TOPSwitch does not start

Fig 12 in the datasheet describes the cycle-by-cycle Drain current limiting. It is active all the time, not at any specific point in Figure 5. Note that Fig 12 shows 1 us per division, whereas your plots are ms / div.

As a quick experiment, pls. try placing a 10 uF "soft-finish" cap right across the TL431 cathode to anode, see if this affects your start-up issue.

Thanks for the support.
The board start directly with the 10uF capacitor.

We have found the reason to the startup problem.
The control capacitors (C208) have degraded from originally 47uF down to around 5uF on the boards with startup problems. The capcitor type is SANYO OSCON SA series (se file attachment,the cap is marked 47/25 but the datasheet does not include 25V ?). Is this capacitor of the wrong type or maby bad quality for this job?
These boards had been in service for about four years when this happens.
I know that some designers add a small capacitor (100nF) parallell to electrolytic capacitors to ease their life and improve the lifetime, is this relevant?
In the AN-16 you do not recommend low ESR capcitors in this position but is that still valid if we have 10ohm resistor in series with it ? What capacitor type and brand do you recommend ?

Regards Åke Larsson

What is the internal ambient temperature profile of the application? (e.g. 55°C 50% of the time, 65°C 255 of the time), or at least, the max temperature and average temperature? Specifically, the temperature of C208? In 4 years, how many hours was the PSU operating?

High temperature shortens the life of electrolytic capacitors - every 10*C reduction vs. rated temperature, doubles the rated life (e.g. 5000 hours rated life at 105°C, doubles to 10,000 hrs at 95°C), for Al electrolytic capacitors. For Oscons it's more than 3x for every 10°C reduction.

If C208 is mounted right next to a hot heatsink that will shorten its life.

Do all the capacitors in all the boards exhibit the same loss of capacitance? (assuming the same ambient temperature profile and time). Or is only a small % affected? If they have the same loss of capacitance, it suggests a life time issue. If only a small % are affected, it suggests a long term reliability issue.


The paralleled 100 nF will not help the capacitor life for C208 - what it will do is provide a shunt for high frequency noise.

In the C208 position a low ESR capacitor (Oscons are *very* low ESR), is a waste of money. A low cost 47 uF / 16V Al Electrolytic capacitor is usually used. The temperature selection depends on the ambient temperature and life spec

Adding a 10Ω resistor if using a low ESR capacitor is good practice as it can prevent control loop oscillation. The 100 nF capacitor in parallel is also good practice for the noise reason explained.

Hi
Thanks for the support !
I had some problems when I responded 20120130 so I send it once again.

The board is mounted in a aluminium box fitted to a water cooling system with a temperature between 40-60 degrees celcius. The board has been running continuously for four years with one 24hour stop every year.
The 10uF soft-finish capacitor we experimented with made the board start despite the degraded 5uF(originally 47uF) capacitor. Is it possible to replace the control capacitor electrolytics with an approx. 10uF ceramic or wima MKP 10 and add the soft-finish capacitor to start up safe anyway. Is this possible and do you have design guidelines to do it without electrolytics ?

Our equipment is supposed to work for at least for 25years so a redesign to get rid of the electrolytics will be profitable. The big electrolytic decoupling capacitors on the board seems o.k. so far but can also be replaced by smaller capacitors of other types to improve lifetime.

Regards Åke Larsson

You can redesign to remove electrolytics, but you have to watch out for:

1) Reduced ESR may affect circuit function - e.g. in output capacitors, this will affect loop response. Same with the C pin capacitor.

2) Reduced capacitance can affect circuit function - again in output capacitors, this will affect loop response. Output capacitors are sized for ESR and not capacitance, so a ceramic substitute will have less capacitance than an electrolytic. One trick to keep stability without using a large low ESR electrolytic in the output capacitance when you substitute to ceramics as the main output capacitor, is to use a single modest sized electrolytic after the output filter inductor. For the trick to work, the inductor value has to be kept on the low side (1- 2 uH). The e-capacitor can then see less ripple current (therefore longer life), and be smaller than if you used all electrolytics. This e-cap will provide the loop damping (from its capacitance) that you lost when switching to ceramics for the main capacitors.

3) Large ceramic SMD capacitors require care in PCB layout, wave soldering and positioning, and they will often crack when the PCB is flexed a certain way in production or in the application. This cracking can be very troublesome by showing up intermittently or later in life.

Check out the various manufacturer e-cap app notes. Perhaps a higher temp (e.g. 120*C) e-cap family, or extra long life (e.g. 5k, 10k hours), will work for you.

Consider other polymer and tantalum capacitors.