TFS7708 high idle temperature.

18 posts / 0 new

Hi to all.
I have a design using a TFS7708 switcher. Input is 380VDC ( from a PFC) and output is 26V @ 15A.
I have noticed that at no load the TFS is getting quite warm. It is on a 8dec C/watt heatsink and it is at about 60degC
with no load. Ambient @ 28deg C.
I have not used this chip before so I'm not sure what is "normal" :0)
If I disable the main switcher and only leave the small flyback running , the temperature drops to almost ambient , so it's definitely
the main converter generating the heat.
If I run the main converter at high load( about 350W) I can run for a few minutes before the heatsink gets to 80deg C so I think
the basic design is ok. I tried increacing my transformer magnetising inductance ( I'm now at 20mH) and that seemed to help a little
but I'm now in danger of saturating under high load.
If this is not normal , where can I start looking to find what is causing the heating of the switcher?
I'm fairly new to this , so I may be missing something obvious :0)

Can you post schematic?

Relevant part of schematic attached. Output diodes are STPS30150 's. Output inductor is 47uH.

ForwardConv.JPG137.18 KB

First off, I note that your design is set for 132kHz operation. If you haven't done so already, you will need to gap the transformer to reduce the primary magnetizing inductance. This will allow the drain to ring down sufficiently to recharge the high side bias. and prevent pulse skipping On our 132 kHz designs, we used 0.5 mil (1 mil = 0.001 inch) transformer tape for the gap.
I also note that you snubber cap is on the large side. (4.7 nF) try cutting the value in half, as I suspect the transformer may not be resetting properly. You can check the switch waveforms to make sure of this. Gapping the transformer will also help with transformer reset.

Thanks for the tip. I am actually running at 66Khz , I didn't update the cap on the schematic , sorry.
I'll try and reduce the 4n7 to 2nF to see what it does , thanks.

Reduced the snubber cap to 2nF. Still running hot at idle. About 60deg C , which translated to about 3-4W under no load. This may be normal , I don't know?
At about 350 - 400W , it seems to run fine , but I need a fan then obviously.
Is 3-4W idle normal? I may be chasing a "problem" that is in fact not there :0)

I had another look at you schematic, and I don't see any resistor feeding the BP pin from the flyback primary bias supply. Please check the application circuit in the TFS data sheet to see what I'm talking about.

The resistor is connected at the junction of the 15V zener and the 10uF cap. There is a net lablel "POWER ON" which is connected to the power on circuit in the top left corner of the schematic.

I do not see a reset circuit from TVS Diode. Is that possible?

OK, comment on bypass pin supply noted. I suspect at this point that you might be losing transformer reset at light load. Your snubber circuit varies fairly considerably from what PI uses in their standard designs. It might be fruitful to check the drain and clamp voltages at light to medium loads to see if the transformer resets properly. If the voltage on the snubber collapses at light load, you may be losing transformer reset. I'm including a link to one of our standard designs so that you can see the difference in the two snubber designs.

Thanks Pi-Wrench.
I was originally using the circuit that PI uses , but the Tranzorb was running VERY hot , so I changed. It may of course been an incorrect transformer turns ratio , resulting in the reflected voltage exceeding the 160V ( I think) of the tranzorb.
What exactly do I look for to see if the transformer is resetting properly. I'm not very experienced with this stuff , so I may be missing something. I assumed any issues with transformer resetting would be more pronounced at heavy load , not no load :0)

going to upload some screenshots of various drain waveforms under different load conditions.
No load , very light load(200mA) , 1A , 5A , 10A , 15A and 18A loads. Bus voltage is 380V from the PFC. This can be seen as indicated by the 1 cursor on the screen.

various drain waveforms under different loads. No load , very light load 1A , 5A , 10A , 15A and 18A loads. output voltage is 27.6V.

no_load.png17.16 KB
very_light_load_200ma.png19.96 KB
light_load_1A.png57.01 KB
5A_load.png43.15 KB
10A_load.png40.74 KB
15A_load.png43.52 KB
18A_load.png43.58 KB

The overheating of the zener clamp option indicates that you might have either too much leakage inductance, too little primary magnetizing inductance, or both. Since you are operating at 66kHz, there is no need to gap the core to reduce magnetizing inductance. If you haven't already done so, splitting the primary winding and placing the secondary between the to primary halves will greatly reduce the leakage inductance. Your high power drain waveform shows a big leading edge spike that may be indicative of excessive leakage. It would be useful to see a pdf of your transformer design file.

Thanks Pi-Wrench.
The transformer was just a quick design using TI's power stage designer tool. I wound it myself using whatever wire I had available in the lab :0) , so it's far from a perfect design/build.
I wound the primary first and then the secondary. When we get back out of lock down I'll try and do the split winding.
Just out of interest. I had a transformer with a higher primary inductance. 24mH I think.
I noticed that under higher load the the drain voltage after the fet's switch orff (the reset stage I think) would start to clime very high , approaching 700V. Could this have been due to the transformer saturating? I don't have enough experience to be able to tell exactly what is happening just by looking at the waveforms :0(

With a proper clamp design, the reset voltage shouldn't get anywhere near 700V.

I think it would be useful at this point if you re-ran your design using the PI software. This will not only give you a suggested transformer design, but also, design specs for the output inductor and values for the components around the TFS-2 chip.

I was wondering what could cause this behavior though. At about 10A load , the drain voltage is fine , but above this it starts to move up with increasing load. What causes this behavior?
I will be redesigning the xfmr , but I'd like to understand what is causing this issue so I know what i'm looking at if I ever see it again. I'm assuming it's the transformer starting to saturate , but I'm not sure.

If the output inductor starts to saturate, this could be cause issues. You can't saturate a forward transformer with current like a flyback, as you only store a minimal amount of energy in the core, with the current simply passing through the transformer interface. Saturation will happen via volt-second imbalance rather than excessive current. Excessive current will cause overheating instead.

However, you can saturate your output inductor with excessive current. Saturation onset is especially sudden if you're using a gapped ferrite inductor. Powdered iron or Kool-mu/Sendust (better efficiency, higher flux capability) are good choices for output inductor material - the PI design tools address these options.

I still advise loading your design parameters into PIExpert or PI XLS to check what you're doing. This is especially important as the PI design uses an atypical implementation of a two-switch forward converter that allows >50% duty cycle operation, with volt-second imbalance protection inside the IC.