TNY287 restart issue

Hi Willie,

Please help to provide the following information especially the schematic (I was unable to open the file format, your power supply requirement and transformer design.
1. The schematic you attached can't be open, is it ok if you will just print screen it then share it again or if you can have it in pdf file it is also ok.
2. Can you also provide your power supply requirements. Like the input voltage range, output voltage, output power, the target efficiency, switching frequency of operation and others.
3. What is your issue with the transformer leakage? We know that the leakage inductance is significant in the operation of the power supply.
4. Can you also share your transformer design, I assume you apply a gap in the core in order to get the primary inductance of the transformer after calculating the number of turns. Please provide the number of turns both on the primary and secondary, the primary inductance and the measured leakage inductance. If transformer is ungapped you will experience oscillation problem which could lead to audible noise and saturation.
5. The drain voltage must be switching, the waveform you've shown looks like it was aliases. What is your target switching frequency requirements? It's either we have to capture the drain voltage in usec/div or if your oscilloscope have a better zoom functionality, you can set the time/division in msec and zoom a portion of the switching drain voltage.
6. 2W at the snubber resistor seems higher. What is the IC package your are using for TNY287 IC or what part number of TNY287, is it P, D or K?

I hope you can provide your power supply requirements, the current schematic in .jpg or .pdf format and your transformer parameter and design.

Thanks!

Hello Marcus,

Thank you for your reply. I have uploaded the same document in PDF format, sorry about that! It will have all details on the transformer voltage range etc. In any case, I will also answer your questions here.

Input Voltage: 120 VAC supplied by VARIAC
Output Voltage: 5 V
Output Power: 10W
Target efficiency: 0.72
Switching Frequency: 137kHz

Leakage issue: Original target leakage by PI design was ~50 uH, I could not meet this with my transformer construction, the best I could get was 112 uH so I had to adjust the RCD snubber clamp values which I made so it could handle up to 200 uH of leakage inductance.

Primary turns: 143
Secondary: 6 turns, I used quadrifilar instead of trifilar to help with leakage, but wire is not T.I.W it is just regular enameled wire.
Primary inductance: 2.05 mH
Leakage inductance: 112 uH

Package TNY287D

Hi Willie,
Thanks for your prompt response.

As you've said you get this value from PIExpert.
In PIExpert, there's a tab Design Results. Can you share to me the spreadsheet. The whole sheet just print screen and share it.
The calculated leakage inductance must also be followed. Increasing the leakage will have an impact on maximum flux density of the core to be exceeded. This is most probably the problem of your power supply. During start-up the current that were sensed by the IC already exceeded the limit causing the IC to be in auto-restart mode every 2.5 seconds.
Maybe some value in the spreadsheet that needs to be change that might cause "Design Failed" and this requires to correct, such as the KP is very low or the VOR is also low that needed to increase. A lower VOR will cause the BM or Maximum FLux Density to be exceeded.
Increasing the leakage inductance will have an impact on flux density, VOR and other parameters so we need to follow the calculated values from PIExpert.
I will also review the spreadsheet there might be some problem with some values that were entered like the VOR causing the primary inductance to be that high.

For flyback transformers, minimizing leakage inductance is very important, since the leakage energy dissipates in external snubber or clamp network, the reason why you are increasing the caps and the wattage of the resistor. It is important to minimize the ratio of leakage to the magnetizing inductance in order to minimize the amount of magnetizing energy lost.

Leakage inductance depends largely on the physical winding geometry. You can reduce the leakage inductance by: 1. Interleaving the layers; 2. Using a wider bobbin winding width, 3. Minimizing number of turns; 4. sum of the heights of all winding layers; 5. sum of the heights of the spacing gaps between winding layers.

So I suggest go back to the spreadsheet, check each of the parameters KP, VOR, BM (maximum flux density), optimize the primary inductance or LP, make sure that the leakage inductance is minimized and do the suggested winding construction to reduced and follow what's calculated in the spreadsheet.

Hi Marqus,

Thank you so much for your detailed response. As requested I have attached some screen shots of the design calculated by PI expert. I will try to play around with the values on the sheet and see if I can reduce the number of turns/primary inductance.

Thanks,
Will

Hi Willie,

Thanks for sending the design results spreadsheet. I already reviewed it.
The output power is 12W as entered in the spreadsheet but this should not pose a problem.

The major problem that I am seeing is again the leakage inductance, we have to minimized it as I've said in my previous suggestion and analysis. Physical geometry can help you achieve the desired leakage inductance.
1. Interleave winding (Primary-Secondary-Primary). Are you using EF20, 10 pins? Just asking this because this is easy to wind an interleaved winding as you can cut the first or second layer of the primary winding then solder to one pin then the second or third layer will be wind after the secondary winding. (I am assuming you will can have the primary winding in two to three layers).
2. Change the secondary wire AWG to higher gauge let say #28 trifilar to reduce the diameter and spread the winding within the width of the bobbin.
3. Go back to the design spreadsheet, you can change the parameters such as the VOR, NS and etc.
4. You can also reduce the number of layers by reducing the primary number of turns. This requires design optimization.
5. You can also use PIXLS so you can play around the parameters such as the minimum input voltage (VMIN).
6. Try also set the DMAX around 0.6 or less, this can be altered by changing VOR, VIN_MIN, number of turns.

Thanks! I will definitely try to interleave, as well as attempt to change several of the parameters. I am using an EF core so it shouldn't be too difficult. I am also curious about things like core material, and winding methods. I am using N27 core material which I attached the data sheet, and for winding I made the wire quadrifilar instead of trifilar, and twisted it together using a drill. In your experience, could these parameters be contributing to leakage inductance?

Hi WIllie,

If you will use quadfilar at the secondary what is the wire gauge?
If you twist the wires (4 wires) in the secondary, your total wire diameter will increase hence the height of secondary winding layer will increase. The objective is to decrease the height of each layer to help in decreasing the leakage inductance.
I suggest if you will make it as a quadfilar, do not twist it just do it in parallel and it will also decrease the gauge wire that you will use. I am assuming that you will increase the number of filars at the secondary winding for better utilization of the width of the bobbin, spreading the whole turns on the bobbin width.
Since you are working with a low power and efficiency is around 70% to 75%, N27 material should be fine.
Interleaving or in layman's term sandwiching the secondary winding in between primary winding will greatly help in decreasing the leakage. Less number of layers, so you need to optimize the number of turns.
Thanks!

Hello Marqus,

I have taken your advice for interleaving, and messing with numbers to bring the primary turns number down and have reduced the leakage inductance down to ~55 uH which is under the permissible leakage inductance of 70 uH. I had to go 5x filar since I was still getting a little bit too much even with 4x filar.

In any case it seems to have solved the issue, but created a new one. As seen on the scope image (see attached "Output"), the output voltage shoots up to 56V and quickly comes down, to the point where the TNY stays off. What could cause this? I have tested for shorts in the transformer and there isn't anything.

Thanks,
William

The spreadsheet you've sent is I think an old one.
Can you share to me the new one? What is the primary inductance, number of turns both at the primary and secondary? Best if you can share to me your new spreadsheet as you mentioned you alter some of the parameters.
Then what is Channel 1 in the scope is it the Vds of the primary MOSFET?
What is Channel 2, it is in 5V/div but the coupling is in AC.

Hi Willie,
Kindly check the winding orientation at the secondary might be unsymmetrical with the primary.
Winding orientation means that if you do counterclockwise winding at the primary, you must also do this at the secondary.

Then I design your power supply using PIXLS.
Instead of Standard Current Limit wherein you used 0.1uF capacitor at the BP pin, I utilized Increase Current Limit. Changing the 0.1uF to 10uF ( in your schematic it is C4).
Then using this current limit this is the design parameters I design for you.
Vin:85 to 265VAC
Vout: 5V
Iout:2.5A
Pout: 12.5W
Efficiency: 72%
Transformer Core: EF20 (or what core is available to you?)
Lpri (primary inductance): 1050 uH
Npri: 69turns
Primary Number of Layers: 2
Nsec: 4
Winding Technique: Split winding, meaning at 1st layer wind the first primary layer then tape, secondary layer, then tape, then remaining half of primary layer.
Then your optocoupler what is the complete part number you re using?
R9 and R10, please change to 10kohm.
If the leakage inductance is already low, please change the snubber RCD values to: Rsnub: 100kohm, Csnub: 2.2nF; Rdamp: 100ohm; Diode: 1N4007GP (this is better for EMI).

Also please check your layout and connection there might be some problem.
You can also put a 1uF to 2.2uF across the anode and cathode of TL431.
Also check each components if there are wrong values.

You can also refer to DER-121, this is almost the same as your design.
The only difference is this utilized TinySwitch 3 IC and there some differences between this IC but utilization is still the same.

Thanks!

Hi Marqus,

So sorry! I had posted the wrong document set by mistake. In any case, I have reviewed your message regarding the design and will try to piece it together and test it. Thanks for the help.

Will

Hi Willie,
PI1 waveform is the output?
If this is the output, is this normally running or during start-up?
The voltage/div is 2V/div so average output voltage is around 10.5V and the time/div is 50ms, the frequency of oscillation seen in the waveform is very low.

If you are in auto-restart mode, this will happen if there's a fault conditions such as output overload, output short circuit or open-loop. You will enter into AR mode for 64ms, meaning the IC will try to operate within 64ms, you can see this when you probe the Vds of main switching MOSFET. During this time output voltage will be observed at the output but at a very low voltage and only within 64ms. Then AR will be on off mode for about 2.5sec until the fault is removed.
Actually if you have a good transformer design and circuit connections are all good, you can do a gradual increase in AC input voltage, you can start at 0V and gradually increase the input. The output will rise when you already surpassed the EN/UV pin requirements. So even in open loop your output would not reach your desired output voltage as AR will act.
Kindly check your transformer constructions. I mentioned a DER from my previous comment you can review the transformer construction. You will only can a higher output if turns ratio are not followed and if the winding configuration at the secondary is interchange.

There are also application notes available for TinySwitch-4 IC which available in PI website you can check this.

Thanks!

Hi again Marqus,

I had replaced the TL431 and in my doing so switch up the anode/cathode which explains why it was not working. After fixing this issue, the TNY blew up at turn on using the design I had. I redesigned with the specs you gave me, and the same thing happened. I want to ask if you could review my step by step transformer design and let me know if I am doing anything wrong here.

I used an N87 core, E20/10/6, with the bobbin that comes in this document:
https://www.tdk-electronics.tdk.com/inf/80/db/fer/e_20_10_6.pdf

Magnet wire on primary side was 30 AWG (not 32 AWG) from BNtechgo:
http://bntechgo.com/bntechgo-30-awg-magnet-wire-enameled-copper-wire-enameled-magnet-winding-wire-4-oz-0-0315-diameter-1-spool-coil-red-tempera…

Magnet wire on secondary side was 26 AWG:
http://bntechgo.com/bntechgo-26-awg-magnet-wire-enameled-copper-wire-enameled-magnet-winding-wire-4-oz-0-0157-diameter-1-spool-coil-natural-tem…

Step 1: I scraped off the enamel on primary side wire, added flux, and soldered to pin 2 of the bobbin, then proceeded to wind a little over half of the turns (~42) perfectly side by side on the bobbin going clockwise starting closes to pins 1 and 2 and moving towards pins 3 and 4 (see attached for transformer bobbin pinout).

Step 2: At 42 turns I left the primary side wire hanging, (did not cut it), and taped 1 layer over it. I soldered the tip of 4 lengths of the 26 AWG wire together and soldered to pin 4. Proceeded to wind clockwise 4 turns of the quadrifilar winding. At the end of these turns since I was near the other side of the bobbin, I made a bend in the wire and brought it back to solder on pin 3. If you can see from the attached image (transformer1) it creates an awkward bulge in the coil.

Step 3: I tape, and wind the remaining turns of the primary up to 69 going back to solder on pin 1. Then finish with 1 more layer of tape.

Step 4: I grinded the core down to where there is a very small gap to be able to get my primary inductance. I measured the inductance using a low budget LCR meter from Proster:
https://www.amazon.com/Proster-Multimeter-Capacitance-Resistance-Inductance/dp/B071WNNYQT
The inductance reading was 1050 uH.

Step 5: I shorted the output winding by soldering a ~1 inch piece of stranded hookup wire to it and measured the primary side, leakage inductance reading was around 75 uH.

The inlet to the transformer was pin 1, and outlet connected to the drain was pin 2. The outlet pin connected to the output diode was pin 4.

My thoughts are first that my LCR meter is giving readings with error of +/- 20%, I have ordered a new B&K 880 LCR meter to help confirm this issue. My solder joints are not very good (enamel is difficult to remove), which could be contributing much higher parasitic inductance (I did not think this would be much of an issue, but maybe?). I can't think of anything else other than there is something blatanly wrong with my design procedure, help is very much appreciated!

Hi ,
Ok, so you have a wrong connection on TL431?
After you found out that you have a wrong connection I assume you did not replaced other parts and used the same transformer you design.
When you construct the transformer based on my suggestions, did you replaced the TNY287 IC? Then replaced the BP capacitor to 10uF and the snubbers value that I mentioned?
Is the optocoupler still good?
Can you share to me a photo of your power supply and its PCB layout?
TNY switch will be damaged if prior to your testing if you performed lots of power up with a wrong connection. Maybe a shorted drain to source that happened in your previous test or an open connection that left the source ungrounded.
Have you check the output diode rectifier if still good and correctly connected with the secondary winding? For high side connection, the anode of the output diode rectifier must be connected at the start pin of the secondary winding.

Correct me if I am wrong, when you wind the secondary you did not remove the bobbin from Mandrel or the winder and just wind the secondary clockwise starting at Pin 4? If you did this you might wind it in counter clockwise as opposite with the primary winding and this will make this as a forward converter which increases your voltage upon turning on causing a large voltage across the drain to source of the MOSFET of the Tiny Switch IC and that's probably the reason why the IC blew up. You can check the drain-to-source.

Just a recommendation after constructing the transformer to check if you wind it correctly. This test need a Signal Generator and a Oscilloscope. From the signal generator set it at pulse waveform 10V pk-pk is ok and frequency at 100kHz. Put a oscilloscope probe both at the primary winding and secondary winding. For example, Channel 1 of the scope probe, put the positive probe tip at the start pin of the primary winding in your transformer at Pin 2 and the negative tip to Pin 1. Then Channel 2 probe at the secondary winding, positive tip at PIn 4 or where you start the secondary winding and negative tip at Pin 3 or the end point of secondary winding. Then connect the probe tip coming from the Signal Generator to the primary winding with the positive tip at the starting point of the primary winding, in your example at pin2. Turn on the signal generator, you will see an approximately pulse waveform with your set voltage and frequency. You will also notice a pulse waveform at the secondary having a lower amplitude based on your turns ratio. If you put the probe tips correctly both primary and secondary windings, the waveforms must be in-phase. If they are in-phase means you have at least winded the wires correctly.

You can also check DER or design example of transformer winding. And let me see if we can simulate it here transformer design with the proper checking at the oscilloscope.

Thanks!

Hi Willie,
As promised we design and build the power supply based on your specification helping you to design your own power supply using TinySwitch IC with ease.
Vout = 5V; Iout = 2.5A; Pout = 12.5W; Efficiency = 80%

We utilized TNY288DG as this is the IC currently available in our lab.
But the design is just the same with TNY287D.
Using TNY288D, we set the current limit configuration to Standard Limit so a 0.1uF capacitor is needed across BP pin of the IC and Source (S) pin of the IC. Compatible design using TNY287D is to set the current limit configuration to Increased, and a 10uF capacitor will be used across BP and SOURCE pin of the IC.

The transformer used is EF20 10 pins vertical (horizontal is not available) but the design is just the same. We've also done two sets of transformer one is with shield and the other without shield. The leakage inductance with shield is 17.4uH while without shield leakage inductance is 16.7uH. This is with primary inductance of 930uH.
The NPRI or Primary Number of Turns is 70 turns (AWG#30) while NSEC or Secondary Number of Turns is 4T (Using triple insulated wire Bifilar #25).

Included files here for your reference are PIExpert Spreadsheet, Schematic, BOM, Transformer Construction Procedures (with shield, just remove the shield if will not be used), breadboard (to hasten the design we utilized different evaluation board with reworks to accommodate the design) and some electrical data.

Hope this would help you design your own power supply.