
control part of of the PSU. This is what I have so far, can you double check this for me when you have time?

the power stage will be configured in a Sziklai pair....
I am Also using the soft start from the TI datasheet....

Thanks.
Regards
Charles

[Zero999]

An alternative to cope with the slight overvoltage for the simple LM317 circuit would be to add one or more diodes between the bridge rectifier + and the reservoir capacitor for additional voltage drop.

There won't be any over-voltage The LM317 is a robust IC, which is more than capable of handling spikes. It's not a CMOS op-amp which will ignite, at the hint of mV over the absolute maximum rating for a ms.

Whilst it's good to beware of how stressed a part is, there such thing as going over the top, when it comes to over-engineering, like fretting over the current through a potentiometer wiper & discharge pin, in a 555 PWM circuit, when the non of the data sheets recommend a current limiting resistor. All it does is introduce more complexity, with no benefit.

Have I got my calculations wrong? Please check them.

If you're really paranoid, use a TL431 set to 40V and transistor to shunt any excess voltage. The transformer's impedance will be more than sufficient to limit the current.  A pre-regulator is a bad idea, because there isn't enough voltage headroom.

@Ian.M

This is what I have come up with, circuit attached, the 470 Ohm will give about 15mA across the TL431, the TL431 has 0.1% resistors so they are not bang on a set resistance I have a multiturn above them to be able to fine adjust the setpoint out to the LM317.

I have left out the transformer and negative charge pump for simplicity.

Vunreg will go directly to the power stage, there will be constant current I have also left this out for simplicities sake. I have followed Lindsay Wilsons directions for the pot connections.

There is also an LM334 to ensure a constant current on the output, this will be behind the sens resistor for the current  control part of of the PSU. This is what I have so far, can you double check this for me when you have time?

the power stage will be configured in a Sziklai pair....
I am Also using the soft start from the TI datasheet....

Thanks.
Regards
Charles

You won't get 30V out from that circuit. There isn't enough voltage headroom.

Either use a higher voltage transformer, or get rid of that pre-regulator and pass transistor.

Going back to the start.

You want 0V to 30V..

One problem you might encounter is component tolerances. To get round this, add a trimmer resistor in series with the resistor between the output and adjust pin.

Oh and please turn the grid off and preferably also anti-aliasing, if possible and make the schematic a little larger before attaching it. It makes it much easier for others to read and edit to make suggestions.

Finally please allow the forum to wrap your text, rather than adding line breaks. The browser window can be resizes, if it makes it easier to edit or read.

[CharlesGibbsNam]

Hi Zero999

Well, ok I hear you, all good, but, I want 5Amps , That is why there are pass transistors?

I am being cautious, as already said, stuff is expensive out here, so, if I can save stuff then all the better.  Is it not better to leave in the LM334 to make sure that there is a minimum current draw? For R14, the lowest value pots I have are 500 Ohms, 10K and 50K. I am adding current limiting. This will be done with a 10x gain opamp and then a voltage follower with a multi turn pot a 0.1 Ohm resistor will be the sense resistor value.


I'm using easyeda where do I disable antialiasing?


Thanks for reviewing.
Regards.
Charles

[Ian.M]

A 30V 5A PSU isn't easy, even without current limiting.  A LM317 is nearly no help at all as the recommended min. 3V Input to Output drop across it + the Vbe drop to turn on the boost transistors result in it needing a minimum of 3.7V headroom above the output voltage to the trough of the ripple.  That's at least 16W more  unnecessary dissipation at max. output voltage vs a floating OPAMP + pass transistors  operating down to as low as 0.3V Vce, with maybe an additional 0.2V drop across their emitter resistors (for current sharing, but also used for current sensing for the current limit loop).  The OPAMP is 'floating' in the sense that it is referenced to the positive output terminal and  running from an auxiliary supply of e.g +/-9V relative to the output voltage.  This removes the need for high voltage or RRIO OPAMPs.

When adding an OPAMP to allow a lower value current sense resistor, be careful of the phase margin of the current feedback loop.  Even if you get that right, transitions from CV to CC and back are likely to be messy, with overshoots and possible ringing even with simple resistive loads, mostly due to 'windup' of the inactive control loop + the delay due to the OPAMP's limited slew rate.

[CharlesGibbsNam]

So basically, I'm screwed either way?

The reason I am trying to do this is that in my money I would be looking at 30K for a 5 amp 30 volt linear supply.....

It seems like I will have to look a discrete components then......

Thank you.
Regards
Charles.

[iMo]

You may try this as your first PSU.
Perhaps 0-30V/1.5A max, with big heatsink on the 317, and a smaller heatsink on the 337..

[CharlesGibbsNam]

No good, my requirements are 0-30V @ up to 0-5amps.....

Thank you for posting.

[iMo]

No good, my requirements are 0-30V @ up to 0-5amps.....
Thank you for posting.

I know, indeed, but that would be your second PSU.
With your first PSU start small.
A 0-30V/0-5A PSU is quite difficult to build, do read dozens of threads here about it..

[CharlesGibbsNam]

I have small 2 amp supplies, they are totally inadequate for high power electronics, this is where my whole problem is currently..  I repair "car" amplifiers in my spare time as a hobby. 2 amps does not cut it, switch modes inject way to much noise, I've had amps that run up all the way to 24V as an input voltage hence the 30V limit and 5 amps range. The cc will be so that if there is a short I can try and trace without popping anything else.

[xavier60]

This is an idea I have tested for adding adjustable current limiting to a boosted LM317 regulator.
https://www.eevblog.com/forum/projects/lm317t-stability-under-bypassed-load/msg4980517/#msg4980517

[Zero999]

I have small 2 amp supplies, they are totally inadequate for high power electronics, this is where my whole problem is currently..  I repair "car" amplifiers in my spare time as a hobby. 2 amps does not cut it, switch modes inject way to much noise, I've had amps that run up all the way to 24V as an input voltage hence the 30V limit and 5 amps range. The cc will be so that if there is a short I can try and trace without popping anything else.

Why do you need to go down to 0V?

A big problem is the excessive power dissipation at low output voltages.

I would just get a couple of switched mode power supplies. Adjustable 12V power supplies are quite cheap and two in series would cover quite a wide range.

[CharlesGibbsNam]

Why do you need to go down to 0V?

Because that is my requirement.... 
All "store bought" power supplies can go down to zero. I would like this one to do the same. Besides I need to free up some much needed realestate on my very small bench

A big problem is the excessive power dissipation at low output voltages.

I understand this hence the use of the tapped transformer. 15-0-15.

I would just get a couple of switched mode power supplies. Adjustable 12V power supplies are quite cheap and two in series would cover quite a wide range.

Switch modes create to much noise, I did elude to this a few posts up, and, unfortunately this noise is carried over onto the DUT. I can see it on my scope and I can hear it, this is why I am trying to build a linear supply.  Not only that but I have limited workspace, having one bench supply would mean a simpler setup, not having to rearrange my desk all the time for one or more PSU's that have to take up valuable workspace.

Today I have a different question, I have a built up a distribution PCB, schematic attached, this has a 15-0-15 transformer one part of the board has a dedicated sub voltage delivery and a main voltage delivery, when the first tap is selected everything works as intended, but, when the relay switches tap C7 (25V/1000mfd) goes bang. Only the ground on this schematic is "commend" this can only be because of back feeding voltage from the higher voltage tap, my question is: Why, how and can this be prevented with out having to use a separate transformer?

Would be nice if there is someone that could explain this please?
Thank you again.
Regards
Charles.

[Ian.M]

Draw out the individual diodes of both bridge rectifiers and their reservoir capacitors, with their negatives commoned, the discrete 1N4007 bridge fed from AC0, AC15 and BR1 fed from AC0, AC30.  Then consider what voltage the capacitors charge to on positive and negative half cycles, and which diodes pass current, and you'll see that your idea for tap switching cannot work.

The rule of thumb for multiple bridge rectifiers is: You can interconnect the DC terminals, or the AC terminals, *NEVER* both!

The fix (which only works for a CENTER tapped transformer) is to get rid of the discrete bridge rectifier and permanently connect BR1 to AC0, AC30.  With BR1- as COM, you'll get the full 38.6V DC on BR1+, and a DC voltage of half of it on AC15! 

The same center tapped transformer feeding a bridge is more commonly used with the center tap as COM to produce symmetrical positive and negative rails

You will also need to rearrange your reservoir capacitors  so you have equal ones connected between BR1+, AC15 and AC15,BR1- (COM).  The relay can then switch between the full and half DC voltages without welding its contacts, or exploding capacitors!

[CharlesGibbsNam]

Good morning @Ian.M

Thank you, I do think I understand what you are saying, let me go draw this out and I will revert back later.

Thank you.

[CharlesGibbsNam]

Hi @Ian.M

Is this what you meant? Will this work?

Thank you,
Regards
Charles

[Zero999]

Why do you need to go down to 0V?

Because that is my requirement.... 
All "store bought" power supplies can go down to zero. I would like this one to do the same. Besides I need to free up some much needed realestate on my very small bench

The problem with such a low output voltage and a linear regulator is the excess voltage is converted to heat, so even if you use tap changing, you'll be burning 60W with a 5A load, which will require a large heat sink and a fan.

A big problem is the excessive power dissipation at low output voltages.

I understand this hence the use of the tapped transformer. 15-0-15.

I would just get a couple of switched mode power supplies. Adjustable 12V power supplies are quite cheap and two in series would cover quite a wide range.

Switch modes create to much noise, I did elude to this a few posts up, and, unfortunately this noise is carried over onto the DUT. I can see it on my scope and I can hear it, this is why I am trying to build a linear supply.  Not only that but I have limited workspace, having one bench supply would mean a simpler setup, not having to rearrange my desk all the time for one or more PSU's that have to take up valuable workspace.

Then add filtering. Plenty of commercial audio amplifiers use swiched mode power supplies, whih are not noisy, with appropriate filtering. Regarding space: linear power supplies are huge. You could put two or three swiched mode power supplies connected in series and a switch to select between taps and it would still be smaller, than a big linear power supply, with a huge transformer.

Today I have a different question, I have a built up a distribution PCB, schematic attached, this has a 15-0-15 transformer one part of the board has a dedicated sub voltage delivery and a main voltage delivery, when the first tap is selected everything works as intended, but, when the relay switches tap C7 (25V/1000mfd) goes bang. Only the ground on this schematic is "commend" this can only be because of back feeding voltage from the higher voltage tap, my question is: Why, how and can this be prevented with out having to use a separate transformer?

Would be nice if there is someone that could explain this please?
Thank you again.
Regards
Charles.

Why are there two bridge rectifiers?

It appears to be too complicated.

All you need to do is monitor the output voltage from the regulator and switch to a higher voltage tap, when it exceeds a certain level. Here's an example of a crude circuit I designed awhile ago. It's not optimal (I would make a few changes if I did it again) and will need modifying to get 5A.

[CharlesGibbsNam]

Why are there two bridge rectifiers?

It appears to be too complicated.

All you need to do is monitor the output voltage from the regulator and switch to a higher voltage tap, when it exceeds a certain level. Here's an example of a crude circuit I designed awhile ago. It's not optimal (I would make a few changes if I did it again) and will need modifying to get 5A.

The above circuit has been redrawn in my answer to Ian.M

Good Morning! @Zero999

There are different voltage requirements on his PCB as well as a floating ground potential and a sub negative rail. There is also a need to switch between a higher voltage and a lower one on his output supply. There is an Arduino that needs a regulated 5V on Vin and also attached peripherals that need a steady 5V. He does not want the lower voltage electronics to be switched to the higher voltage tap. Could you please explain how your circuit switches voltage taps? I have not seen such a circuit before and would like to understand how it works please?

Thank you,
Regards
Charles

[Ian.M]

Your redrawn circuit is still wrong.  *ONE* bridge rectifier, fed by a CT transformer, gives DC on the center tap, half way between BR+ and BR-.

[CharlesGibbsNam]

Hi Ian.M

Thank you, but I am unsure then how to correctly connect up the transformer to the rectifier.
I have never done it that way so am unsure about those connections.
I will try once more, thank you.


Thank you for your time.

Regards
Charles

[Ian.M]

That's looking better, but it has one significant problem - when the relay switches from 38V unreg to 18V unreg, it connects the fully charged C2+C3 in parallel with C7, transiently overvoltaging C7 to about 36V.  As I said earlier, reservoir capacitors should be before the relay.

There are also some issues with C9, C10 which probably need resistors in series to make RC snubbers - see https://sound-au.com/articles/relays2.htm#s42

[CharlesGibbsNam]

Hi Ian.M

Thanks. I'll Have to pick it up tomorrow, I do not have anymore time now, I will also read your comments carefully to try and implement the suggested changes.
I fear my brain is not yet capable of grasping all you are trying to convey, I have to go and do the audio at a conference now so will only revert back soonest tomorrow.

Again, thank you for your time.
Regards
Charles.

[CharlesGibbsNam]

Good Morning @Ian.M

Another day, let us hope what I have to offer will now be able to meet with approval? 
After reading the link you gave I settled on higher values for the resistor s and caps at the relay contacts. I do not know if I will be able to get that high a value non polarised cap but "woolC" I have effected the changes needed for the relay circuitry, please tell me if you see any other problems. Thank you kindly.

Regards
Charles.

[Ian.M]

The tap switching (unreg voltage selection) now looks sane and workable.  I'm not commenting on your choices for regulating auxiliary rails/supplies etc, as I do not know what precision is required or what parts availability constraints you have, except to say, I'd probably use more three terminal regulator ICs and fewer discretes!

[Zero999]

Good Morning @Ian.M

Another day, let us hope what I have to offer will now be able to meet with approval? 
After reading the link you gave I settled on higher values for the resistor s and caps at the relay contacts. I do not know if I will be able to get that high a value non polarised cap but "woolC" I have effected the changes needed for the relay circuitry, please tell me if you see any other problems. Thank you kindly.

Regards
Charles.

That looks much better.

I doubt you need R13,14 and C9 and 10. It's not switching an inductive load.

Why not simply use the TL431 for the entire regulator?

Can't you use the LM78L05 or LM317L for those smaller regulators?

[Terry Bites]

It might help to think about the 317 like this.

The Adj terminal sits at 1.25V below Vout.
Its important that any opamp control ciruit takes account of this.

[CharlesGibbsNam]

@Zero999

That looks much better.

Thank you.

I doubt you need R13,14 and C9 and 10. It's not switching an inductive load.

That I do not know, this is not for me but a client that has specific needs, this has been designed to suit his needs.

Why not simply use the TL431 for the entire regulator?

So, as there are different sources and voltages that are not all standard eg (6.5V) and a floating isolated (5V) he asked me to use the same design core through out, according to him, in the field they want repeatability. I also said that there are other ways in which to "float" the LM78XX range of analoge regulators but, they already tried that and have had "an unacceptable amount" of failures.

Can't you use the LM78L05 or LM317L for those smaller regulators?

I could use the LM3XX range but this is what was requested for field application.

Thank you kindly for your time and suggestions.
Regards
Charles.

[CharlesGibbsNam]

Hi @Terry Bites

Thank you for the explanation.

Much appreciated.

Regards
Charles.