It seems a little safety warning is in order.
Potential very dangerous... don't even think about it !
Suitably rated diodes are the circuits safegaurd. It MUST be safe under all conceivable conditions, fault or otherwise. What about startup. What about one cap changing value or going high ESR ?
It must have been linked to several times but this covers it all,
Mains DC and Transformers
I agree. Unfortunately electrolytic capacitors in reverse mode can work with max. 1VDC, independend of the reading rated voltage. Therefore I am looking for the best topology for an DC filter.
What would you prefer from the circuits found by the links at my post 223 ? go to
http://www.diyaudio.com/forums/solid-state/2080-dc-filter-23.html
Why isn't a wire between the two positive legs of the electrolytics and the diodes at figure 8 about follow URL ??
http://sound.westhost.com/articles/xfmr-dc.htm
I have heard, the serial mode of caps represents a risk as long no resistors in parallel mode are in use.
About the PDF
http://www.epcos.de/web/generator/W...__nn.pdf;/PDF_GeneralTechnicalInformation.pdf
I read follow on page 28:
6.2 Series connection of aluminum electrolytic capacitors
When designing series circuits with aluminum electrolytic capacitors, care must be taken to en-
sure that the load on each individual capacitor does not exceed its maximum permissible voltage.
Here, the fact that the total DC voltage applied is divided up among the individual capacitors in
proportion to their individual dielectric insulation resistances (figure 23) must be taken into consid-
eration.
Since the dielectric insulation resistance of the individual capacitors may differ quite strongly, the
voltage distribution may also be non-uniform, which may lead to the permissible voltage of individ-
ual capacitors being exceeded. For this reason, forced balancing of the voltage distribution is rec-
ommended. The safest method of achieving this is to use electrically isolated voltage sources for
the individual capacitors as shown in figure 24.
If this is not possible, external balancing resistors RSymm (see figure 25) can be connected to the
individual capacitors. The balancing resistances must be equal to one another, and must be sub-
stantially lower than the dielectric insulation resistance of the capacitor.
Experience has shown that it is preferable to choose balancing resistance values that will cause a
current of approximately 20 times the leakage current of the capacitor to flow through the resis-tors. The equation for calculating the resistance value is "Rsymm = 100MegOhm x uF x 1/Cr"
The balancing measures described above may be omitted in cases where the total DC voltage to
be applied is substantially lower than the sum of the rated voltages of the capacitors to be used.
Experience has shown that this is possible for n = 2 to 3 single capacitors in series without any
considerable risk if the total voltage does not exceed 0.8 n VR. However, this solution can only
be implemented if the series circuit consists of matching capacitors (same type, same capaci-tance), so that the dielectric insulation resistance of the capacitors, which is the only factor deter-mining the voltage distribution in this case, will not vary too greatly from one capacitor to the next.
Last edited:
Very quickly 
I prefer the Rod Elliot version. There's no wire between the caps and bridge because to do so would allow one cap to be reverse biased if there were an offset. I must admit that if I were building something like this I would consider two high voltage (and they MUST be high voltage specified) resistors to equalise any imbalance caused by unequal capacitance. Is that whats in the pdf ? The similar circuit that does have the wire link also has four rather than two caps. The one with the transorb device looks as though it could reverse bias the cap.
The one with a 4.3 volt transorb device seems a bit obscure and looks like it could reverse bias the cap.
I prefer the Rod Elliot version. There's no wire between the caps and bridge because to do so would allow one cap to be reverse biased if there were an offset. I must admit that if I were building something like this I would consider two high voltage (and they MUST be high voltage specified) resistors to equalise any imbalance caused by unequal capacitance. Is that whats in the pdf ? The similar circuit that does have the wire link also has four rather than two caps. The one with the transorb device looks as though it could reverse bias the cap.
The one with a 4.3 volt transorb device seems a bit obscure and looks like it could reverse bias the cap.
Very quickly
I prefer the Rod Elliot version. There's no wire between the caps and bridge because to do so would allow one cap to be reverse biased if there were an offset. I must admit that if I were building something like this I would consider two high voltage (and they MUST be high voltage specified) resistors to equalise any imbalance caused by unequal capacitance. Is that whats in the pdf ?
No, but the imbalance of ESR (Z) and dielectric insulation resistance of different capacitors - if I understand this right.
interesting news there are to read from post #88 about
http://www.diyaudio.com/forums/powe...buzzing-toroid-transformers-what-right-9.html
http://www.diyaudio.com/forums/powe...buzzing-toroid-transformers-what-right-9.html
I am also planning to use figure 8 from Rod Elliot's article.
And I am going for Nippon electrolytics, KY series, rated at 4700uF/25V, 10.000 hours at 105 C, 4.220 mA ripple at 105 C. Or even the GPA series equivalent, having 3.730 mA ripple capability at 125 C.
Would the aforementioned caps suffice?
And I am going for Nippon electrolytics, KY series, rated at 4700uF/25V, 10.000 hours at 105 C, 4.220 mA ripple at 105 C. Or even the GPA series equivalent, having 3.730 mA ripple capability at 125 C.
Would the aforementioned caps suffice?
Depends on what transformer you have audiostrat.
I will also go for the figure eight, but ends up with 2X10000uF 6,3V times two.
Will be used in my Goldmundclone.
Getting to be a quite nice build, two 625VA transformers should be good as gold.
And after having tested them today, I find that the DC-component on the powerline actually have something to do with the hum and noise. And the idle powercvonsumption.
Never concidered it a problem earlier. Strange.
But as I always state: One learns something every day, the whole life through.
I will also go for the figure eight, but ends up with 2X10000uF 6,3V times two.
Will be used in my Goldmundclone.
Getting to be a quite nice build, two 625VA transformers should be good as gold.
And after having tested them today, I find that the DC-component on the powerline actually have something to do with the hum and noise. And the idle powercvonsumption.
Never concidered it a problem earlier. Strange.
But as I always state: One learns something every day, the whole life through.
Well, after having looked around I find that Panmasonic have a nice cap for this. FM series
2700uF 16V. 12,5mm diameter, 30mm high. Lifespan 8000Hrs/105°C
Ripplecurrent 3630mA When putting four of theese in paralell, adding another bank of four caps in series as figure 8 indicates, and there it is. Whopping 14520mA ripple current.
And this for a price not exceeding $8,- Nice.
Rectifierbrigdes of 35A is in my workshop since long ago.
So there we are, ready built at a price close to a fraction of most DC-filters on the market I will have my toroids working under the very best conditions.
2700uF 16V. 12,5mm diameter, 30mm high. Lifespan 8000Hrs/105°C
Ripplecurrent 3630mA When putting four of theese in paralell, adding another bank of four caps in series as figure 8 indicates, and there it is. Whopping 14520mA ripple current.
And this for a price not exceeding $8,- Nice.
Rectifierbrigdes of 35A is in my workshop since long ago.
So there we are, ready built at a price close to a fraction of most DC-filters on the market I will have my toroids working under the very best conditions.
Last edited:
Well, the transformer will be a 160VA-200VA toroid.
I know, a bit small! So these caps do for it, right?
In any case, shouldn't the ripple rating be the maximun steady current the primary could draw? In my case, roughly 0.7-0.9A - say 1A.
As already said, You shouldn't have any hum-problems caused by DC on your powerline. Excactly where the limit is I am not sure, but around 500VA I brelieve.
If You build according to Elliots figure 8 the diodebrigde will be the one actually carrying the power. Already at about 1% on the climb on the sinecurve the diodes are "on the case" and they stay there until the voltage is below their triggervoltageX2.
So Your choise of components are adequate in any case.
But check what DC-resistanse your primary windings have.
Will easily determine wether You will have benefit of this filter or not.
vdi_nenna: I don't have any problems yet since I have not constructed the build. I was thinking to put it on, mainly for trial, to take measurements. And of course, to reduce off-load consumption - on Rod Elliot's site, it is clear that you don't need the dc component not only for audible reasons, but also for unnecessary magnetising currents. Even if small, bigger than normal.
TANDBERGEREN: This is what I am planning to do!
I was thinking to put it on, mainly for trial, to take measurements. And of course, to reduce off-load consumption - on Rod Elliot's site, it is clear that you don't need the dc component not only for audible reasons, but also for unnecessary magnetising currents. Even if small, bigger than normal.TANDBERGEREN: This is what I am planning to do!
Last edited:
Notice that this is about applying DC voltage. This application is AC voltage.
Has anyone tried one of these for removing DC?
L C Audio Technology / Front Page
I would prefer to buy a premade unit I think, save flaffing around
L C Audio Technology / Front Page
I would prefer to buy a premade unit I think, save flaffing around