SMPS - TNY 284 - 4W design - Regulation distorted by Loading.

[mariush]

See this for clearance and creepage information : http://blog.optimumdesign.com/clearance-and-creepage-rules-for-pcb-assembly

Also this: http://www.smpspowersupply.com/ipc2221pcbclearance.html

In your board, the easiest way to increase the clearance is to not put traces under the optocoupler and the capacitor, anything you put between the terminals of those components reduces the clearance distance.
The distance between the pins of the optocoupler (if it's a DIP stile package) should be enough for this application (in my opinion). If you go further and create a slot between the pins, it's even better.

Something else i forgot to mention .. it's probably not needed since your product is low current (only 5v 0.8a) but for higher currents some designers also like to drill a series of small holes (like the holes for a regular chip pins or slightly larger) on the center line where the transformer would be placed (at equal distance between the terminals of the transformer).  Those holes would allow for a bit of air movement around the transformer potentially cooling it a bit. I don't think it's needed in your design.

Oh and some answers to your reply to my post...
For bridge rectifiers ... I'm not saying it's better to use bridge rectifiers instead of individual diodes, i'm just saying it looks nicer and takes less space on pcb and separates AC voltages from DC voltages much easier. Unless you go for the cheapest possible BOM (bill of materials) it's worth thinking about it.
Most cheap chinese designs use separate diodes because manual work is super cheap, they buy 100k pieces 1n4006 and some worker will be paid a dollar a day to shove diodes on the pcb. These days bridge rectifiers are not more expensive than 4 separate diodes if you buy them in some small volume (like let's say 50-100 bridge rectifiers, you buy them at less than $0.2 each)
If you want to build a lot of boards, replacing those diodes with a bridge rectifier may allow you to reduce the height of the pcb which in turn may allow you to put more boards on a single pcb panel and therefore you'd get more boards for your money.
Why 470uF for the second capacitor? 470uF is just a standard value and very common.. it's used in lots of products so chances are lots of stores will have it in stock in large quantities so at cheap prices. Also, it can probably be found at the same height and diameter as smaller capacitance capacitors like 330uF or 270uF and the price differences would be pennies if you buy a bunch of them. If you go over 470uF, you'll probably get taller capacitors and for 5v 0.8A, it won't make much of a difference.
Once you get the other problems sorted, it may be worth investigating if you could use 2 x 470uF or 2x560uF capacitors there, i mean use the same capacitor instead of the 1000uF capacitor, simply as a way to reduce the number of different components (and get better price by ordering a bigger quantity of same component) .. only if you plan on making more than a few of these boards.

[T3sl4co1l]

Zener diodes under 6V do not have a sharp (avalanche) characteristic.  They are true zener diodes: a very soft knee.  The rated voltage is only true under the test current; expect wildly different results for other conditions.

Use a TL(V)431 instead.

Tim

[wraper]

BTW it is strange to use all through hole and just 3 SMD parts. It makes no sense to mix two technologies in such a way. I would understand if there was a lot of both, but just for 3 parts  .

I was trying to do it with the parts already available. Is that a bad practice in electronics scenario. Are there any bad aspects to it. would love your inputs on the same.

You introduce 2 different manufacturing processes for no reason. It should be avoided, as increases production time and cost. Even if you see SMD components in production PSU, then usually they are put on the bottom of the PCB and glued with epoxy. So everything goes through wave soldering only once and reflow process is avoided.

[getfast_kiran]

Second Part :
Could you tell me what is the clearance expected between the primary and secondary tracks. And I will edit the pcb without the trace as you told. But could you tell me what could be possible effect of not having that pad there.

As much clearance as optocpoupler pins allow. If the device is not grounded (double insulated), I would aim for about 8mm, not less than 6mm, and then put a slot under the part (optocoupler, Y cap) which requires me to make lower distance.
No pad there won't make any effect. But if there is a pad, you may have an electric arc over it and small bang if, say, humidity is high.

Okay.

[getfast_kiran]

This was the layout suggestion the software gave. I know I made a it so huge compared to it. I was trying to make it more safer. I must have but instead made it worse. Could you suggest how could I make it safer. And also will it be the reason the voltage sinked so much.

Don't forget that it is a layout on a single layer. But you managed to make much dirtier layout even on two layers while it should be completely opposite.

Now you guys have reviewed it. It definitely feels dirtier. Will try to improve it as per suggestions.

[getfast_kiran]

BTW it is strange to use all through hole and just 3 SMD parts. It makes no sense to mix two technologies in such a way. I would understand if there was a lot of both, but just for 3 parts  .

I was trying to do it with the parts already available. Is that a bad practice in electronics scenario. Are there any bad aspects to it. would love your inputs on the same.

You introduce 2 different manufacturing processes for no reason. It should be avoided, as increases production time and cost. Even if you see SMD components in production PSU, then usually they are put on the bottom of the PCB and glued with epoxy. So everything goes through wave soldering only once and reflow process is avoided.

That was some good information. Now I understood why SMD parts are alone kept underneath. Did not think about it before. Thanks.

[getfast_kiran]

Does anyone think the use of following (SB140)
http://www.vishay.com/docs/88715/sb120.pdf

instead of the IN5819 Diode
http://www.diodes.com/_files/datasheets/ds23001.pdf

could result in the regulation problem and drop in the output voltage.

[getfast_kiran]

See this for clearance and creepage information : http://blog.optimumdesign.com/clearance-and-creepage-rules-for-pcb-assembly

Also this: http://www.smpspowersupply.com/ipc2221pcbclearance.html

In your board, the easiest way to increase the clearance is to not put traces under the optocoupler and the , anything you put between the terminals of those components reduces the clearance distance.
The distance between the pins of the optocoupler (if it's a DIP stile package) should be enough for this application (in my opinion). If you go further and create a slot between the pins, it's even better.

Something else i forgot to mention .. it's probably not needed since your product is low current (only 5v 0.8a) but for higher currents some ers also like to drill a series of small holes (like the holes for a regular chip pins or slightly larger) on the center line where the transformer would be placed (at equal distance between the terminals of the transformer).  Those holes would allow for a bit of air movement around the transformer potentially cooling it a bit. I don't think it's needed in your .

Oh and some answers to your reply to my post...
For bridge rectifiers ... I'm not saying it's better to use bridge rectifiers instead of individual diodes, i'm just saying it looks nicer and takes less space on pcb and separates AC voltages from DC voltages much easier. Unless you go for the cheapest possible BOM (bill of materials) it's worth thinking about it.
Most cheap chinese s use separate diodes because manual work is super cheap, they buy 100k pieces 1n4006 and some worker will be paid a dollar a day to shove diodes on the pcb. These days bridge rectifiers are not more expensive than 4 separate diodes if you buy them in some small volume (like let's say 50-100 bridge rectifiers, you buy them at less than $0.2 each)
If you want to build a lot of boards, replacing those diodes with a bridge rectifier may allow you to reduce the height of the pcb which in turn may allow you to put more boards on a single pcb panel and therefore you'd get more boards for your money.
Why 470uF for the second ? 470uF is just a standard value and very common.. it's used in lots of products so chances are lots of stores will have it in stock in large quantities so at cheap prices. Also, it can probably be found at the same height and diameter as smaller capacitance s like 330uF or 270uF and the price differences would be pennies if you buy a bunch of them. If you go over 470uF, you'll probably get taller s and for 5v 0.8A, it won't make much of a difference.
Once you get the other problems sorted, it may be worth investigating if you could use 2 x 470uF or 2x560uF s there, i mean use the same  instead of the 1000uF , simply as a way to reduce the number of different components (and get better price by ordering a bigger quantity of same component) .. only if you plan on making more than a few of these boards.

Thanks for the valuable information. I will try to take care of these aspects in the future..when I do design. Thanks a lot.

[T3sl4co1l]

Does anyone think the use of following (SB140)
http://www.vishay.com/docs/88715/sb120.pdf

instead of the IN5819 Diode
http://www.diodes.com/_files/datasheets/ds23001.pdf

could result in the regulation problem and drop in the output voltage.

Basically the same thing.

I already gave the solution,

Zener diodes under 6V do not have a sharp (avalanche) characteristic.  They are true zener diodes: a very soft knee.  The rated voltage is only true under the test current; expect wildly different results for other conditions.

Use a TL(V)431 instead.

Tim

[getfast_kiran]

Zener diodes under 6V do not have a sharp (avalanche) characteristic.  They are true zener diodes: a very soft knee.  The rated voltage is only true under the test current; expect wildly different results for other conditions.

Use a TL(V)431 instead.

Tim

This is the only option left to try. I will try to check that while I revise the board. Currently when I measure the voltage across the zener at full load was 3.7V. So do you think
it will result in the regulation problem ??

[getfast_kiran]

Does anyone think the use of following (SB140)
http://www.vishay.com/docs/88715/sb120.pdf

instead of the IN5819 Diode
http://www.diodes.com/_files/datasheets/ds23001.pdf

could result in the regulation problem and drop in the output voltage.

Basically the same thing.

I already gave the solution,

Zener diodes under 6V do not have a sharp (avalanche) characteristic.  They are true zener diodes: a very soft knee.  The rated voltage is only true under the test current; expect wildly different results for other conditions.

Use a TL(V)431 instead.

Tim

Will surely try it and get back to you. It is very hard to get approval for making testing board. Hence Why I was waiting. I will try it now and get back to you. Thanks again.

[T3sl4co1l]

Test board?  Easy enough to air-wire on existing pads!

Tim

[tatus1969]

So I went with it since I did not know much about the safety part of it.

Could you suggest how to make it more safer if you could or point me to any suggestion in this forum else for 230v 50Hz circuit.

Is it a commercial product that you are developing? If yes, then please consider that you are developing a product that deals with lethal voltages and must be designed to protect the consumer in a professional, repeatable, certifiable and tested way. The regulation standards are the first place to go. They explain how to handle things like creepage, clearance, different levels of insulation (functional / basic / reinforced), which one to use when, and many more things.

Nevertheless, your questions and the level of your circuit design suggests that your design experience is (not yet) at expert level. Receiving feedback from a forum like this should not be your only source of learning. Developing mains powered devices requires a whole lot of experience, and I would not dare going this way myself without good backing from a good source of expertise, for example a senior expert within my company.

Just trying a shot in the dark: do you know the safety requirements for capacitor C4 in your design?

To the feedback circuit: a) it is most certainly the source of your bad load regulation. b) don't make the mistake to just replace the zener with a TL431, use the entire reference design from the TNY284 datasheet.

[getfast_kiran]

Test board?  Easy enough to air-wire on existing pads!

Tim

I was not having through hole component hence why. But since you told I am trying to do the air-wire thing to max possible on the existing pads. Thanks for sharing your views of an experienced person.

[getfast_kiran]

So I went with it since I did not know much about the safety part of it.

Could you suggest how to make it more safer if you could or point me to any suggestion in this  else for 230v 50Hz .

Part 1:
Is it a commercial product that you are developing? If yes, then please consider that you are developing a product that deals with lethal voltages and must be ed to protect the consumer in a professional, repeatable, certifiable and tested way. The regulation standards are the first place to go. They explain how to handle things like creepage, clearance, different levels of insulation (functional / basic / reinforced), which one to use when, and many more things.

Nevertheless, your questions and the level of your   suggests that your  experience is (not yet) at expert level. Receiving feedback from a  like this should not be your only source of learning. Developing mains powered devices requires a whole lot of experience, and I would not dare going this way myself without good backing from a good source of expertise, for example a senior expert within my company.

Part 2:
Just trying a shot in the dark: do you know the safety requirements for  C4 in your ?

Part 3:
To the feedback : a) it is most certainly the source of your bad load regulation. b) don't make the mistake to just replace the zener with a TL431, use the entire reference  from the TNY284 datasheet.

Part 1:
I was handed over this work when the experienced guy left suddenly. And he was the one who usually does this high voltage stuff. I presently hence do not have someone to ask to..about it.
This is for a prototype which after few iterations will be added to a board ( In long future). I am totally new to this high voltage circuit and as you told I told my boss I need someone with experience to assist me which I  was not allocated till now.
As you told apart from this forum am trying to read on it more. If you could suggest some good materials. That would be great.

Part 2:
I am not sure.  I am only learning. Is it that the capacitor has two different voltages on its both legs. So the clearance here is the distance between its legs. I have made a mistake by taking the trace go beyond the legs.
Is that the safety concern... Am I correct. If I am wrong Could you please tell me the correct safety requirements.

Part 3:
As you told and also as told by all others in the forum. I am trying to redesign a circuit as per the datasheet I was shared here.

 Thanks for you time. Please do advice me on the queries above. Thanks.
 

[mariush]

C4 must be Y-class capacitor (they're usually blue).

See this : http://www.kemet.com/Lists/FileStore/900%20Series%20Product%20Training%20Module.pdf  and this http://electronics.stackexchange.com/questions/216959/what-does-the-y-capacitor-in-a-smps-do

Y capacitors are constructed in such a way that they're supposed to never fail in short circuit. If they fail, they're open, safe for the end user.

Other ceramic or electrolytic capacitors could fail short and then a major point for using transformer (isolation) would be lost.

[tatus1969]

A bad situation that you are in. I would go for 1:1 copying an existing reference design in your situation. For further reading, I suggest to go for the safety regulation standards. Google "psu safety regulations" gave me this article as a good starting point: http://www.de.cui.com/catalog/resource/power-supply-safety-standards-agencies-and-marks.pdf Purchase the relevant safety standards, and read them carefully.

Re C4: mariush gave the answer that I was hoping for  Remember, your design effort can affect the safety of other people. I hope, this example shows it clearly. If you design C4 to be something else than an Y capacitor, people can be put into lethal danger. You *must* know what you are doing, otherwise... I don't know... maybe it would be better to talk to your boss and explain him your situation?

[mariush]

I'd also be concerned about the transformer itself.
These ac-dc switcher chips require custom transformers and usually the datasheets are pretty low on the details about the number of windings, the wire thickness, the ferrite core type and so on. Do you guys know what you're doing when it comes to manufacturing the transformers, or are you ordering them from somewhere?

I've seen the datasheets for TNY** chips, I've also seen the datasheets for ST's Viper series and they're also not very detailed - Viper22 looks like a nice IC especially when paired with ST's TSM101 chip (voltage reference + voltage/current monitor, does the feedback for the chip through an optocoupler)... but again, it's still low on the technical details. Would be hard to use by a beginner.

One of the most detailed datasheets I know of is for the TEA172* chip from NXP ... : TEA1721 (max 5w) , TEA1723 (max 11w)

Lots of detail, lots of formulas to determine the components, it's a cheap IC which also uses a primary auxiliary winding to determine the output on the secondary side so it doesn't need a voltage reference and an optical isolator... you're saving money by not using those, but you may be spending a bit more money on the transformer..  the transformer would have to be custom made as it has the auxiliary winding between the primary and the secondary winding, and the secondary winding MUST have triple insulated wire to be absolutely safe for users.

As there's no feedback from secondary side, they may be less accurate with the output voltage but if the cost of manufacturing the transformer is the same as with other chips, you could just set the output voltage a bit higher than desired and spend 0.1-0.2$ (less than the voltage reference and optocoupler costs) on a tiny 5v linear regulator that would drop the 5.x volts to exactly 5v..

They also have an application note which explains their demo board and has lots of useful information for someone like you. Here's the application note : http://www.nxp.com/documents/user_manual/UM10521.pdf
And another one : AN11060  TEA172X 5 W to 11 W Power Supply/USB charger : http://www.farnell.com/datasheets/1760795.pdf?_ga=1.189448363.1081003429.1454881204

later edit: and they also have an online calculator sort of thing for these chips : http://nxp.transim.com/fb/design.aspx#General_InputSpecifications

Unless you guys have some investment in using those TNY chips, it may be worth looking at alternatives before committing to using them.

[tatus1969]

@mariush, you are right, there is a ton of things that need to be taken into account. i have a 500 page book on this topic. safety rules, winding construction, skin effect, leakage inductance, topologies, core types, different types of power losses, output filter design, input capacitor design, emi filtering, and countless other topics.

normally you go to one of the transformer manufacturing specialists and let them wind your custom part. but for a number of specific chips they have stock parts that are matched to the corresponding reference design.

[getfast_kiran]

Hi,

Thanks for all your suggestion.

I have redrawn the board based as per your suggestions. Can you all please look at it and help me improve it.

[getfast_kiran]

@mariush, you are right, there is a ton of things that need to be taken into account. i have a 500 page book on this topic. safety rules, winding construction, skin effect, leakage inductance, topologies, core types, different types of power losses, output filter , input  , emi filtering, and countless other topics.

normally you go to one of the transformer manufacturing specialists and let them wind your custom part. but for a number of specific chips they have stock parts that are matched to the corresponding reference .

Could you please tell the book that you were referencing.
And currently my new board is showing output of 4.85 - 5.00 but it will glitch to 8-9V for a second and comeback to 4.85 -5V regularly with one transformer.

And it will not glitch but generate a voltage of 12V at the output with another transformer of same rating.

do you think it is due to problem of transformer winding

[getfast_kiran]

C4 must be Y-class capacitor (they're usually blue).

See this : http://www.kemet.com/Lists/FileStore/900%20Series%20Product%20Training%20Module.pdf  and this http://electronics.stackexchange.com/questions/216959/what-does-the-y-capacitor-in-a-smps-do

Y capacitors are constructed in such a way that they're supposed to never fail in short circuit. If they fail, they're open, safe for the end user.

Other ceramic or electrolytic capacitors could fail short and then a major point for using transformer (isolation) would be lost.

That was a new knowledge to me. Plus I heard they are good at reducing common mode noise by coming across the transformer even though there is a contact between
the primary and secondary...Do you know if its that effective...

[getfast_kiran]

A bad situation that you are in. I would go for 1:1 copying an existing reference design in your situation. For further reading, I suggest to go for the safety regulation standards. Google "psu safety regulations" gave me this article as a good starting point: http://www.de.cui.com/catalog/resource/power-supply-safety-standards-agencies-and-marks.pdf Purchase the relevant safety standards, and read them carefully.

Re C4: mariush gave the answer that I was hoping for  Remember, your design effort can affect the safety of other people. I hope, this example shows it clearly. If you design C4 to be something else than an Y capacitor, people can be put into lethal danger. You *must* know what you are doing, otherwise... I don't know... maybe it would be better to talk to your boss and explain him your situation?

Yes thanks for giving reference to that safety standard. I remember them having huge cost. Does every company purchases such document for standardization or is available anywhere in the internet for the download.

As you told I understand there is great problem with safety of the problem..I am trying to make a documents with the books and papers you guys suggested to make him understand
its dangerous for a newbie like me...(Although I am worried he may fire me). So till that he allows I am really hopinh you guys would help me with the same.

Thanks again..

[getfast_kiran]

I'd also be concerned about the transformer itself.
These ac-dc switcher chips require custom transformers and usually the datasheets are pretty low on the details about the number of windings, the wire thickness, the ferrite core type and so on. Do you guys know what you're doing when it comes to manufacturing the transformers, or are you ordering them from somewhere?

I've seen the datasheets for TNY** chips, I've also seen the datasheets for ST's Viper series and they're also not very detailed - Viper22 looks like a nice IC especially when paired with ST's TSM101 chip (voltage reference + voltage/current monitor, does the feedback for the chip through an optocoupler)... but again, it's still low on the technical details. Would be hard to use by a beginner.

One of the most detailed datasheets I know of is for the TEA172* chip from NXP ... : TEA1721 (max 5w) , TEA1723 (max 11w)

Lots of detail, lots of formulas to determine the components, it's a cheap IC which also uses a primary auxiliary winding to determine the output on the secondary side so it doesn't need a voltage reference and an optical isolator... you're saving money by not using those, but you may be spending a bit more money on the transformer..  the transformer would have to be custom made as it has the auxiliary winding between the primary and the secondary winding, and the secondary winding MUST have triple insulated wire to be absolutely safe for users.

As there's no feedback from secondary side, they may be less accurate with the output voltage but if the cost of manufacturing the transformer is the same as with other chips, you could just set the output voltage a bit higher than desired and spend 0.1-0.2$ (less than the voltage reference and optocoupler costs) on a tiny 5v linear regulator that would drop the 5.x volts to exactly 5v..

They also have an application note which explains their demo board and has lots of useful information for someone like you. Here's the application note : http://www.nxp.com/documents/user_manual/UM10521.pdf
And another one : AN11060  TEA172X 5 W to 11 W /USB charger : http://www.farnell.com/datasheets/1760795.pdf?_ga=1.189448363.1081003429.1454881204

later edit: and they also have an online calculator sort of thing for these chips : http://nxp.transim.com/fb/.aspx#General_InputSpecifications

Unless you guys have some investment in using those TNY chips, it may be worth looking at alternatives before committing to using them.

Hi actually the guy who was doing ordered a bucket full of TNY284 chips almost a 1000 of them. So basically my boss want me to stick to it.

As you told I am worried if the transformer is problem cause now it started glitching even though I changed the feedback back to a TL431 based circuit with
1 transformer and out put goes from 4.8 - 5 v to suddenly 8v or 9v...
With another transformer....The glitch will be gone when I use another transformer but here the problem is the output will be at 12-11V...

do you know what might be the problem...behind it.

[tatus1969]

@mariush, you are right, there is a ton of things that need to be taken into account. i have a 500 page book on this topic. safety rules, winding construction, skin effect, leakage inductance, topologies, core types, different types of power losses, output filter , input  , emi filtering, and countless other topics.

normally you go to one of the transformer manufacturing specialists and let them wind your custom part. but for a number of specific chips they have stock parts that are matched to the corresponding reference .

Could you please tell the book that you were referencing.
And currently my new board is showing output of 4.85 - 5.00 but it will glitch to 8-9V for a second and comeback to 4.85 -5V regularly with one transformer.

And it will not glitch but generate a voltage of 12V at the output with another transformer of same rating.

do you think it is due to problem of transformer winding

i'm afraid but that one is in german, but i can check at home maybe there are translations. what is the winding ratio of your transformer? plus please post your circuit drawing, that allows us to better help you.