Solution Finder Get Tech Support

RDR-839 Shuts off with load

Posted by: Ninjabean on

I have copied the RDR 839 design onto my custom board and using all the same components on the recommended BOM with the exception of the Power IC and transformer. I am using a TNY288D and custom made transformer made according to the reference design.

I am testing with a 100 ohm resistive load on the DC rail.

I have bought up 5 custom boards, 3 are fine but 2 of them would raise the DC rail to just over 12V and then immediately shut down. The power supply will not auto restart. When I remove the mains supply the DC rail will stay at about 20mV for a minute or so and it will not be able to restart it until the DC rail collapse to 0V, which takes about a minute.

The power supply will be able to sustain 12VDC if there is no load connected.

What failure mode is this and where can I start looking to debug this problem?

Comments

Submitted by PI-Wrench on 12/10/2021

There might be an error in the bias winding causing the OV protection(R3, VR1, D4) to trigger - either that or VR1 might be reversed, which would cause a low OV trigger point. The circuit description also mentions tuning the OV trip point by varying R3. This resistor value may not be optimized for your implementation.

Submitted by Ninjabean on 12/14/2021

Thank you PI-Wrench your observation was correct.

After some more digging into datasheets and testing I found that the PSU went into shutdown due to tolerance stack up issues, i.e. Output OverV protection threshold (R3, VR1, D4) was on the low side and OutputV feedback threshold was on the high side (VR2, R7, U2, R8).

On the failing unit, with 900mA resistive load, start-up Output OverV protection kicked in around 12.3V and Output OutputV feedback threshold was 12.5V. I was able to tune the failing power supplies back into range by increasing R3 to 100ohms or reducing R7 to 200ohms. This gave me a start-up Output OverV protection threshold of 13.7V(DC output rail) or Output OutputV feedback threshold of 12.3V.

In the RDR-839 reference design report it only shows steady state Output OverV protection triggering at around 14.5V (0% loading) and 14.3V (100%) (which is in line with my calculations of between 14.1 and 15V).

In light of all the above information am I correct in assuming that the "start-up" Output OverV protection threshold is a lower voltage than the "steady state" Output OverV protection threshold, and depends on how heavy a loading the PSU has, a lower load will give a higher start-up Output OverV protection threshold, hence, without modification to a failing board, I was able to boot the PSU up and keep it running on 0% loading and steadily add more loading just fine, but would shut down when I boot up with a load attached?

 

When I increased R3 from 10ohms to 100ohms will my new steady state Output OverV protection threshold will be increased by 0.45V? (V = 90ohm x 5mA)

 

Submitted by PI-Wrench on 12/14/2021

One  thing you might want to try is to reference your OV protection network to the filtered output of the bias winding rather than directly at the bias winding output. This is the way that OV protection is usually implemented in our reference designs.

Submitted by Ninjabean on 12/14/2021

I am using the reference schematic. And have replicated the layout as close as possible. And even with a reference board with a reference transformer (only change is removing R7 to disable the output feedback) I am getting a peak of 13.7V before shutdown occurs.

After testing many of my custom transformers with the reference board I found that they tend to have 12.3V to 12.9V peak output voltage before shutdown. I assume that is down to construction variation so the bias windings, I suppose they are picking up more current, and they are consistent enough that I can compensate by increasing the value of R3.

 

I tried monitoring the BP/M pin direct but do not see much difference between properly regulating units and units that go into OV shutdown.

Attachment Size
PSU test 3 BP pin_211215_155525.pdf 15.97 MB
Schematic_7.png 109.2 KB
Submitted by PI-Wrench on 12/15/2021

Keep in mind that this OV protection scheme will never be as precise as one that directly monitors the output on the secondary side, and it will be dependent on manufacturing variables, particularly transformer construction . This will be mostly due to differences in coupling between the bias and output windings. The OV protection shown in this particular design example is different than what is usually presented in our other design examples, where the OV protection is connected at the rectified and filtered bias supply rather than ahead of the bias rectifier as in this particular design example. This may have been an attempt to get around the variation of bias voltage vs. output load by looking directly at the voltage  from the bias winding, which in a perfect world would be a reflected replica of the output voltage scaled by the difference in turns between the two windings.. 

Submitted by Ninjabean on 12/15/2021

Understood, can you recommend a reference design that makes use of OV protection connected at the rectified and filtered bias supply?