Large surges (like load dump) also supply a lot of energy to the fuse itself. Take the peak current and voltage ratings seriously: if exceeded, the fuse will still clear, but it will clear once and for all, not resetting itself afterwards... If your surge voltage has a known maximum, you may be able to select the fuse so that its minimum resistance is enough to limit current to a safe value, at the peak voltage... Polyfuses aren't very practical beyond 30V and some number of amps, depending on situation. The fault energy is just too much for the polyfuse to handle, or it becomes impractically large (and slow). You could still use an active circuit at that point -- though things get increasingly bulky, as the circuit has to deal with both inrush and fault energy, and by a few hundred volts of capacity, you're looking at a handful of transistors...
Exactly. Voltage spikes on the input power line would exceed the Vmax spec of most any PTC, especially of the PTC is placed first in the circuit, before the TVS can limit the spikes. I personally put a single-blow fuse externally on the power wire of the wiring harness, thereby eliminating the need for a PTC fuse on the PCB.
The inline diode used in the circuit is to protect against reverse polarity, but to ensure a lower voltage drop (and less wasted power) a Schottky diode would be best. However, Schottky diodes like a B270 (70V, 2A) have lower voltage ratings than say a 1N4004 (400V, 1A). So if a Schottky is used, it would be prudent to place it
after the TVS, ensuring the Vc of the TVS is lower than Vmax of your chosen Schottky. But when you put the Schottky AFTER the TVS, it means you need to use a BI-DIRECTIONAL TVS diode so that reversal of polarity will not adversely affect the TVS. Putting the Schottky first allows use of a UNIDIRECTIONAL TVS, but again, a Schottky's Vmax is too low for that in a 24V application, so you would either need to use a regular 1N4004 diode first with a UNIDIRECTIONAL TVS, or use a Schottky AFTER a BIDIRECTIONAL TVS.
Vc of TVS diodes is another consideration. 600W TVS diodes are often used in automotive applications, but there is more than one Vc on some datasheets. Consider this:
http://www.st.com/content/ccc/resource/technical/document/datasheet/1a/3b/f2/a9/96/72/49/ca/CD00000720.pdf/files/CD00000720.pdf/jcr:content/translations/en.CD00000720.pdfLook at the
P6KE47A in that datasheet, which one might consider for 24V vehicle use. Vc@10/100us =
64.8V. That would seem to be an acceptable TVS choice when paired with a robust Buck Regulator like a Linear
LTC3637 which can handle up to
76V on Vin (absolute Max is
80V). But looking more closely at the TVS datasheet we also see Vc@8/20us =
84V, which would exceed Vmax of the LTC3637. If one is designing for a worst case scenario, then using the 8/20us Vc would be more prudent than simply relying on the 10/1000us Vc. That would mean choosing something like the
P6KE39A which has a Vc@8/20us=
69.7V, which would work fine with the LTC3637.
On the other hand though, your TVS choice depends on the maximum "continuous" (for a length period of time lasting many seconds or minutes) voltage you expect to see in your 24V vehicle. In some situations a 24V truck could see recurring voltages rise up to
36V. So if you account for that with a TVS, you would want to make sure your breakdown voltage
Vbr-min is 36V or higher. But perhaps the biggest consideration is
jump starting. What if someone tries to jump start with
48V? The
P6KE47A TVS diode has a Vr=40V (which is higher than your 36V case, so you're covered there) and
Vbr-max = 49.4V. So if someone jump starts the 24V truck with 48V, it's not exceeding the maximum breakdown voltage of Vbr-max, but the TVS will be conducting a significant amount of current during the duration of the jump start. For that same TVS diode, Vc@10/1000us =
64.8V, but Vc@8/20us =
84V. So if you use a buck converter like the
LTC3637 which has a Vin-max of
76V and an absolute max of
80V, you would only have a problem in the event of an 8/20us load dump pulse type. So you would then ask which is more likely -- the 48V jump start case or the 8/20us case? Perhaps the former, which means your TVS diode pick would be the P6KE47A when used with a Linear LTC3637 buck regulator.
The
wattage rating of the TVS is yet another consideration. In most automotive 12V and 24V designs I see 600W TVS's being used. I believe the designer is using that lower wattage value not merely to cut cost but going on the assumption that the device will be used in newer vehicles that have protection diodes in the alternator. Without the diodes in the alternator, your Load Dump in a 12V vehicle can be 100V to 125V. In a 24V vehicle the Load Dump could be 202V. But with protection diodes, a Load Dump on a 12V car is usually capped to about 60V and on a 12V vehicle it usually won't exceed 70V or 80V. So with that in mind, a 600W TVS may suffice, versus a more expensive 1500W or 2200W version. But to do the math and be sure, you would need to know the internal resistance of the alternator, which really does require some guesswork. Even so, there is a 24V example calculation given on page 11 of the following PDF:
http://www.newelectronics.co.uk/article-images/82298/automotive_circuit_protection.pdfThe absolute best engineering solution is to use
"active" transient suppression rather than a TVS, but cost, complexity and board space requirements make active suppression a more difficult choice than a TVS:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/4240