RDK-131 Load vs No-Load Audible Noise
I've built a prototype RDK-131 circuit (without the AC rectifier circuit because we are designing it to be powered from 110Vdc) using the same or equivalent components, for powering LEDs.
When the circuit is fully loaded (300mA), there is a very quiet humming that is barely noticeable, but when the load is disconnected there is quite a lot of noise.
This may be a problem because we require the lights to flash, so when the lights are off, the circuit will have virtually no load.
I noticed that in another thread with a similar problem the inductor seemed to be "the culprit", and as far as I can tell, the noise on my board sounds like it is coming from the inductor, but it is a little hard to tell. Is this a common problem?
I'd also like to know if electrically disconnecting the load would be the best way to switch the LEDs on and off, or if it would be better to do it (somehow?) with the feedback pin on the LNK306 chip?
Thanks for any help.
댓글
Thanks for the reply and sorry for not replying sooner. I thought I would wait until I had built a smaller layout surface mount board before replying so I could see if that helped. Unfortunately the surface mount boards are noisy too. One design switches the LED load on and off, and another few designs have a constant 300mA load, but all of them have the noise.
The circuit I am using is the the same as the RDK-131 reference design circuit, except with a different inductor and equivalent transistors. RDK-131 reference design document (with circuit diagram):
RDR-131 DocumentThe transistors I am using are the On Semiconductor equivalent transistors:
Farnell MMBT3906WT1G
Farnell MMBT3904WT1G
I have tried two different inductors with the same problem. The first I used for prototyping was an Epcos 1000uH 0.33A inductor, found here:
Farnell Epcos B82464A4105K
Datasheet
The second type was a Coilcraft RFB0810-102L inductor, datasheet here:
Coilcraft RFB
Datasheet
Both circuits with different inductors have approximately the same amount of noise. However, I'm still not sure that the noise is coming from the inductor. It may be coming from the IC. It is hard to tell with all the components so close together.
Our PCB layout, while quite small, does not really attempt to minimise "the loop formed between the LinkSwitch-TN, inductor (L1), freewheeling diode (D1), and output capacitor (C2)" as stated in the LinkSwitch-TN datasheet. It was overlooked in the rush to get a PCB built. Could this contribute to some sort of noise problem? That's the only other thing I can think of.
Thanks.Thanks for that Zedman, I'm interested to see your layout.
Here is the basic circuit and PCB layout for our prototype design. Note that some of the ideas, such as minimising loop sizes weren't really considered for this design. It would be interesting if anyone could tell me if the loop size is large enough that you think it would cause problems.
I also saw that the copper connected to the source pins should be minimised within thermal constraints, but since our boards will be on aluminium, the copper area has been kept very low. Would this be a problem?
The circuit is being fed with +110Vdc.
Hi All,
I'm doing some measurements in the lab to confirm the source of noise (see attachment) and will report back later (hopefully today).
Cheers
PI-Chekov
Here are the results of measuring the audible noise of the RD-131 board.
With an LED load audible noise was very low, below 35 dBr A, our internal goal for audible noise (measured at 2 cm using the setup shown in the earlier posting). This is shown in figure 9 and 10.
Sweeping the load showed some conditions where the noise just exceeded this 35 dBr A level but it was not objectionable at a distance, basically inaudible at 1 m. This is shown in figure 7 and 8.
I'm now going to try to confirm that the inductor alone is the source.
Cheers
PI-Chekov
Good news
It appears that the bulk of the noise is generated by the output capacitor C3 but only at light loads (when driving a LED load at 300 mA the design is silent).
Forum - RD-131 Audible Noise (6).jpg shows the no-load noise level comparison between ceramic and electrolytic cap.
Forum - RD-131 Audible Noise (7).jpg shows the swept load noise level comparison between ceramic and electrolytic cap.
Forum - RD-131 Audible Noise (8).jpg shows that the inductor can change the audible noise generated.
Cheers
PI-Chekov
Please try changing the output capacitor (C3) to an electrolytic or tantalum type and see if that solves your noise problem.
I've posted comparative plots of noise levels to illustrate towards the top of this topic.
Cheers
PI-Chekov
Wow, thanks a lot for those tests!
I'll see if changing the capacitor does anything, but if you say that the noise only appears at light load, then I'm not sure if it will completely solve the problem.
A few days ago I built the circuit in Figure 6 on page 11 of the Application Notes AN-37, "High-Side Buck-Boost Constant Current Output
Configuration" (including the optional feedback diodes). This circuit seems to have much less noise at full load, and no noise at all at low load, even though the circuit is just a simple prototype built on veroboard.
The design uses electrolytic capacitors (and a single ceramic for the BP capacitor), so maybe that's part of the reason.
I plan to re-design the original circuit's layout and have both topology circuits made on PCB in a week or so, and I'll post results once I've built them.
We've done some testing of various components, and it appears that the noise is coming from two components.
Firstly, the output capacitor (C3, 2.2uF) is the noisiest component, but only when it is soldered to the board; its vibrations are amplified by the PCB when it is soldered on. The board is operating at relatively high current (almost the maximum of the LNK306), so it seems to be behaving differently to your audible noise tests. Removing it from the PCB and connecting it via wires makes it silent. We've tried the circuit without the capacitor connected at all, and the voltage ripple is fine for our application, so we are thinking about leaving it off completely.
Is there any reason we should not do this? We also plan to test several other capacitor types and values, so we shall see if that helps.
Secondly, we did the same thing with the inductor, removing it and connecting it with wire, and some noise is still there (in the inductor, not the rest of the circuit), but it is much more reduced. When the inductor is placed against the board, again, the noise is amplified. If the inductor is pressed hard against the board, the noise disappears, so it appears that the inductor is vibrating the PCB (either vibrating against the PCB, or shaking the whole PCB and making the noise). We've tried several different inductor brands, and all the ones we've tested have this noise.
With the inductor we're looking to:
1) test with more different types/brands
2) test with different values of inductance, and
3) change something mechanical about the PCB/inductor/mounting (not exactly sure what would work).
Any more suggestions would be appreciated.
Hello rowand
Please post your schematic and ideally the datasheet of the inductor you are using and we'll take a look on both the noise and on/off questions.
You will need always need a pre-load on the output (>= 3mA) to maintain regulation - see post for explanation:
http://www.powerint.com/en/forum/ask-pi-engineer/lnk306-failed-open-loaded-problem
Regarding audible noise, no before the forum posts I had not heard of audible noise using discrete inductors and LinkSwitch-TN. I'm intrigued because maybe there are some inductors out there that are just not suitable for this application. It looks like the other user (zedman) with similar issues solved the issue during a redesign but the reason is not clear.
http://www.powerint.com/en/forum/ask-pi-engineer/lnk304-audible-noise
Cheers
PI-Chekov