[SOLVED] Advice on power supply schematic

Hi all,

I need advice on the following schematic:

https://electronics-diy.com/high-current-power-supply-lm338.php


It's my third power supply, previous projects were '' small '' projects, with no more than 5A.

I only want to know:

- schematic with no mistakes?
- if adding 4 more LM338K with SK88 heatsinks from Fisher, mandatory to place them in a box with 2 cooling fans?

4 x SK 88 Length: 250 mm Rth : 0,7

**broken link removed**


Pcb is already drawn with Kicad, high amperage being withstood by cooper plates and 10 AWG - 6 mm 2 wires -


Advices about this project would be great.


Thanks a lot.


jm

Make output caps low ESR type.

Not very efficient but simple.

Hi all,

Thanks SunnySkyguy for your reply.

Is there a more efficient way, without using Switch Mode Power Supply schematics?


Thanks a lot.

jm

here's a few ideas

- drop the 0.3R to 0.1R if you mount all active devices on a CPU cooler with low speed fan.
- use a series pass transistor as PWM controlled charger to maintain a lower Voltage drop on the regulators or use a stock SMPS to maintain a 2 V drop on the linear regulator.
- use a lower LDO regulator.

Lambda used another method on 100V 100A linear PSU's and smaller... was use an SCR bridge on the primary to pre-regulate a voltage sufficient to regulated secondary at minimal voltage drop. Ramp Phase control sampled using a reference and regulating voltage drop.

Hi all,

Thanks SunnySkyguy for your fast reply and complete advice.

About LDO, like LT1084V from STMicroelectronic, by example?

Now after crude maths for cooling, finding right heatsink is really difficult!!!!


Power: 20 A x 24 V = 480W



Rth Case to Junction: 3° C/W

T junction limited to 110 ° ( My choice )

T ambiant: 40 ° ( Inside a box with cooling fans )

8 x LD1084D 480 / 8 = 60 Watt per item


First trial with Latex - must be great opensource :grin:...... - :

\[ Rth _ {jc} = {3^0}C/W \]


\[ T _ {case-maxi} = {110^0}C\]


\[ T _ {ambiant-maxi} = {40^0}C\]


\[ P _ {dissipated} = ({{T _ {case-maxi}-T _ {ambiant}}) / {Rth} \]


Some basic maths:

With 8 x LD1084D so 60 Watt by regulator .....

\[ Rth={(110-40)}/{60} \] \[ \Rightarrow \] \[ Rth _{total} = {1,2^0}C/W \]


With these results, must ask Pacific Ocean to perform the job ???????????????? :sad::sad::sad:

Any idea ?:idea:


Regards.

jm

It is not uncommon to have very large heatsinks for linear type high current power supplies, often one or more fan(s) is used to reduce h/sink size.

It is easy to do power conditions for worst case dissipation from (Vin - Vout) x current

e.g. Vin typ 30V say, Vout min (controlled) = 12V say, I = 50A, dissipation in series pass devices = 18 x 50 = 900W or most of the power being supplied by your 1.5kVA transformer.

If you have 26V ave in and 24VDC out at 50A then the diss will be 100W. Quite a difference.

Transistors need to be seriously derated to be inside their SOA curves for this type of application, emitter res a must! 30 -40W per TO3 can on a decent heatsink.

Hi Orson,

Thanks a lot for your reply, now I'm going to google to find TO3 can Low Dropout Regulator.

Regards.

jm

Brute force linear power supplies are inefficient. Why use them? You can use LC filters to reduce the noise on SMPS so why ignore SMPS?

Hi SunnySkyguy,

To be honest after googling and googling alot, about pro and cons of smps versus linear ones, i'm going to build a smps.

Up to now my main concern is about winding transformer, but found many links about this.

There's a great site about smps,mainly directed to hi-fi guys:

**broken link removed**


Others:

https://www.smps.us/


With a homemade built transfo, how to be sure 3KV isolation is really effective by exemple????


Thanks a lot for your reply, as usual it makes sense.


Jm

mylar tape has good hipot performance

http://www.pulseelectronics.com/download/3836/switch_mode_transformer_platform_overview_2013

Hi SunnySkyguy,

As usual thanks for your reply.

Regards.

jm

I heard some guyz to say that, they've destroyed their MCUs supplied from smps. I guess linear is more jitter free than smps. Well, a personal problem. I need to power up two peltier coolers (each rated 12v, 5A, 60W) from a transformer rated at output (12v-0v-12v, 6A). what shold I do? paralleling the coolers on (12v/10A) or connecting them in series with (24v,5A) ?? any idea?

Hi rocket scientist,

Read this link:

https://www.eleccircuit.com/regulator-12v-10a-by-ic-7232n3055/

To short 12 x 1,44 = 17,28 V minus 2 V from bridge rectifier so 15 V before 7812 voltage regulator.

Should be sufficient enough for 7812 ,for low current I think. View attachment 7812-datasheet.pdf

But as you see in above link, there is other stuff for more than 1,5 A....



Others links to know why there is a dot on transformer: - read carefully -

- https://en.wikipedia.org/wiki/Transformer
- https://www.electronics-tutorials.ws/transformer/transformer-construction.html
- https://www.smps.us/magnetics.html



For short:

- paralleling 2 transfos but 12 V AC from trany for 12 V DC after rectifying stage, safety margin ?????
- paralleling 2 coolers but see line above?????
- apply above schematic


I am a beginner too, so gurus in this field, will help you; and if i made mistakes they help me too :wink:




Hope this helps.

jm

Hi Jim, thanks a lot for the circuits. nice stuff. Well I just wanted to make a very very simple DC power supply, so I think maybe I should gor for the series connection. Also the secondary has current limitation according to the wire gauge.
well It's nice to have a 10 Amp power supply.

RS..

Dont forget old PC ATX power supplies have plenty of 12V capacity and are easy to make work. Some old ones require a 10% preload on the 5V, to make the 12V work, but not the newer ones in last 5 yrs.

**broken link removed**

Hi SunnySkyguy,

Thanks a lot for your numerous replies and advices, i'm now on modifying one schematic found on web site.

jm

Hi all,

As planed, i 've just posted a new one '' Advices needed for modifying smps power supply '' , Power Electronics forum.

I don't know, if the best practice is to continue this post, or to begin a new one?

Thanks.

jm