PCB component noise reduction for mri application, help needed

Thread Starter

drew_holm

Joined Sep 26, 2019
6
Hi all,

We are using an off the shelf pcb for application in an mri. We are shielding the device using copper mesh which is removing a large part of the 64 mg hz noise that is picked up by the mri (and displayed as discrete image artifacts) but we still have a small amount of noise, again showing up as discretes. Our next step is to start looking at the components of the pcb. We do have the OEM schematics. We would like to talk to/hire someone that has had experience in this area to assist us that this point. Or possibly get some posts as to what direction we might go to mitigate the noise yet keep the pcb functioning. Thanks in advance for your help.

Steve/SI
 

Picbuster

Joined Dec 2, 2013
982
Hi all,

We are using an off the shelf pcb for application in an mri. We are shielding the device using copper mesh which is removing a large part of the 64 mg hz noise that is picked up by the mri (and displayed as discrete image artifacts) but we still have a small amount of noise, again showing up as discretes. Our next step is to start looking at the components of the pcb. We do have the OEM schematics. We would like to talk to/hire someone that has had experience in this area to assist us that this point. Or possibly get some posts as to what direction we might go to mitigate the noise yet keep the pcb functioning. Thanks in advance for your help.

Steve/SI
It's not easy!

The most difficult part is to measure the strength interval and frequency mandatory to find a solution.

Trial and error method:
First thing to look at are the electrolytic capacitors and coils if used.
PCB tracks act as an antenna/coil/ capacitor or combination.
Wire wound resistors.
Capacitor sound strange however two layers with isolation in between is a capacitor with a potential difference is between the two.
This field could be disturb by a magnetic field hence noise.
(Electrolytic) capacitors same problem effect can be reduced by using tantalum. ( not always possible)

All cables & wires: avoid ground current loops and loops in wires use shielded twisted pairs. ( including (MPU) power lines)
Device shielding with copper will work on higher frequencies.
In a magnetic environment special metal (mu metal) should be used and as far away from your MRI as possible.
Again, measure is the only way to produce a solid answer.


Picbuster
 

Thread Starter

drew_holm

Joined Sep 26, 2019
6
It's not easy!

The most difficult part is to measure the strength interval and frequency mandatory to find a solution.

Trial and error method:
First thing to look at are the electrolytic capacitors and coils if used.
PCB tracks act as an antenna/coil/ capacitor or combination.
Wire wound resistors.
Capacitor sound strange however two layers with isolation in between is a capacitor with a potential difference is between the two.
This field could be disturb by a magnetic field hence noise.
(Electrolytic) capacitors same problem effect can be reduced by using tantalum. ( not always possible)

All cables & wires: avoid ground current loops and loops in wires use shielded twisted pairs. ( including (MPU) power lines)
Device shielding with copper will work on higher frequencies.
In a magnetic environment special metal (mu metal) should be used and as far away from your MRI as possible.
Again, measure is the only way to produce a solid answer.


Picbuster
Thanks Picbuster,

you have given us something to start working on, thank you for that! We would like to find someone with some experience but we are not sure what specialty we would search for (and where)...Might this be an rf design ee or someone that works on cell phone pcb design, maybe? Sorry to bother you with this but any help appreciated. Steve
 

Thread Starter

drew_holm

Joined Sep 26, 2019
6
thanks Picbuster for the above response, I have posted this on many forums and you are the only one to offer advice so thanks again!. We are located in the US, in a city with a lot of military (and support businesses) and cell phone software/hardware companies, most notably Qualcomm. I would assume both of these industries would have similar needs and there might be someone local that can help us. What do you think? and what particular jargon might we use when placing a help wanted ad? I assume we would want an EE but, again, not sure what jargon would help us in our search efforts, again any help is extremely appreciated.

ty!
 

Ylli

Joined Nov 13, 2015
766
The MRI is picking up the radiation from your added board or the added board is being affected by the MRI transmit? I'll assume the former.

Shortly before I retired, we were developing a switching power supply to feed our equipment cabinet. We had switching harmonics well into the 3T frequency range (128 MHz). Between the harmonics of the power supply switching frequency and the receiver LO, we were getting image artifacts that we traced to as high as 100+th order.

Solution was to carefully select the power supply switching frequency such that all the interference fell outside of the imaging bandwidth.

Bottom line is that the simplest solution would be to change the clock frequency of your added board so that the harmonics do not fall inside your nominal receiver bandwidth.
 

Thread Starter

drew_holm

Joined Sep 26, 2019
6
The MRI is picking up the radiation from your added board or the added board is being affected by the MRI transmit? I'll assume the former.

Shortly before I retired, we were developing a switching power supply to feed our equipment cabinet. We had switching harmonics well into the 3T frequency range (128 MHz). Between the harmonics of the power supply switching frequency and the receiver LO, we were getting image artifacts that we traced to as high as 100+th order.

Solution was to carefully select the power supply switching frequency such that all the interference fell outside of the imaging bandwidth.

Bottom line is that the simplest solution would be to change the clock frequency of your added board so that the harmonics do not fall inside your nominal receiver bandwidth.
 

Thread Starter

drew_holm

Joined Sep 26, 2019
6
Thanks Ylli, wow you worked in this area, we are very fortunate that you replied as this is a very quite niche area!

I have mentioned the switching issue you encountered to our design team and they said that we are using batteries to power the pcb/device and every thing is running on DC, so hopefully there are no switching power supplies are involved. This is a small portable device that will be brought into the mri bore, btw. However do you think there might be some other clocks to worry about? This is off the shelf device we are trying to modify and the OEM is working with us but not sure if can modify a clock speed but we could shield and/or replace components with less noisy components if we can identify them. We are taking noise readings in our faraday cage then making shielding adjustments. We felt that we had eliminated all the noise but we still are getting image artifacts when we tried the device in an mri. I am not sure if our noise measurements are not being done well via our analyzers or and we not using our noise analyzers correctly. A main goal of my posting here is to find help, I know you are retired but might you know anyone that we could contact (maybe you?) and possibly hire to assist us?

thanks again!
 

Ylli

Joined Nov 13, 2015
766
The technology has advanced since I retired, and EMI was really not my area of work. But I do know that an MRI system finds signals well below thermal noise and if there is any interference to find the system will find it. Again, if it is a digital device that uses a clock, then best bet would be to move the clock frequency. If it is just digital, then configuring it so that harmonics of any signal transitions fall outside the imaging bandwidth. Beyond that, it is hard to make any detailed suggestions without knowing what the device is and having some hands-on experience with it. Your challenge here is a big one, hope you find an effective solution.
 

Picbuster

Joined Dec 2, 2013
982
thanks Picbuster for the above response, I have posted this on many forums and you are the only one to offer advice so thanks again!. We are located in the US, in a city with a lot of military (and support businesses) and cell phone software/hardware companies, most notably Qualcomm. I would assume both of these industries would have similar needs and there might be someone local that can help us. What do you think? and what particular jargon might we use when placing a help wanted ad? I assume we would want an EE but, again, not sure what jargon would help us in our search efforts, again any help is extremely appreciated.

ty!
For the US;
this might help;

Picbuster
 

Thread Starter

drew_holm

Joined Sep 26, 2019
6
The technology has advanced since I retired, and EMI was really not my area of work. But I do know that an MRI system finds signals well below thermal noise and if there is any interference to find the system will find it. Again, if it is a digital device that uses a clock, then best bet would be to move the clock frequency. If it is just digital, then configuring it so that harmonics of any signal transitions fall outside the imaging bandwidth. Beyond that, it is hard to make any detailed suggestions without knowing what the device is and having some hands-on experience with it. Your challenge here is a big one, hope you find an effective solution.
Thanks Ylli, in your reply you mentioned "and EMI was really not my area of work." Should we be looking for someone who is well versed in EMI reduction on pcb's? One of our challenges is having the correct industry jargon to do a search for some help. We believe we need an electrical engineer and there are lots of defense contractors and cell phone companies in our city but we are not sure what jargon to use when searching, any help appreciated. And thanks for your previous reply...ty!
 

JMW

Joined Nov 21, 2011
122
Hi all,

We are using an off the shelf pcb for application in an mri. We are shielding the device using copper mesh which is removing a large part of the 64 mg hz noise that is picked up by the mri (and displayed as discrete image artifacts) but we still have a small amount of noise, again showing up as discretes. Our next step is to start looking at the components of the pcb. We do have the OEM schematics. We would like to talk to/hire someone that has had experience in this area to assist us that this point. Or possibly get some posts as to what direction we might go to mitigate the noise yet keep the pcb functioning. Thanks in advance for your help.

Steve/SI
Remember one thing, every component you use must be certified by the manufacturer for use in medical equipment. This includes non active components such as your PCB. I would contact the OEM, and advise them of the situation you are trying to remedy
 
Top