I'm going to compare the use of the transformer in the EHT post: https://forum.allaboutcircuits.com/...-and-construction.113504/page-69#post-1248540

to one of the Antek transformers.

As I understand it, the transformer in the EHT post (I'll refer to it as transformer X) was to be a 1:1 isolation transformer modified to become a 2:1 transformer. The voltage at its secondary would be 1/2 the voltage applied to the primary by the variac.

Capacitor input power supplies output a DC voltage about equal to the peak of the sine wave voltage applied to the bridge rectifier, minus about 2 volts drop in the rectifier (I'm going to neglect voltage drops due to copper losses in the variac and transformer). The peak voltage is 1.414 times the secondary AC voltage, so if 120 VAC (that's RMS, and the variac can actually output about 140 VAC when turned all the way up) were applied to the primary of transformer X, we would get 60 VAC at the secondary and this would result in about (60*1.414) - 2 = 82.84, or about 83 volts applied to the filter capacitors. This is way more than the 50 volt rating of those capacitors, and also way more than the desired maximum 50 volts out. The variac would have to output less than 120 VAC to the input of transformer X so the voltage applied to the bridge rectifier would be less than (50 + 2)/1.414 = 36.8 VAC, which means that the output of the variac should be less than 73.5 VAC. The users would need to be careful about this. Inevitably somebody will turn the variac up too high, and blow up their filter caps if they are rated for 50 volts; Aleph mentioned this problem. Also, this arrangement underutilizes transformer X and the variac.

Using a transformer with the secondary wound to output the desired voltage when the full 120 VAC is applied to the primary is a better option. The Antek tranformers cover a suitable range of output voltages and power capabilities.

One of the Antek 1000 VA transformers would seem to be just about right; the price is right, too. The AN-10435 is rated for a nominal 35 VAC out and about 31 amps. The AN-10440 is rated for a nominal 40 VAC out and about 25 amps; this will give a little over 50 volts DC when unloaded. The AN-10435 wouldn't give a full 50 volts DC out with 120 VAC in, but the variac can be turned up to apply more than 120 VAC to the primary, and the manufacturer suggests that 20% more than rated output can be drawn "without any problem". Using the 35 VAC AN-10435, we could get about 55 volts DC on the caps under no-load conditions if the variac is turned up all the way. It might be necessary to use 63 volt rated capacitors if the users can't be trusted to not turn the variac up all the way with no load.

Only HP knows if the full 50 volts DC will be needed, and how often, or for how long.

Because the output of a suitable Antek wouldn't be able to apply way more than 50 volts to the filter capacitors, we probably can safely use 50 volt (or 63 volt to be extra safe) rated filter capacitors. This would lead to the happy circumstance that the voltage applied to those filter capacitors would be more nearly equal to their ratings, and we wouldn't have to worry about them de-forming due to too low applied DC voltage.

Now I'm getting more interested in this project, so I'm going to order an AN-10435 to make some measurements.

 

Again, many thanks for your continued, and, so it seems, burgeoning interest

Please be advised that we will be (and are) strongly recommending the 'Antek route'...
That said, for the 'benefit' of those already possessed of suitable transformers (and inveterate, impecunious or 'dangerously masochistic' DIYers) I will present a brief illustrated rewind example (Spec completion of Aleph(0)'s 'project') -- if only to illustrate the inadvisability of DIY radical modification.

But to your post:

As I understand it, the transformer in the EHT post (I'll refer to it as transformer X) was to be a 1:1 isolation transformer modified to become a 2:1 transformer. The voltage at its secondary would be 1/2 the voltage applied to the primary by the variac.

Capacitor input power supplies output a DC voltage about equal to the peak of the sine wave voltage applied to the bridge rectifier, minus about 2 volts drop in the rectifier (I'm going to neglect voltage drops due to copper losses in the variac and transformer). The peak voltage is 1.414 times the secondary AC voltage, so if 120 VAC (that's RMS, and the variac can actually output about 140 VAC when turned all the way up) were applied to the primary of transformer X, we would get 60 VAC at the secondary and this would result in about (60*1.414) - 2 = 82.84, or about 83 volts applied to the filter capacitors. This is way more than the 50 volt rating of those capacitors, and also way more than the desired maximum 50 volts out. The variac would have to output less than 120 VAC to the input of transformer X so the voltage applied to the bridge rectifier would be less than (50 + 2)/1.414 = 36.8 VAC, which means that the output of the variac should be less than 73.5 VAC. The users would need to be careful about this. Inevitably somebody will turn the variac up too high, and blow up their filter caps if they are rated for 50 volts; Aleph mentioned this problem. Also, this arrangement underutilizes transformer X and the variac.

Using a transformer with the secondary wound to output the desired voltage when the full 120 VAC is applied to the primary is a better option.

Agreed! -- Hence my 're-consideration' of appropriate expedient filter working EMFs (CIP 250VDC) (as discussed in THIS post -- relevant excerpt reposted below) -- while said selection provides a broad 'safety margin' under maximum unloaded EMF conditions (which --neglecting rectifier offset-- being ≈ 85V [variac wired for 'coextensive' operation] or ≈ 99V ['boost' operation]) -- I must confess my 'lingering' dubiety as regards protracted 'sub-working EMF' operation of electrolytic capacitors...


*****Beginning of excerpt Re: LVPSU filter capacitor selection*****
Having reflected upon Aleph's well considered suggestion --and my pathetic expedient <<please see original post for context>> - I know I can do better

Following significant (non-original) research and consultation with several manufacturers - it seems 15 millifarad @250V electrolytics will be wholly acceptable in this application -- That said, I nonetheless strongly advise allowing the caps to 'soak' at Ca. 85-100V (i.e. max PSU OC output) occasionally...

Example of acceptable capacitors -- inexpensively purchased new or readily salvageable (in good condition) from multi-killowatt power conversion systems (e.g. inverters/UPSs).

****End of excerpt****

Only HP knows if the full 50 volts DC will be needed, and how often, or for how long.

Ideally I'd like to 'see' 0-150V @ 30A continuous (DC output) -- while such capacity might readily be realized via use of two appropriately connected isolation transformers preceded by a 'muscular' variac - it is my 'sense' that the cost (to say nothing of the weight) of such an arrangement would 'disappoint' a significant portion of our readership --- As an aside - While I'm certain all active participants on this thread are aware that non-abusive implementation of a variac requires observance of its maximum current specification without regard to output EMF adjustment (a consideration likely to prove burdensome to those restricted to asymmetrical [i.e. 120V only] service)-- I strive to be 'lurker friendly'

For all that - up to 50V @ 25A (intermittent) and up to 50V @ Ca. 15A continuous will be quite sufficient for the nonce

For the benefit of the curious: While employment SMPS techniques would be with the advantages of increased efficiency, readily implemented load EMF regulation as well as significantly lower cost, size and weight - experience has shown such to be dubiously applied to development of EHT systems... Supplying finished designs is, of course, quite another matter! -- By way of analogy (Re: the former point) - 'flack vests' are all of heavy, uncomfortable, ugly and... Oh yeah! Bullet-proof!

Now I'm getting more interested in this project, so I'm going to order an AN-10435 to make some measurements.

Thank you! Your interest, insight and assistance are greatly appreciated!

I'm afraid Powervar transformers' haphazard winding patterns look downright sloppy!

Oh my! -- Saints preserve your sanity whilst beholding my 're-wind' of @Aleph(0) 's transformer! -- Which being almost as 'tidy' as your hair (as worn in your avatar)!

FWIW some of the 'sloppiness' apparent in even industrially-wound toroidal products owes to the fact that --would you believe?-- the inner circumference is less than the outer circumference -- Over-sized cores might help that - but...


Very best regards
HP

 

The technology is rather prost:
1 step - calculate the Focault effect layer deep. The wire must be less that double of this figure in diameter.
2 step - calculate the approx turn count and find the approx wire length. Allpy roughly 20% more wire as calc shows.
3.step - find the damn long place where people are absent and beat in the two nails in both wall ends. Crincle the wire of chosen diameter between nails to form the optimal total wire cross section for J=2...3 A/mm2
4 step - one end of bucket put into screw-machine and turn it on. Wait while there will happen about 1 loop per centimeter or slightly more. Dont be excessive, better less as too much.
5 step - wind those litcendrath on the textolyte plank, about inch wide and 40 cm long with V-shape gap in both ends. Dont know how to label this instrument, my language is names SAIVA.
6 step - wind by this few kilograms heavy thingy the turn number into ferrite ring according the calculation
7 step - switch the all hell-machine on, and measure the voltage. According results take some minor tuirns off (or by bad fate - add on. Then insulate everything very well and live happy.

 

P.S. NEVER use the previously used wire for rewinding aim. It is useable exclusively for rope the parcels anymore. Just old wire have hardened insulation paint, at bending it give the cracks so the voltage leaks out and amperage goes straight at it`s tail. Just throw out the unwinded wire.

 

Example of acceptable capacitors -- inexpensively purchased new

Do you know about this place? https://www.surplussales.com/homenew.html#Capacitors That is the link to their capacitor page, they also have many other things that may be of interest to the project.

 

The technology is rather prost:
1 step - calculate the Focault effect layer deep. The wire must be less that double of this figure in diameter.
2 step - calculate the approx turn count and find the approx wire length. Allpy roughly 20% more wire as calc shows.

Yeah! Given a chance Mr. "Eddy" Foucault can be a total d..k! So.. I imagine sidestepping some of the complexity was Julie's rationale for leaving the primary alone.. Well.. I'm new to all this so I just go for the lowest magnetizing current vs turn-count tradeoff. In the absence of a viable compromise I try again with a bigger core.. Yeah! I get it! But honestly! People who know me know I'm NO empiricist! But what ya gonna do in the absence of comprehensive published core data?

3.step - find the damn long place where people are absent and beat in the two nails in both wall ends.

Well.. Julie and HP have huge sheds on their properties. I'm stuck with a condo and a HOA holding a no doubt dim view of "parking lot tholianism"

4 step - one end of bucket put into screw-machine and turn it on. Wait while there will happen about 1 loop per centimeter or slightly more. Dont be excessive, better less as too much.
5 step - wind those litcendrath on the textolyte plank, about inch wide and 40 cm long with V-shape gap in both ends. Dont know how to label this instrument, my language is names SAIVA.
6 step - wind by this few kilograms heavy thingy the turn number into ferrite ring according the calculation
7 step - switch the all hell-machine on, and measure the voltage. According results take some minor tuirns off (or by bad fate - add on. Then insulate everything very well and live happy.

Are you talking the sort of winding machine @HP linked in post 8? She calls out three manufactures; "Swant", "Vergara" and "Waterfield". SEs return loads of hits on those terms but nothing related to transformer manufacture Well.. No matter! Youtube knows a lot of toroid winder vids by whatever names.

P.S. NEVER use the previously used wire for rewinding aim. It is useable exclusively for rope the parcels anymore. Just old wire have hardened insulation paint, at bending it give the cracks so the voltage leaks out and amperage goes straight at it`s tail. Just throw out the unwinded wire.

With any other manufacturer I'd agree! But having seen Julie's Powervar secondary winding wire I can say their insulation is basically indestructible! You'll see what I mean when HP uploads the images.

Ok! Let's be clear! We're NOT recommending rewinding! Without special equipment and experience garnered skill, the outcome will be a lot of work for an inferior transformer! Antek (sp?) looks like your best solution and an excellent one at that! So please let's run with it! I'd hate to see the EHTPSU courses quagmired again! So in @HP's words: "not often, but sometimes the easy way is the right way"!

@Janis Blahins In case my attempt at humor (Re. eddy currents) doesn't translate well, I want you to know that I find your infusion of science into this drearily empirical conversation most welcome and refreshing! Thanks!

Do you know about this place? https://www.surplussales.com/homenew.html#Capacitors That is the link to their capacitor page, they also have many other things that may be of interest to the project.

Thanks! I've forwarded the link to @HP and Julie (aka @Aleph(0) )

Saints preserve your sanity whilst beholding my 're-wind' of @Aleph(0) 's transformer! -- Which being almost as 'tidy' as your hair (as worn in your avatar)!

Yeah well.. That's apropos! As you may be aware, in my avatar photo I'm endeavoring to depict a psychotic tweeker with anger issues. IMO, as it turns out, a likely outcome of one attempting to manually wind power transformers

Thanks!

 

I received the AN-10435 transformer from Antek today: http://www.antekinc.com/an-10435-1000va-35v-transformer/ and http://www.antekinc.com/content/AN-10435.pdf

My first measurement is of the turn-on surge. Using a standard shunt to sense primary current and displaying applied voltage and resulting current on an oscilloscope, here's the result. The secondary is open circuited and 120 VAC is repeatedly applied to the primary until it happens that the voltage is applied just as the sine wave is crossing zero in the positive direction.

The yellow trace is applied voltage and the blue trace is resulting current. Notice the scale on the blue trace--50 amps per division! A peak current of 300 amps results. The lights in the room flicker! The applied voltage (yellow) never reaches its normal peak; the high surge current drops a considerable voltage in the resistance of the house wiring! However, subsequent current pulses are much smaller; the second is only about 20 amps peak.



Next, the same test but with a 10 ohm resistor in series with the primary of the transformer. Notice that the current scale has changed to 10 amps per division.



Finally, with a 5 ohm resistor in series with the primary. Here the maximum surge current reaches 30 amps peak. This would be perfectly safe, without risk of damage to the on-off switch.

 

I received the AN-10435 transformer from Antek today: http://www.antekinc.com/an-10435-1000va-35v-transformer/ and http://www.antekinc.com/content/AN-10435.pdf

My first measurement is of the turn-on surge. Using a standard shunt to sense primary current and displaying applied voltage and resulting current on an oscilloscope, here's the result. The secondary is open circuited and 120 VAC is repeatedly applied to the primary until it happens that the voltage is applied just as the sine wave is crossing zero in the positive direction.

The yellow trace is applied voltage and the blue trace is resulting current. Notice the scale on the blue trace--50 amps per division! A peak current of 300 amps results. The lights in the room flicker! The applied voltage (yellow) never reaches its normal peak; the high surge current drops a considerable voltage in the resistance of the house wiring! However, subsequent current pulses are much smaller; the second is only about 20 amps peak.

View attachment 156151

Next, the same test but with a 10 ohm resistor in series with the primary of the transformer. Notice that the current scale has changed to 10 amps per division.

View attachment 156152

Finally, with a 5 ohm resistor in series with the primary. Here the maximum surge current reaches 30 amps peak. This would be perfectly safe, without risk of damage to the on-off switch.

View attachment 156153

Electrician huge tnx! I totally like how your scientific approach to independent research and demonstration shows how theory and empiricism can totally augment each other! Which I say is vry important point for ppl (like me) who come from basically abstract background (and so perspective)

So since major component lineup of low voltage PSU is basically: Line→Variac→Power xfmr→Rectifiers→Filter Caps→Load. I say worst case scenario for inrush is power-up during crossing with variac set for max output voltage! So o/c variac inrush is same no matter wiper position (so practically determined just by input phase angle) but higher power-up voltage to power transformer means more pwr transformer inrush! Also higher output voltage at PU means greater inrush to filter caps and load capacitance!

Now since it's sometimes totally important to be able to safely power-up under load at voltage, I say best compromise for inrush mgmt is basically what HP's saying here:

Hence my proposal of a basic timing circuit consisting of 'bone-generic' components (e.g. a 'tweakable' monostable lm555 circuit, a 2n222 relay driver, and a real [i.e. electro-mechanical] relay as the bypass switch) -- Note that the requirement of an 'active timer circuit' owes to the desirability of 'zero' off-time latency...

So I think @Hypatia's Protege prolly meant 2N2222 BJT but I wouldn't DARE alter her text w/o permission!

Also @Hypatia's Protege just so u know I was 100% sincere saying I totally get how your phase-synchronized switch (like you designed for the Alexitron) is totally more _elegant_ solution but since it's more complex and doesn't help filter cap and load inrush I say for this application it's basically like gilding inside of aqua regia vessel!

Also LM555 and TO-220 BJTs are easy to find for foreseeable future so when $#!t happens it's an easy repair w/o redesigning circuit! Cuz let's face it! Those LVPSUs are going to take major abuse in hands of (now I'll be nice) _novices_ powering 100kV+ generators!

 

Also LM555 and TO-3 BJTs are easy to find for foreseeable future

So o/c I meant TO-220 or sometimes just TO-92 packaged BJTs Cuz relay coils aren't that _hungry_ (except maybe old FMC starter relays)! So thanks to lurking HP for heads up!

 

Continuing the saga. Getting involved in all this has now given me an excuse to buy a bigger variac which I've been wanting for a while.

I got one of these: http://www.mpja.com/2KVA-0-130VAC-Variable-Power-Transformer/productinfo/15163+TR/

First thing I did was to measure the maximum turn-on surge as I did with the transformer; here's the result:



The peak current is about 275 amps; this is only a little less than the 300 amps peak for the Antek transformer. I don't show a scope capture, but adding a 5 to 10 ohm ballast resistor decreases the surge to a safe value just as with the Antek transformer.

For the benefit of non-EE readers: regarding the value of these peak currents I should mention that the DC resistance of the primary of the Antek AN-10435 transformer is .218 ohms. Since the peak voltage of the 120 VAC line is 170 volts, one might expect that the peak current ought to be 170/.218 = 780 amps. So why is it only 300 amps? Because the resistance of the house wiring is added to the resistance of the primary in determining the value of the peak current. When the core of the transformer (or variac) saturates, the inductance of the primary drops to a negligible value, and it's only the total resistance of the circuit that determines the peak currrent.

The surge that occurs when the AN-10435 is connected to the output of the variac, and the variac wiper is set to the position on the winding just where the line input is connect is essentially not much different than the transformer alone.

 

The next thing I wanted to measure is the surge due to the charging of the big electrolytic capacitors following the bridge rectifier. I obtained a couple of 34,000 uF 50 volt capacitors and a heavy duty bridge rectifier. I connected everything in the manner of this post: https://forum.allaboutcircuits.com/...-and-construction.113504/page-69#post-1248540 except for some of the details such as the line filter and the over voltage protection devices.

There is no ballast resistor in series with the on-off switch for the following captures.

The line is connected to the variac at a point most of the way up the winding. When the wiper of the variac is set to that same place, the insertion impedance of the variac is at a minimum, and the electrolytic charging surge would be expected to be a maximum there.

When the turn-on surge of the toroidal transformer (AN-10435) by itself is measured, what determines the peak current is the resistance of the house wiring and the resistance of the primary of the transformer; note that the resistance of the secondary is not involved. But for the case of the charging surge of the filter capacitors, the resistance of the variac and the secondary of the transformer come into play, and we shouldn't expect such a large surge.

Setting the variac wiper to the same point where the line is connected and applying line voltage the the input of the variac repeatedly, I captured the maximum peak current into the bridge rectifier. This is the same as the current out of the secondary of the AN-10435 transformer. The first image here is with a single 34,000 uF capacitor on the output of the bridge. The green trace is the current into the bridge rectifier, and the purple trace if the voltage across the filter capacitor:



The peak surge current is nearly 200 amps and you can see the voltage across the filter cap increase rapidly during that first current pulse. Subsequent current pulses gradually decrease and the voltage also gradually approaches its final value. Since this current is the current out of the secondary of the transformer, we could expect that the primary current surge would be less by a factor equal to the turns ratio of the transformer, which is 120/35 = 3.43. This would give a primary peak surge current of 58 amps. Of course, this is separate from the transformer turn-on surge due to saturation of the core.

Next I connected another 34,000 uF capacitor in parallel with the existing one. Again line voltage was applied repeatedly to the input of the variac until I captured the maximum charging surge. Here's the scope capture:



Notice that the peak current of the first pulse is the same as with only one capacitor. This shows that it's the resistance in the variac and the primary and secondary of the AN-10435 that limits the current, not the capacitance (total microfarads) of the filter. However, one difference is that the capacitor voltage (purple trace) increases more slowly.

 

Now for the dirty little secret. The "heavy duty" bridge rectifier I used for these measurements was this: https://www.digikey.com/product-detail/en/on-semiconductor/GBPC3502/GBPC3502-ND/1057390

Having a look at the spec sheet: http://www.onsemi.com/pub/Collateral/GBPC3510-D.pdf we see in figure 2 that the non-repetitive surge rating for the 35 amp bridge is 400 amps. Since HP wants the power supply to supply 30 amps, for some extra margin of safety I used two such bridges in parallel. Since the peak surge current I measured is 200 amps, and since I have two bridges in parallel (I'll discuss sharing in a moment), the actual peak surge seen by each bridge is only 100 amps, well below the allowable 400 amps.

Adequate sharing of current in the two bridges is accomplished by using longer than needed (12 inches) hookup wire to each bridge from the secondary of the AN-10435 transformer, and making the lengths of the "longer than needed (12 inches)" identical. Measuring the actual currents into each bridge showed that sharing is so good one can hardly tell any difference.

And finally, all this so far has been without any ballast resistor in series with the line at turn-on. With a 10 ohm ballast resistor the initial charging current surge from the secondary of the transformer into the bridges is greatly reduced. Here's a capture of the turn-on current onto the filter caps with a 10 ohm ballast resistor. Notice that the scale for the purple current trace is 10 amps/division rather than the earlier 100 amps/division:



Using a ballast resistor and a switch with make-before-break contacts tames the surge due to saturation of the variac and transformer, and also the filter cap charging surge.

 

The setup (kluge?) so far:



By the way, the little clamp-on meter seen in the photo would be a good thing for a builder to have. It's a UT210E and can be had on eBay for less than $50. It's a voltmeter and a clamp-on ammeter. The clamp-on ammeter feature not only measures AC current as clamp-ons have done for years, but it can also measure DC current. This is a feature that is fairly new in a low-cost clamp-on.

 

In an earlier post: https://forum.allaboutcircuits.com/threads/rewinding-2dary-of-toridal-pwr-xfmr.149589/#post-1283322

I discussed the effect in capacitor input power supplies like this where the secondary RMS current is larger than the DC output current due to the very peaky current pulses drawn from the transformer.

It's difficult to model and calculate the ratio of secondary RMS current to DC output current without knowing the many parameters involved, such as DC resistance of windings, ESR of the filter caps, apparent impedance of the line at the service entrance, resistance of the house wiring, etc. So we resort to measurement.

For the following captures, the variac was set to output a voltage equal to the grid voltage of 120 VAC which is applied to the transformer primary. The filter capacitance was 68,000 uF (two 34,000 uF in parallel) and for a load I used a 4 ohm, 120 watt rated power resistor (the green thing in the picture above) and severely overloaded it to 400 watts for 5 seconds at a time. The variac was turned up until the DC output voltage was 40 volts and the current in the resistor was 10 amps.

The measured ratio of secondary RMS current to DC output current is 18.6/10 = 1.86. This is not as high as 2 or more as I speculated it might be for a large toroidal transformer, but a value of 1.86 means we have to significantly derate the transformer for continuous duty.

A linear extrapolation tells us that if we want 30 amps DC out of this supply, the secondary RMS current would be 55.8 amp, almost double the rated 30.8 amp output for the AN-10435 transformer. Antek says that their transformers can be overloaded 20% without any problems which helps a little.

Another matter of concern is the ripple current in the filter caps. With 30 amps DC out, the total filter cap ripple current would be 48.6 amps, a value not to be ignored.

This capture shows the current in the secondary of the transformer (green trace) and the ripple voltage riding on the 40 volt DC output. The ripple voltage was about .85 volts peak-to-peak. The secondary RMS current was 18.6 amps:



This next image shows the ripple current in the parallel combination of the two 34,000 uF filter capacitors:



And, finally, the current in the primary of the transformer:

 

Because the resistance of the house wiring is added to the resistance of the primary in determining the value of the peak current. When the core of the transformer (or variac) saturates, the inductance of the primary drops to a negligible value, and it's only the total resistance of the circuit that determines the peak currrent.

Electrician that's totally correct and vry observable looking on difference between nasty results of zero-crossing connection to 15A circuit (which is served by 14 AWG) vs 20A branch (10 AWG)!

The surge that occurs when the AN-10435 is connected to the output of the variac, and the variac wiper is set to the position on the winding just where the line input is connect is essentially not much different than the transformer alone.

Electrician tnx! Cuz that's vry good point I hadn't even thought of! Paralleling vry, vry small variac and xfmr primary wdg resistances won't make much difference to inrush cuz current limiting is mostly by greater resistance of wiring from load center! Fwiw I say house wiring inductance is too small for XL to come into it vry much (but I admit not calculating it)?

I used two such bridges in parallel. Since the peak surge current I measured is 200 amps, and since I have two bridges in parallel (I'll discuss sharing in a moment), the actual peak surge seen by each bridge is only 100 amps, well below the allowable 400 amps.

Adequate sharing of current in the two bridges is accomplished by using longer than needed (12 inches) hookup wire to each bridge from the secondary of the AN-10435 transformer, and making the lengths of the "longer than needed (12 inches)" identical. Measuring the actual currents into each bridge showed that sharing is so good one can hardly tell any difference.

Electrician it's vry good to have practical confirmation of that! Cuz we just use huge industrial rectifier diodes (with surge current spec of like 400kA) connected as bridge. But since a lot of ppl dont want to spend extra for over-design it's vry nice 2b able to confidently tell them paralleling bridge modules is doable! Tnx to your research!

In an earlier post: https://forum.allaboutcircuits.com/threads/rewinding-2dary-of-toridal-pwr-xfmr.149589/#post-1283322

I discussed the effect in capacitor input power supplies like this where the secondary RMS current is larger than the DC output current due to the very peaky current pulses drawn from the transformer.

It's difficult to model and calculate the ratio of secondary RMS current to DC output current without knowing the many parameters involved, such as DC resistance of windings, ESR of the filter caps, apparent impedance of the line at the service entrance, resistance of the house wiring, etc. So we resort to measurement.

For the following captures, the variac was set to output a voltage equal to the grid voltage of 120 VAC which is applied to the transformer primary. The filter capacitance was 68,000 uF (two 34,000 uF in parallel) and for a load I used a 4 ohm, 120 watt rated power resistor (the green thing in the picture above) and severely overloaded it to 400 watts for 5 seconds at a time. The variac was turned up until the DC output voltage was 40 volts and the current in the resistor was 10 amps.

The measured ratio of secondary RMS current to DC output current is 18.6/10 = 1.86. This is not as high as 2 or more as I speculated it might be for a large toroidal transformer, but a value of 1.86 means we have to significantly derate the transformer for continuous duty.

A linear extrapolation tells us that if we want 30 amps DC out of this supply, the secondary RMS current would be 55.8 amp, almost double the rated 30.8 amp output for the AN-10435 transformer. Antek says that their transformers can be overloaded 20% without any problems which helps a little.

Another matter of concern is the ripple current in the filter caps. With 30 amps DC out, the total filter cap ripple current would be 48.6 amps, a value not to be ignored.

This capture shows the current in the secondary of the transformer (green trace) and the ripple voltage riding on the 40 volt DC output. The ripple voltage was about .85 volts peak-to-peak. The secondary RMS current was 18.6 amps:

View attachment 157092

This next image shows the ripple current in the parallel combination of the two 34,000 uF filter capacitors:

View attachment 157093

And, finally, the current in the primary of the transformer:

View attachment 157094

Electrician I totally agree cuz my IEC fusor chain's low-voltage/high current section (which is now just modified Lambda Alexitron which is basically a continuously variable 1MVAR 0V-400V, 2500A phase-control regulated {by which I mean thyristor _modulated_ } line xfmr followed by rectifiers and filters) was blowing up filter caps like coconuts in a microwave! So when I studied problem it was all cuz of cap temp rise from high ripple currents. So I solved problem with combination of pulse rated caps and temp sensors for good measure!

But I say our LVPSU project should be ok with just standard industrial electrolytic Caps cuz of vry low high current demand d/c But we're definitely planing to warn builders that electrolytic caps need 2b in safety shield in case of bursting! So that's always good idea anyhow but even moreso with paralleled caps! So IMO _wet_ electrolytic caps are just inherently unreliable and delicate with disappointing ESR and related electrical characteristics. But sad fact is ppl can be vry stingy with hobbies! So keeping our pledge of meeting them half way sometimes means sacrifice of best practice except when it's safety issue! Cuz safety is totally non-negotiable!!!

So electrician I say HUGE tnx cuz there's no substitute for _field testing_! And we totally appreciate your systematized, intelligent (IOW scientific) approach to it!

 

By the way, the little clamp-on meter seen in the photo would be a good thing for a builder to have. It's a UT210E

Electrician how do you feel abt Fluke 337? That's what we've been recommending cuz for inexpensive AC/DC clamp ammeter it gives nice stable reading and holds calibration vry well!

Just asking your opinion!

 

Electrician how do you feel abt Fluke 337? That's what we've been recommending cuz for inexpensive AC/DC clamp ammeter it gives nice stable reading and holds calibration vry well!

Just asking your opinion!

I'm biased; Fluke Corporation is about 25 miles up the road from where I am, and several EE friends work there. I've used Fluke meters during most of my career and they are top notch. The UT210E was recommended on another forum and I got it to check it out. It would not be my everyday go to meter, but it's perfect for a hobbyist with limited budget.

 

@The Electrician

So electrician I say HUGE tnx cuz there's no substitute for _field testing_! And we totally appreciate your systematized, intelligent (IOW scientific) approach to it!

I wish to add my thanks to @Aleph(0)'s!

we totally appreciate your systematized, intelligent (IOW scientific) approach to it!

Indeed! -- Such is all of informative, interesting and most refreshing! -- Many sincere thanks!

Very best regards
HP