Calculate with maximum current the standard breakers in homes can handle and a margin. It is not the load that counts, it is the maximum current that the installation may deliver when something goes wrong. It would be good advice to study contemporary DC blockers and the design choices that are made.
Last edited:
Quick calculation example to start with:
if taking a common 16A circuit to start from, and using those beefy schottky rectifiers with .55V drop, we're talking about 0.034375ohms to have across those diodes before they start to conduct.
So the goal would be to have enough capacitance to have no more than that much in impedance.
And that would be about 93mF (milliFarads) or so. That is using just big caps in parallel with the diodes.
I would use 2 sets of those diodes, 2 by 2 back to back (4 total). They have a current handling of 15A (200A surge), so 2 together is 30A, each direction (2 diodes each way).
That is quite a margin above whatever usage.
Even a 2kVA toroid doesn't come close to 16A draw when fully loaded in normal regime.
Only during the start up sequence the current could be somewhat above that, but here again, we're using a soft start that pretty much removes all of the inrush, and we can make it so it never allows much more than the max full load current anyway.
Well within any normal circuits and a 16A would be no problem.
So of course some 93mF is quite a bit to come up with, and would be rather large on pcb, which is why looking at supercaps becomes interesting.
The caps will never be exposed to more than those .55V across the diodes, and with this much capacitance, this should not even happen, and the current should remain well below the circuit's 16A max.
This should be enough margin as is to not be put in trouble. Even in case of some abuse, the breaker would trip and the surge would not last very long and is within the capabilities of the diodes. And the caps don't care that much, as the diodes are doing the work in case of such abuse.
What could I be missing here?
if taking a common 16A circuit to start from, and using those beefy schottky rectifiers with .55V drop, we're talking about 0.034375ohms to have across those diodes before they start to conduct.
So the goal would be to have enough capacitance to have no more than that much in impedance.
And that would be about 93mF (milliFarads) or so. That is using just big caps in parallel with the diodes.
I would use 2 sets of those diodes, 2 by 2 back to back (4 total). They have a current handling of 15A (200A surge), so 2 together is 30A, each direction (2 diodes each way).
That is quite a margin above whatever usage.
Even a 2kVA toroid doesn't come close to 16A draw when fully loaded in normal regime.
Only during the start up sequence the current could be somewhat above that, but here again, we're using a soft start that pretty much removes all of the inrush, and we can make it so it never allows much more than the max full load current anyway.
Well within any normal circuits and a 16A would be no problem.
So of course some 93mF is quite a bit to come up with, and would be rather large on pcb, which is why looking at supercaps becomes interesting.
The caps will never be exposed to more than those .55V across the diodes, and with this much capacitance, this should not even happen, and the current should remain well below the circuit's 16A max.
This should be enough margin as is to not be put in trouble. Even in case of some abuse, the breaker would trip and the surge would not last very long and is within the capabilities of the diodes. And the caps don't care that much, as the diodes are doing the work in case of such abuse.
What could I be missing here?
How about that one: FM0V104ZF
On mouser.
It's really small for its value. 100mF in such a tiny footprint. Not bad. Only 3.5V, but that's more than enough for this.
It's polarized, so 2 of them needed back to back.
On mouser.
It's really small for its value. 100mF in such a tiny footprint. Not bad. Only 3.5V, but that's more than enough for this.
It's polarized, so 2 of them needed back to back.
I guess the real question about using supercaps for a dc blocker is: can they handle the continuous strain from that ac current flowing through?
Looking at the established normal regime, even full load, a 2kVA toroid would draw something like 8.5A or so. Would a supercap handle this sustained regime properly without risking failure?
If not, I wonder what the real use for supercaps would be then.. If they can't handle such usage, what are they good for?
Looking at the established normal regime, even full load, a 2kVA toroid would draw something like 8.5A or so. Would a supercap handle this sustained regime properly without risking failure?
If not, I wonder what the real use for supercaps would be then.. If they can't handle such usage, what are they good for?
No need for proportionality with a triac (or SCR): in fact, it would rather be the opposite: if the load is insufficient (or inductive), the device might not hold properly, which is why it is necessary to apply a generous trigger, both in duration and strength
35A Graetz bridges are normally chosen in case something goes wrong. Again, the load is not that important. It is the current the installation can deliver. The home installation can deliver way more current than 16A for a short time too... That is at least how I learnt to deal with matters. Normally one counts with 10 to 15 times I nominal inrush current with transformers as well.... That is more than the "somewhat" you describe and what if the soft start circuit fails?
About parallel diodes, when overloaded one likes to fail first as with high currents a small difference counts. I think the classic circuit with a 35A diode bridge was not invented just like that.
Anyway I see I am old fashioned but that is because I see failures a lot. May I suggest you ask the manufacturer of the supercaps if your application is a wise idea?
About parallel diodes, when overloaded one likes to fail first as with high currents a small difference counts. I think the classic circuit with a 35A diode bridge was not invented just like that.
Anyway I see I am old fashioned but that is because I see failures a lot. May I suggest you ask the manufacturer of the supercaps if your application is a wise idea?
Last edited:
Many old household electrical installs are still using fuses instead of breakers, and those do have a more serious reaction delay to surges.
The issue I see using a bridge for the dc blocker is it provides 2 diode drops, not 1, and the drop is more like 1V per diode, so we're talking about having some 2V threshold for dc removal, and 2V is more than enough to make some toroids very unhappy and grouchy.
I was only trying to use an arrangement using only 1 diode drop, and using schottky rectifiers, the drop is even lower than 1V.
The schottky rectifiers I found have .55V drop, so that lowers the dc removal threshold quite a bit, hopefully bothering less toroids.
Now using a bridge also does put 2 diodes in parallel anyway, which is what I was doing with the single diodes, same thing.
Turns out, those schottky diodes are rated for a surge current of 300A, not 200A, so quite sturdy to handle some short term abuse. Having 2 such diodes, each way, and we're at some 600A of short term surge handling.
The issue I see using a bridge for the dc blocker is it provides 2 diode drops, not 1, and the drop is more like 1V per diode, so we're talking about having some 2V threshold for dc removal, and 2V is more than enough to make some toroids very unhappy and grouchy.
I was only trying to use an arrangement using only 1 diode drop, and using schottky rectifiers, the drop is even lower than 1V.
The schottky rectifiers I found have .55V drop, so that lowers the dc removal threshold quite a bit, hopefully bothering less toroids.
Now using a bridge also does put 2 diodes in parallel anyway, which is what I was doing with the single diodes, same thing.
Turns out, those schottky diodes are rated for a surge current of 300A, not 200A, so quite sturdy to handle some short term abuse. Having 2 such diodes, each way, and we're at some 600A of short term surge handling.
You could consider a transformer type not as sensitive and broadband as toroids. The toroids need a DC blocking circuit and definitely a mains filter ... Choosing a better type will reduce hum as toroids always hum a little, eliminate the need for a DC blocker, reduce the need for a mains filter (although I would still use one)..
Sorry but your calculation of parallel diodes is off. As diodes are not 100% equal one (series pair) of them will take the load. Still, 300A is a lot. You see, I work with mains circuits a lot and heavy devices connected to mains. The power of a short circuit when something fails can be impressive. We consider the load a black box and calculate the other way around.
Sorry but your calculation of parallel diodes is off. As diodes are not 100% equal one (series pair) of them will take the load. Still, 300A is a lot. You see, I work with mains circuits a lot and heavy devices connected to mains. The power of a short circuit when something fails can be impressive. We consider the load a black box and calculate the other way around.
Last edited:
This thing about parallel diodes applies to the ones in bridges too. So, same thing there. I don't think there are that many bridges where the effort was made to pair up the diodes inside, so there will be differences between diodes inside bridges just as much as single discrete ones.
And as mentioned, using a bridges puts 2 diode drops there, and 2 sets of 2 diodes in series, that's all it does. The only reason why bridges are so popular for this, is the convenience of a single package rounding up everything in it.
And we all know that even 1V of dc can make some toroids grouchy.
Even a pretty big toroid like a 2kVA for example, only has a full load current in the order of some 8.5A or so, which is quite within the 16A carrying capacity of the most common electrical circuit, and so if aiming for the 16A circuit, and adding a comfortable margin on top of that, it "should" do the job. In any case, although a short might be brutal, it's still short lived and the parts chosen can handle that.
The only thing I'm wondering about is how supercaps compare with electros when it comes to that ac current handling. Electros are bulky, while supercaps quite tiny in comparison.
And as mentioned, using a bridges puts 2 diode drops there, and 2 sets of 2 diodes in series, that's all it does. The only reason why bridges are so popular for this, is the convenience of a single package rounding up everything in it.
And we all know that even 1V of dc can make some toroids grouchy.
Even a pretty big toroid like a 2kVA for example, only has a full load current in the order of some 8.5A or so, which is quite within the 16A carrying capacity of the most common electrical circuit, and so if aiming for the 16A circuit, and adding a comfortable margin on top of that, it "should" do the job. In any case, although a short might be brutal, it's still short lived and the parts chosen can handle that.
The only thing I'm wondering about is how supercaps compare with electros when it comes to that ac current handling. Electros are bulky, while supercaps quite tiny in comparison.
Yes it applies to Graetz bridges too that is why only one (series pair) is used in calculations. This is not the same as you suggest, it is you that calculate both paths. The beauty of AC is that only 1 series pair will conduct depending on the sinus. A short circuit will also work out differently but it will be hefty nonetheless.
And no I don't think the currents involved will work out OK with tiny supercaps when something goes wrong.
And no I don't think the currents involved will work out OK with tiny supercaps when something goes wrong.
Last edited:
As a DC blocker I use two antiparallel 47m/10V elcos protected by diode bridge. Followed by soft-start, mains filter and fuses.
It absolutely does matter a lot where you put fuses. From mains side seen it should start with a fuse, then a switch/mains filter (most effective to use the filter as a first stage) and then the rest. This protects the cabling of the house installation and it secures a blown fuse when anything in the chain fails like switch, mains filter, DC blocker or soft start. If you put fuses last in the chain a defective circuit will possibly overload/short the house installation and we then rely on the circuit breakers only. Thankfully regulations forbid this and no sane person builds stuff like that (I hoped) till I saw an installation last week that did not have a switch or a fuse.
Last edited:
fuse in iec inlet (or panel), two-pole mains switch (rear panel), rest on pcb: dc-blocker, soft-start, filter, fuses for tr primaries
Last edited:
One nice byproduct of putting the fuse as far "upstream" as possible, is that it allows a downstream clamp device like a MOV or a GDT to rupture the fuse (and protect the equipment) when an overvoltage surge occurs.
That is a very good idea as now both the home installation and the device are optimally protected. I think a GDT is to be preferred as MOV's used to be noisy (at least last time I used them). Coincidentally I have a GDT laying in front of me on my desk 🙂
* I would not mind, just like Elektor used to do, that there would be a safety page with a minimum of required safety measures when dealing with mains related DIY projects at diyaudio.com. A standard IEC inlet PCB with GDT/MOV provision, pads for a 16A mains filter and connector for mains switch or maybe an IEC inlet also with power switch as a minimum would make the hobby safer. I see the strangest of dangerous contraptions of which audio hobbyists apparently feel not ashamed of. I don't understand that one can build mains connected devices without even knowing what PE is.
That PCB could be sold with enclosed isolated spades and the required parts (with the crimping tool for crimping as an option). Maybe exaggerated but it seems safer than the current average. Mains filtering is no luxury anymore as well.
* I would not mind, just like Elektor used to do, that there would be a safety page with a minimum of required safety measures when dealing with mains related DIY projects at diyaudio.com. A standard IEC inlet PCB with GDT/MOV provision, pads for a 16A mains filter and connector for mains switch or maybe an IEC inlet also with power switch as a minimum would make the hobby safer. I see the strangest of dangerous contraptions of which audio hobbyists apparently feel not ashamed of. I don't understand that one can build mains connected devices without even knowing what PE is.
That PCB could be sold with enclosed isolated spades and the required parts (with the crimping tool for crimping as an option). Maybe exaggerated but it seems safer than the current average. Mains filtering is no luxury anymore as well.
Last edited:
There is one point about dc blockers that needs to be made:
Many seem to think any of the "normal" current MUST be passing through the caps, and only anything beyond that would go through the diodes. So basically meaning the diodes would only be meant to conduct fault currents.
This isn't quite right. The diodes can very well conduct some of the normal current, well below anything fault or surge related. So this would mean many are aiming a little too high and are thus using oversize caps, for nothing, except to make the circuit bigger and more expensive.
And I went looking again at ESP's pages about this stuff, and it's quite clear he's not advocating using such very large caps and the diodes should conduct with normal current.
There is no reason that I can see against this. So there is no good valid reason to use overly large caps.
Using beefy diodes is definitely recommended, but most are quite capable of handling very high short term surge current that are well beyond their nominal carrying capacity, long enough to let a fuse blow or breaker trip.
And one more thing, about the limits of mains circuits, which are pretty much set by the breakers or fuses: the actual limits to look at aren't the mains circuit's but rather the device or unit which should have its own built-in breaker or fuse, which should be chosen according the the unit and not the mains circuit's limits.
In case of fault, the mains circuit's breaker or fuse doesn't have to be tripping or blowing, as it's the one in the unit that should go first, as the caliber should be set right for the unit and should be well below the limits of the mains circuit.
Many seem to think any of the "normal" current MUST be passing through the caps, and only anything beyond that would go through the diodes. So basically meaning the diodes would only be meant to conduct fault currents.
This isn't quite right. The diodes can very well conduct some of the normal current, well below anything fault or surge related. So this would mean many are aiming a little too high and are thus using oversize caps, for nothing, except to make the circuit bigger and more expensive.
And I went looking again at ESP's pages about this stuff, and it's quite clear he's not advocating using such very large caps and the diodes should conduct with normal current.
There is no reason that I can see against this. So there is no good valid reason to use overly large caps.
Using beefy diodes is definitely recommended, but most are quite capable of handling very high short term surge current that are well beyond their nominal carrying capacity, long enough to let a fuse blow or breaker trip.
And one more thing, about the limits of mains circuits, which are pretty much set by the breakers or fuses: the actual limits to look at aren't the mains circuit's but rather the device or unit which should have its own built-in breaker or fuse, which should be chosen according the the unit and not the mains circuit's limits.
In case of fault, the mains circuit's breaker or fuse doesn't have to be tripping or blowing, as it's the one in the unit that should go first, as the caliber should be set right for the unit and should be well below the limits of the mains circuit.