How to get 5V from a 12V SMPS without using Buck and not heat up LDO or 7805

[nctnico]

I'd use a multi-tap transformer. A couple of zeners at the output can take car of no-load condition. Since the SMPS basically is a buck converter the Vin/Vout ratio is determined by the pulse width and transformer ratio.

[jbb]

Like nctnico and chris_leyson, I suggest a dual output; say 6V and 12V out.  The 6V out can go through an LDO to make 5V with little loss.

Following is the signal I got across SB1100 Schottky diode at the rectifier stage.  Can anyone please tell me if it has any problems. This waveform is at no load.

OK, here's what I see on that waveform:

• The trigger might give a less messy response if set to 20V
• The pulse width might be a bit uneven - I suggest you check with trigger @ 20V
• The fast ringing when the diode turns off around t = -2us suggests oscillation between diode capacitance and tx leakage inductance.  It's around 10MHz and probably more prone to sneaking into other stuff. Maybe you could damp it better than this by adjusting R26. Also, I expect that it will get worse as the load increases.
• The slow ringing off to the right (circa 1MHz) is, IMHO, between transformer magnetising inductance and assorted capacitances (TNY286, D1/D2, D6, C20 etc.).  The R26/C20 snubbers won't be very effective against it.  However, this may change a lot at higher loads (e.g. you might go into continuous conduction mode), so try it at operating current.

In general, try it at a few load points to see what changes.

On the schematic:

• I suggest leaving room for an extra RC snubber across D1/D2; it might be helpful
• The feedback loop formed using R18, D4 and OK2 won't be very well controlled (but could be fine in many applications)
• The output filter L4, (C18+whatever output load) could resonate at an inconvenient frequency.  You should check out what happens with the real system capacitance and assorted load steps. For extra fun, it could interact with the main control loop too.

HOWEVER, the hard switched nature of the flyback converter will always generate some high frequency energy (could be from 10MHz - 200MHz+) due to the fast dV/dt and dI/dt of the switching circuits.  With careful design (and good PCB layout), you can try to attenuate as much of this as possible before it leaks out of the switching supply and into the analogue stuff.

But a better way might be to use a soft-switching converter and generate less noise in the first place.  These converters use resonance or auxiliary switches to help with the switching, and inherently produce less noise than a more violent hard switching converter.  They normally wouldn't be used for a low power supply, but might help with your particular requirements.  Possible contenders:

• Active clamp flyback converter. (TI has just started sampling the UCC28780 which could be interesting.)
• Active clamp forward converter. (LT has several controllers. Probably too many parts for this application...)
• LLC converter. This one is nice because it uses the transformer leakage inductance; you can use a split winding bobbin to get really good primary-secondary isolation. Power Innovations makes the LCS700 which is a bit overpowered but might do the trick.

As a final note: doing your own off-line supply does give you a lot more safety worries than just buying a qualified one.  Be careful, and prepare to spend extra time reviewing everything to check creepage, clearance, component ratings etc.  And be very thorough with your transformer design and winding...

Edit: fixed format boo-boos

[chris_leyson]

One example of a tapped secondary http://katalog.we-online.de/ctm/datasheet/750871630.pdf
or separate windings katalog.we-online.de/ctm/datasheet/750316585.pdf

[getfast_kiran]

I'd use a multi-tap transformer. A couple of zeners at the output can take car of no-load condition. Since the SMPS basically is a buck converter the Vin/Vout ratio is determined by the pulse width and transformer ratio.

Hi, Yes I was thinking of this but I  am trying to stay away from it since my loads on the two taps are so different so regulation will take me a good amount of time to test again and settle on. If everthing else fail I will definitely go as you said.

[getfast_kiran]

Like nctnico and chris_leyson, I suggest a dual output; say 6V and 12V out.  The 6V out can go through an LDO to make 5V with little loss.

Following is the signal I got across SB1100 Schottky diode at the rectifier stage.  Can anyone please tell me if it has any problems. This waveform is at no load.

OK, here's what I see on that waveform:

• The trigger might give a less messy response if set to 20V
• The pulse width might be a bit uneven - I suggest you check with trigger @ 20V
• The fast ringing when the diode turns off around t = -2us suggests oscillation between diode capacitance and tx leakage inductance.  It's around 10MHz and probably more prone to sneaking into other stuff. Maybe you could damp it better than this by adjusting R26. Also, I expect that it will get worse as the load increases.
• The slow ringing off to the right (circa 1MHz) is, IMHO, between transformer magnetising inductance and assorted capacitances (TNY286, D1/D2, D6, C20 etc.).  The R26/C20 snubbers won't be very effective against it.  However, this may change a lot at higher loads (e.g. you might go into continuous conduction mode), so try it at operating current.

I will do a more comprehensive analysis on it based on your notes.

In general, try it at a few load points to see what changes.

On the schematic:

• I suggest leaving room for an extra RC snubber across D1/D2; it might be helpful
• The feedback loop formed using R18, D4 and OK2 won't be very well controlled (but could be fine in many applications)
• The output filter L4, (C18+whatever output load) could resonate at an inconvenient frequency.  You should check out what happens with the real system capacitance and assorted load steps. For extra fun, it could interact with the main control loop too.

Yes surely will try it .Can you please elaborate on how to test this second and third point and if you could tell me how to analyze changes it can create on main loop it would be
great because I think this what is happening with it since the clamp diode is dying on long time operation due to excess heating.

HOWEVER, the hard switched nature of the flyback converter will always generate some high frequency energy (could be from 10MHz - 200MHz+) due to the fast dV/dt and dI/dt of the switching circuits.  With careful design (and good PCB layout), you can try to attenuate as much of this as possible before it leaks out of the switching supply and into the analogue stuff.

But a better way might be to use a soft-switching converter and generate less noise in the first place.  These converters use resonance or auxiliary switches to help with the switching, and inherently produce less noise than a more violent hard switching converter.  They normally wouldn't be used for a low power supply, but might help with your particular requirements.  Possible contenders:

• Active clamp flyback converter. (TI has just started sampling the UCC28780 which could be interesting.)
• Active clamp forward converter. (LT has several controllers. Probably too many parts for this application...)
• LLC converter. This one is nice because it uses the transformer leakage inductance; you can use a split winding bobbin to get really good primary-secondary isolation. Power Innovations makes the LCS700 which is a bit overpowered but might do the trick.

As a final note: doing your own off-line supply does give you a lot more safety worries than just buying a qualified one.  Be careful, and prepare to spend extra time reviewing everything to check creepage, clearance, component ratings etc.  And be very thorough with your transformer design and winding...

Edit: fixed format boo-boos

Thanks for such a comprehensive explanation for what was going on. I will have to sit on it and do as you have told.  Also for suggesting the alternative IC's. Do you think those are better than the IC's from Power integrations in stability. Also duly noted the Final note.

[getfast_kiran]

One example of a tapped secondary http://katalog.we-online.de/ctm/datasheet/750871630.pdf
or separate windings katalog.we-online.de/ctm/datasheet/750316585.pdf

Thanks for giving the reference.I will look into multi tap once this solution seems unsolvable cause it will take more time for me to build the circuit and test it again.

[getfast_kiran]

Solved one of the problem.

The heating in the clamp circuit was reduced by reducing the number of turns in the primary and seconday of the transformer keeping the turns ratio the same. I think my circuit since it was designed for universal input supply. The duty ratio of Tiny switch at 230 V was at the other end of the spectrum which means it was off for most of the time. So add this to problems in leakage inductance...(btw which I
do not know yet how to link them perfectly) resulted in heating my clamp diode.

Now reducing the turns in both the winding might have reduced the spikes coming at low duty ratio( hence helping in minimizing the effect of leakage inductance - according to friend) hence causing less voltage spikes and relaxation for the clamp diode.

Now my only problem is the heat being dissipated in the LDO (which also has come down drastically) as per calculation given to me here before. Now I will try to design the heatsink as per the specs and get back to you with the result of the cooling effectiveness of the same.

[james_s]

There seems to be a common misconception around LDOs. They are linear regulators, the *only* thing special about the LDO type is that the dropout voltage is lower than a standard linear regulator. That is you need less overhead, the input voltage to the regulator can be lower for a given output voltage, reducing the voltage that the regulator has to drop and that is what makes the regulator run cooler. Under equal circumstances a LDO will get exactly as hot as a standard regulator. A resistor, diode(s), or any other linear dropping device will dissipate the same power and produce the same heat. Dropping a voltage with linear methods always converts all that excess voltage into heat. If you want to avoid making this heat you either need to use a lower voltage secondary on the transformer or use a switching regulator to drop the voltage. There is no magic here, those are your options.

[getfast_kiran]

There seems to be a common misconception around LDOs. They are linear regulators, the *only* thing special about the LDO type is that the dropout voltage is lower than a standard linear regulator. That is you need less overhead, the input voltage to the regulator can be lower for a given output voltage, reducing the voltage that the regulator has to drop and that is what makes the regulator run cooler. Under equal circumstances a LDO will get exactly as hot as a standard regulator. A resistor, diode(s), or any other linear dropping device will dissipate the same power and produce the same heat. Dropping a voltage with linear methods always converts all that excess voltage into heat. If you want to avoid making this heat you either need to use a lower voltage secondary on the transformer or use a switching regulator to drop the voltage. There is no magic here, those are your options.

Yes james. Thanks for the info.I was not able to drop the voltage using switching voltage regulator due to issues with my following analog IC and former cannot be used cause of my circuit needs. I understand now. LDO was heating up unusually for a 3.14W power dissipation before the turns where reduced. Now its heating up reasonably. As you said I will try to dissipate this heat off with standard technique. Let me verify it for few weeks and I will mark this thread as solved.