Stacking Caps versus Stacking Supplies

[luvdunhill]

So, I had a bit of fun last night and experimented with stacking a small (valued) cap with a large valued cap to get extra WVDC capacity from the two caps in series. Works like a champ and it looks that my little psu scales up quite nicely. Of course, now I see where the danger lies, in that the sleeve of one capacitor can be near lethal to touch. Also, unequal values seemed to work better than a pair of equal values, which kinda makes sense.

So, I started thinking... Would just strapping a pair of HV supplies, say two 400v supplies to make a single 800v supply be a better approach (or four 400v supplies to make a +- 800v supply, etc.) This way I can use 500v caps and still have the flexibility of scaling up to higher psu voltages.

Thoughts?

[Pars]

Maybe I'm not thinking thru this correctly, but if you stack two supplies to achieve your 800V, won't the "ground" (i.e., the filter cap - and case, etc.) of the upper supply effectively be at +400Vdc? Just like stacking caps? Or will the - terminal of the upper supply short out the bottom supply?

Need more coffee

[JBLoudG20]

Maybe I'm not thinking thru this correctly, but if you stack two supplies to achieve your 800V, won't the "ground" (i.e., the filter cap - and case, etc.) of the upper supply effectively be at +400Vdc? Just like stacking caps? Or will the - terminal of the upper supply short out the bottom supply?

Need more coffee

Yes it will effectively be at 400v, but the voltage potential at that point will still be 400v when referenced to the top supplies' rails. Same for the lower supply. So when you reference the top supply's + and the lower supplies gnd, the potential is 800v.

I'm sure Kevin can chime in to decide if it is a better approach though, luvdunhill

[luvdunhill]

heh, I have no idea.... hence the question.

Also, what are the disadvantages for stacking supplies? series output impedance is one I can think of...

[n_maher]

I thought that I remembered someone talking about doing this with TREADs or STEPS before and that for those applications it wouldn't work. Is it possible that this ability is application specific?

[JBLoudG20]

I thought that I remembered someone talking about doing this with TREADs or STEPS before and that for those applications it wouldn't work. Is it possible that this ability is application specific?

Its a distinct possibility. Assuming you want to reference ground, one possibility for error that I can think of would be psu's which have a virtual ground. 2426, or cap referenced.

[Pars]

Thinking about it a bit more, I think you need floating supplies for this (where the - is not ground referenced). Kevin could show you how these are done, I'm sure. I could do it with my bench supply (well, not the 400V part), but as Nate mentioned, your normal Tread/Steps/sigma are ground referenced, so the - of the top supply is going to short out the + of the bottom one, leading to some potentially spectacular results

[Pars]

Well, they aren't 800V, but I thought one of you could make something really dangerous with some of these...

[kevin gilmore]

You can make a supply the same way as my original ac coupled tube

amp. That is use a center tapped secondary, or in fact two seperate

windings each of which rectified makes something like 400 volts with

450 volt capacitors. Then stack the two and make the bottom ground.

Only issue is whether or not the transformer has enough isolation for

the top winding to sit 400 volts above ground. Normally these transformers

are rated at 1kv or more, so it is not a problem.

[justin]

unless it is a plitron, then you have tell them

[luvdunhill]

Justin/Kevin:

I am using two 350VAC Plitron toroids at the moment.

If you're referring to primary to secondary insulation, looks like I'm covered. According to the Plitron website:

"Meets the test requirement of >4 kV for one minute. 60 Hz transformers, three layers minimum, >2.5 kV for one minute"

So, what about stacking four toroids (and their respective psus) together to get +- 800v?

So, is this level of insulation something that one can get from a EI-type transformer as well?

[kevin gilmore]

Justin/Kevin:

So, is this level of insulation something that one can get from a EI-type transformer as well?

Absolutely. But also custom. I have filament transformers for xray systems that

sit at -60kv. The higher voltage they can withstand, the larger they get.

[luvdunhill]

Absolutely. But also custom. I have filament transformers for xray systems that

sit at -60kv. The higher voltage they can withstand, the larger they get.

Kevin:

with two supplies in series per rail (so 4 supplies), each with n smoothing capacitors each of size C, would the capacitance per rail seen by the circuit would be (C1+C2+...+Cn)/n?

[luvdunhill]

wrong...

trying again, the total capacitance seen by the circuit per rail will be as if the all capacitors for that rail are in series, thought there will not be the issues with the live cases if each of the 4 supplies has their own secondary and rectification.

[kevin gilmore]

Capacitors in series are CAPACITORS IN SERIES no matter how you wire

the power supplies that feed them. So 2 x 100uf /450 volt caps in series

is the equivalent of 1 x 50 uf 900 volt cap.

If all the caps are value C, then the total capacitance is C/n

[luvdunhill]

thanks!

I may be getting some giant polypropylene in oil caps to use not as cool as your teflon caps, but we have to start somewhere:

[img width=150 height=113 alt=PICT0096]http://www.head-case.org/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=2831&g2_serialNumber=2&g2_GALLERYSID=245ffaedcdd434c3e031cadbc1e3669e