Voltage Control Feedback

[williamjtadeu]

Hello,

Our DCDC Design has passed every test before applying different loads on the output, however when the fan is turned on, just the load of the fan causes a output voltage drop of 6-10V down so that the output is 14-18V, and when the fan is turned off the output voltage goes to 28-35V depending on the input high voltage. This suggests after the load is removed the control loop stops working and a fixed gain is applied. We also noticed the fan could not be plugged in on start up as this would lead to the fixed gain scenario on startup.

The minimum Voltage we operate in our system is 400V, so as per build guide instructions, RV2 was configured to just reach 24V output at 380V input. The min switching frequency in this scenario of 380V is 124kHz, which does not align with your reccomened 95kHz :/

Any suggestions on how to procede?

[Rootthecause]

Hi!
First of all, I need to ask (because you wrote "our DCDC Design"): Did you change anything from the original design or do you mean just your own build (replica) of the original design?
If you've changed anything, please provide your files - otherwise I can't really help.

From what you've described I assume an issue in the control loop.
Have you checked the build notes at 260324?
It might be, that U7 is the wrong type from the old PCBA BOM.

Regarding the recommended 95 kHz: That's for Vin(min) = 200 V.
At 400V the switching frequency without load was 123.8 kHz in my case. Due to tolerances your 124 KHz at 380 V is very reasonable :) 
Does the frequency stay the same when you connect a load?

Could you please explain what you trying to show with the picture?
I would not recommend connecting an Analog Discovery anywhere to the DCDC, because it's not insulated unless you use a USB-Insolator. And even then, the transient voltages on some nodes of the DCDC could be way higher than the ±25V Input range - Be cautious!

[williamjtadeu]

Hi

We kept the same design as yours and adapted the OV/UV to accomodate our TS voltage range.

(260324) we have the correct pinout footprint for U7 according to datasheet so that is not our case.

What we are probing is net "HS" out of the UCC25600 and yes we have insulation.

When the fan load is activated, at 460V input we see the output drop down to 16V and net HS has this strange oscillating behaviour which does not look correct:

[Rootthecause]

Okay, thanks for the clarification!

If your measured signal is indeed the signal on the HS pin, that would pretty unusual.
Where did you connect the negative Probe potential from the Analog discovery? Did you use the BNC Adaper or just some wires to hook it up to the DCDC? And where did you measure the HS signal? I would recommend connecting a 10x Probe to U12 Pin 8 (HS) with a ground spring on Pin 6 (GND). If you dont use a probe, it might be, that the input capacitance of the AD3 and the wire inductance create a resonant circuit and ring like shown here. If I see it correctly, the "shadow" behind the waveform shows the outliers (option "Acquire noise" enabled). This would indicate, that the peaks are even at 10 V.
I would highly recommend using a trustworthy scope for comparison.

Now after a second look I realised something else: Those pulses are at 170 Hz. But they should be in the range of 100 to 350 kHz.
If you zoom into the range of 5 µs/dif, there should be individual pulses visible. The oscillation you see in your previous answer might be entirely due to antialiasing.
This would shift the issue from measurement to the loop control.
Your loop control might overshoot the output voltage and then wait until it reaches the threshold again, causing this low frequency oscillation. Either some capacitors are not the right value (check C26, C29, C30, C32) or the power supply for the control loop is faulty. Can you confirm, that there are clean 9V (without ringing) between the U9 pin 1 and GND? Also please check +12V_sw as well. Maybe some component is drawing too much current on the 12V rail, leading to a whole bunch of other issues.

[williamjtadeu]

Hi, so we investigated the power supply of the control loop and identified that +12V_SW and consequently 9V was behaving abnormally when the fan was activated. We then noticed that you suggested to power the fan with either 12V or 3V7 and decided to change the power to 3V7, which fixed the problem !

[Rootthecause]

Hey, glad you found a fix!
But this is still weird. Which Fan where you using on the 12V rail?
One of the main reasons for the existence of the 9V rail is to avoid noise issues (such as those caused by fans) on the control loop. I had the same issue as you before, but it was much milder (the output remained at 24 V, but there was slight noise).

[williamjtadeu]

Hi, we used the fans (MakerHawk) you suggested on the build guide. We noticed that the 9V rail dropped to 2V-3V and the 12V rail drops to around 7V (when the fan is powered by 12V). We have validated that applying a much larger load (15W) on the 24V output does not cause this effect. What do you think is causing this as the only thing i think of was a vreg's component failure either U2, U3, or U4.

[Rootthecause]

we used the fans (MakerHawk) you suggested on the build guide

But this Fan is for max. 5V 😅 ... it worked at 12V ?!

As you've set it to the 3.7V rail now, is there still an issue with the output regulation? What is your output voltage now?

[williamjtadeu]

As you've set it to the 3.7V rail now, is there still an issue with the output regulation? What is your output voltage now?

It now fully works with 3V7 yes

But this Fan is for max. 5V 😅 ... it worked at 12V ?!

Ah yes that makes sense oops

[Rootthecause]

:D I'm really curious now - Could you power it again with a 12V supply and tell me how much current it draws? It probably spins like crazy 😅

[williamjtadeu]

We discovered that 400mA is the limit before it dies XD

Here are some of our experiments:

IMG_1543.mov

IMG_1509.mov

[Rootthecause]

Whoaaa 😅
Sad, it's without sound.

Yeah, seems a bit above 400 mA @ 12 V. So 4.8 W is definetely more than the recommended max. 2 W Fan load for the DCDCv9-3 from my datasheet :D

[Rootthecause]

Will close this topic as resolved :)