“Lab” PSU design help

[CountChocula]

Hello—

Absolute beginner here; I am trying to learn a bit about analog electronics by designing and wiring up a simple CC/CV power supply. My goal is for it to take a regulated 12VDC input and provide a variable 0-12V / 0-1A output.

Using several reference circuits available online as a guide, I ended up designing the schematic below. I breadboarded it, and, somewhat to my surprise, it mostly works—I can set the voltage from 0 to ~11.95V, and the constant current limit works correctly between 0 and 1A as seen by R10.

However, the load regulation is poor; for example, at 8V 1A, I see a drop of around 250mV, which seems excessive. I thought that perhaps there could be a problem with the input, but it stays at a solid 12V, so there is clearly something wrong with my circuit; just to be absolutely sure, I replaced R2 with a 5.1V zener reference, because I thought that perhaps a drop in VIN could skew the voltage set point, but that did nothing. I would be grateful for any suggestions!

While I'm at it, I would also love some suggestions on how this circuit can be improved in general (or whether it even can be improved at all… most of the other circuits I've seen seem much more complex and go well over my head, so I suspect that I'm drastically underestimating the difficulty of the task at hand :-) ).

For example, am I correct in assuming that the Q1/Q2 and Q4/Q5 darlingtons are superfluous? I wanted to be careful about not exceeding U1's 50mA output current (not the least because these TLC2272s cost $8 a pop!), but, if my calculations are correct, there should be no risk of that—even if either Q2 or Q5 were to drop all 12V across R7 or R8, they would be sinking ~60mA, and the worst-case beta of the 2N2222 is 30, which would mean that the circuit should never draw more than 4mA from the op amp).

In any case, thanks in advance for any help. Cheers!

[ledtester]

Do you have a scope? The poor load regulation you're seeing might be an indication that your control loop and thus your output is oscillating. Such oscillations could be in the kilohertz and above range -- something a multimeter wouldn't pick up.

... not the least because these TLC2272s cost $8 a pop! ...

octopart.com is a convenient site for searching several major electronics distributors at one go. It looks like the SMD versions of this chip are significantly cheaper than the through-hole package. And with adapter boards you can get from ebay/aliexpress they can be made just as easy to use.

SMD adapter boards: https://www.aliexpress.com/item/2251832772375001.html

[magic]

General remark: you don't need two power transistors in series, the two opamps could control one power transistor.

General remark: I never built such a PSU, but others did and you should be able to find schematics somewhere on the forum or maybe somebody else could post links.

It seems that you want rail to rail output. This is somewhat unique and it precludes a simple emitter follower output. It's very likely to be unstable without compensation because the output stage has voltage gain. Stability will depend on the value and ESR of the output capacitor - low resistive impedance provided by the capacitor reduces voltage gain of the common emitter output stage.

Note that your current sense resistor reduced maximum output by 1V at full load.

The datasheet says that your opamp can sustain output short circuit to either supply rail indefinitely.

[strawberry]

C5 small as possible (can destroy sensitive devices )
temperature stable resistors in voltage/current sense and reference circuits
TL084 should be good enough down to couple mV levels (rail-rail OP amp wont go down to some  mV anyway)
differential amplifiers to rule out wire voltage drop (0.01ohm * 5A = 0.05V)
zener stability is aprox -1..+10mV/K (depends on zener voltage and circuit design) compensated zener circuit less than ~0.1mV/K and LM399 1ppm/K
TL431 with current source would get more stable (LM336 bit better bandgap reference ~20ppm)

[magic]

Output capacitance may destroy devices connected to the PSU by temporarily exceeding the preset current limit, but it is also essential to stabilizing LDOs, which OP apparently tries to build.

[srb1954]

My goal is for it to take a regulated 12VDC input and provide a variable 0-12V / 0-1A output.
Using several reference circuits available online as a guide, I ended up designing the schematic below. I breadboarded it, and, somewhat to my surprise, it mostly works—I can set the voltage from 0 to ~11.95V, and the constant current limit works correctly between 0 and 1A as seen by R10.

You are going to need more than 12V I/P if you want the O/P to go up to 12V. Using an I/P voltage of 13.8V would be a good choice as there many power supplies available for this voltage. However, be aware you are getting close to the 16V maximum supply rating of the TLC2272.

However, the load regulation is poor; for example, at 8V 1A, I see a drop of around 250mV, which seems excessive. I thought that perhaps there could be a problem with the input, but it stays at a solid 12V, so there is clearly something wrong with my circuit; just to be absolutely sure, I replaced R2 with a 5.1V zener reference, because I thought that perhaps a drop in VIN could skew the voltage set point, but that did nothing. I would be grateful for any suggestions!

Part of the problem is that you are taking the tap-off point for the O/P voltage at the bottom end of the current sense resistor. The circuitry is attempting to maintain the constant O/P voltage across the series connection of the output terminals and the current sense resistor. As the output current increases some of that constant voltage will be lost in the current sense resistor leaving less across the O/P terminals.

The way around this is to move the current sense resistor closer to the input terminals and refer the bottom end of voltage control dividers R2/RV2 and R3/R4 to the negative O/P terminal. You will then definitely need to put a Zener diode across RV2 to ensure a stable reference voltage as you can no longer rely on the I/P voltage as a reference source.

While I'm at it, I would also love some suggestions on how this circuit can be improved in general (or whether it even can be improved at all… most of the other circuits I've seen seem much more complex and go well over my head, so I suspect that I'm drastically underestimating the difficulty of the task at hand :-) ).

For example, am I correct in assuming that the Q1/Q2 and Q4/Q5 darlingtons are superfluous? I wanted to be careful about not exceeding U1's 50mA output current (not the least because these TLC2272s cost $8 a pop!), but, if my calculations are correct, there should be no risk of that—even if either Q2 or Q5 were to drop all 12V across R7 or R8, they would be sinking ~60mA, and the worst-case beta of the 2N2222 is 30, which would mean that the circuit should never draw more than 4mA from the op amp).

The configuration of using a separate series pass transistor for current control and another for voltage control is unusual. More normally only a single series pass device is used and the controls for current and voltage are combined together to control this single device. This results in a more complex control circuit but is usually less expensive to manufacture and has the benefit of a lower drop-out voltage.

The Darlingtons could be deleted and replaced by a single transistor. You could use a lower spec and cheaper op amp in this application as you don't really need the rail-to-rail inputs and outputs provided by the TLC2272.

Lastly I would put a resistor, say 1k , between the base and emitter of each TIP42. This will bypass any leakage current from the transistors and ensure that they can be fully turned off when there is little or no load on the output of the power supply. Otherwise, you might find that the output voltage rises uncontrollably, particularly if it has been running at full load for while and the load is suddenly removed.

[CountChocula]

Thank you all for the kind replies. I will try some of these suggestions out over the next few days and report back. Cheers!

[xavier60]

This is an example of the use of ORing diodes to allow the CV and CC loops to control one pass element, https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg3582664/#msg3582664

[CountChocula]

Hello again! I managed to make things work in the end. As srb1954 suggested, the key was to move the current sense resistor at the input, and then reference everything else to it. It took me a while to understand what was going on, but I think I got there in the end, and I only released the magic smoke once 

I also managed to eliminate one of the pass transistors as suggested, though I'm afraid that the circuit that xavier60 posted goes well above my head, so I don't know if, in the end, I landed on the same solution. I have added that schematic to my “dissection list”; I want to spend the time and redraw it to understand it better.

I was able to swap in an LM358 for voltage control, but I could not, for the love of me, make it work as a differential amplifier for current regulation. For some reason, I would always get a ~0.65V offset on the output (suspiciously close to a diode drop, though I wouldn't understand the significance of that, tbh). I don't know if it is because I wired things incorrectly, or because of some other reason, but, once I substituted in a '2272, everything started working as expected. I will have to look more into this, I guess… right now, it doesn't make any sense (I got these from a reputable source, so I don't think there is anything wrong with them).

In any case, these changes leave me with a well-behaved circuit. At 10V/1A draw (full scale with the 5.1V reference), I get a voltage drop of around 50mV, which represents a 0.5% error—I'll take it! There's a 100mV ripple, which is fine for my purposes and I think is caused primarily by the switched input line.

One final question: am I correct in calculating that the sense resistor dissipates 1W per amp of current drawn by the circuit? Right now, I have 10 1/2W 10-ohm resistors wired in parallel, but they are getting mighty toasty :-)

Thanks again for all the help—this has been a great learning experience. Updated schematic below; I welcome further suggestions and ideas for improvements. Cheers!

[srb1954]

Hello again! I managed to make things work in the end. As srb1954 suggested, the key was to move the current sense resistor at the input, and then reference everything else to it. It took me a while to understand what was going on, but I think I got there in the end, and I only released the magic smoke once 

You have also put the current sense resistor in the negative supply rail as long as the input supply is floating and doesn't share an output common with any other circuitry.

I also managed to eliminate one of the pass transistors as suggested, though I'm afraid that the circuit that xavier60 posted goes well above my head, so I don't know if, in the end, I landed on the same solution. I have added that schematic to my “dissection list”; I want to spend the time and redraw it to understand it better.

I was able to swap in an LM358 for voltage control, but I could not, for the love of me, make it work as a differential amplifier for current regulation. For some reason, I would always get a ~0.65V offset on the output (suspiciously close to a diode drop, though I wouldn't understand the significance of that, tbh). I don't know if it is because I wired things incorrectly, or because of some other reason, but, once I substituted in a '2272, everything started working as expected. I will have to look more into this, I guess… right now, it doesn't make any sense (I got these from a reputable source, so I don't think there is anything wrong with them).

The reason why a 358 doesn't work in your circuit is that, unlike the TLC2272, its allowable input voltage range doesn't go all the way up to the positive supply. If either input exceeds Vcc - 1.5V the 358 is not guaranteed to function correctly. However, the 358 could be used as the current sense amplifier if the sense resistor was in the negative rail as the 358 operates correctly with its inputs all the way down to 0V and even a little below.

 
 In any case, these changes leave me with a well-behaved circuit. At 10V/1A draw (full scale with the 5.1V reference), I get a voltage drop of around 50mV, which represents a 0.5% error—I'll take it! There's a 100mV ripple, which is fine for my purposes and I think is caused primarily by the switched input line.

One final question: am I correct in calculating that the sense resistor dissipates 1W per amp of current drawn by the circuit? Right now, I have 10 1/2W 10-ohm resistors wired in parallel, but they are getting mighty toasty :-)

The 1 sense resistor dissipates 1W at 1A output but the power dissipation varies as the square of the current, P=I²R, so at 2A output current the dissipation would be 4W. You could reduce the sense resistor value  a little to reduce its power dissipation to a more acceptable level. The rest of the current sense circuit could be adjusted to compensate for the lower voltage across the sense resistor.

You can expect the resistors to get toasty; normal film resistors typically have a surface temperature of 125-150C when operated at their rated power. A power wirewound resistor can be up to 250C when run at its rated power!

You might want to modify the 358 circuit for the LED driver a little. With the configuration as shown it might not indicate reliably depending on the input offset voltage of the 358. By biasing the -ve input up to several hundred mV you will ensure that the TLC2272  can drive the 358 +ve input well above and below the comparison point and eliminate the uncertainty due to the input offset voltage.

[magic]

However, the 358 could be used as the current sense amplifier if the sense resistor was in the negative rail as the 358 operates correctly with its inputs all the way down to 0V and even a little below.

Only a little below and then the circuit's behavior becomes very poorly defined, with transistors saturating and base-collector junctions forward-biasing and stuff like that, the output may go to either rail and it's hard to guess which one without testing. I wouldn't even be surprised if the exact outcome varies from manufacturer to manufacturer or with temperature.

Don't use LM358 with inputs below ground
If you want to use it here, connect the chip's power supply to the left of the current sense resistor.

[Kleinstein]

The power dissipation of the shunt is a tricky point and a principle problem:  one wants enough resistance to also get a stable / low noise regulation also well below the full range. So making the resistor much smaller option is limited. A little smaller may work, if the amplifier directly following is a relatively good one. So it may be worth having something better than the LM358 there.
For good accuracy one wants to operate the shunts well below there rated power. So 10 x 1/2 W is not yet overly large and when mounted close together the cooling is often reduced.

As shown the compensation (BW limit) for the current limit is rather slow. Though not much room to saturation it will take quite some time until the current limit would react.
The speed would also depend on the set current, as the resistance at the pot changes. So at least one should add a resistor in series with the wiper and chose a correspondingly smaller capacitor for the integrator.

The current for the zener diode should come from before the shunt of cause.

[CountChocula]

You have also put the current sense resistor in the negative supply rail as long as the input supply is floating and doesn't share an output common with any other circuitry.

Thank you again for your reply. I guess I didn't quite “get it” as much as I thought :-)

Could you please help me understand this point better? I don't think I see how I have accidentally created a negative rail by placing the resistor where it is. Where do you think it should go? Maybe between D2 and R15, where it is unaffected by the current drawn by the zener reference and the bulk discharge resistor? (I meant this PSU to run from a floating supply, so maybe it doesn't matter, but I still would like to learn what you're trying to teach me here!)

The reason why a 358 doesn't work in your circuit is that, unlike the TLC2272, its allowable input voltage range doesn't go all the way up to the positive supply. If either input exceeds Vcc - 1.5V the 358 is not guaranteed to function correctly. However, the 358 could be used as the current sense amplifier if the sense resistor was in the negative rail as the 358 operates correctly with its inputs all the way down to 0V and even a little below.

Right, but then we're back to the same problem with voltage sensing as before, correct? Or did you perhaps mean that I should have kept the current shunt where it was originally and connected the voltage sensing divider between the two output terminals?

 
The 1 sense resistor dissipates 1W at 1A output but the power dissipation varies as the square of the current, P=I²R, so at 2A output current the dissipation would be 4W. You could reduce the sense resistor value  a little to reduce its power dissipation to a more acceptable level. The rest of the current sense circuit could be adjusted to compensate for the lower voltage across the sense resistor.

You can expect the resistors to get toasty; normal film resistors typically have a surface temperature of 125-150C when operated at their rated power. A power wirewound resistor can be up to 250C when run at its rated power!

Ouch—OK, well, at least I know my calculations were right. I think there's also a big airflow problem in the current configuration, as Kleinstein also suggests; the 10Ω resistors are bunched together on the breadboard due to space limitations and probably heat up even more as a result.

You might want to modify the 358 circuit for the LED driver a little. With the configuration as shown it might not indicate reliably depending on the input offset voltage of the 358. By biasing the -ve input up to several hundred mV you will ensure that the TLC2272  can drive the 358 +ve input well above and below the comparison point and eliminate the uncertainty due to the input offset voltage.

Thank you, that's a great suggestion. I was pretty iffy on this part of the circuit, but also too lazy to bias the non-inverting input; I should have just done that from the beginning.

Cheers!

[Kleinstein]

The comment about the shunt position was for the original circuit. In  the first version the problem was not the position of the shunt, but the position of the negative side voltage sense.

There are several possible positions for the shunt resistor, all with some downside:
1) the shunt at the low side, as in the 1st. version of the regulator has the problem of getting voltage sensing right. A possible solution is to have the reference voltage also above the shunt and thus have the reference current also flowing through the shunt. This solution is used in some cases.

2) the shunt directly at the positive supply, before the power transistor. This version has the problem of also including the base current for the power transistor. So the accuracy for the current is slightly limited. Another somewhat tricky part is getting the current signal down from the high side to the low side where the regulation is working relative to ground.

3) the shunt directly at the positive side output. The difficutly here is that the shunts common mode voltage can move around quite a bit and this make the circuit prone to also react to common mode signal and it can get tricky to get a suitable common mode range for the amplifier. This is a somewhat tricky way, but it is done in some commercial supplies. The usual LM723 circuits current limit works this way, but this not made for precision, more like damage avoidance.

[srb1954]

You have also put the current sense resistor in the negative supply rail as long as the input supply is floating and doesn't share an output common with any other circuitry.

Thank you again for your reply. I guess I didn't quite “get it” as much as I thought :-)

Could you please help me understand this point better? I don't think I see how I have accidentally created a negative rail by placing the resistor where it is. Where do you think it should go? Maybe between D2 and R15, where it is unaffected by the current drawn by the zener reference and the bulk discharge resistor? (I meant this PSU to run from a floating supply, so maybe it doesn't matter, but I still would like to learn what you're trying to teach me here!)

The reason why a 358 doesn't work in your circuit is that, unlike the TLC2272, its allowable input voltage range doesn't go all the way up to the positive supply. If either input exceeds Vcc - 1.5V the 358 is not guaranteed to function correctly. However, the 358 could be used as the current sense amplifier if the sense resistor was in the negative rail as the 358 operates correctly with its inputs all the way down to 0V and even a little below.

Right, but then we're back to the same problem with voltage sensing as before, correct? Or did you perhaps mean that I should have kept the current shunt where it was originally and connected the voltage sensing divider between the two output terminals?

Yes, I didn't make it clear that the current sense resistor was to remain where it was in the negative output leg with the bottom of the voltage sensing divider to go to the top end of that resistor. The bottom end of the voltage reference chain would also have to go to the top end of the current sense resistor as well so that you are comparing the divided output voltage only against the reference voltage and not the reference voltage plus the voltage drop across the current sense resistor.

[CountChocula]

Yes, I didn't make it clear that the current sense resistor was to remain where it was in the negative output leg with the bottom of the voltage sensing divider to go to the top end of that resistor. The bottom end of the voltage reference chain would also have to go to the top end of the current sense resistor as well so that you are comparing the divided output voltage only against the reference voltage and not the reference voltage plus the voltage drop across the current sense resistor.

OK, understood. Thanks to you both—I'm away from home for a few days, but I will try implementing your suggestions and report when I'm back. Cheers!

[CountChocula]

Hello again! I had some time to work on this tonight and came up with the updated schematic below. I ended up moving the current-sensing resistor back to the low side of the output, and then referenced the voltage sensing divider across the output as discussed. This meant using a different zener for the current limiter, but it also got rid of one of the op amps, so I suppose that's good.

It seems to work well enough, so maybe I'll try actually putting it on a PCB to see how it behaves once off the breadboard. Curious what everybody thinks… does this approach look reasonable?

Thanks!

[magic]

You probably meant D4 to go to ground in parallel with RV2.
You probably didn't mean to connect Q2/R4 with ISENSE.
You haven't drawn U1 power supplies, be sure to connect it to ground and not ISENSE.

Q4 can't turn on as drawn; if anything, put D2/R5 at its collector, not emitter.

The ICONTROL circuit could be simplified further. For example, put the LED between R6 and Q1 and maybe add 1kΩ from Q1 base to ground to increase LED current. Then Q4 can be removed. Alternatively, if you increase R4 to force Q2 base voltage higher, U1B could steal R2 current away from Q2 base directly trough a diode (its inputs need to be swapped in such case). Then the LED with its resistor could be connected simply between U1B output and 12V.

The 100nF compensation caps are a little heavy-handed, but OTOH regulation may go unstable if they are reduced too far. Currently the time constant is 10kΩ·100nF=1ms and I believe such will be the time to go from no output current to 63% of applied load when you suddenly apply load. Then another millisecond to cross 63% of the remaining 37% and so on. During that time, the output capacitor will discharge maybe a volt or two per amp of applied current. After removal of load current an output overshoot will occur, for the same but opposite reason.

edit
Sorry, it's 5kΩ·100nF and hence 0.5ms. So its twice faster and half the under/overshoot as I described.

[Kleinstein]

As the current redulation is also rather slow, it would be a good idea to have some kind of faster acting secondary current limit. This could be at the power transistor or the transistor before (e.g. limit the base voltage).

The voltage regulation should work OK, though a bit slow.
The cross over between CC and CV mode may still have some tricky points and not so ideal transients.

3.3 V zener diodes are often not very stable. A TL431 ref chip with 2.5 V or a 5.1 or 5.6 V zener would be the more obvious choice.

With the 5.1 V zener current also flowing through the current shunt, ripply from the input supply would be coupled to the measured current. It may be a good idea to add another layer of filtering / stabilization here. So split the resistor in 2 and have another zener (e.g. 10 V) and / or capacitor to ground to first.

[CountChocula]

You probably meant D4 to go to ground in parallel with RV2.
You probably didn't mean to connect Q2/R4 with ISENSE.
You haven't drawn U1 power supplies, be sure to connect it to ground and not ISENSE.

Q4 can't turn on as drawn; if anything, put D2/R5 at its collector, not emitter.

Thank you! That was some really bad capture on my part. U1's power supply is shown as a separate unit at the bottom—that's just the way Kicad does it—but the rest were all mistakes. I'm not yet at the point where I can look at a diagram and intuitively grasp its intention (let alone its correctness), so I tend to make a lot of these.

The ICONTROL circuit could be simplified further. For example, put the LED between R6 and Q1 and maybe add 1kΩ from Q1 base to ground to increase LED current. Then Q4 can be removed. Alternatively, if you increase R4 to force Q2 base voltage higher, U1B could steal R2 current away from Q2 base directly trough a diode (its inputs need to be swapped in such case). Then the LED with its resistor could be connected simply between U1B output and 12V.

Interesting! Is the thinking here that, if I add a 1kΩ resistor from the base of Q1 to ground, I force U1B to supply more current in order to overcome it and effect the regulation, thus also increasing the voltage and providing enough Vf to turn on the LED?

While I was mucking around last night, I noticed that there wasn't enough voltage between Q1 and R6 to turn the LED on, and so I just added another transistor, but this would be a much more elegant solution (and also very counterintuitive to me, so good learning experience).

The 100nF compensation caps are a little heavy-handed, but OTOH regulation may go unstable if they are reduced too far.

As the current redulation is also rather slow, it would be a good idea to have some kind of faster acting secondary current limit. This could be at the power transistor or the transistor before (e.g. limit the base voltage).

Thank you both for this—I picked 100nF because that's all I had on hand :-)

I mostly play around with digital electronics, and so I tend not to have a large selection of passives in my stock. I'll drive by my local store and grab some soon; I did notice the transients, and figured that smaller values would be more appropriate for the compensation.

3.3 V zener diodes are often not very stable. A TL431 ref chip with 2.5 V or a 5.1 or 5.6 V zener would be the more obvious choice.

Yep, agree. Again, that's what I had in my drawers :-)

I'll pick up some TL431s as well—I've been meaning to play around with it as well, as it seems like a very versatile chip.

With the 5.1 V zener current also flowing through the current shunt, ripply from the input supply would be coupled to the measured current. It may be a good idea to add another layer of filtering / stabilization here. So split the resistor in 2 and have another zener (e.g. 10 V) and / or capacitor to ground to first.

Ah, very interesting; I hadn't thought about this at all. I'll give it a try. Thanks again!

[CountChocula]

Interesting! Is the thinking here that, if I add a 1kΩ resistor from the base of Q1 to ground, I force U1B to supply more current in order to overcome it and effect the regulation, thus also increasing the voltage and providing enough Vf to turn on the LED?

Answering myself: Nope, that wasn't the thinking

I moved the LED around and added the 1k resistor to ground as suggested. Works a treat!

I also procured some more caps and settled on 100pF for compensation, and 100uF as output bulk; this seems pretty stable with good transients and reasonable regulation.

I also added buffered outputs for the voltage and current sense circuits, so that I can connect some kind of display, and external VSET/ISET inputs so that the circuit can be controlled externally (I would love to try using an MCU for that). I haven't dealt with improved voltage references yet… the store didn't carry TL431s, unfortunately, so I will have to mail-order them.

New schematic attached; once more, I welcome a sanity check and all suggestions! I'd like to make a proper PCB out of this, even just to see the difference between it and breadboarding.

A couple more questions if anyone can spare a few more thoughts:

• What additional protection circuits does it make sense to implement? Again, this is primarily a learning circuit, and so I'm curious to learn what other potential sources of disaster a professional would look for.
• I'm uncertain about the sizing of R6. My thought was that, in a worst-case scenario, it will have to drop all 12V of VCC to ground, which would dissipate 0.72W (so, even 1/2W is undersized, I suppose), but no matter what load I throw at the circuit, it hardly gets warm, which makes me wonder whether my assumptions are incorrect, I am not thinking about the right scenario, or both.

Thanks!


Edit: I should point out that, so far, I have only tested the PSU with purely resistive loads. I am a little hesitant to try something like a motor for fear that I haven't really thought through the consequences, but maybe I will give it a try. Worst case, it's just a couple cheap op amps

[rstofer]

What happens to the poor TIP42 when it see maximum voltage drop and maximum current.  Say the output is 1V at 1A and you actually have 12VDC input.  The transistor is dropping 11V at 1A or 11 Watts.  That may be ok with adequate heatsinking but without heatsinking, the transistor is only good for 2 W at 25 C ambient.  With huge heatsinking, maybe the case temperature can be held to 25C and then the dissipation limit is 65W.

https://www.onsemi.com/pdf/datasheet/tip42c-d.pdf

It is always the case that max current at lowest voltage creates the most heat for the poor transistor.

[ledtester]

For an exploration of other design ideas, you might be interested in this video:

Building a Lab Power Supply with a Discrete Linear Regulator -- The Post Apocalyptic Inventor
https://youtu.be/_CFIovMkRyg

It's more sophisticated, but the techniques are widely used in other designs.

[CountChocula]

What happens to the poor TIP42 when it see maximum voltage drop and maximum current.

Well, I suppose the answer to that is to either use a tracking pre-regulator to reduce the dissipation by making the supply always a little higher voltage than the desired output, or… use a big heatsink, which is what I am doing. I suppose you could even throw in active cooling if you prefer to keep things smaller, but all the “traditional” lab PSUs that I have ever seen typically sport some rather impressive chunks of metal to keep the pass transistors from melting into slag

It would be neat to add a pre-regulator to this circuit… I wonder how much harder it would be.

Thanks!

[CountChocula]

For an exploration of other design ideas, you might be interested in this video:

Thank you! Conveniently, I have to kill some time before a flight today, so I will check it out. Cheers!