PFS726 drops PF after temperature rises
Hello,
I have noticed, that PF drops after temperature rises. This drop is almoast linear with temperature rise. For example we
have about 400W load at 220VAC and PF at startup is 0.998, when temperature rises to 54C on PFS726 body and inductor temperature is 60C, capacitor has around 52C PF drops to 0.965. When temperature rises for another 20C, so the
average temperature is about 80C, the PF drops to around 0.86 or near by. At this temperature none of the components
change its initial value so much, that effect will be on PF.
Just for clarification we have:
- 105C/68uF capacitors
- inductor core T157-60 (1800uH initial with 2.5A calculated current)
I also noticed, if I remove filter capacitor CFB on feedback pin, PF is lower around 0.88, but it is temperature stable. Then I solder 2.2nF instead of
10nF proposed by the application note I get starting PF 0.99 and when temperature rises to about 80C PF drops to 0.90 and it is dropping, but
slower then with 10nF. If no capacitor is soldered, then PF stays around 0.88. I also noticed, that phase angle increase from initial 5 degree to
18 degrees at higher temperature with CBF=2.2nF
Could you please give me some guidance what to check, that PF drops.
regards
Comments
Hello,
we are using LQA08TC600 boost diode and they share the same heatsink. Unfortunatelly I can't test with only warm
diode, and PFS/inductor are cold. Paralell with diode we have 27pF NPO capacitor.
If for example I disconnect power from the circuit for about 10-20 seconds, and then plug it again, the power factor rises
even if temperature remains at the same level. In this case PFS, diode, inductor does not drop any temperature. Also heat
sink is quite big.
The most strange is, if I remove CFB completelly, then PF stays around 0.9, regardless of temperature rise. If I put small
capacitor like 470pF (NPO) PF rises to 0.99 and it takes more time to fall below 0.9.So I guess, that some issue might be
with the feedback or I am wrong.
Can you post the layout file?
Also pls. post the input current and/or choke current waveforms of the different conditions with different PF.
Thanks
Hello,
I attached you layout with description about component location. Currently I can't send you waveforms, because at the
moment device is under test. I will send you as soon as possible. But I attached you input current waveform with PF
around 0.98 at room temperature. All test done at 230V AC main supply.
I will be very pleased if you could check this and give me some conclusion.
regards
Right away I see that the copper traces of the very noisy node, the PFS Drain node, flows, under the control pin parts (D12, C54, etc). This is a violation of our layout guidelines. Pls. study them carefully, in the datasheet and Appnote. As a start I suggest swapping the positions of the boost diode and the PFS. This will reduce the noisy loop area formed by: source - Drain - boost diode anode - boost diode cathode, and the "loop-shrinking capacitor" which connects form the Source pin to the diode's cathode. These 3 parts should be close together, with short traces connecting them, this loop area should be minimized, and no other components inside of the this loop. The small signal parts should be kept away from the Drain node. The surface area of the traces connecting to the Drain should be kept small. The last 2 tips are to minimize stray capacitive coupling between the Drain node, which has huge dv/dt, with the small signal nodes.
With a poor layout the PF can drop as the 2 transistors in the feedback loop warm up.
Hello,
thank you for suggestions. Unfortunatelly I can't change or swap output diode and PFS, because of heat sink design and special clipping system. But I can change location of some componets if this will help. As I known temperature rise will increase transistor colector current, so to be more temperature independent usualy Re is added, which provides negative feedback and stabilise transistor against temperature or supply change. But of course this decrease the voltage gain.
So do you have some suggestions regarding this solution?
regards
I don't see an easy way without swapping the diode and PFS positions.
Hello,
I have changed diode position on our prototype board. The same issue still remains.Then I done
further tests with temperature rise. I have discovered, that when I rise temperature of ambient to
around 90C then PF drops to 0.62 and voltage drops to 300V. If I turn power OFF and then emmediatelly
on PF rises to 0.97 and Vout to 381V DC. So I assume, that there MUST be something in feedback or
maybe some internal limit is reached and with hysteresis PFS turns ON even if temperature is high.
Could you please check this conclusions!
When I mentioned temperature rise, I did't warm up PFS726, only feedback and components related
to the feedback!
regards
Sound like the pullup BJT is turning on. Pls. double check for wrong component values and board surface contamination.
Yes I did check it. I also changed both transistors with others complementary pairs and I can see, that temperature has great affect on feedback regulation. I assume, that the problem is in Collector Cut-off current, because at high temperatures this current increased more than 10 times. And because of low 500nA feedback input current a small difference between high and low transistor collector cut-off current can have great impact on regulation. If I disconnect power for a few seconds and because some load is always present compensation capacitor of 4.7uF is discharged and when power is ON again, output voltage is regulated regardless of temperature. So I assume, that temperature above 90C increases collector cut-off current and this capacitor is overcharged and PF and Vout fails. When capacitors discharges PFS starts working.
Are my assumptions correct?
When you tried reversing the positions of the diode and PFS, did you keep the Drain node and traces away from the BJTs? The BJT circuit is very sensitive to stray-capacitively coupled noise from the Drain node.
Yes I did. Both BJT are far away from the drain node. Only GROUND track on PCB bottom side is below transistors.
Can you show the input AC waveforms when the output voltage is sagging?
What boost diode are you using?
Are the boost diode and PFS sharing a heatsink? Is it possible to do a test where the diode is warmed up while the PFS and the choke are cool?
This may be a noise/layout related issue; many boost diodes' reverse-recovery 'snappiness" and Qrr get worse at high temperature, so they generate more noise.