Simple design of a 12V to 5v supply questions

[_c32_]

While building my z80 computer i wanted to make a proper power supply for it.
The circuit will be using about 2A max.
i searched around and my current circuit uses a lt1085 with a bd536.
So far all my experience about building power supplys was using 2 caps on a 7805 so I am a little bit afraid that something is going to go wrong and send 12V to the output.
Should i try to add a dropper resistor on the input?
Have i made any mistakes or is there something to improve?
Also how does the pass transistor not mess up the voltage from the regulator ?
does the regulator compensate by adjusting the current it takes on the input?
Thanks

[Zero999]

The pass transistor is unnecessary, because the LT1083 is already capable of 2A, with no additional components.

The LT1083 is overkill because a low drop-out regulator is not required to 7V. The old LM317 is perfectly adequate for this application and will be more stable too.

I'd recommend the L78S05 with a 2R >5W resistor on the input to reduce the power dissipation of the IC.
https://www.jameco.com/Jameco/Products/ProdDS/889461.pdf

EDIT:
My previous comments assumed the 12V rail is reasonably stable.

Where are you getting the 12V from? If it's a car electrical system, then a LDO isn't such a bad idea, as it will work when the battery voltage drops, when the engine starts. Some over-voltage protection will also be required because high voltages can be generated, due to a load dump.

[_c32_]

i used the pass transistor to spread the heat and make it cool easier
My supply is from a wall plug so stable enough i guess.
since i don't have an l78s05, will a 7805 with a pass transistor be as good?

[SiliconWizard]

Are you strictly against a switching step-down regulator? This would be much more efficient and would generate very little heat compared to a linear approach.

[_c32_]

Yes, i want to get a linear supply on this project because it looks more old school (fitting for the project) and also it's an opportunity to learn more about the subject

[Wolfgang]

Why not try an LM723 with pass transistor ?
Thats probably the most stable variety for little money.

https://electronicprojectsforfun.wordpress.com/power-supplies/

[SiliconWizard]

Yes, i want to get a linear supply on this project because it looks more old school (fitting for the project) and also it's an opportunity to learn more about the subject

Ok. Your call of course.

For the record, actually, quite a few computers of the 70's and 80's, apart from the small home computers, already had switching power supplies. I think the Apple II did, for instance. Even the goold old BBC Micro was initially using an all linear PS but due to severe heating problems, they switched (no pun intended) to a switching PS in a later revision.

And there's just as much to learn (probably even more) building a switching PS (maybe you were thinking I was suggesting to use a ready-made one, which of course would also be an option, but you can design one yourself without much difficulties for this range of current).

[Ian.M]

If the board is valuable, (and IMHO a home-built fully featured Z80 system qualifies), to prevent blowing all the chips on it if your 5V rail looses regulation and goes high, add a crowbar circuit across the output of your 5V regulator.  See T.I's TL431 shunt regulator datasheet, Figure 22. Crowbar Circuit.
Set it up using a variable bench supply so it crowbars at 5.3V, and make sure your regulator circuit has short-circuit protection and isn't over 5.15V.

[Zero999]

i used the pass transistor to spread the heat and make it cool easier

But now it has no over-current protection.

My supply is from a wall plug so stable enough i guess.
since i don't have an l78s05, will a 7805 with a pass transistor be as good?

Yes, but I repeat, it will have no over-current protection, with an LM7805+pass transistor, unless additional components are added.

Since you already have the LT1085, just use it as it is, without the additional pass transistor. The 2R series resistor will reduce the power dissipation. It just needs to be rated to 4W, but I wouldn't recommend going below 5W.

[David Hess]

I might use a couple of LM317s in parallel with operational amplifiers to enforce current sharing.  Multiple devices have the advantage of spreading the heat around making cooling easier.  A big pass transistor as shown is simpler though.  The transistor could also be placed in series to absorb some of the voltage drop instead of amplifying the current.  Or multiple regulators could be used in series.

For safety, add an SCR crowbar circuit to the output and a fuse to the input.  Then in the event that the output rises to a dangerous level, the SCR shorts it and blows the fuse.

[mariush]

Considering you want to keep it linear, my advice would be to simply chain two linear regulators, so that each one will dissipate a part of the voltage drop.

For example, get a LM1085  (3A LDO) and take the voltage down to ~ 8-9v, and the second LM1085 can take it down to 5v  - this way at peak 2A of current, the regulators will dissipate ~ 6-8w of heat.

you could also put a couple of diodes in front of the two regulators to lower the voltage by around 1.5v but I don't think it would make much difference.

With LM1085 you'd have to be careful because the tab is connected to Vout so for example if you want to screw two LM1085 to a copper plate / heatsink or something like that, you'd have to insulate at least one of them.

[NiHaoMike]

Just get a car charger for a smartphone. They're cheap and handle all the details for you.

[HB9EVI]

I used for my Z80 devboard a LM2576: it's readily available, not too noisy and doesn't require who knows what heatsink for 12V -> 5V

[_c32_]

long reply, thanks a lot everybody for helping ! I learned quite some stuff today

Why not try an LM723 with pass transistor ?

Thanks! That looks quite interesting , i will most likely get one and play around with it .
I also like that it comes in a to-100 package 

maybe you were thinking I was suggesting to use a ready-made one

that's exactly what i was thinking. I am now considering a switching supply too but i have a sound chip, video and composite encoder on 5V. Could that introduce too much noise? i guess not if i implement it correctly.  idk

to prevent blowing all the chips on it if your 5V rail looses regulation and goes high, add a crowbar circuit across the output of your 5V regulator. 

Thanks a lot! That's exactly what i was looking for about overvoltage
i have some stuff that i really don't want to damage

i used the pass transistor to spread the heat and make it cool easier

But now it has no over-current protection.

i hadn't though of that, the lm723 seems to sense current after the pass transistor which makes it a bit more appealing

Since you already have the LT1085, just use it as it is, without the additional pass transistor. The 2R series resistor will reduce the power dissipation. It just needs to be rated to 4W, but I wouldn't recommend going below 5W.

that sounds like the best option for now. I'll use that and then maybe check out the lm723 or the  LM2576 mentioned by HB9EVI

Use a couple of LM317s in parallel with operational amplifiers to enforce current sharing.

that sounds quite interesting but it sounds complicated and i have no idea how i would go about implementing it

The transistor could also be placed in series to absorb some of the voltage drop instead of amplifying the current

so using a constant current on base-emitter and then the collector would go to the input of the regulators ?
wouldn't a resistor do a better job?

My advice would be to simply chain two linear regulators, so that each one will dissipate a part of the voltage drop.

That sounds great, maybe i can use my lt1085 to drop to something like 7 volts and then feed that to a 78S05 or whatever else.
wouldn't a dropper resistor be as good though?
is it to ensure that the voltage doesn't drop too much?

With LM1085 you'd have to be careful because the tab is connected to Vout

I use separate heatsinks for the regulators and transistors and they are also coated with something black that makes them non conductive but thanks, that could have ruined my day

Just get a car charger for a smartphone.

that aint no fun
I want to build the supply myself 

I used for my Z80 devboard a LM2576: it's readily available, not too noisy and doesn't require who knows what heatsink for 12V -> 5V

but it doesn't look nearly as cool as the to-100 package of the lm723
seriously though, that seems like a good option but it supports up to 1.5 amps. Maybe i could cut down some stuff and use that

lots of good options to choose from

[Wolfgang]

Hi, for a computer supply its maybe good enough. I'll try this and I have ordered some of those. Thanks !

[_c32_]

Hi, for a computer supply its maybe good enough. I'll try this and I have ordered some of those. Thanks !

if you are trying it for me, thanks but I have already tried it and it works.
if you want to use this yourself, depending on the current consumption of the circuit it might be an overkill and also the current protection of the regulator is useless due to the pass transistor

[Wolfgang]

Dont worry, I will try it out just for my own curiosity and to learn something.
3A output is OK for me, the LM2576 can handle this without a pass transistor.

[m3vuv]

could you not use an old pc desktop psu and tap off of the 5v rail,ready made cheap solution!.

[David Hess]

Use a couple of LM317s in parallel with operational amplifiers to enforce current sharing.

that sounds quite interesting but it sounds complicated and i have no idea how i would go about implementing it

It requires an additional operational amplifier so it is probably not worth the extra complication.  It would be more suitable for a higher current power supply where one regulator is not enough.

The transistor could also be placed in series to absorb some of the voltage drop instead of amplifying the current

so using a constant current on base-emitter and then the collector would go to the input of the regulators ?
wouldn't a resistor do a better job?

This idea is similar to using two regulators in series except that the first regulator is a power transistor which divides the needed voltage drop by half.  In practice it might be as simple as a power transistor, resistor, and zener diode.

[Zero999]

I don't see the need for two regulators in series/parallel, the LM723 or a pass transistor, when 2R plus the LT1085 will do. It's simple and effective. The series resistor will dissipate some of the heat, so it isn't all being dissipated in the linear regulator. The LT1085 also has safe operating area protection, current limiting and thermal protection, which solutions using pass transistors lack.

The only other improvement is a switched mode power supply, which will reduce the power dissipation to a minimum. The LM2576 is a bit old and isn't the most efficient IC around, but is sufficient for the application.

[HB9EVI]

If it can cost a little more, there is also the LT1171; has even some current reserve and is available in non-smd case

[_c32_]

could you not use an old pc desktop psu and tap off of the 5v rail,ready made cheap solution!.

i want to make my own supply and also a pc one is big and noisy

If it can cost a little more, there is also the LT1171; has even some current reserve and is available in non-smd case

i'll check it out if i make a switching one but i think i can get away with a linear and i already have the parts

My current circuit uses just an lt1085 with a dropping resistor and a tl431 with a BTA140 triac for crowbar. I'm now wondering if i should keep the pass transistor with a switch that will enable it and disable the dropper to allow for higher current but sacrifice some features, if i use it for another project. Also i could put a temperature sensor for the cpu to read with the extra space since it's a square board.
This is a test fit and some stuff are not the correct values since i don't have them yet
Thanks a lot everyone !

[Zero999]

To switch the pass transistor on and off, connect as per the previous schematic, but with a switch in series with the base connection.

R5 and R6 are excessively low for a potential divider for the TL431. Their values could be multiplied by 10 or even 100, to save a bit of power.