Designing dual linear power supply

[Alex Eisenhut]

jahimmelhergottsakra

(a bavarian curse ;-)

LOL I heard similar stuff growing up! From my Dad when he was tooling around his power tool repairs and projects...

[Alex Eisenhut]

I must reiterate that I do not wish to use a negative regulator in this design.

And that's perfectly fine. It's not different than two positive lab supplies hooked together to make +/-. You've just put them in one box and freaked some people out  .

The problem is that your schematic shows a line under your diodes (D1 and D4) , those dots are showing a connection to that green line under the diode symbols. You may not have intended that, so you don't see it, but other people will point it out but you will now be talking at cross purposes. If you ever make a netlist from this schematic to go to a PCB, the short circuit WILL end up on the PCB unless you notice it. Then you've got problems.

The diodes are there to protect the regulators because usually when Vout>Vin+.7V, they fry. The diodes prevent that, they are cheap.

I would also reduce that 1500uF output cap. Maybe check the transformer specs at max current, and the filter cap.

The only thing is I'd get rid of is the connection between the two supplies, and make that connection externally, for future flexibility. Banana jacks and cables are cheap. And if you are going to put this thing in a box, maybe some input protection on the AC side like a fuse and a common mode filter, with a connection for earth to chassis.

You've basically built my student project from college, a dual linear supply with a Hammond transfo, and a sheet metal box. The thing is robust and I still use it today.

Of course I am old and lazy and if I wanted a new supply, I would buy an already made supply, if only for the transformer and input is already taken care of.

https://www.ebay.ca/itm/International-Power-IHBAA-40W-100-240V-Input-5VDC-Output-Power-Supply/274523214819?epid=1933579616&hash=item3feadbd3e3:g:qP0AAOSwewdffNlL

These are cheap and great. I have one of those, and if you were in Montreal we could arrange a pick up since I rarely use it. 20$?

You could then use the chassis to hack away to your heart's content, or just use it as-is... 20$ doesn't even get you a decent transformer IMO.

[networkingdude]

I now see this line through the diodes and have removed it, thanks for that!

I will be reducing output capacitor to100uF, increasing input filter caps to 10000uF, adding 2.2K 1/2 watt bleed resistor next to filter capacitor, removing 1uF output capacitor as the 100uF is good by itself. Not pictured is the IEC input which will be filtered for noise and all 1uf, 10uf and 150uf caps will all be aluminum polymer for lower ESR.

Also the transformer will be upgraded to 18v 50va pcb mount.

I will also separate the connection to make it flexible.

Thank you everyone for the suggestions!

[bdunham7]

I will be ... increasing input filter caps to 10000uF,

Whoa!  Where did that come from?  I have large, very powerful audio amplifiers (for speakers, not headphones) that have filter caps that are smaller than that!

[networkingdude]

I will be ... increasing input filter caps to 10000uF,

Whoa!  Where did that come from?  I have large, very powerful audio amplifiers (for speakers, not headphones) that have filter caps that are smaller than that!

Rule of thumb is 2000uF per amp, I have 3a so 6000uF works, so I figure 10,000uF even better! I'm probably way overkill here. I figure extra capacitance removes even more ripple?

[Labrat101]

Doctor Frankenstein I presume .
  A lighting Bolt will Have a better effect .
     A Tad more Power ..

 PS .. I have some 100,000 UF caps they weigh about a KG ..

    Interested .

[networkingdude]

Doctor Frankenstein I presume .
  A lighting Bolt will Have a better effect .
     A Tad more Power ..

Yes very much so.

[networkingdude]

Is there a better rule to follow here? Should I be sizing the caps at 2000uf per amp?

[Labrat101]

 A big heat sink and a concrete Bunker .
 O you might think of doubling up those LM,s to take the Extra PoWer

 I have some 100,000 UF caps they weigh about a KG ..
 But it wont help .. but it will be fun to watch

[networkingdude]

A big heat sink and a concrete Bunker .
 O you might think of doubling up those LM,s to take the Extra PoWer

 I have some 100,000 UF caps they weigh about a KG ..
 But it wont help .. but it will be fun to watch

Might take some time to charge up eh.

[Labrat101]

 Nop . I can charge them up within 10ns .
 But it still wont solve Your Problem 
 I really Suggest you take up A new hobby  or Read some good books on Basic Electronics .

     

[Vovk_Z]

Yes. 1A AC, which is about 0.7 A DC.

Ignorant question: is this a rule of thumb, or is it a calculation?

A fast calc ( +- 10...20%).

[networkingdude]

Nop . I can charge them up within 10ns .
 But it still wont solve Your Problem 
 I really Suggest you take up A new hobby  or Read some good books on Basic Electronics .

     

Sure, can you recommend some good ones?

[bdunham7]

Rule of thumb is 2000uF per amp, I have 3a so 6000uF works, so I figure 10,000uF even better! I'm probably way overkill here. I figure extra capacitance removes even more ripple?

Where are you getting 3 amperes?  You have 25VA per winding, a power factor of probably 0.6 under load, I think you'll struggle to go much over one amp.

Yes, extra capacitance removes more ripple, but the VR will be very good about blocking 120Hz input ripple so I wouldn't worry about that.  I think the rule of thumb for capacitors ahead of VRs is more like 1000uF/A, but if you want to double it and use 2200uF, all the better.  10mF would be a lot.  Go ahead if you like, I don't think it will burn anything up on turn-on.  Use a robust bridge rectifier.

[networkingdude]

Rule of thumb is 2000uF per amp, I have 3a so 6000uF works, so I figure 10,000uF even better! I'm probably way overkill here. I figure extra capacitance removes even more ripple?

Where are you getting 3 amperes?  You have 25VA per winding, a power factor of probably 0.6 under load, I think you'll struggle to go much over one amp.

Yes, extra capacitance removes more ripple, but the VR will be very good about blocking 120Hz input ripple so I wouldn't worry about that.  I think the rule of thumb for capacitors ahead of VRs is more like 1000uF/A, but if you want to double it and use 2200uF, all the better.  10mF would be a lot.  Go ahead if you like, I don't think it will burn anything up on turn-on.  Use a robust bridge rectifier.

I was going by the size of the regulator and I had mentioned earlier that I would be making the transformer larger. 50va should get me up to 1.5 amp ish, so then would 3000uf be good enough?

[Vovk_Z]

and all 1uf, 10uf and 150uf caps will all be aluminum polymer for lower ESR.

There is no need to use polymer caps here. Their impedance is too low for IC regulator output. And they won't really help at the IC input.

[Vovk_Z]

so then would 3000uf be good enough?

You may use any value you like, or any value you can fit into the place. Either 3000 uF or 30 000 uF.
3000 uF is only a minimum.

[networkingdude]

and all 1uf, 10uf and 150uf caps will all be aluminum polymer for lower ESR.

There is no need to use polymer caps here. Their impedance is too low for IC regulator output. And they won't really help at the IC input.

I was misunderstanding the datasheet I guess, it says to use between 1uf to 1000uf tantalum or electrolytic.

[bdunham7]

I was going by the size of the regulator and I had mentioned earlier that I would be making the transformer larger. 50va should get me up to 1.5 amp ish, so then would 3000uf be good enough?

Sure.  3300uF will be easier to find.

[Vovk_Z]

it says to use between 1uf to 1000uf tantalum or electrolytic.

Days of tantalum caps are gone. You may use them if you have them in your hands yet and don't know where to use them, or you have some temperature requirements (tants are more stable with temperature). Otherwise in all similar circuits just use usual aluminum caps. That's it.

[networkingdude]

Revised schematic

[Wolfgang]

Revised schematic

Some comments:
- LM350 is capable of delivering 3A. At 18V winding, we have 3*18*1.4 = 75VA per winding.
- I dont see a current limiting mechanism except the limiting in the LM350 (plus overtemp shutdown).
- For 3A the cap is way too small.

I guess you had 1A in mind. Then:
- Total transformer rating would be 2*18*1*1.4 = 50VA. 1.4 is loading factor for full bridge rectification.
- Filter cap is OK for 18V secondaries.
- You should have a current limiter.

Regards
  Wolfgang

[networkingdude]

Revised schematic

Some comments:
- LM350 is capable of delivering 3A. At 18V winding, we have 3*18*1.4 = 75VA per winding.
- I dont see a current limiting mechanism except the limiting in the LM350 (plus overtemp shutdown).
- For 3A the cap is way too small.

I guess you had 1A in mind. Then:
- Total transformer rating would be 2*18*1*1.4 = 50VA. 1.4 is loading factor for full bridge rectification.
- Filter cap is OK for 18V secondaries.
- You should have a current limiter.

Regards
  Wolfgang

The capacitor was much larger but I was advised it was too large given the 1.4 amp limit of the transformer. I've not put much thought into current limiting as of yet.
The transformer is the largest size I could find that was pcb mounted on digikey. 2 or 3 amps would be nice but I don't need that much power. lm317 could have been used but this gives good margin and much lower noise.

[bdunham7]

- You should have a current limiter.

Why?  The 3A 'capability' of the regulator is meaningless and there's no more issue oversizing the regulator than there is oversizing the rectifier or the wires.  If there is a concern about shorts or overloads, a fuse would be a sufficient answer. 

[Wolfgang]

Revised schematic

Some comments:
- LM350 is capable of delivering 3A. At 18V winding, we have 3*18*1.4 = 75VA per winding.
- I dont see a current limiting mechanism except the limiting in the LM350 (plus overtemp shutdown).
- For 3A the cap is way too small.

I guess you had 1A in mind. Then:
- Total transformer rating would be 2*18*1*1.4 = 50VA. 1.4 is loading factor for full bridge rectification.
- Filter cap is OK for 18V secondaries.
- You should have a current limiter.

Regards
  Wolfgang

The capacitor was much larger but I was advised it was too large given the 1.4 amp limit of the transformer. I've not put much thought into current limiting as of yet.
The transformer is the largest size I could find that was pcb mounted on digikey. 2 or 3 amps would be nice but I don't need that much power. lm317 could have been used but this gives good margin and much lower noise.

For 1A the component values are OK. If you dont want to think about current limiting use a fuse.´ or a "weaker" regulator chip.
If you need better regulation, less noise, ... you could go for a bit more complex design using op amps or a 723.