Development of 2xMOPP galvanic isolator for USB-connected device

sergs

Registered
Hello, dear Experts!

Currently I am developing educational system for pupils\students that can reads different biosignals such as: EMG, ECG, PPG, EEG. It is not medical system, only for visualisation signals for educational needs.

I have a question about connection of my system to Personal Computer (PC). Due to some parts of sensors have direct electrical connection to the human body, I must provide corresponding level of safety: 2xMOPP (corresponding to IEC 60601).

Configuration of my system is following: there is a special module (Hub), that connects to PC throw USB-cable. Galvanic isolation of this module from PC consists of two components:
1) ADUM4160BRIZ for isolated USB-communication between my Hub and PC. It provides 2xMOPP by IEC60-601 (as I right understand, at voltage grater than 250 Vrms)

2) REM1-0505S for power supply of my Hub and sensors from PC. It provides 2xMOPP at 250 Vrms by IEC60-601 .

All my sensors (ECG, EMG,...) has direct electrical connection to Hub (no additional galvanic isolation in this modules) and connected directly to human body.

By attached files I am sending You photos of system (Hub with connected ECG-module) and Hub.

Is this configuration enough to provide safe utilization of my system, during it connected to PC? Can this system brings to user elecric shock from PC?


Also, I have a question concerning selection of the case of the hub and connectors of the biosignal sensor to the hub device.
Now 3.5 mm stereo jack connectors are used for connection between sensors and hub, but as I was told it must be replaced with more specific connnectors (see "First variant" and "Second Variant" pictures in attached files).

Is it right connectors? And should I replace female miniUSB to female USB Type B in order to prevent user touching the open metal parts?

Thank You in advance!
 

Attachments

  • Hub.png
    Hub.png
    129.8 KB · Views: 276
  • system.png
    system.png
    157.1 KB · Views: 257
  • First variant.jpg
    First variant.jpg
    12.5 KB · Views: 242
  • Second variant.jpg
    Second variant.jpg
    8.9 KB · Views: 215
M

MedMartin

Hi Sergs,

the configuration provides enough isolation from the PC with 2 MOPP for 250 VAC.
One further important consideration in addition to the isolation are leakage currents as defined by your applied parts. As long as only the parts you mentioned are used and there are no capacitors over the isolation distance the leakage currents for 50 Hz are well within limits (probably applied part class BF?).

The idea behind special connectors is that not any house hold connector (like e.g. a 3,5 mm stereo jack) is plugged into your device. This is especially important if you have only one applied part and several connectors.
The new connectors should be fine. Those are used on many ECG devices.

The USB connector may be touched as it is isolated by 2 MOPP for 250 VAC.

Do you want to certify the device?

Best regards
Martin
 

Pads38

Moderator
Until Peter can post I shall try a few comments.

Component Choice
Both the isolator and the power supply module are reasonable choices and could provide the required separation.
But you need to be very careful over things like board layout. This is to ensure that the required creepage and clearance distances are maintained and not bridged.
From your photos I can see a clear zone, particularly by the digital isolator. But it is less clear by the power supply. I would suggest that you aim for creepage and clearance of 8mm and 5mm respectively (that is for a working voltage of 250V RMS and 2 x MOPP). This is probably excessive - but better safe than sorry!
One thing I noticed was that you have a wire link bridging a component (L6?). This appears to be intruding into the 'separation zone' so be careful there.
There also seems to be mounting screw points within the separation zone, these also could affect things.

Connector Choice
The advice to use other than a 3.5mm jack is sound and the grey plastic type in your photo would be a sensible choice. And it would be good to use these throughout your system and not just on the hub.
What you are trying to satisfy here is clause 8.5.2.3 of 60601-1.
Basically, it MUST NOT be possible, in any and every way, for the end of a patient lead to be able to be inserted into any mains connector. (Yes - it has happened). A 3.5mm jack could be forced into the socket of a IEC60320 lead so is not suitable.

For the USB connector choice - you need to prevent your 'patient' from being able to touch any part where touch currents may exceed 100uA - which might include the metal shell of a USB connector. In clinical settings the 'patient' environment is regarded as being a)the patient bed/couch and b) the space surrounding it to a distance of 1.6m (in 3 dimensions). If your computer equipment (including your 'hub') is outside this zone then the chance of patient contact is considered low and is acceptable.
But in your situation it is quite likely that the 'paitent / student' will also be the Operator so this distance is not going to be kept. I would look to add insulating material around the mini-USB on the hub and provide a locking mechanism for the lead so that no contact is possible.
And I think that best practice, in an educational setting, would be to replicate a clinic setting where one student would be the patient and another would act as the Operator. Good practice would be that the patient does not touch the computer and hub and the Operator does not touch the patient whilst also touching the computer and hub.
 

Peter Selvey

Leader
Super Moderator
Pad38 has got all the highlights!

For full 601 compliance, in the first photo it looks like under the text ADUM4160 might be creepage/clearance issue with the PCB traces as well.

While it's good to comply with the standard, it's also good to keep in mind that the requirements in 601-1 are based on several worst case assumptions including for example long term use, unconscious patient, small contact area etc. These may not apply in this case. And to be honest strictly speaking the standard itself does not apply in this situation (there is no patient).

For this environment, my thoughts would be:

- galvanic isolation of at least 1 MOP for 250V between the electrodes (student) and the PC is a must;
- galvanic isolation between the electrodes and other accessible parts should exist but not necessarily meeting 1 MOP, just don't connect the student directly to bare accessible metal parts
- electrode connectors should be female type or shrouded (insulated against direct contact)

That said, explaining how the system is designed to fully meet IEC 60601-1 could also be a great part of the educational experience, even if it is overkill for the student environment.
 

sergs

Registered
Dear @MedMartin, @Pads38 and @Peter Selvey! Firstly, I would like to give a thanks cause You spend a time to help me!
Especially thanks to @Marc for managing of our discussion!

@MedMartin:
One further important consideration in addition to the isolation are leakage currents as defined by your applied parts. As long as only the parts you mentioned are used and there are no capacitors over the isolation distance the leakage currents for 50 Hz are well within limits (probably applied part class BF?).
Yes, applied part is BF-class only.

I have some questions about leakage currents:
1) Isolation capacitance of DC-DC-converter (REM1-0505S) is 25 pF: it is high enough cause it can lead to high 50 Hz noise. I think that decreasing isolation capacitance can lead to decreasing of 50 Hz noise.
How can I decrease It?
(I think, that no way, cause It is intrinsic property of DC-DC: I can only supress 50 Hz in power supply nets from the DC-DC output).
I would like to note, that each sensor, connected to the hub, has its own built-in LDO-regulator (from 5 to 3.3 V) with all needed filter-capacitors. Additionally I made active referense electrode to compensate 50 Hz noise (for EMG, EEG, ECG).
At most cases It is enough, but sometimes I have troubles with noisy signals (when hub connected to PC with noisy power supply).
Assume, that noise goes through power supply nets (we do not get in consideration EMC-disturbance from such power supply).

2) How can I make a proper current leakage limiting on a body?
- In case of connection only one sensor to hub I should to limit current flowing into human body. Am I right understand, that for EMG, ECG, EEG It shouldn't exceed 100 uA? Now I use current limiting resistors on each line to human body.
- The same question, but in case of simultaneous connection a several of different sensors. Currently, each sensor's input of the hub has no galvanic isolation between each other. Should I provide It?

3)
The USB connector may be touched as it is isolated by 2 MOPP for 250 VAC.
In my case USB connector has direct electric contact with PC (it is placed on non-galvanic isolated part of PCB), so I should to prevent touching It by "patient".

4)
Do you want to certify the device?
It is very complicated question... I think, as a minimum, this devise will be tested to correspondence of electrical safety by 60-601.

@Pads38:
But you need to be very careful over things like board layout.
Absolutly agree with Yours comments, layout will be changed. I want to post a picture with a new layout, and I would be very happy to have a feedback on it.

There also seems to be mounting screw points within the separation zone, these also could affect things.
Can I leave theese screws If I displace them from separation zone? What is the distance should be between screw and the closest conduction part? (Is >6 mm enough?)

And I think that best practice, in an educational setting, would be to replicate a clinic setting where one student would be the patient and another would act as the Operator.
Very sound Idea! Of course, there are situations when "paitent" simultaneously acts as an Operator. But It can be non-standart application, not including in the educational course.

@Peter Selvey:
- galvanic isolation of at least 1 MOP for 250V between the electrodes (student) and the PC is a must
Dear Peter, are You mean 1xMOPP? In general case I didn't know if the power supply of PC has corresponding electrical safety certification, so I think we should lead to worst case (PC itself doesn't provide additional 1xMOPP) and use own DC-DC with 2xMOPP?

- galvanic isolation between the electrodes and other accessible parts should exist but not necessarily meeting 1 MOP, just don't connect the student directly to bare accessible metal parts
Is I right understand, this recommendation connected to my question 2) for MedMartin about galvanic isolation (see above)? To be honest, I don't want to make 4 different isolations for each input (it is expencive). As an applicable compromise, I can isolate only GSR-sensor (cause it breaks signals from the other sensors because of applying potential differ from EMG, ECG). Is this very bad from safety point of view?

- electrode connectors should be female type or shrouded (insulated against direct contact)
Finally, I decided to choose plastic grey connectors as I posted in my first message.

That said, explaining how the system is designed to fully meet IEC 60601-1 could also be a great part of the educational experience...
Very good Idea! I will take it into account.

Dear experts, additionally I would like to ask You about the necessity of using 1 GOm resistor (with enough creepage distance) between grounds of galvanic isolated and non-isolated circuits of the Hub. Professional in medical devices told me that this is recommended part of desigh real ECG (to prevent static discharge, especially against defibrillator).
How this part necessary to use in my case?

Thank You in advance!
 
M

MedMartin

Hi Sergs,

to 1)
Actually 25 pF is pretty low. The leakage currents can be estimated with I_max = ω x C x f_max x V_max which results in about 2,4 µA if I got it right (with 60 Hz, 250 VAC). Lower must not be better, this depends on your circuit and what kind of noise you have. The leakage currents are measured with a frequency dependent measuring device.
Sometimes noise is avoided with higher capacitance over the isolation barrier. This depends on your circuits and noise sources. You have to measure and tweak it, if necessary.

to 2)
The AC patient leakage current for BF is 100 µA in normal case and 500 µA in case of a first failure.
The normal case can easily be measured. In case of first failure you may have to perform analytical worst-case evaluation, maybe simulation and measurements in the error case. If the different sensors are not isolated they belong to one applied part, that is fine for 60601-1 and has to be observed for the patient leakage current measurements.

to 3)
Yes, would probably be better but may also be ok in your case.

to 4)
Ok, that is fine.

Best regards
Martin
 

eldercosta

Involved In Discussions
@sergs,

With 25pF, the leakage current between applied part would be around 2uA@50Hz for a 250Vac test voltage. Usually the isolation resistance of these modules start at one GigaOhms so you may consider leakage is caused only by the capacitive reactance . It takes around 125pF to reach the 10uA limit of CF and that is a relatively high capacitance. Of course, you have parasitics of the PCB and electronics too, not only the isolators but still you would have a very safe margin even if your requirements were CF. You do not need to add resistors high value resistors to the inputs for safety. I have designed ECG amplifiers and usually a few tens of kOhms plus a few hundreds pF to the floating GND were added to filter RF.

About the 50Hz noise, one way of doing it is reducing the capacitance between floating and "real" GND (by "real" GND I mean your secondary circuit, a PC in your case, your desk and everything else in the environment including your body. However this is neither practical, as you probably know (physical limits) nor effective. The poor CMMR you are observing is caused by imbalances on the parasitic capacitances between your signal inputs and the aforementioned "real" GND. If you consider a 120V mains and do some math, you will see that even a very tiny fraction of differencial parasitic capacitance between two inputs will cause a difference that degrades CMRR.

You have a few ways to improve that:
1- Use a reference electrode connected to the floating GND. Probably you are already doing that in your ECG module
2- Shield your ECG module. The shield must be connected to the floating GND inside the ECG module. What the shielding does is reducing the effect of those parasitic capacitance difference. Even if the capacitance between your floating ECG circuit and the shield is higher, the voltage drop is minimum.
3- Add an active reference. Do a search with "Right Leg drive" words.

One link I found is this one. I did a cursory read but from what I saw it covers the three topics above a little deeper. It might be also worth taking a look at application notes of Analog Devices or TI ECG front ends even if your solution is discrete: they could provide some useful insights.

I hope this helps.
 

sergs

Registered
Dear Experts,
as I told above, I would like to ask You to to criticize my PCB's traceroute for Hub.
1) Pic1 - here You can see traceroute for schematic, directly connected to PC through USB-B.
PC Galvanic Isol.png
U7 - ADUM4160BRIZ,
U5 - Rem1-0505S,
R12 - high voltage 1 GOhm resistor (MHR0417SA108J20)
There I pointed the clearance distance between the closest conductors of isolated\non-isolated parts of PCB. Additionally was made rectangular cutting (but I think this is useless).

Pic2 - Here You can see the same picture in 3D:
PC Galvanic Isol 3D.png

2) Pic3 - here the traceroute for 4 chanel (for sensors connection (EMG, ECG, ...)).
Chanel Galvanic Isol.png
Isolation for each chanel consists of ADUM1201 and B0505S-1WR2. This components provide only galvanic isolation each chanel from the others, so I didn't bother with the choice about its MOPP, isolation voltages, etc. Additionally, were laid footprints for 1 GOhm resistors (R46-49), but I think it is overkill to install them.

And the general view of the PCB:
General View.png General View 3D.png

Questions:
Is there any mistakes in the traceroute for Pic1?
Do I keep the necessary footprint clearance distance (mm) for Pic1?
Is it needed to make rectangular cutting in the PCB for Pic1?
Is there any mistakes in the traceroute for Pic3?

Thank You!
 
Top Bottom