Solution Finder Get Tech Support

Frequency change with changing load

Posted by: Dodger741 on

My design provides 6W at 5 volts. I have noticed that on some boards the TNY277 thermal characteristics are much worse than others. What I have seen is that when powered on, everything works as expected. The current ramps to the threshold and the device turns off. If I disconnect the load and reconnect (with power on), the on cycle goes a smaller value and the frequency increases. Since most of the losses are switching related, the thermals degrade drastically. Once in the mode, connecting and reconnecting the load does not revert back to the preferred operating mode. What would causing the device to switch to a lower current threshold and remain there??? see attached photos

Comments

Submitted by PI-Sarek on 04/09/2012

Hi,

The waveform that you havce provided looks abnormal. There is a huge leading edge spike of current. This suggests one of the three possibilities below:

1. Either the converter has got wired as a "Forward" instead of "Flyback". Could you please check your circuit for these conditions? You will need to verify the start and end connections of the primary and all the secondary windings.

2. Second possibility is that you are using a slow recovery output diode

3. Third possibility is that you have a possible error in the snubber circuit.

If there is another reason why you have this behavior with your circuit,I will need the schematic, board layout and transformer construction and design details to help you further.

Regards

PI-Sarek

Submitted by Dodger741 on 04/11/2012

Unfortunately I cannot post the schematic as the customer has marked it proprietary. The transformer is wired correctly, the waveform at the TNY drain matches the waveform at the rectifier diode anode. The diode is an ON Semiconductor MBR230LSFT1G. I to am concerned as to where the leading current spike is originating from but I don't see how it can influence the current ramp threshold which triggers much later. There are other components which need to be investigated related to this. Regardless I cannot understand why the TNY can essentially change its current threshold instead of maintaining it and changing the duty-cycle which is how I understand the device works. Do you have any explanation. The transformer p/n is TSD-2270 EEL19

Submitted by PI-Sarek on 04/11/2012

Hi,

The Tiny Switch family parts have a feature wherein the current limit is selected by the controller based on the load level. This feature ensures that the switching frequency remains high even at light load levels. This operation is described on page no.5 of the datasheet. I am providing a link below to the TNY-III family datasheet:

http://www.powerint.com/sites/default/files/product-docs/tny274-280.pdf

The change in current limit has some hysteresis to ensure that the current limit does not change continuously. So if the controller decideds to use one current limit level, it will not change to the next current limit level unless it finds that a significant number of consecuitive pulses are either being conducted or skipped. There are four different current limit levels that the controller may select from. The device will operate at the specified current limit level only at full rated load. At lower loads it will operate at one of the other three lower current limit levels.

Sometimes it may so happen that the controler will select one current limit level at start however when the load is changed, it may operate at the next higher or lower current limit level and this is due to the hysteresis between the different current limit levels.

This feature is designed such that you will not notice any significant change in efficiency or thermal performance and works very well.This suggests that a detailed review of your design needs to be conducted

It is not clear therefore why you are experiencing much difficulty and get a performance which is distinctly different when the controller decides to operate at a different current limit level due to load changes.

If you are unable to provide more details of your design, I suggest you contact one of our field offices in your area who may be able to sign a confidentiality agreements and have one of our Field Applications engineer review your circuit. I am providing a link below to help you find the nearest office in your area:

http://www.powerint.com/en/company/sales-information/worldwide-sales-support-locations

Regards

PI-Sarek

Submitted by Dodger741 on 04/27/2012

PI-Sarek: I am trying to work with the field office. I did determine that my layout was effecting the performance and have modified it temporary to resolve that issue. The thermals are much better now. There is a large current spike still at the leading edge but its duration is less than 100nsec. I have changed the snubber and rectifier diodes to fast recovery diodes with little effect. Some of this could still be the layout and the balance must be the transformer.??. I noticed that there only appears to be two TNY on time durations, 800ns and 1700nsec, regardless of load. In a quest to increase the on time and lower the frequency, I changed from a TNY277 to a TNY279 and finally to a TNY280 with little effect on the max current ramp or duration.. Why is that. In the attached photo the TNY280 waveform does not change duration (at all) between full load at 1.3A and no load. Red current trace is 0.5A/div. The frequency changes in all cases by why the constant/semi-constant on duration?

Submitted by PI-Sarek on 05/01/2012

Hi,

I think there is some confusion. I suggest you review the datasheet for the part in detail. The Tiny Switch ICs use a technique which uses skipping of cycles as a means to regulate output voltage. This method is therefore called ON-OFF control. Unlike PWM control, in this method, the current limit for each cycle is fixed depending on how many cycles are being skipped. The slope of inductor current (di/dt) is set by the input voltage and primary inductance. Every cycle the current will ramp up to a predetrmined value and then the MOSFET inside the IC will turn off which will result in transfer of energy to the secondary.

So in each cycle the current limit is the same if the load on the output is fixed. As the load is increased only the number of cycles per unit time will change in order to maintain the output voltage constant. the di/dt and current limit will not change if the input voltage is still the same. Once the load level drops sufficienctly such that the number of cycles that the controller has to skip in order to regulate the output voltage increases, the controller will then reduce the current limit level thereby reducing the size of the "energy packet" being delivered so that it does not have to skip too many cycles in order to regulate. I hope this clarifies. This is different from the classical "PWM".

Hope this helps.

Best Regards,
PI-Sarek

Submitted by Dodger741 on 05/02/2012

I do understand the skipped cycles and how that regulates the output. What confuses me is that in my test conditions I have a constant input voltage, load and inductance. My current ramp is approx 390mA/us with a 0.1uF BP/M capacitor. The current limit as you change device types from 277 to TNY280 increases according to the datasheet by 60%. This should increase the ON time of the TNY but it doesn't in my test conditions with the different TNYs. The TNYs turn off at approx 300mA regardless of the TNY series. This happens in about 0.7-1.4usec at a 15 - 10 usec rate. What I interpreted from your earlier comment indicated that the internal state machine can lower the "Standard Current Limit" if the pulse rate is too slow which I don't think should apply in my case (15-20usec rate) but something is lowering the current threshold. The devices are not over temperature so what am I missing.

Submitted by PI-Sarek on 05/02/2012

In reply to by VCastrellon

I believe the state machine is doing its job of lowering the current limit. If you choose a bigger device in the family, the state machine lowers the current limit depending on the operating conditions.

If you believe this is not allowing you to meet any performance requirements, it may be best for you to contact one of our field application engineers at this time.

In order to help you further, we really need to look at the board and the operation as we sweep the load. I suggest you verify the different current limit levels as you sweep the load while you keep the input voltage constant. You should use the lowest rated input voltage for your design when conducting this test.

Regards

PI-Sarek