Power supply based on TFS7704H.

14 posts / 0 new

I am making a power supply based on the TFS7704H, which I plan to mass-produce. Input voltage 160-250V AC. The output voltage is 14V 8A. Without power factor corrector. Earlier, I wrote to you.
Now the first layout is assembled, I am engaged in its adjustment.
The first thing I noticed was the lack of stability of the form of voltage at the drain in the absence of load.
Can you give me a direction in which to look for the cause of instability?
The source gives 6A with an efficiency of 86%



Since you are running at 132 kHz, you will likely need to gap the transformer core to reduce primary inductance and increase the magnetizing current, to allow the top-side driver supply to charge. Otherwise, the topside driver supply can sag and cause the TFS to skip pulses at low/zero load. This is detailed in the report for DER-368. referenced below: The gap can be very small - in the DER-368 we used 0.5 mil (0.013 mm) thick polyester tape.


Hello. Thanks for the answer. Of course, I immediately made a gap. The problem was different. I used a slow bootstrap diode. I replaced it with US1M and everything began to work stably. But I still need some help. As I wrote above. Power supply without PFC. Is it possible to ensure stable operation in the range of 160-250V AC?

33.PNG1.9 MB
44.PNG1.15 MB

Here is a reference to a TFS-based supply that runs off 90-132 VAC using a doubler circuit.. It is a small step to reconfigure the input to run from 160-250 VAC. The output is configured as a current limited battery charger. This could be changed for a typical TL431-based CV output control.


If I make an input voltage doubler. That maximum rectified voltage will exceed the permissible voltage for transistors. In your example, low AC voltage. Or I don’t understand something. Please explain.

In my previous post, I mentioned reconfiguring the input. To use 160-250 VAC input, you would convert the input rectifier circuit from a doubler to a standard full wave bridge with a single B+ filter cap of appropriate rating.

Hello. How to change or decrease VUV ON (max)? The converter turns on when the input voltage is too high.


You will need to change the values of the resistors feeding the L pin and the R pin on the TFS chip. The L pin and R pin resistor strings need to be the same value in order for the internal engine to properly limit the volt-second product delivered to the output transformer for proper transformer reset during off time. Possibly the best approach would be to re-run the design spreadsheet with your lower desired turn-on input DC voltage. The spreadsheet assumes as a default that the TFS is being fed a nominal 380 VDC from a PFC stage. You will need to override the default input voltage parameters to get what you need. You will also be best served by using a clamp-to -ground snubber rather than a clamp-to-rail snubber on the forward stage in order to optimize transformer reset at minimum line voltage. The two snubber types are covered in the TFS application notes. If I remember correctly, the design example I referenced for you uses a clamp-to-ground snubber scheme for the forward converter stage.

I still need help ... I can not get the required voltage. I’m quite happy with 180-250VAC. I made a new calculation. On the layout, the switching voltage is 196VAC. It's a lot. It seems to me that the inductance of the transformer is too small. The real value is 1.06 mH. Please check my calculation.

2020-04-22_03.58.37.jpg6.13 MB

What do you mean that you cannot get the required voltage? Please describe what you see on your prototype in more detail.

Hello. I wanted to say that I can’t get a wide input voltage range. I need stable operation in the range of 180-250VAC. I do not use PFC. I’m upset that this power supply is not working, I spent 3 months on prototyping. there were plans in the summer to release a batch of power supplies.

When you say that you can't get a wide input voltage range, what does the power supply do to misbehave, and at which end of the voltage range?

I took a close look at your spreadsheet, and you are using a transformer core size quite a bit larger than needed for your output power and operating frequency. You could actually operate at 66 kHz with that size transformer and save some headaches (meaning, you don't need to change, but you can...).
Also, since you are running with an unregulated B+, the actual input Vmin will be lower than 1.4 X Vacmin due to ripple. Depending on your bulk capacitor value, you will need to subtract at least 20V or so from your nominal value for min DC (that was using a 220 uF bulk capacitor). Since you are operating with a fairly wide range input voltage, you will need to use a clamp-to- ground snubber to get extra Dmax and reset margin for that low input voltage..

Just as a reminder, the TFS-2 IC is based on an implementation of the 2-switch forward with an asymmetric clamp voltage rather than the standard diode to DC rail, allowing operation in excess of 50% duty cycle. With this in mind, don't be surprised when you see a high off-time voltage on the transformer, especially at low input voltage. The IC has a proprietary engine inside to manage the transformer volt-seconds.

The source simply turned on at a voltage of 196VAC. Now I have changed the parameters of the transformer. To get a turn-on voltage of about 180VAC, I reduced the resistors R and L to 2.82M. Now the turn-on voltage is 183VAC and the turn-off voltage is 172VAC. I have a 300uF input capacitor. 3 * 100uF. But I get warnings in the table. How do I choose the correct voltage for the reset circuit diodes? Now I have three 150V diodes. In your examples, there are 150 + 150 + 170 diodes, and in the table the value of VCLAMP is 530V. Thank you very much for your help!

111.JPG272.07 KB
222.JPG65.19 KB
333.JPG56.91 KB
HiperTFS2_Forward_Design_12-8.pdf717.83 KB

You can change the value of the clamp in the left-hand side of the line. I boosted it to 540V, which is 3 X 180V TVS in series. I'm not getting any warnings in my spreadsheet, but then I chose 230VAC X 1.4 for my nominal input voltage. I also shifted Vmin to 245 VDC, as there is considerably less ripple with the larger bulk cap value.

I did have to mess around with the turns on the output choke to get a value that would satisfy the spreadsheet. I needed a powder turns multiplier of 5.

Topic locked