Battery Powered ECG Holter Dielectric Test

[bardia_barai]

Hi everyone,
We've developed a battery powered body-worn ECG Holter. It uses an internal single-cell LiPo battery which can be recharge when it is not connected to the patient's body. We sent the device for IEC60601-1 certification. The lab did this experiment for the dielectric insulation test (please check the attached image):
1- They opened the enclosure and disconnected the internal LiPo battery to it's PCB.
2- The negative (GND) and the positive pins of the battery on the circuit were shorted.
3- They give high voltage between this shorted pins and one of the ECG electrodes (the device is single lead and it has 2 electrodes)
4- They rise the high voltage to 1kV and it fails around 400V (the current rises suddenly more than allowed)

Is it mandatory for this device to pass such a test?

 

[Peter Selvey]

Short answer: no.

Dielectric strength is an ageing test of solid insulation (e.g. plastic on wiring, sleeving, terminal blocks, interleaving insulation between windings in a transformer interwinding layers, transformer bobbins, or in components such as optocouplers, EMC capacitors, PCBs). That's all. The theory behind the test is to prove that the solid insulation can survive say 15 or 20 years under normal voltage stress without actually waiting 15 or 20 years. So 1min at 1kV might be the same as 20 years at 50V. It's an exponential relationship, derived many decades ago.

Also, insulation tests should be performed according to an isolation plan, from which various components and spacings can be identified that form the safety insulation. In some cases, it can be possible to perform a dielectric strength on a fully assembled device or sub-assembly, but it has to be done with care to make sure that (a) all the solid insulation that needs to be tested is in fact tested, and (b) that parts not intended to provide safety insulation are not stressed.

For example, if your design used a small isolated circuit for the ECG with three components that bridge the insulation barrier: a small dc/dc converter, and optocoupler and Y2 capacitor. Then you could do a 1kV test between the ECG circuit and the non-isolated circuit which should stress all the solid insulation in the three components and not damage any other components.

But you would only do this after inspecting the design, and making sure there is a clear isolation barrier, and noting the components bridging the barrier are designed for 1kV and that the spacings (creepage, clearance) meet the requirements.

You would not just blindly attach a high voltage test to two points in a circuit and wind up the voltage! That's professional negligence.

Now, the issue on the manufacturer side is: is isolation needed (possibly not for body worn equipment). If yes, what is the isolation plan? Did your company provide this information to the lab? Was the insulation plan discussed and agreed before testing?

 

[bardia_barai]

Thank you so much for your answer. No the isolation is not needed. We put 50kohm resistors (for both electrodes) to make sure about the current leakage. The voltage of the battery will be less than 4.2V (between GND and positive pin of the battery), and there are some step-down converters for ECG AFE supply. So, we do not think that we need isolation. The lab insists that you need to pass the previously mentioned test. Do you think that ECG patch electrodes need this kind of isolation?

 

[Peter Selvey]

The ECG should be designed to limit the "patient auxiliary currents" in normal and fault condition through circuit design e.g. series resistors such as the 50kΩ. From experience, 50k could be a bit low but I'd need to inspect the circuit e.g. these days it's typical to have ±1.5V supplies for the front end but the chip might also have 3V somewhere, so 3V/50k = 60µA, but if the device has only two leads and no right leg drive it might be OK, or the right leg drive often has 100k or 470k. Anyway, it's not complicated.

If the device is body worn there should be no isolation required.

The test lab may struggle with this, but it's a matter of consistency and logic. A typical patient monitor will have an isolation barrier with a dc/dc converter and then in the floating circuit there will be rail voltages like 5V, ±3V and so on with respect to the GND in the isolated circuit. No reasonable test lab on the planet would require 1kV between the rail voltages in an isolated ECG circuit and the ECG electrodes. If it is agreed that 1kV isolation is not required there, then equally it should not be required here.

Requirements for dielectric strength and spacings (creepage and clearance) are intended to be applied between isolated circuits i.e. between mains and secondary, mains and earth, secondary to an isolated patient circuit. It is not intended to be applied within a circuit, and in particular within a electronic circuit that is operating at very low voltages such as 12V or less. There is nothing within the circuit that would justify huge test voltages (1kV) or huge distances like 3.4mm, they make no sense to be applied within an electronic circuit.

 

[bardia_barai]

I checked IEC60601-1 but I couldn't find any phrase to confirm "If the device is body worn there should be no isolation required". Can you address that?
By the way, the device is single-lead. So, it has 2 electrodes and both of them has a 50kohm resistor and we don't have right leg drive electrode. The supply voltages of the ECG AFE is 1.8V and GND and the Lead Bias Voltage is around 0.9V. So the maximum current will be 1.8V / (50kohm + 50kohm) = 18uA. Also, our AC to DC adaptor passed the test (220Vac to 5Vdc). The only problem is about the isolation between electrodes and the detached battery pins.

 

[Peter Selvey]

For a non-Holter ECG (e.g. a patient monitor), there would normally be several insulation barriers to be assessed:

A: mains to patient (2MOPP)
B: patient to earth (1 MOPP @ mains voltage, due to CF rating)
C: patient to secondary (1 MOPP @ mains voltage, due to CF rating)
D: patient to other parts (1 MOPP @ mains voltage, due to CF rating)

There are two other paths:
E: secondary to patient (2MOPP @ secondary voltage)
F: internal voltages inside ECG circuit (e.g. rail voltages) to patient (usually ignored)

Insulation E is theoretically applicable, but it's not really important for safety, it's really an artefact of the IEC 60601-1 approach which considers even 0.1Vdc as live and dangerous. Fortunately, Insulation C already covers the same area and has stronger requirements. So there's no need to argue any point for insulation E.

Insulation F is theoretically applicable but for a full patient monitor it would usually be ignored. Actually the theory is correct in that all parts that act for safety should be reliable, the problem is the the distances and test voltages in IEC 60601-1 are totally out of place. For example, a distance of just 0.2mm is fine for 1.8Vdc, instead of 3.4mm. And a dielectric strength of say 10V would be fine for 1.8V. So you could then select the two 50kΩ as being your "safety isolation barrier" between the 1.8V electronics and the patient, and make sure that the PCB traces on each side of the resistor are not more than 0.2mm and the resistor selected can handle 10V. Obviously it should be possible to do this with a lot of overkill. But going to the extreme 3.4mm and 1kV is just too much.

Why are the values in IEC 60601-1 so high? The answer is that they are intended fully isolated circuits. In this case, between isolated circuits you can still have voltage distributions and overvoltage transients from the mains that can justify the higher values. There's more to the story but there are cases where secondary circuits can need higher insulation levels than you would expect from the voltages in the circuit. But none of these situations exist within the circuit, as can't get significant overvoltage transients or residual mains voltages within a circuit. The circuit would break.

The thing is, with a patient monitor there lots of isolation barriers to assess (A, B, C, D) so the test lab is happy to close their eyes to insulation F, in fact, most test labs wouldn't even realised they are ignoring F. And this doesn't just apply to the ECG circuit, it applies to any components within a circuit that is there for a safety. In a typical power supply, for example, there are circuits for overcurrent, overvoltage and overtemperature protection that rely on electronic components, and these can also be shorted according to a strict reading of Clause 8.1 if they don't meet the isolation requirements in the standard. But nobody does this, it wouldn't make sense.

OK, so in general test labs usually overlook insulation requirements for components within a circuit. It's kind of natural.

For a Holter ECG that is fully worn on the patient, insulations A, B, C, D and E don't exist.

This leaves only F.

Suddenly, the test lab wakes up to F and decides to apply 1kV in place that is obviously not designed for 1kV.

 

[bardia_barai]

Thanks a lot. Our device is an ECG Holter. What do you suggest? We may need to change our PCB design and components, but Is there any clause which we can negotiate with the lab to ignore this test?

 

[bardia_barai]

Thanks a lot. Our device is an ECG Holter. What do you suggest? We may need to change our PCB design and components, but Is there any clause which we can negotiate with the lab to ignore this test?

By the way, ECG AFE is supplied from a Buck converter (from a single-cell LiPo battery to 1.8V). The lab reported that 1kV is necessary because of 8.5 and 8.8.3 clauses of IEC60601-1. I double-checked IEC60601-1 and IEC60601-2-47 (which is specifically for ambulatory ECG monitoring) and I didn't find a clear answer for my last questions (I replied them).

 

[Peter Selvey]

The first line of Clause 8.8.3 is:
"The dielectric strength of solid electrical insulation ..."

See also 8.8.3 (c) which discusses precautions that have to be taken if the test is applied as a black box test (fully assembled device). However, in principle, the test should be applied just to the solid insulation only. Also check the rational for 8.8.3, in particular the 3rd paragraph.

Modern test labs do not understand this, are not familiar with the theory or actually familiar with the text of the standard, in particular the reference to "solid" insulation. They typically apply the test as a black box and assume that it is a test intended for a fully assembled device. But it is actually just an ageing test for solid insulation, nothing more, nothing less.

I think it's necessary to see your or the test labs "isolation diagram" or isolation plan before discussing further. That should be the starting point. You don't just blindly apply voltages to a fully assembled device, regardless of what it is.