EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: Psi on July 24, 2022, 12:33:19 am
-
I see TVS diodes sometimes where the Imax clamping voltage is lower than the 1mA breakdown voltage.
Which, all things being equal, doesn't seem to make sense unless something else is going on.
I did some googling but didn't find a satisfactory answer as to why this occurs.
Do thermal effects of the power dissipation change its properties and lower the breakdown point?
eg, are Vclamp and Vbreakdown exactly the same property just measured at active vs inactive temperature.
-
Some high-end ESD protection diodes for high-speed data lines use the thyristor principle to create negative resistance - once triggered, they remain in conduction until the voltage drops to a level much lower than the initial trigger voltage. These are marketed as "snapback TVS diodes".
There's also a related technology called Thyristor Surge Suppressor (TSS), commonly used for telecom applications to clamp huge surges in telephone lines. Large TSS can compete with Gas Discharge Tubes (GDT).
But if you are talking about ordinary TVS diodes, I don't know the answer.
(https://www.eevblog.com/forum/chat/quick-tvs-diode-question-vclamp-lt-vbreakdown/?action=dlattach;attach=1547140;image)
(https://www.eevblog.com/forum/chat/quick-tvs-diode-question-vclamp-lt-vbreakdown/?action=dlattach;attach=1547146;image)
(https://www.eevblog.com/forum/chat/quick-tvs-diode-question-vclamp-lt-vbreakdown/?action=dlattach;attach=1547152;image)
-
Here's an example of what i mean. It doesn't seem to be a snapback, or it doesnt say that anywhere.
The graph in fig 3 on page 4 looks normal, ie no inversion/hysteresis like in your example graph.
But the clamp is still less than the breakdown which seems to go against what fig3 shows
https://assets.nexperia.com/documents/data-sheet/PESD5V0C1USF.pdf (https://assets.nexperia.com/documents/data-sheet/PESD5V0C1USF.pdf)
(https://www.eevblog.com/forum/chat/quick-tvs-diode-question-vclamp-lt-vbreakdown/?action=dlattach;attach=1547392;image)
-
Yeah, snapback:
https://assets.nexperia.com/documents/leaflet/Nexperia_TrEOS_ESD_protection_for_USB_Type-C_leaflet.pdf
Some high-end ESD protection diodes for high-speed data lines use the thyristor principle to create negative resistance - once triggered, they remain in conduction until the voltage drops to a level much lower than the initial trigger voltage. These are marketed as "snapback TVS diodes".
Just to clarify -- snapback isn't a thyristor (4-layer) mechanism, it's... punch-through, I believe? So, a 3-layer structure, like a BJT without a base connection, and with a light enough doped base that it fully depletes at some point, effectively shorting from C to E. (I might have this wrong, but ESD protection features are notoriously secret-sauce, so that's not necessarily my fault. :P ) When this is tuned for just the right breakdown voltage, it can happen that a fair amount of current is already flowing (avalanche/zener mode) by the time the negative resistance effect pulls in, and the negative resistance effectively compensates for the bulk resistance of the device, hence flattening the curve -- or at least not having such a severe "snapback" as for thyristor devices.
They can be made in much lower voltages than avalanche types (which are zeners below 5V, so, basically useless for surge purposes!), and are the only type that offers effective protection for systems 3.3V and below.
they're also faster, as it only depends on the motion of carriers, rather than diffusion of minority carriers (thyristors are fast enough for surge, which they excel at, but aren't suitable for ESD, AFAIK).
Tim
-
It seems that the "clamping voltage" is the rise above the "breakdown voltage" related to current.
-
Yeah, snapback:
Also, Figure 7 of the datasheet.
Here's an example of what i mean.
Max Standoff > Typical Conduction
Nothing fishy about that, even without snapback.
-
4 layer, snapback, are not TAZ.
See extensive papers and app notes from TAZ Mfg like MicroSemi, Claire, General Semiconductors from 1970s..1980s.
I gave a paper on this tipoc at PCIM 1983 Paris with a TAZ mfg.
Indeed there's a tempco.
jon
-
Last month a FET here failed during a thunderstorm.
It was on a HF antenna longwire, downstream of a toroidal isolating transformer followed by resonant LC.
I think I need OVP with capacitance > 20 pF and fast acting ~ 50 nsec hopefully to minimise ringing and Hi Z below -10dBm, to be placed across antenna input terminal. Already have a 1 A 3AG fuse which did not fail.
Any suggestion for a device would be appreciated. I know various PL 259 insertion OVP are available from ham suppliers but I want to do it at component level.
-
Last month a FET here failed during a thunderstorm.
It was on a HF antenna longwire, downstream of a toroidal isolating transformer followed by resonant LC.
I think I need OVP with capacitance > 20 pF and fast acting ~ 50 nsec hopefully to minimise ringing and Hi Z below -10dBm, to be placed across antenna input terminal. Already have a 1 A 3AG fuse which did not fail.
Any suggestion for a device would be appreciated. I know various PL 259 insertion OVP are available from ham suppliers but I want to do it at component level.
The HP engineers solved this brilliantly in the legendary 34401 DMM.
A TVS in series with a GDT suppressor works great. It has extremely low capacitance due to the GDT and will not interfere with your signal.
But the combination will take a lot of abuse.
-
Last month a FET here failed during a thunderstorm.
It was on a HF antenna longwire, downstream of a toroidal isolating transformer followed by resonant LC.
I think I need OVP with capacitance > 20 pF and fast acting ~ 50 nsec hopefully to minimise ringing and Hi Z below -10dBm, to be placed across antenna input terminal. Already have a 1 A 3AG fuse which did not fail.
Any suggestion for a device would be appreciated. I know various PL 259 insertion OVP are available from ham suppliers but I want to do it at component level.
The HP engineers solved this brilliantly in the legendary 34401 DMM.
A TVS in series with a GDT suppressor works great. It has extremely low capacitance due to the GDT and will not interfere with your signal.
But the combination will take a lot of abuse.
Varistor and GDT:
"RV100-RV102 Diode-Varistor 1.1KV"
"E100-E101 Tube Electron Surge Arrestor 1500V/10A"
https://xdevs.com/doc/HP_Agilent_Keysight/service/34401A/34401A%20service.pdf
I guess the reason is only to limit the clamping voltage (glow voltage) of the GDT, as that can be too low to the point of blowing a fuse somewhere (<150V).
-
Thanks Benta and Thm_w
I had been thinking of GDT and TVS in parallel, but now I will do series. I can get some GDT from Surplus Sales.
I have put in a station ground close by, per ARRL recommendation consisting of two driven copper plated rods 2.5 m (8ft) long spaced 1200 mm.
Before brazing on the 4AWG leads, I measured the DC resistance between them with tractor battery, at 36 Ohm.
Then when trying later with a Ohm meter, the ground had charged up between them to 0.4 V DC and stayed like that.
So I don't know if the grounds will conduct a surge.
-
Series is when you need a high turn-off voltage, such as for mains voltage protection. For signal purposes such as telephone (POTS), antennas probably, Ethernet, etc., a raw GDT or thyristor will do.
Hmm, I wonder if parallel connection is reasonable anyway. Let's see, they break down at a minimum voltage, but it takes time (some µs), and it's faster as you go up from there, I think up to an upper limit where breakdown is very fast. Presumably, you could clamp that initial excess, given a TVS beefy enough (which isn't too bad, it's only a few µs), then the GDT kicks in and clamps the rest.
Looks like at least a few others have thought so:
https://itecnotes.com/electrical/electronic-two-gas-discharge-tube-in-parallel-a-fast-acting-gas-acting-tube-and-a-gas-discharge-tube/
so if that's a good enough reference, I guess that's something. The case here with a series inductor between them, is nice: under the surge's dI/dt, the GDT gets much more voltage, triggering sooner and dumping less energy into the TVS.
Note if this is on something like feedline, treat the shield (ignore the signal inside it) as a single-ended line with respect to earth. Mind you only get as good suppression as you have ground impedance.
Tim
-
Thanks Tim,
I will get some devices and test with a 200 V DC supply.
However, I ran some qucs on the circuit. If a pulse of only 50 V and 30 ns on antenna gets into the tuned circuit,
it rings, being high Q and is all over for the fet which is rated 35 V between all.
If a TVS on input fires some time later to an ideal short it (depending on timing) increases the ring up voltage as Q of tank is raised by shorting the antenna Z.
So I will also add back to back zeners at output of tuned circuit before fet.
Later Edit: For HF radio application, the MicroSemi SA5.0CA Bi-Directional series might be suitable.
No snapback, < 5 nanosec, with a nominal 500 Watt at 1 millisec. in DO-41
SA14CA 14 V standoff for receivers and SA150CA for up to 100 Watt transceivers.
https://download.datasheets.com/pdfs/diode/mcs/msc1400.pdf
-
I've been replacing damaged MOVs in Tripp-Lite Isobar power strips with the GMOV series from Bournes:
https://media.digikey.com/pdf/Data%20Sheets/Bourns%20PDFs/GMOV_Series_DS.pdf
-
Good thread!
And I am about to make a bunch of mistakes. Can you spot them ? :D
So I'd like to perhaps implement the solution from the link below:
https://itecnotes.com/electrical/electronic-two-gas-discharge-tube-in-parallel-a-fast-acting-gas-acting-tube-and-a-gas-discharge-tube/
so if that's a good enough reference, I guess that's something. The case here with a series inductor between them, is nice: under the surge's dI/dt, the GDT gets much more voltage, triggering sooner and dumping less energy into the TVS.
for automotive device protection. I am going to take a nasty load dump for the first impulse to mitigate. Attached its specs. Instead of a GDT I shall use this guy, an GMOV-20D450K - which by the way is a 49J part so I see no way for it to be able to absorb the load dump energy; let me assume I am wrong here, moving on:
I've been replacing damaged MOVs in Tripp-Lite Isobar power strips with the GMOV series from Bournes:
https://media.digikey.com/pdf/Data%20Sheets/Bourns%20PDFs/GMOV_Series_DS.pdf
Does V_c in the datasheet table stand for V_clamp ? I am going to assume that as well. For the GMOV to *only* begin conducting the inductor should build up a voltage spike of 150V
150V = U_ind < L * di/dt
For the impulse rise, worst case: di = (U_s - U_alternator) / R_i = (101 - 14) / 0.5 = 174A and dt = 5ms
150 < L * 174 / 5
L > 150 * 5 / 174 mH
L_min = 4.45 mH
For the impulse fall, worst case:
U_ind = 4.45 * 174 / 40 = 19.36V
- which, with a proper TVS, will not cause problems. But I would probably want more L so the GMOV will actually have time to conduct that energy away, perhaps a tenfold value, 44.5 mH? Any possible issues with the TVS or the rest of the circuitry to be protected?
Now, please correct me. Thank you!
-
L what? Show an equivalent circuit.
If you're thinking about the source: load dump isn't an RLC equivalent circuit -- well, maybe you could design a pulse generator that way, but the real deal is an equivalent due to a positive feedback effect in the alternator; there will be some ESL still (winding inductance) but it isn't dominant, and it's modeled as a waveform from an ideal Thevenin source.
Tim
-
Ugh, L is the inductor in between the two protection devices in https://itecnotes.com/electrical/electronic-two-gas-discharge-tube-in-parallel-a-fast-acting-gas-acting-tube-and-a-gas-discharge-tube/
L_min would be my calculated minimum value for it.
Thanks
-
Ah, that way.
That's only needed for fast-ish rise times, where the GDT breakdown delay matters. You're well past that here. Just pick a GDT low enough to work.
More important is the GDT needs a TVS in series with it, so it doesn't short out the supply after the load dump. It's a discharge, it stays latched on until current falls below holding.
Which is a TVS that still needs to handle the full current of said dump, so, doesn't save much compared to a full size TVS just clamping the whole thing. The GDT also needs a quite generous energy rating, as its voltage drop during breakdown is some 10-20V or whatever. I haven't checked but I'm guessing they're neither recommended for, nor available for, load dump application.
...You're also spark-discharging any capacitors on the bus, which, whatever, it's going to be less energy than load dump itself, but still, it's more stuff going on, and that extra backwards current might do weird things to attached devices, who knows.
Tim
-
GDT in series with TVS would clamp the voltage way to high for automotive.
Back to the linked diagram, I would not use a GDT, but, as said, a MOV (more precisely the GMOV that @Tomorokoshi has mentioned - https://media.digikey.com/pdf/Data%20Sheets/Bourns%20PDFs/GMOV_Series_DS.pdf)
Which is a TVS that still needs to handle the full current of said dump, so, doesn't save much compared to a full size TVS just clamping the whole thing.
Huh? Tim sir, can I possibly get a practically sized TVS diode that would actually handle that whole blast of energy by itself? Oh, and within budget too would be nice :D - say max $10 a pop.
-
Just use the MOV straight up? MOVs are just sloppier TVSs...
Huh? Tim sir, can I possibly get a practically sized TVS diode that would actually handle that whole blast of energy by itself? Oh, and within budget too would be nice :D - say max $10 a pop.
I'm sure the name brands get theirs for less; but then, I suppose your production quantity is somewhat less than 10 million/year, too, huh? ::)
:-DD
Tim
-
I shall start with, uhm, 10 per year :)
Which diode would that be anyway ? I am calculating 4 kJ of energy in that impulse; approximating rise and fall curves with straight lines
E = (U^2 / R_i) * 1/2 *(t_r +t_d)
Worst case: U = 101V, R_i = 0.5 \$\Omega\$, t_r + t_d ~= t_d = 400ms
Now, there is a bit of resistive wiring before 12V hit my PCB as well, and the TVS diode would present some resistance too. I should factor those in too, not sure what values to estimate them at. Does my calculation look right ?
-
That would be the energy dissipated by the source, assuming it has real resistance, into a short circuit; at most your power is 1/4 that, because of power transfer theorem. But your load or TVS won't be 0.5 ohm: the fact that Vc < Us/2 means E is less as well.
If we take more like Vc * (Us/Ri) * td, that's a better approximation of the energy absorbed by the load. So, you can see TVS are superior for having Vc ~ 2 Vnom, whereas MOVs can be more like 3-4x (and they're worse in low voltages, besides).
If you're still not convinced that a voltage limiter (switch or regulator type) is viable in many situations... :)
Tim
-
Of course! it is V_c * I, not U_s * I...
Ok I have found one that I guess would work, although marginally, perhaps not at all if I get unlucky - https://www.mccsemi.com/pdf/Products/15KP17(C)(A)-15KP280(C)(A)(R-6).pdf (https://www.mccsemi.com/pdf/Products/15KP17(C)(A)-15KP280(C)(A)(R-6).pdf)
Extending the P_pp - t_d graph yields some 1kW+ able to withstand over 200ms.
Do you know how would a 1A reed switch behave if battered with 2A? Will its life just shorten (to whatever, even 1% will be ok) or will it fail short circuit , or something else?
-
IIRC, 15kW parts are okay for lower levels of load dump, maybe not the worst case you've selected. There are 30kW parts out there too. The lower voltage rating helps.
Note that Fig.1 must eventually flatten out: the 8W continuous rating will take some 10s of s to stabilize at. The sub-sqrt slope of the plot suggests limited thermal diffusion (exactly, diffusion would go as t^(-1/2)). It's not clear if that would get steeper over package-scale time scales (i.e., it takes ~10s ms for heat to diffuse from die to lead frame and package). There's nowhere to go beyond the package, so the curve should steepen at that point, until lead conduction and convection dominate, then finally flatten out at the 8W figure (in the ~10s s range).
If we assume the curve remains flat, then we can take its slope of ~ -0.443, and... let's see, the rated pulse is a 10/1000us waveform, but that's the 50% figure; helpfully, the curve shows it reaching 10% at about 4ms. The load dump waveform takes 400ms to reach 10%. If we simply equate these, then we need a 100x time figure, which would be 100^(-0.443) = 7.7 times lower power, or 1.95kW.
The load dump peak power is Vc * (Us - Vc) / Ri or 4.8kW, so it seems this part is undersized by ~2.5 times.
Which hey, that meshes with my expectation that it's fine for lower ratings, but not the full thing. And in fact an even bigger one is needed then (probably two 30kW in parallel would be safe enough? -- three 15kW would be too marginal I would say, but four or more is probably fine).
Contrast with MOVs, for example:
https://www.yageo.com/upload/media/product/productsearch/datasheet/cpc/mov/20D_1.pdf (https://www.yageo.com/upload/media/product/productsearch/datasheet/cpc/mov/20D_1.pdf)
Nothing particular about this brand, I think they're all the same materials; ratings/specs are more or less industry standard, curves always look the same.
Notice 18-68V types are only rated 20Apk, and the 82V part clamps at the same voltage as the 68V but at 5x the current! (LV MOVs rather stink.)
Looking at the clamping curves, the 18V and 22V parts clamp at 100A and 40/50V respectively, which is... useful enough, but the energy ratings are tiny. (A much bigger disc would afford more energy, but likely not enough.) The derating curves suggest for N = 2 about a t^(-1/2) slope, and the pulse is the same type so let's assume 10 times the highest (10ms) point or sqrt(10) less current, or about 13A. Or 22A for the J version ("high surge").
They don't say here BTW, but the maximum current spec I believe is for 8/20us, which, let's see... lines up on the derating plots for single event, yeah. So that's why that figure is so high.
So, for about a 100A surge at this level, you'd need more like a stack of 5 of these in parallel, preferably 7-10 for better sharing/reliability. Or larger discs, but you'll still need multiple in parallel I think.
Note they aren't rated for leakage; that's probably a bad thing for automotive.
As for relays, you're not thinking of using one to disconnect the load, are you? They don't respond fast enough (~1 to 10s of ms).
Tim
-
Wow, I need to revisit college notes...
>> then we can take its slope of ~ -0.443
How did you derive this number out of that logarithmic graph in fig.1 ?
>>The load dump peak power is Vc * (Us - Vc) / Ri or 4.8kW,
Of course (2)! Also, reading the note on load dump impulse in my attached picture a few posts back:
a
If not otherwise agreed, use the higher voltage level with the higher value for internal resistance, or use the lower voltage level with
he lower value for internal resistance.
Does that mean that under worst circumstances, for the diode in question, P_peak yields about 29 * 50 / 0.5 =~ 3 kW ?
>>probably two 30kW in parallel
Would that not require perfectly matched V_clamp ?
These 15 kW diodes are over budget anyway, so, if MOVs are that bad... reed switches! Yes I would actually use a comparator to trigger the reed relay(s). Many of them are specc'ed at 1ms max engage time, which would be great I think. Most of the automotive impulses are ~ uS long, so the reed switch would not even engage them - a smaller TVS diode may be employed for them. And when the big one hits, then TVS takes out a bit of their energy and then 1ms later reed switch relays the rest of it into an 1M resistor or such. But... their carry / switching current ratings are quite lousy. What are they going to do, weld into place ?
-
Wow, I need to revisit college notes...
>> then we can take its slope of ~ -0.443
How did you derive this number out of that logarithmic graph in fig.1 ?
Take the rise and run. Take the log of each, then take the ratio of logs. That is:
\[ a = \frac{ \ln \frac{y_2}{y_1} }{ \ln \frac{x_2}{x_1} } \]
for pairs of points on the curve (x1, y1), (x2, y2) such that \$y \propto x^a\$. I noticed that, for 1 decade (10 times) of y, there was about a 180 times span in x -- you don't need absolute points, you can read the ratios off the axes directly.
Then, say, knowing that you need to extrapolate 10 times further than the plot shows, go \$10^a\$ times further along the other axis.
>>The load dump peak power is Vc * (Us - Vc) / Ri or 4.8kW,
Of course (2)! Also, reading the note on load dump impulse in my attached picture a few posts back:
a
If not otherwise agreed, use the higher voltage level with the higher value for internal resistance, or use the lower voltage level with
he lower value for internal resistance.
Does that mean that under worst circumstances, for the diode in question, P_peak yields about 29 * 50 / 0.5 =~ 3 kW ?
That or 29 * 72 / 4, which is lower, so yeah. Or inbetween values, which I guess would be at customer request (if someone (OEM, say?) is looking for a particular combination).
>>probably two 30kW in parallel
Would that not require perfectly matched V_clamp ?
Not so much: Vclamp is moderately resistive, and has a positive tempco. They might not share evenly at first, but once the die heats up, both will be bearing the load.
Which is why you need extra capacity: they won't track perfectly, so you lose some difference in capacity. The result is strictly more capacity than a single, but how much more does depend.
I wouldn't have a problem using several in parallel at say... 30% derating? Or less in greater numbers.
These 15 kW diodes are over budget anyway, so, if MOVs are that bad... reed switches! Yes I would actually use a comparator to trigger the reed relay(s). Many of them are specc'ed at 1ms max engage time, which would be great I think. Most of the automotive impulses are ~ uS long, so the reed switch would not even engage them - a smaller TVS diode may be employed for them. And when the big one hits, then TVS takes out a bit of their energy and then 1ms later reed switch relays the rest of it into an 1M resistor or such. But... their carry / switching current ratings are quite lousy. What are they going to do, weld into place ?
Engage, sure, what about disengage time? What's a 1M going to do? (Did you mean 1m?) You don't have many ms left before it has to open into the full 100A surge.
Welded? Melted? Absolutely. The contacts are quite small.
If only we had some kind of technology, that could switch faster... ???
Tim
-
Got it, thank you! Actually t_d will be 400/2 = 200 ms, not 100 ms. That takes the diode ability further down to 1.45 kW, so 3 of these in parallel is probably what I need. Too expensive...
Reed switches will be cheaper if I can make them equally share current. Is the 1M resistor set up like below not suited for absorbing the load dump energy ?
-
What energy, it's a meg? It doesn't need to go anywhere, you can just open-circuit it...
You know that 100V 5A MOSFETs are cheaper than reed relays..?
Tim
-
Right, of course...
Hmm, there are also (shorter) pulses defined for +150V and -220V... And what about their driving BJTs, these will need some sort of protection as well.
-
What about BJTs? Like for general purpose load switching (lamp/solenoid/etc. driver)?
A small TVS or MOV handily absorbs the higher voltage, shorter, and importantly: higher impedance, pulses. With the inductance trick if you like. :) A typical MOSFET load dump switch won't respond fast enough for those, passing such pulses easily to a TVS.
Tim
-
Hmm... perhaps that could work. How about adding an TVS diode (+ maybe inductor & MOV trick) to this diagram
(https://i.stack.imgur.com/fsb6r.png)
With some caveats:
- substitute U1 and U6 with at least 101V able P-mosfets. Actually just U1 I think, U6 should be disengaging -220V spikes, I wonder if fast enough...
- parallelize U1 and U6 into a number of P-mosfets each in order to avoid heat issues and voltage drop
- work these resistors to adjust for +150V spikes on the small Zeners + fat one (-220V too?), and maybe not screw up whatever balance the current values achieve
What do you think ?
-
Something like that.
Note that TL431 can't saturate less than a Vbe below VREF, so won't turn off a 2N7002. Some voltage offset could be added to M1 source, or a P-type inverter used.
Also note you want hysteresis to prevent chatter as it goes on and off. Your load is going to be pretty significant after being off for a few 10s-100s ms. :)
Tim
-
Your load is going to be pretty significant after being off for a few 10s-100s ms. :)
Tim
Could you elaborate on this ? Not sure what to make of it.
Also, does a 1uH inductor look right for the clamping voltages involved in 12V automotive ?
-
I mean when it switches back on, charging your input caps will surely make it chatter.
1uH is probably fine, also helps with general filtering, put some small caps around it (couple uF?).
Tim
-
I mean when it switches back on, charging your input caps will surely make it chatter.
Tim
Ok... but why ? And which part is going the chatter to form at, the fat Zener ?
-
The... TL431 and switch, it's being used as a comparator?
-
Err... right! Already over my head. How about this other one - trying to protect the FETs Q2 and Q3 with a kit for reverse spikes made of
Q1 (a and b) R1 (a and b) R1 and D1. Normally each of the Q1 should carry around 1A.
A positive spike (up to 150V per ISO) will either pass through Q1 and Q3 ending up at R2 and at the lamp if it happens when it is already lit
A positive spike (up to 150V per ISO) will either pass through Q1 ending up at Q2 ad Q3 when the lamp is not already lit
A negative spike (up to 220V per ISO) will always be clamped by the Q1 D1 R1 kit in some, perhaps, 10us then be kept at Q1 drains.
MCU branch of the circuit, due to its low current draw, has its own protective circuitry consisting of fast TVS and rectifier diodes + limiting resistor.
Taking care at the P-FETs for rds_on and current, does this stand a chance in real life ?
Thanks!
-
Not sure about your definition of "spike" here: those PMOS might be 10nF+ equivalent, so that 47k gives a time constant of some 470us, not 10! But that's easily solved with a diode clamp from GND to G. Maybe with a clamp diode to S as well, to avoid dumping surge current through the 6.8V zener there, or a (much smaller) limiting resistor on said clamp diode (>10 ohm?).
Likewise Q3 seems to be exceeding Vgs(max). Just an oversight I guess.
Adding a separate path for an MCU is a poor idea for both efficiency and EMC, but I suppose can be done.
Tim
-
Adding a separate path for an MCU is a poor idea for both efficiency and EMC, but I suppose can be done.
Tim
I'll deal with that later :)
I so wanna use one N channel MOSFET instead of these parallel high Rds_on P-MOSFETs for the negative spikes protection
They say:
(https://img.electronicdesign.com/files/base/ebm/electronicdesign/image/2016/03/electronicdesign_com_sites_electronicdesign.com_files_uploads_2015_02_0216_TI_RevPolarity_F5.png?auto=format,compress&w=1300&h=730&fit=max)
These gate driver ICs seem to all have this drawback - they are tied to V_in = 12V, the one I am trying to protect my circuitry from. As far as I can tell, not an option without a never ending rabbit hole.
Solution: a charge pump which has V_in = 5V already protected from MCU side of circuit. PWM will come form MCU as well, +5 / 0V
(https://circuitdigest.com/sites/default/files/circuitdiagram/Charge-Pump-Circuit-Diagram.png)
I don't get it... is the pump above supposed to charge C20 to, what, 3x Vcc - 6x Vf ? I think it does not. How do I charge the final cap to 4 or 5 times Vcc ?
Oh, and... how to protect Q3's GS junction ? :-[
-
Don't make an absolute rail, make a relative rail.
For example, use a gate driver supplied from +12V (or since not much current should be required here, a CD4xxx gate would also do), make a square wave say of some 10s to 100s kHz, and connect a capacitor in series from its output to the middle of a series diode pair, anode to source, cathode to +12BS (bootstrap). Also put a zener from S to +12BS to drain extra current in case the charge pump is doing "too well" (mainly due to voltage swing on the source).
Level shifting is required, for which a regular bootstrap type driver (IR2101 etc.) is fine, leave the low-side channel pulled to GND (or, use it to power the charge pump!) and drive the high side from logic.
But this is a lot of bother when you can just get an LT surge stopper chip ready to go or whatever. They're spendy but that's a lot of faffing around you save.
Oh, and... how to protect Q3's GS junction ? :-[
The... same way Q1 is??
Tim
-
I don't get it... is the pump above supposed to charge C20 to, what, 3x Vcc - 6x Vf ? I think it does not. How do I charge the final cap to 4 or 5 times Vcc ?
The diode Vdrop is proportional to current and usually you would use schottky diodes for this circuit to minimize loss.
Something like 1n5819.
The Vf would then be more like 0.2V per diode, especially once everything has charged and current is small.
Also, 10uF caps is too large. The MCU GPIO wont be able to handle that except at very slow switching speed which will make it perform very bad.
I would run between 1u to 100n caps and use a much faster speed, like 10-100khz.
Then you'd probably get around 25V from that. ( 4.8 + 4.6 + 4.4 + 4.2 + 4.0 + 3.8 )
Or even better, just get one of those little isolated 5V to 5V DCDC brick for $2 and you can have 5V referenced to whatever point you want. https://www.digikey.co.nz/en/products/detail/gaptec-electronic/1S4E-0505S1-5U/13691708 (https://www.digikey.co.nz/en/products/detail/gaptec-electronic/1S4E-0505S1-5U/13691708)
-
Lower caps, Schotkies, got it.
Then you'd probably get around 25V from that. ( 4.8 + 4.6 + 4.4 + 4.2 + 4.0 + 3.8 )
Or even better, just get one of those little isolated 5V to 5V DCDC brick for $2 and you can have 5V referenced to whatever point you want. https://www.digikey.co.nz/en/products/detail/gaptec-electronic/1S4E-0505S1-5U/13691708 (https://www.digikey.co.nz/en/products/detail/gaptec-electronic/1S4E-0505S1-5U/13691708)
So that pump does go beyond 2x Vcc, interesting.
Why are those ethernet bricks better ?
-
So that pump does go beyond 2x Vcc, interesting.
Ops, sorry. I'm wrong. it's only 3 stage not 6 stage.
(The 3 stages will be around 8.5V then 11.5V then output 15V with maybe 1-2mA loading. 100khz 1uF caps)
I should go to bed instead of posting on here.
Why are those ethernet bricks better ?
Because they are isolated so you can generate 5V relative to any point you want. Excellent for generating some amount of voltage above the voltage at the source pin of an n-channel mosfet (high side switch)
-
But this is a lot of bother when you can just get an LT surge stopper chip ready to go or whatever. They're spendy but that's a lot of faffing around you save.
Ok, having read through their datasheets only reaffirmed what I already knew. I am way to clueless to make this work. So yeah, a surge stopper, totally!
BUT!
Best there is: "Reverse Input Protection to –60V". In the meanwhile, ISO 7637-2-2011 talks about -220V spikes...
BUT(2)!
Such bidirectional TVS on input makes me think the thing can actually do -220V as well, as long as the TVS diode can handle the energy left after spike has been clamped:
[attach=1]
Oh, and... how to protect Q3's GS junction ? :-[
The... same way Q1 is??
Tim
Sort of like this ? I think 47k resistance should be fine for Q3. V_z = 20V as Q3 will likely have Vgs_max at that point.
[attach=2]
Thanks a bunch!
-
Best there is: "Reverse Input Protection to –60V". In the meanwhile, ISO 7637-2-2011 talks about -220V spikes...
BUT(2)!
Such bidirectional TVS on input makes me think the thing can actually do -220V as well, as long as the TVS diode can handle the energy left after spike has been clamped:
I already?... :palm: It's a low energy pulse! It's not plugging into the wall, a TVS is fine!
Sort of like this ? I think 47k resistance should be fine for Q3. V_z = 20V as Q3 will likely have Vgs_max at that point.
Uh, check the drain current before you wire that up...
Also, you've already got a 6.8V zener in there, why not like...
...Or you didn't understand the choice of that, before, either, but, why not ask instead of, I guess copy-pasting something else--?
Perhaps, I've overestimated your level; these examples seem roughly reasonable, but then your understanding of them seems quite lacking. Should this [subthread] actually be in the beginner section, say?
Not to belittle you, mind -- well, admittedly, maybe a little bit, in as much as giving a bite for this apparent discrepancy :P -- but still not much, after all, fake it long enough, you will eventually make it! :D -- but just from the point of: if you have a particular, immediate need, then please lay out those requirements, and let someone solve it for you, don't worry about understanding it; or if you have a general need for which understanding and knowledge is required, then show us where to start from and we can build upon that.
And these are not exclusive options [practical vs. theoretical], the one can motivate the other; but make it clear what your aim is, and be open to examples alongside, but different from, your particular need, as a means of illustrating how the outputs (component choice, values, topologies) can vary with the inputs (requirements).
Or more generally still: learning how to learn. Use your existing knowledge to guide the acquisition of new knowledge. This circuit could be discussed top-down (it has to do X, Y and Z, using T, U and V components in the process; which in turn operate as....), or bottom-up (we're inevitably going to need diodes, transistors, etc., and they operate in this and that way....), or anything inbetween. Figure out what works for you and guide the conversation as best suits you. :-+
Cheers!
Tim
-
So the thing will protect against negative 220 spikes as well, looks like a winner then! Why don't they make it more clear in the datasheet... :D
Just one more question about these surge stoppers, will they also protect fast Schotky diodes in series with them, when such diodes are placed before the protector IC, in line with the two power mosfets ?
Sir, copy pasting is what all great engineers do, not only the rookies; perhaps the source of confusion. I thought it was pretty obvious my level was not that high with all these questions.
P.S. - would have never thought to arrange Q3's Zener that way :-+
-
What do you need a schottky for if you have back-to-back MOSFETs?
Tim
-
I want some energy to run the MCU a bit after 12V went away. I guess I can put the storage cap + diode after surge stopper too, but probably same deal since they are in series.
-
Oh in a separate path, that's fine. The diodes won't care, but you get the full load dump out there so probably need a depletion MOS style voltage limiter.
Tim
-
Hello,
Tim, sir, can i have a bit more help please ?
I want to take this diagram:
[attach=1]
And adapt it to also guard against +100V load dumps besides negative -220 spikes. Peeking at:
[attach=2]
i would make the following changes:
- replace FETs with some 250V capable ones (https://www.tme.eu/Document/b1f0417c1825165ff0291afca4867b2a/IXFP60N25X3M.pdf?b244=b244)
- keep the SMAJ58CA but use it like the SMAT70A from the second diagram, together with an 1W resistor
- replace Rsns with a 6 mohm value and a max current of 50 mV / 6 mohm ~= 8.33 A
Any obvious faults ?
Thanks
-
Hmm, minimum rating is -60V, you'll have to do something about that. Can you not afford a series diode?
Tim
-
Is the LT4356 not going to survive -220 shocks because of its GND, TMR and FB connections ?
Series diodes get hot really quickly at my amperage, and device will be mounted in crammed hot place already.
-
Maybe just use the "passive" PMOS polarity protection thing in front, not integrating it with the LT chip.
Also, again, you're worrying about 220V like the terminal is going to reach -220.00V with respect to ground? You know what impedance means right? It's not plugging into mains here.
Tim
-
220V PMOS @8A would be worse than series diodes... best I could find was Rds_on of 0.5 ohm (typical value)
Hmm, impedance, let me check - Ri = 50 ohm and Us = -220V, in series with all of the grounded surge stopper IC pins... no, in series with... :-//
But why do they say protection up to -80V if the chip itself can only do -60V ? Page 21 here (https://ro.mouser.com/datasheet/2/609/LT4356_1_4356_2-1269492.pdf), first diagram titled "Overvoltage Regulator with Reverse Input Protection Up to –80V"
Side note, those IRLR2908 are rated to 80V.
-
Yeah like that. Note also 2N3904 rating. Or MMBT3904 more likely. 1N914 for that matter.
But why do they say protection up to -80V if the chip itself can only do -60V ?
Vin doesn't directly touch any IC pins...
Tim
-
1n914 ?
Hmm, minimum rating is -60V, you'll have to do something about that
Ok, so... if chip does not touch Vin, where is the problem ?
-
Well, I ain't hell bent on understanding what's the matter. Would be nice if I would though.
In the meanwhile, how about using the MCU for protecting the bigger load ?
[attachimg=1]
Q1 = 250V N channel MOSFET
Q2 = logic level N MOSFET
D1 = standard variety 0.5 - 5 kW TVS
D2 = probably not needed, some very low value Zener
MCU has a 330 uF reservoir so it will still operate even on a negative spike; its 5V is protected separately. This ought to work, right ?
LE: actually Q1 will be two back to back MOSFETS