I tested the basic reverse polarity protection circuit:
12V battery, 100A fuse, diode.
If the TVS (1.5SMC24AT3) is used, it gets destroyed into the short circuit. The fuse is blown, of course; but the device is damaged. This happens at the very first attempt.
However if the trivial 1N5404 is used, it blows the fuse and survives the test. Tried it several times.
Interesting. I didn't expect the TVS to be weaker then a diode.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
I'm a bit lost here. I assume you are using the TVS to protect for overvoltage. That is, you have a 24V breakdown to protect the 12V rail. Then you depending on the forward voltage of the TVS to protect for reverse battery application.
Assuming that is the circuit, you are depending on the VF of the TVS to provide the reverse protection. But that isn't the target parameter of the TVS. So it seems to me a "normal" diode is a good idea for reverse protection.
I have a power distribution box that I use for my own use (not a product designed for fee). What I did is use some Schottky diodes for reverse protection. The reasoning being the schottky will take the current flow before the diodes in the device you are trying to protect. Being paranoid, I also use a silicon diode too, figuring if the Schottky blows, the next path for the current is in the silicon diode.
Not surprising if you look at the specs. Same basic 200A surge rating (1/2 sine 8.3ms) for both parts, but the TVS is in a 165degC/W package and the 1N5404 is in a 15degC/W package.
Dave.
--
================================================
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
Interesting... Perhaps the diode has a bigger piece of silicon in it and can take more energy before failing. What about if you use both the TVS and diode... the old "belt and suspenders" (braces for you brits.) approach to design.
How about a simple solution with something like a starter-relay, the coil of it powered via a diode? The relay will then not come in if the voltage is reversed. Saves fuses...
Cost and size does matter. There must be an external fuse, the TVS is also necessary. An obvious solution could be a healthy TVS in the TH package, but they are expensive and not very easy to get.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
Uae the 1N5404. A TVS isn't optimized for reverse voltage protection - it's optimized for absorbing transient OVERvoltage faults. In the forward direction, it's just like a Zener or any other semiconductor diode, with nowhere near the forward current capability of something as mongo as a
1N5404. As we all know, under those circumstances, semiconductors make excellent fuse protectors! ;-)
The TVS quoted and a 1N5404 have exactly the same abs max forward surge current rating of 200A half cycle sine, 8.3ms.
The OP is obviously 'operating' both of them outside abs max ratings.
I am also a little surprised the TVS appears to be more fragile. The TVS is still a diode and being designed to take very high power dissipation pulses you would expect it to be a large lump of silicon with good thermal bonding to its leads and so take high power dissipation pulses from forward currents.
Maybe the diode has a bit more intrinsic resistance helping to spread forward current across the die while the TVS aims for minimum resistance.
Yes, interesting. Given the very similar 200A ratings it might be interesting to see if two fuses in parallel will destroy a 1N540x.
Can you use the conventional series MOSFET polarity protection?
I imagine the several volts that will be developed across the 1N540x at several hundred A will be enough to damage things.. maybe it would protect an e-cap.
It's not irrelevant. It shows that the dissipation ability of the devices are not equal. That would be of some relevance when it comes to overloading the thing, and hints that it's not apples vs apples here.
I got my 15deg J-A figure from Diodes Inc:
formatting link
They do not specify the conditions, but you'd probably assume it's a mounted figure.
Vishay show 20degC/W for a board mounted 0.375" lead length, less than half the figure on-semi quote. So it looks like this varies a far bit between brands.
formatting link
In either case it's still a LOT lower (20-50deg) than the really wimpy
165deg J-A figure of the TVS. Doesn't that hint at a little something?
Correct. It likely comes down to how beefy the internal silicon and bonding is etc, and the J-A figure would be a fair bet to be representative of that, at least in the absense of any other firm data. If you have a better insight into this then please share it with us Phil.
And both devices have the same 200A surge figure, yet one dies and the other doesn't. IMO it's fairly obvious that the diode is likely to be beefier internally than the TVS, so would be more likely to survive an (unspecified in this case) out of spec overload better.
Obviously it's above the 200A/8.3ms figure (or equivalent), otherwise it wouldn't go splat.
Dave.
--
================================================
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
Perhaps not weaker. What's the TVS voltage drop when it operates for reverse polarity protection? What is the voltage drop for the 1N5404?
A higher voltage drop, even if it reduces the current somewhat, might land closer to the maximum power point that the source is capable of delivering. The 1N5404 will be closer to a dead short, dissipating less power. On top of that, if the TVS voltage drop reduces the current somewhat, it'll slow the fuse down and have to dissipate even more energy.
All the figures for ratings at 8.3 mS look suspiciously like half cycle ratings. These might assume that its the zero crossing that actually causes that open fuse to stop conducting. DC can be much more difficult to clear.
--
Paul Hovnanian mailto:Paul@Hovnanian.com
------------------------------------------------------------------
All wiyht. Rho sritched mg kegtops awound?
" > > However: 8.3mS is NOT enough time for the mounting method to > > come into play >
Yeah, every hunk of stuff, has a thermal time contant. How long does it take for heat made inside to get out. I believe a friend of mine told me that the time constant goes as the square of the linear dimension. (double all the dimensions and the TC goes up by four.) But don't take my word for it you can easily work it out for yourself. (Time constant =3D heat capacity X thermal resistance). 8ms strikes me as a small piece of stuff.. but I'm not sure how big.
Would be rather interesting to decapsulate the two devices and see the physical difference. Also the difference between different brand 1N5404's Likely to be a few factors at play here though...
Dave.
--
================================================
Check out my Electronics Engineering Video Blog & Podcast:
http://www.eevblog.com
It can't be a "diode" as it is bidirectional. As to thermal properties remember that in either polarity the conduction of the TVS is at a much higher voltage than the forward voltage of a diode of similar ratings. Makes it more like a MOV but with better clamping and substantially better applicability at low voltage.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.