Ragooman
New Member
I'm a little rusty on noise problems from switching power supplies. And I hope someone can offer some advice. There's this new system which relies on a small +12v smps. This system has a separate controller board mated with a motor controller using board connectors. And the 12v smps connects directly to the controller board. There's a very large amount of power supply management and regulator hardware to control the voltage sequencing. And I'm getting what appears to be a significant amount of switching noise from this power supply. And this is prevalent on every power rail and all of the LVDS signaling. Also, we recently experienced a large rampant case of component failures. I'm not sure yet if the two are related.
I can see noise spikes on my o'scope at a periodic frequency. The voltage amplitude is approx 800mv to 1v. And the period is approx 15usec to 17usec, it's hard to get a definitive measurement since it bounces around. So this is roughly 60Khz, what amounts to the switching frequency of this power supply. I used a 2nd o'scope probe with the scope tip insulated to use as a inductive sensor. So then I can observe the switching pulse in the power supply as I hover the scope probe over the power supply. and I confirmed the frequency matched between the noise and the switching pulse.
This noise is rather large, as I mentioned above, approx 800mv to 1v. And this system makes use of many parts with LVDS signals. These signals are only about 300mv the most. The components which are failing the most are these new 16bit ADC's using LVDS signaling.
Could this be a problem for some of the LVDS parts and potentially damaged them ?
I checked the various LVDS signals in this system and noticed the noise spike is destroying the clock and data signals on every component. I would trigger the scope at the same levels of the noise spike, which is higher than the LVDs signal, to see when this occurs. I took many screenshot images to view how the signals are getting mangled by the noise spikes.
Since I can see this happening on the scope, would the noise spikes actually be interfering with these components ?
I also noticed the 12v smps is not isolated, the ground on the mains line is not isolated from the ground on the 12v output. I thought any industrial grade system would necessitate this isolation.
Wouldn't this be required to help alleviate the noise problem ?
Also, since the noise spikes are higher than the LVDS signaling, I think these LVDS parts have a limited voltage tolerance.
Wouldn't the noise spikes emanating from the power supply be considered excessively large ?
There's this other little check that somebody showed me about testing for noise from the smps. Basically place the scope probe on the same ground reference testpoint. And if you still see the noise on the ground reference testpoint, this supposedly means that the noise spikes are irrelevant and not the cause of the component failures. I tried this step using several diff ways. First with the o'scope on a ground-lift adapter from the mains as usual. Then with the o'scope grounded without using the ground-lift adapter on the mains. Then with another 0'scope that is isolated, both from the mains and the scope grounds, by using an internal battery.
Again, since I'm very rusty on these noise problems, I'm not sure how this applies, could someone elaborate more on this ?
thanks
Dan
I can see noise spikes on my o'scope at a periodic frequency. The voltage amplitude is approx 800mv to 1v. And the period is approx 15usec to 17usec, it's hard to get a definitive measurement since it bounces around. So this is roughly 60Khz, what amounts to the switching frequency of this power supply. I used a 2nd o'scope probe with the scope tip insulated to use as a inductive sensor. So then I can observe the switching pulse in the power supply as I hover the scope probe over the power supply. and I confirmed the frequency matched between the noise and the switching pulse.
This noise is rather large, as I mentioned above, approx 800mv to 1v. And this system makes use of many parts with LVDS signals. These signals are only about 300mv the most. The components which are failing the most are these new 16bit ADC's using LVDS signaling.
Could this be a problem for some of the LVDS parts and potentially damaged them ?
I checked the various LVDS signals in this system and noticed the noise spike is destroying the clock and data signals on every component. I would trigger the scope at the same levels of the noise spike, which is higher than the LVDs signal, to see when this occurs. I took many screenshot images to view how the signals are getting mangled by the noise spikes.
Since I can see this happening on the scope, would the noise spikes actually be interfering with these components ?
I also noticed the 12v smps is not isolated, the ground on the mains line is not isolated from the ground on the 12v output. I thought any industrial grade system would necessitate this isolation.
Wouldn't this be required to help alleviate the noise problem ?
Also, since the noise spikes are higher than the LVDS signaling, I think these LVDS parts have a limited voltage tolerance.
Wouldn't the noise spikes emanating from the power supply be considered excessively large ?
There's this other little check that somebody showed me about testing for noise from the smps. Basically place the scope probe on the same ground reference testpoint. And if you still see the noise on the ground reference testpoint, this supposedly means that the noise spikes are irrelevant and not the cause of the component failures. I tried this step using several diff ways. First with the o'scope on a ground-lift adapter from the mains as usual. Then with the o'scope grounded without using the ground-lift adapter on the mains. Then with another 0'scope that is isolated, both from the mains and the scope grounds, by using an internal battery.
Again, since I'm very rusty on these noise problems, I'm not sure how this applies, could someone elaborate more on this ?
thanks
Dan
Last edited:
