PTC performance measurements

I’m going to be posting some information over the next couple of weeks with real world measurements of the PTC components in the VEX products.

First a refresher, PTC is an acronym for Positive Temperature Coefficient, and is a passive component used to protect electronic circuits from over current conditions. This device is actually a type of thermistor, current through the device causes a small amount of resistive heating. If the current is large enough to generate more heat than the device can lose to its surroundings, the device heats up, causing its resistance to increase, and therefore causing even more heating. This creates a self-reinforcing effect that drives the resistance upwards, reducing the current and voltage available to the device. (thanks wikipedia).

The PTC in the VEX power expander is the HR16-400, this part is also (as far as I can tell) used in the cortex and perhaps the PIC micro controllers. The VEX 2-wire motors also use a PTC but with a lower current rating, perhaps I will evaluate these in the future but for this post I’m going to concentrate on the HR16-400. As with many passive components it’s not always possible to find the exact part used in a design, for the following tests I purchased a number of equivalent parts from digikey.

The key parameters for this part are as follows.

Hold current, maximum current at which the device will not trip at 25 deg C in still air. This is 4 Amps for the HR16-400

Trip current, minimum current at which the device will always trip at 25 deg C in still air. This is 6.8 Amps for the HR16-400.

I wanted to verify these numbers and also see how the device behave after repeated tripping. There is a theory that after repeatedly being tripped theses devices become “weak” and perhaps wear out, the plan is to try and prove this one way or the other.

Test setup

To test the PTC devices I’m using a bench power supply that can operate in a constant current mode, the particular unit used is an Agilent E3633A. The power supply is set to the desired output current and connected to the PTC, an oscilloscope is used to measure the voltage accross the PTC as a means of easily determining the time from starting a test to the PTC tripping. A typical test will create a waveform on the scope as follows.

TEK00001_PTC.jpg

Although this trace shows voltage against time, it is a convenient way to measure when the PTC trips.

Test 1 - Can the PTC sustain a 4 Amp current.

The first tests were done with a new device, a 4A current was set on the power supply and the time to trip the PTC was recorded. Power was then removed from the PTC which was then given time to recover. The test was repeated for various recovery times.

The results are as follows.

New, never been tripped, 4A current, 59 seconds to trip
1 minute recovery time, 4A current, 29 seconds to trip
5 minutes rest time, 4A current, 40 seconds to trip
30 minutes rest time, 4A current, time to trip 42 seconds

So it’s interesting to note that with this PTC the 4A current could not be sustained for more than 1 minute when the PTC was completely cold. This is different than the datasheet would have us believe.

Test 2 - Can the PTC sustain an 8 Amp current.

Similar to the first test but with 8 Amps of current, a different new PTC was used for this test.

New, never been tripped, 8A current, 6.5 seconds to trip
1 minute rest time, 8A current, 5 seconds to trip

As expected the time to trip is much shorter, however, even with a short recovery time it does not change much.

Test 3 - Can the PTC withstand a 3A current for an extended length of time.

The same PTC as used in test 2 was setup with a 3 Amp constant current, the device was still working after 10 minutes so the test was aborted with the conclusion that 3A is not going to trip the device.

Test 4 - Are all PTC devices made equally.

For this test three different devices were timed with a 3.5A current.

PTC-1, 3.5A current, 123 seconds to trip
PTC-2, 3.5A current, 57 seconds to trip
PTC-3, 3.5A current, 48 seconds to trip

So it appears that some devices will withstand a given current near to the rated minimum hold for longer than others.

Test 5 - As test 4 but run at 5 Amps.

PTC-1, 5A current, 15 seconds to trip
PTC-2, 5A current, 13 seconds to trip
PTC-3, 5A current, 13 seconds to trip

This time they are closer in performance but PTC-1 still lasts longer.

So what are the conclusions from this first series of tests.

  1. Contrary to the datasheet, none of the devices tested could withstand the minimum hold current for more than 60 seconds.

  2. Not all devices perform the same when near the minimum hold current.

  3. After the device has been tripped, it takes a significant amount of time before the original performance is regained.

  4. These devices seem to be better suited to a constant 3A current than the specified 4A.

In a few days time I will post results after these devices have been tripped at least 100 times over a period of several days.

I know it’s not nearly as much fun as murdomeeks reveal, but hopefully is informative :slight_smile:

.

3 Likes

Does the Power Supply drive a Particular Load, or is it a Dead Short??

The original plan was to use a resistive load in series with the PTC, thats how we would normally test this type of thing, however, the only low value high power resistors I had on hand were 1.3 ohm 10W, obviously with 5A flowing the power dissipated would be around 32W and they were going to heat up far to quickly.

So the power supply is driving an almost dead short in constant current mode, set the desired current and maximum voltage (in this case 7.5V) and let the power supply handle things by controlling voltage to maintain current. The PTC has a small resistance even when conducting, as it heats up the resistance increases and the PSU increases voltage accordingly to maintain the set current. At some point I will pick up a 50W 1 ohm resistor and repeat the experiment but I would expect the same results. I’m really more interested in what happens after I trip the PTC many times over a period of days, does the performance degrade or remain constant. I need to setup some automation to do this as I don’t have time to keep tripping it 20 times a day to simulate extreme abuse.

1 Like

That was what I was thinking too… :wink:

It will take a little longer to trip, is what I would expect…

Thanks for doing this kind of testing… It gives all those potential Engineers and Scientists a little idea of how to go about answering questions about how thing work… And for those that don’t got the Engineering or Science route, general knowledge of How Things Work, is just a Good Life skill… :wink:

1 Like

The heat the device can lose to its surroundings depends on the heatsinking effects of mounting, airflow, thermal mass, etc.
How does your test circuit compare to Vex circuit for heatsink effects to the PTC?

I like to brainstorm alternatives. Alternatives to the PTC are likely to be pricier than ~25c (10ku).
Once you have a current sensor for feedback, the supervisor cpu could do closed loop current control by scaling back the drive value;
(as long as the supervisor cpu process doesn’t get hung up, which is one risk of active control)

Many competition users would likely appreciate a graceful auto-throttle vs current status of instant-cut-off.
If there were user-accessible current monitoring, users could characterize their own current limits,
and do user-cpu auto-throttle, if they wanted to.

1 Like

Thanks jpearman!
This looks to be a very useful and helpful test, which should disspell some myths, and establish some scientific precedence in the field of overheating/breakers tripping. Now I am curious, I thought the Cortex and the Power Expander had a breaker like device in them that was different then the thermisters used in the vex motors. Am I incorrect? (I hope so!)
Again, thanks for running this test jpearman!

@jgraber My dad had this idea a while ago of monitoring the battery voltage level with the Cortex, and if a dramatic decrease in voltage occurs, since current draw is inversely linearly proportional to available voltage, we could effectively trap for too much current being drawn by the motors, and scale back respectively on the motor power so they don’t burn out. What do you think of that?

1 Like

First off, great job! Very professionally done. These findings should be very helpful to take into account. The times you found seem to correlate perfectly with problems I experienced with drive motors this year.
Note to everyone on forum**** read these types of things, they are helpful!!

1 Like

jg: numbers added

1 is possible, you can also display battery voltage value to the LCD or terminal
2 speculation not yet demonstrated
3 reversal of cause and effect?
4,6 Maybe. Sounds like a good science project to find out.
5 how much is too much?
7 they what? they=motors? s/they don’t burn out/the PTC doesnt trip/

I don’t have a good feeling about monitoring the battery voltage to measure current, related to your #3.

Another idea might be to think about a mathy physics model of each PTC tripping.
A The PTC trips based on heat.
B Heat flows out from PTC per “ohms law of heatflow” and by methods of conduction, convection, and radiation.
C Current flow adds heat into the PTC.
D current flow comes from motor speed settings (known) and motor load.
E motor load can be sensed by comparing motor speed settings with actual motor speed from motor-back encoder, wheel optical quad encoder, or arm pot.

Take a bunch of data and combine it in different ways until you get a metric that seems to match real-world PTC tripping data. The track the metric in real time and scale back motors to avoid tripping the metric rule limit?

1 Like

First of all, this is going to be a long term project, longer than I expected and I doubt we will have any definitive results until after this season is over.

I have been testing a single 4A PTC over the last few days on and off when work permitted. To achieve this I’m using the remote control capability of the power supply and have created a script to exercise the PTC by supplying a 4A current for 60 seconds (generally more than enough to cause the PTC to trip) and then allowing it to cool off for 2.5 minutes before repeating the cycle. So far i’ve put about 100 cycles on the PTC in batches of 20 with a couple of hours rest between each batch. The results so far are not what I expected and are inconclusive to the point that I may start then entire test over.

The summary is that so far I notice two things worth mentioning.

  1. The PTC trip time is increasing with age, this device had an initial trip time of 27 seconds at 4A and after 100 cycles this has apparently increased to 47 seconds. For a 3.5A load the change is from initially 57 seconds to 146 seconds.

  2. The trip time for any given PTC seems to be very variable even when other parameters of the test such as ambient temperature are not, for example, a different PTC that has not been excessively exercised, when tested under the same conditions shows a trip time for a 3.5A current varying between 120 seconds and 180 seconds.

My plan over the next week is to try and put perhaps another 500 4A cycles on the first candidate and see if it can be tested to destruction. At around 15 cycles per hour this will take some time as I’m not comfortable running this test unattended.

To answer simmons 2.0’s question, I have seen code that tries to monitor battery voltage and predict motors stalling. The PTC (the 4A cortex one) has an observed resistance of about 0.12 ohms when conducting even though the data sheet shows much less, this gives a voltage drop of about 480mV when the current is 4A, this rises quickly (as the scope trace shows although this was for a different PTC and current) as the PTC heats up just before tripping, it may be possible to detect this and back off the power to the motors but it will take additional experiments on battery voltage under “normal” high load conditions to see if voltage drop due to increased PTC resistance can be discriminated from voltage drop due to internal battery resistance under high load. As I said at the beginning of this thread, this will take some time to assemble all the data, perhaps I need an intern for the summer :slight_smile:

2 Likes

It would be really interesting if repeated trips increased (from competition view) performance.

Its a good idea to be cautious about leaving it run unattended.
Are you using GPIB or something, and recording all the data too?

re Simmons ideas:
Is the battery voltage sensor before or after the PTC?
I had assumed the feedback was before the PTC.

Hypothesis: If the PTC increases resistance before it trips,
then the through-current goes down,
and the sensed battery voltage (before the PTC) goes up.
The graph shows there might be a ~4 second window of opportunity to scale back motor speed.

For this idea to work, the sensed battery voltage has to be dependent on the current draw. The current draw is dependent on both the motor speed settings and on the PTC tripping, so a monitoring routine has to look at both.

1 Like

The PSU has both GPIB and RS232 for control, I use the RS232 as thats far easier. The protocol used is SCPI (Standard Control for Programmable Instruments) and the control software is my own, it can also record data from the PSU.

The voltage sensor (well analog output) is after the PTC in the power expander but I also assume before the PTCs in the cortex.

This may work with the power expander, less likely for the cortex.

1 Like

I wanted to update this thread as it’s been a couple of months and I finally found time to complete the first phase of testing.

To recap, I was looking at the performance of a 4A PTC which, according to it’s datasheet, is equivalent to that in the cortex and power expander. I had initially though that the time taken to trip the test device was increasing with age, however, now that 500 cycles have been put on this device that does not seem to be the case. The only conclusion I can draw at this stage is that performance has not degraded and it seems unlikely that the device will fail, good news I guess.

At this point I have decided that any future testing will be done using a different setup with a large power resister as the current limit rather than relying on the power supply. I’m still a bit concerned that the device trips at all at it’s stated hold current.

Just for everyones amusement, I shot one cycle using an infrared camera we have in the lab. You can see how the PTC heats up under load to over 100 deg C before tripping. After power is removed it takes about 1 minute before the device has cooled to almost ambient temperature again. This was shot with a 4A current during the extended testing period. The video of the PSU in the background was just recorded on an iPhone and composited in later. The cross hair and number in the middle of the image is the current temperature of the device. The power supply shows the current on the right hand display.

(click to play)

2 Likes

Your analysis of these PTCs is amazing stuff. For this new SkyRise game, I predict a lot of kids are going to be tripping out their motors, Cortex, and power expanders like never before. Intimate knowledge of PTC limitations will be essential. Woe be to those who don’t grok your posts! :smiley:

2 Likes

Great tests. I can say that after designing these into a number of products, their worse/best case tolerances range quite a bit from the ideal specs.

1 Like