SMD power supply. Some questions.

[MASantos]

So I came up with this crazy idea of designing a power supply for my PINT. At the moment I am starting to learn how to use Eagle and I believe that this will be a nice first project that will give me enough experience to do an amp afterwards.

My self imposed requirements are the following:

1. MUST fit a Hammond 1455C801 enclosure.
2. Output:12-14V and at least 150mA of output current.
3. Should perform as good as, or better than a TREAD.

Transformer will be an Amveco 62010, which fits the case perfectly. It has a diameter of 42mm which leaves a PCB area of 38x45mm(LxW). This board area must include the IEC and either a DC jack or strain relief. Qualtec 771W-X2/01 looks like a good choice but I would prefer something less bulky (I know this will be hard to find).

For the filter cap I have two choices depending on board space:

• Panasonic FC 680uF 35v, dimensions:16x15mm(DxL)
• Panasonic FC 820uF 35v, dimensions:18x15mm(DxL)

I am trying to fit 2x the 680uf ones.

So here are the questions:

Diode bridge: What should the minimum specs be? Acording to the specs of the transformer, I shouldn't need more than 50v and 0.5A right? I have found these ones:

MB2S DF04M HD02

Would the SOIC be a good option or should I go the one with higher values to be on the safe side?
Regarding the bipass caps, Tangent says:"The exact value to use here is not terribly critical. 100 pF is a reasonable value. If you choose a different value, I'd say smaller is better." How low can I go and are there any benefits in doing this?

Regulator: I will probably use a LM317( Other options?). Is there any reason why I shouldn't use the SOT-223 package? They are rated at 1A and this circuit won't draw more 0.2A so there shouldn't be any problems right?

Resistors, protections diodes,LED:
SMD's all over.
Resistors are 1206, Diodes are S1A, Led is a through hole 3mm. (I'm still not sure if the Front panel will have a led, but it is always good to have one, even if only to discharge the filter caps)

Bipass Cap: any reason not to go SMD with this one?

Output cap: In the LM317 datasheet refers to a tantalum or electrolitic for this position but doesn't say which is better. I haven't found anything in tangent's pages.

All opinions are welcomed!

Manuel

 

[Garbz]

I can make some suggestions. Firstly 680uf is not a lot to work with. The input ripple to the regulator will still be quite high especially as current draw goes up. See if you can't find any higher values in the 25V range.

Try to build everything to tight tollerances. So rather then use an LM317 which with a 1.25V minimum dropout will need to operate at a lower voltage thus decipating more power pick an LDO regulator. This will need a smaller copper heatsink area as the voltage can be closer to the input voltage.

I have built powersupplies with and without caps on the diode bridge. The ringing was not really that audible but it may be easier to eliminate ringing via 2 beads down the line rather then 4 caps at the diode.

For output capacitor pick a standard Electrolytic. Low-ESR caps cause more resonance with the regulator. Admittedly this limits its effectiveness but the regulator will remain stable and resonance will be reduced.

 

[MASantos]

I have been following Tangent's Steps and tread pages for some of the parts selection. He suggests that the filter cap should be at least 2x the output voltage. 25v is not within those limits, is ti still ok to use it? If yes, there are some 1000uf with the same diemensions as the 680 ones.

I also considered a LDO regulator but I have no idea where to start searching. It LT1026 that tangent recomends a good starting point?

I think that the bridge caps won't be a problem because I will place them on the bottom layer and they don't take much space. I could also place the two ferrite beads don't the line.

What about the bridge itself, any comment on thos I choose?

Quote:

Originally Posted by Garbz I can make some suggestions. Firstly 680uf is not a lot to work with. The input ripple to the regulator will still be quite high especially as current draw goes up. See if you can't find any higher values in the 25V range. Try to build everything to tight tollerances. So rather then use an LM317 which with a 1.25V minimum dropout will need to operate at a lower voltage thus decipating more power pick an LDO regulator. This will need a smaller copper heatsink area as the voltage can be closer to the input voltage. I have built powersupplies with and without caps on the diode bridge. The ringing was not really that audible but it may be easier to eliminate ringing via 2 beads down the line rather then 4 caps at the diode. For output capacitor pick a standard Electrolytic. Low-ESR caps cause more resonance with the regulator. Admittedly this limits its effectiveness but the regulator will remain stable and resonance will be reduced.

 

[Garbz]

For the capacitor there are 2 things to remember. Pick a capacitor for your worst case voltage plus 10%, and for the desired ripple. So 15V transformer is ~20V after rectification. Add 10% for line variance and a further 10% for safety still gets you to 25V.

Ripple voltage is Iload/2fC for full wave rectification. Ideally you want to get the ripple voltage low so the regulator has more to work with. However seeing there are such low current draws ripple for this circuit will already be near 5mV

I have no experience with LDO regulators myself so I can not recomend a part. Mind you this is not strictly needed. You are still left with 16V after rectification of a 12V torroid so you should still be able to produce 12-14V. An LDO would however allow you to get a higher voltage.

Any of those bridges should be sufficient for your needs.

 

[NeilR]

Quote:

Originally Posted by Garbz For the capacitor there are 2 things to remember. Pick a capacitor for your worst case voltage plus 10%, and for the desired ripple. So 15V transformer is ~20V after rectification. Add 10% for line variance and a further 10% for safety still gets you to 25V.

Hi Garbz... correct me if I am wrong but I thought the cap needed to handle the peak AC voltage (not the average), which would be (1.4 X 20.4) or 28V when the supply is unloaded.

 

[tobias_svensk]

Perhaps you could do it something like P-A's QRV04 board that I have build.

In the upperhalf you'll se SMD diodebridge (DF) and SMD LM317/337
http://www.ettnet.se/~tobias/diy/qrv04/QRV04-3.jpg

/Tobias

 

[Garbz]

Quote:

Originally Posted by NeilR Hi Garbz... correct me if I am wrong but I thought the cap needed to handle the peak AC voltage (not the average), which would be (1.4 X 20.4) or 28V when the supply is unloaded.

Good question. Someone else may shed light on this. I am not entirely certain but you may be right.

 

[DaKi][er]

Quote:

Originally Posted by NeilR Hi Garbz... correct me if I am wrong but I thought the cap needed to handle the peak AC voltage (not the average), which would be (1.4 X 20.4) or 28V when the supply is unloaded.

the average (rms) is 15 and the peak is 20.4 in this case, you've multiplyed by sqrt(2) twice

 

[MASantos]

Quote:

Originally Posted by DaKi][er the average (rms) is 15 and the peak is 20.4 in this case, you've multiplyed by sqrt(2) twice

Don't forget that the full wave rectifier bridge also increases voltage x1.41 so :

14Vx1.41(peak voltage)x1.41(full wave bridge)=27.8V in the worst case scenario.

I must go with 35V for safety.

I am still undecided about which regulator to use. If I go LM317 I can use a single sided board and etch it myself. But if I go double sided I think I could fit a Jung Super regulator variation such as this one by Pear Anders:

http://home.swipnet.se/~w-50719/hifi/jsr02/index.html

I am more inclined to the LM317 tough since I don't have lots of experience in PCB layout.

 

[DaKi][er]

Quote:

Originally Posted by MASantos Don't forget that the full wave rectifier bridge also increases voltage x1.41 so : 14Vx1.41(peak voltage)x1.41(full wave bridge)=27.8V in the worst case scenario. I must go with 35V for safety.

Your getting a little confused here, after AC is full wave rectified, the peak voltage will be Vac * 1.41 less the diode drops and that is it, end of story.

The only other thing is that an unloaded transformer will have a higher voltage compared to its listed voltage which is not (easily) able to be calculated and is very dependent on the size of the transformer and its construction, measuring it is the only way to be sure and as mains varies quite a lot that will take that into consideration too, but in the end you are getting an AC voltage value, * 1.41 - diodes is what you'll get

You are wise to pick 35v caps, it’s the safest way, but with careful selection of parts and proper testing you could use 25v

 

[MASantos]

Quote:

Originally Posted by DaKi][er Your getting a little confused here, after AC is full wave rectified, the peak voltage will be Vac * 1.41 less the diode drops and that is it, end of story. The only other thing is that an unloaded transformer will have a higher voltage compared to its listed voltage which is not (easily) able to be calculated and is very dependent on the size of the transformer and its construction, measuring it is the only way to be sure and as mains varies quite a lot that will take that into consideration too, but in the end you are getting an AC voltage value, * 1.41 - diodes is what you'll get You are wise to pick 35v caps, it’s the safest way, but with careful selection of parts and proper testing you could use 25v

The transformer's unloaded voltage is, according to the manufacturer 20.4.

Even if I go with 35V caps I can fit 1640uF ( 2*820UF 35V Pana Fc caps)

 

[Garbz]

Quote:

Originally Posted by MASantos I am still undecided about which regulator to use. If I go LM317 I can use a single sided board and etch it myself. But if I go double sided I think I could fit a Jung Super regulator variation such as this one by Pear Anders: http://home.swipnet.se/~w-50719/hifi/jsr02/index.html I am more inclined to the LM317 tough since I don't have lots of experience in PCB layout.

One of the most important factors for PSU design still remains the implimentation. The Jung Regulators are fantastic when used properly but I question their usefulness when running externally and providing a single voltage. The design itself has provisions for Load sensing and ground sensing, and layout itself is very important to achieve the noise specs. No doubt this is a great regulator but it would be wasted more then a few cm from it's load and not having 2 separate GND and 2 separate Load lines.

That said the LM317 could also benefit from GND sensing. You can impliment this quite simply on the PCB by running separate regulator and true grounds out to a plug on the back. If you have a device which impliments separate ground sense lines then run the additional cable, if not connect the grounds together.

Just an idea to improve things further

 

[MASantos]

I see your point.

The objective of this power supply is to provide the cleanest DC power I can fit inside the smallest hammond 1455 enclosure. It will power a PINT(and charge it's batteries) and that is the only purpose. I want the pint to sit on top of the PS with a very small power cable. Probably only a few cm length.
I have about 5.5 square inches of board available and that's accounting both side of the board. Single layer I have about 2.72 square inches available. I must fit the IEC receptacle in that space. I will probably do a preliminar board design tonight.


Quote:

Originally Posted by Garbz One of the most important factors for PSU design still remains the implimentation. The Jung Regulators are fantastic when used properly but I question their usefulness when running externally and providing a single voltage. The design itself has provisions for Load sensing and ground sensing, and layout itself is very important to achieve the noise specs. No doubt this is a great regulator but it would be wasted more then a few cm from it's load and not having 2 separate GND and 2 separate Load lines. That said the LM317 could also benefit from GND sensing. You can impliment this quite simply on the PCB by running separate regulator and true grounds out to a plug on the back. If you have a device which impliments separate ground sense lines then run the additional cable, if not connect the grounds together. Just an idea to improve things further

 

[MASantos]

As I suspected board design is taking longer than expected because EAGLE is really confusing some times!!

I will change the transformer to the 62012 which is 12v 133mA secondaries. I will wire the secondaries in parallel which gives 12V 266mA. This will allow to use 25v caps with safety margin which increases capacitance. The regulator will need to drop less voltage which decreases heat dissipation.

Any reason why i shouln'd goo ahead with this change?

I am also looking for a smaller AC receptacle( smaller than the one I posted in the 1st post). if anyone know of one please let me know.

Garbs you spoke about placing two ferrite beads "down the line". I supose these should go after the output capacitor of the LM317? What specs should I aim to?

 

[NeilR]

Quote:

Originally Posted by DaKi][er the average (rms) is 15 and the peak is 20.4 in this case, you've multiplyed by sqrt(2) twice

I did not multiply by sqrt(2) twice; I multiplied the UNloaded peak voltage (not the rated voltage at full load) ) for the 62010 (20.4V according to the spec sheet) by 1.4. The cap needs to handle the full unloaded voltage, right? otherwise it might blow up without a load, or under a very light load.