Can PTC Thermistors Oscillate?

[Galgso]

Context:
I was looking at the datasheet for some PTC thermistors (Vishay PTCEL Series, see attached datasheet) that I want to use for a circuit I'm designing to discharge high voltage capacitors (588v Max). The datasheet provides a handy formula for determining the necessary amount of thermistors to absorb a particular amount of energy from a capacitor (or capacitor network) and I have determined that a single thermistor will work well for my application (capacitance < 320uF). I am looking to use the PTCEL17R102UxE404 type with a resistance of 1kΩ at 25 degrees centigrade.

If you think it's odd that I am in charge of designing a safety-critical circuit like this with my limited experience, it is because this is an extracurricular project I am undertaking at university. The circuit shall be inspected by experienced technicians and tested thoroughly in their presence.

Issue:
My circuit involves a discharge contactor and I am required to design a circuit that will not blow up if the discharge contactor (single normally-closed contact) fails to disengage or gets stuck in the closed position. Since this contactor is required to be normally closed for safety, if for whatever reason power fails to reach the contactor coil the contactor will remain closed with 588VDC across the discharge thermistor in series with the contactor. While the datasheet states that the thermistors are "Self protecting in case of overload with no risk of over-heating" (see features section with bullet points on page 1 of datasheet), it seems to me that if 588VDC are applied continuously across the thermistor, it shall first remain in the low resistance state for a while and then increase in resistance significantly. I fear that this will cause a form of oscillation where the thermistor cools down after it heats up initially due to the increased resistance and then overheat again.

Looking at the resistance vs temperature graphs on page 4 of the datasheet, it seems clear to me that it is unlikely the thermistor will reach an equilibrium temperature that it can maintain indefinitely since it seems to be deliberatley designed to avoid this. The graphs are incredibly steep and I don't see how it could actually be self-protecting as the datasheet states unless it can actually handle repeated cycles of heating and cooling continuously.

As for the calculations I made to arrive at the conclusion that the thermistor will overheat and oscillate, one need only look at the dissipation factor (19.5mW per Kelvin or less). With 588v across the thermistor and a resistance of around 1kΩ at 25 degrees centigrade, it is quite obvious that it will heat up pretty quickly.

Summary:
Can a PTC thermistor heat up, increase in resistance and then cool down again leading to a sort of astable oscillation condition that may destroy the thermistor if it persists?

Thanks in advance for any help

edit: specified that I wish to use the PTCEL17R102UxE404 type with a resistance of 1kΩ at 25 degrees centigrade.

 

[Nigel Goodwin]

It will adjust to the voltage provided, and should settle down at a temperature dependent on that voltage - is the voltage changes then the thermistor will alter accordingly. However, with 588V across 1000 ohms you're dissipating 346 watts, I suspect it's unlikely to last very long at all.

If in doubt, get one and try it - simulations aren't real world anyway.

Generally you just use resistors for discharging capacitors, of a suitable value and wattage - usually in the 100's of kilo-ohms, and a couple of watts or so - check what what valve guitar amplifiers use

 

[schmitt trigger]

What I have seen with PTC thermistors they reach a thermal equilibrium point, and should stay there until the end of times.
Well, not really. At that point, the PTC will usually be running hot, quite hot. That excess heat will overheat the surrounding components and that is never good from a reliability point of view. Additionally, some PTC materials, polymers in particular, can and will degrade. I don’t recall the specific details whether this over time could cause a runaway condition. Ceramic PTCs are better but the hot/cold resistance ratio is not as large as the polymers.

Our studies were conducted back in the mid-1990s, the PTC technology may have improved now.

 

[Galgso]

Generally you just use resistors for discharging capacitors, of a suitable value and wattage - usually in the 100's of kilo-ohms, and a couple of watts or so - check what what valve guitar amplifiers use

Yes, my original design which has been tested and works originally used 10W wirewound resistors with a high momentary overload rating (100W for 10 seconds). That design worked fine but was not compliant with the requirements as it would not be able to handle high voltage permanently. The reason I can't simply use very high value resistors that would be able to handle the voltage permanently is due to the requirement that the discharge circuit bring the voltage across the capacitors (320uF as previously mentioned) down to less than 60v in less than five seconds. My only option if I were to use standard resistors would be to use a high-wattage resistor mounted to a metal chassis.

Also, as you can see from the graph below, the thermistor would dissipate three watts with a constant 600v across it which is unfortunately still far too much for it too handle. What's also funny is that if you try to put say two thermistors in series to increase the power handling capability, the current at 300v would be higher than the current at 600v and it would still be too much so you can't merely add more thermistors. If I'm not mistaken, putting thermistors in parallel would have no effect on the individual power dissipation if they are connected to a high voltage source with no series resistance (or negligible series resistance as will be the case in real life).

What I have seen with PTC thermistors they reach a thermal equilibrium point, and should stay there until the end of times.
Well, not really. At that point, the PTC will usually be running hot, quite hot. That excess heat will overheat the surrounding components and that is never good from a reliability point of view. Additionally, some PTC materials, polymers in particular, can and will degrade. I don’t recall the specific details whether this over time could cause a runaway condition. Ceramic PTCs are better but the hot/cold resistance ratio is not as large as the polymers.

Our studies were conducted back in the mid-1990s, the PTC technology may have improved now.

Yeah I guess it's not looking good for my hypothetical thermistor solution. I was hoping these PTCEL series thermistors would work and that I was missing something since the datasheet does seem to imply that they would be capable of handling an overload permanently but I guess not.

 

[Galgso]

For more context this is going to go on an electric formula car (IMechE Formula Student Car). This is why this needs to be failsafe and why it has the aforementioned five-second requirement for the discharge circuit. I wish to avoid having to mount a high-wattage resistor on the chassis as it is quite frankly a huge pain in the ass to make such a design compliant with high voltage isolation requirements. At the moment it seems that the only real way to make this work is to get a high power PTC thermistor with a metal case that can be mounted inside a metal high voltage enclosure on the car that will be fastened to the chassis. Or if any such component exists, a high wattage thermistor that can dissipate say 10W without heatsinking assuming a reasonably low ambient temperature.

Thanks for your replies so far. I'm open to any alternative solutions for this.

 

[Diver300]

I don't see why 3 W continuous would be a problem. The thermistors are self-limiting, so if the cooling isn't as good, there will be less power. You will be meeting the requirements to discharge quickly and to survive the 600 V permanently.

You could even thermally insulate the thermistor. When the normal discharge happens, the thermistor will start cold so the discharge will be almost as fast. If there is a fault, and the thermistor heats up, it will get hotter because of the the insulation so that it will take less current, but nothing else will be heated up much.

To get to 50 V from 600 V, you need about 2.5 times the time constant of an RC network, so you are looking for a time constant of about 2 seconds, and a resistance of 6250 Ohms with a 320 uF capacitor, if using a fixed resistor. That resistor would dissipate a bit under 60 W, and the amount of heat coming out of it will be the same if it is insulated or not. The 3 W or less will be a lot easier to deal with than the 60 W from a resistor.

With any thermistor discharge, the 5 seconds will not be met if the power had been applied continually beforehand, but I don't think that the rules are looking for 5 seconds to discharge in those conditions. If the rules did call for 5 seconds discharge after the power had been applied continually, it would be looking for protection against two faults in very quick succession.

The discharge to 60 V in 5 seconds is to prevent electric shock, so that is someone is unwise enough to touch a high-voltage connection without testing it first, things should be safe.

The requirement to avoid damage is so that discharge resistors aren't burned out. The problem about a discharge resistor burning out is that it could happen during a malfunction and never be noticed. That could lead to no discharging ever, even after the original malfunction had finished or had been repaired.

With a thermistor that continues to take 3 W, or 50 mA from 600 V, a 320 uF capacitor will still discharge in around 30 seconds, so it's still improving the safety. Nothing will be damaged, so once the malfunction is fixed, the thermistor will cool down and the discharge time will return to less than 3 seconds.

 

[rjenkinsgb]

My circuit involves a discharge contactor and I am required to design a circuit that will not blow up if the discharge contactor (single normally-closed contact) fails to disengage or gets stuck in the closed position.

I don't think the PTC thermistor concept will work as you expect, regardless of component ratings.
320uF at 600V means around 58 Joules stored energy to dissipate.


When switched in, the thermistor will heat up rapidly and switch to its high resistance state, and would then have to cool significantly again to get down to a low resistance value and cause a faster discharge - which will not happen within the required five seconds.

During a fault, while the capacitors are charged and the thermistor is continuously connected the thermistor will again be in it's heated state with a resistance roughly 120K Ohms (from your 5mA hot current).


The correct way to do this is use fixed resistors plus a contactor that also has a normally open contact, used to enable the charging side, or better still a multi-pole one that directly switches the incoming charging power as well as the discharge resistors.
(And use a guided type so NO and NC cannot both close in case of a fault).

If this is supposed to be a safety-grade circuit (to prevent injury), strictly speaking every part should be replicated with monitoring so a single failure does not prevent it operating and you cannot re-enable the charging side if any contactor is stuck/welded.

 

[Nigel Goodwin]

If fast discharge is important, then add an active fast discharge circuit to discharge the capacitor as the circuit is shut down, you could still include the thermistor as an additional safety backup, an active circuit will be able to discharge the capacitor a LOT faster.

 

[Diver300]

I don't think the PTC thermistor concept will work as you expect, regardless of component ratings.
320uF at 600V means around 58 Joules stored energy to dissipate.

The data sheet for the thermistor shows it having a thermal capacity of 2.3 J/K so it will only heat up by about 25 deg when absorbing the full energy of the capacitor. That seems fine to me.

It will have to dissipate that heat later, and I don't know how often the system will have to discharge the capacitors, which affects the average power dissipation needed.

 

[Galgso]

I don't see why 3 W continuous would be a problem. The thermistors are self-limiting, so if the cooling isn't as good, there will be less power. You will be meeting the requirements to discharge quickly and to survive the 600 V permanently.

I I'm not mistaken, given the dissipation factor of 14.5-19.5 mW per Kelvin (roughly 50 Kelvins per Watt), a power dissipation of 3W would result in a temperature rise of ~150 degrees centigrade bringing the temperature of the thermistor well above the maximum operating temperature of 105 degrees.

Although not explicitly stated in the rules, I do believe that you are correct in saying that the rules would not require a five second discharge time after a continuous fault. The rules call for a five-second discharge time for three consecutive cycles (assuming no faults occur) and state that the discharge time may then be exceeded within reason.

I don't think the PTC thermistor concept will work as you expect, regardless of component ratings.
320uF at 600V means around 58 Joules stored energy to dissipate.


When switched in, the thermistor will heat up rapidly and switch to its high resistance state, and would then have to cool significantly again to get down to a low resistance value and cause a faster discharge - which will not happen within the required five seconds.

During a fault, while the capacitors are charged and the thermistor is continuously connected the thermistor will again be in it's heated state with a resistance roughly 120K Ohms (from your 5mA hot current).


The correct way to do this is use fixed resistors plus a contactor that also has a normally open contact, used to enable the charging side, or better still a multi-pole one that directly switches the incoming charging power as well as the discharge resistors.
(And use a guided type so NO and NC cannot both close in case of a fault).

The thermistors are capable of absorbing 58J. It is the continuous power dissipation that is the issue in the event of a fault. The charging side (i.e. high voltage accumulator aka battery) is already switched by two isolation contactors, one on each pole of the battery. Both failing simultaneously would be very unlikely. I believe that realistically speaking the only fault (other than both isolation contactors getting welded shut) that could cause the discharge contactor to be closed while both contactors are also closed would be a broken wire to the discharge contactor coil causing it to close (is it NC). The circuit is already set up such that is is impossible (barring faults) for the discharge contactor to be closed if any other contactor is closed. I think your idea of a multi-throw contactor is great but high voltage high current DC contactors are expensive and we have a limited budget. Having to replace one of our contactors should be avoided unless there is no other solution.

If this is supposed to be a safety-grade circuit (to prevent injury), strictly speaking every part should be replicated with monitoring so a single failure does not prevent it operating and you cannot re-enable the charging side if any contactor is stuck/welded.

All the contactors connected to the battery have machanically linked auxiliary contacts which are being monitored by a seperate circuit so we will know if any of these are stuck/welded. The discharge relay does not have an auxiliary contact unfortunately but since the rules already specify that the circuit must be protected if power is applied indefinitely and we have a separate circuit that indicates the voltage on the vehicle side of the accumulator isolation contactors. The indicator circuit is all hardwired electronics (no microcontrollers/FPGAs) as per the rules and monitors all contactor coils and auxiliary contacts in addition to the DC bus voltage. It performs a plausibility check using all these inputs and will reliably indicate that the car is either safe to approach, dangerous (live) or that an error has occured so the circuit does not know. The power source is a battery so there is no way to disable it apart from opening the contactors.

If fast discharge is important, then add an active fast discharge circuit to discharge the capacitor as the circuit is shut down, you could still include the thermistor as an additional safety backup, an active circuit will be able to discharge the capacitor a LOT faster.

I believe the circuit is already active in the sense that it does not rely on small continuous leakage currents for discharge and instead uses deliberate switching of a relay for that. I imagine this is what you mean by active and not active in the sense of containing semiconductors. The issue is that the rules will not be satisfied with this alone (and as mentioned in my reply to rjenkinsgb, it is already set up such that the discharge relay cannot be closed if any other contactor is active) and require the discharge circuit to be able to handle the full system voltage indefinitely.

The data sheet for the thermistor shows it having a thermal capacity of 2.3 J/K so it will only heat up by about 25 deg when absorbing the full energy of the capacitor. That seems fine to me.

It will have to dissipate that heat later, and I don't know how often the system will have to discharge the capacitors, which affects the average power dissipation needed.

I am not worried about energy absorption in the absence of faults. I have already determined that these thermistors will work for that and should be able to handle the aforementioned three consecutive discharges in 15s (i.e. three 5-second discharge periods) required before the discharge time may be exceeded. The issue is that it appears they will melt/fail if the system voltage is applied continuously, and I can't find any PTC thermistors online that are made to dissipate a few watts continuously.

Is anyone aware of any PTC thermistors that can continuously dissipate a few watts? The only issue with the PTCEL series is that they appear to be incapable of this. Please correct me if I'm wrong.

 

[rjenkinsgb]

2.3 J/K

Doh.. I used 2.3C per joule

 

[rjenkinsgb]

I I'm not mistaken, given the dissipation factor of 14.5-19.5 mW per Kelvin (roughly 50 Kelvins per Watt), a power dissipation of 3W would result in a temperature rise of ~150 degrees centigrade bringing the temperature of the thermistor well above the maximum operating temperature of 105 degrees.

With the correct rating as Diver pointed out, the single discharge power is fine.

The continuous connection rating is not a problem - 105'C is the maximum ambient temperature not device temperature. (The resistance vs temperature graphs go up to 250'C).

The Vmax derating graph shows it can handle it's maximum rated voltage for a thousand hours at anything below around 100'C ambient.

 

[Galgso]

With the correct rating as Diver pointed out, the single discharge power is fine.

The continuous connection rating is not a problem - 105'C is the maximum ambient temperature not device temperature. (The resistance vs temperature graphs go up to 250'C).

The Vmax derating graph shows it can handle it's maximum rated voltage for a thousand hours at anything below around 100'C ambient.

Thank you for your response. So you are saying that as long as the thermistor's surrounding are protected and the ambient temperature remains sufficiently low, it should be fine heating up way beyond 100 degrees and can continue doing so for the amount of hours stated in the V max. derating vs T amb. graph? It does make sense for the operating temperature range to be referring to ambient temperature. I am using to looking at semiconductor device datasheets and they usually list the operating range for the semiconductor junction. Glancing at the thermistor datasheet I immediately assumed they were referring to the temperature range for the thermistor itself.

 

[Tony Stewart]

0.5*320 e-6 * 588^2= 55.3 J. How quickly must you drain the caps?
1 s ? 10 s?
I would use a constant current sink and choose max temp as 30 'C rise. < 10 s. A reasonably small CPU heatsink with a TO-220 can do this in 1 second if you want. Or a 25 W sink
sim : https://tinyurl.com/2a7b2hyf

Polyfuses tend to operate 120'C

 

[Tony Stewart]

Which do you prefer?

 

[Galgso]

0.5*320 e-6 * 588^2= 55.3 J. How quickly must you drain the caps?
1 s ? 10 s?
I would use a constant current sink and choose max temp as 30 'C rise. < 10 s. A reasonably small CPU heatsink with a TO-220 can do this in 1 second if you want. Or a 25 W sink
sim : https://tinyurl.com/2a7b2hyf

Polyfuses tend to operate 120'C

View attachment 149409

View attachment 149410
View attachment 149411

Which do you prefer?

That is pretty nice. I'm guessing this is what Nigel Goodwin meant by an active discharge. I did not think of using a current source for discharging. Thanks for sharing as this is pretty interesting but rjenkinsgb made me realise that the operating temperature range referred to the ambient temperature range rather than the temperature of the thermistor itself in the datasheet so my original thermistor solution would work I think as long as I adequately protect the surrounding components from the thermistor's heat. Also the caps need to be drained in less than 5s.

I wouldn't be able to use your suggested circuit without modifications as the circuit I am designing needs to be able to handle the maximum voltage permanently without being fried. A TO-220 MOSFET/BJT constant current sink would fail pretty quickly if exposed to continuous voltage hence my desire to use a PTC thermistor. Thanks for the idea though I'll keep it in mind if I need to design a similar circuit in the future with less stringent safety requirements.

 

[Diver300]

Thank you for your response. So you are saying that as long as the thermistor's surrounding are protected and the ambient temperature remains sufficiently low, it should be fine heating up way beyond 100 degrees and can continue doing so for the amount of hours stated in the V max. derating vs T amb. graph? It does make sense for the operating temperature range to be referring to ambient temperature.

The ambient temperature rating is up to 105 °C. The switching temperature is 130 to 140 °C. It would be a useless PTC if it couldn't get to its switching temperature without damage.

The maximum ambient temperature is set to be low enough that it will be turned off at zero current at maximum ambient temperature. It also needs some margin so that it does not switch with too little current at high ambient temperatures.

 

[Nigel Goodwin]

That is pretty nice. I'm guessing this is what Nigel Goodwin meant by an active discharge.

No, I was thinking of using a thyristor to switch a large wirewound resistor across the capacitor, when you want to discharge it - so as the circuit shuts down it triggers the thyristor, and discharges the capacitor through the resistor. The thyristor will automatically switch OFF, once the capacitor is discharged.

 

[Diver300]

No, I was thinking of using a thyristor to switch a large wirewound resistor across the capacitor, when you want to discharge it - so as the circuit shuts down it triggers the thyristor, and discharges the capacitor through the resistor. The thyristor will automatically switch OFF, once the capacitor is discharged.

The problem with a thyristor design is that if whatever is changing the capacitor is still connected, the thyristor will stay on, so the resistors would need to be rated to run continuously.

 

[Nigel Goodwin]

The problem with a thyristor design is that if whatever is changing the capacitor is still connected, the thyristor will stay on, so the resistors would need to be rated to run continuously.

As I understand it, the requirement is to discharge the capacitor at switch off, for safety reasons, so the thyristor won't stay ON as the capacitor will be discharged.

Or you could turn the thyristor OFF once the capacitor has discharged enough?.