Paralleling Three Full Wave Rectifier Bridges

Maybe one of your rectifiers is bad. Check them with a meter on the diode scale and mayb try them one at a time.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward" 
speff@interlog.com             Info for manufacturers: http://www.trexon.com 
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

I have no idea on the variac fuse issue, but you really can't parallel diodes due to current hogging.

Maybe there is a scheme that isn't too complicated to use power mos devices instead of diodes, i.e. active rectification.

If your mains supply is US style 120 V, it's not common to have more than about 20 A available. You *can* get up to 50 A on the right circuit, but not from a "normal" outlet. For Europe-style 230 V or

240 V, you probably have 16 A or less available.

That size Variac is only going to be good for something like 10 A. Bigger Variacs do exist.

Both the mains supply and the Variac will limit how much current you can get. Neither one has much to do with your fuse-blowing problem, though.

35 A, 700 V bridge rectifier. OK.

Probably OK. If this is made out of multiple capacitors, you might try using just one at a time - maybe you have a bad one.

This may or may not work. The diodes in each bridge will turn on at very slightly different points in the AC cycle. When an AC cycle begins, one bridge rectifier may temporarily be carrying all the current, until the AC voltage rises a little more.

Do you have the bridge rectifier(s) on a heat sink?

Like Sphero said, test each bridge individually. Maybe you have a bad one, or maybe the terminals are incorrectly labeled.

You also have a lot of capacitors to charge up; they will appear as a dead short at the beginning of the first AC cycle and draw a large current for a brief time. (On the other hand, it does work if you use just one bridge - but does it work with each of the three bridges used individually?)

If the Variac has a fast-acting fuse in it, this may be part of the problem; try a "slow blow" fuse of the same current rating and physical size. Be aware that you might blow the next fuse in line (like the fuse or circuit breaker for the building mains) instead when you do this.

I know this *can* work, as I've helped do something similar to charge a 288 V battery pack: 240 V mains, isolation transformer (1:1), Variac, bridge rectifier, small capacitor (around 1000 uF), battery pack. I only used one bridge, though. The control system was a student that watched a voltmeter and an ammeter and turned the Variac knob. :)

Matt Roberds

Are they identical? parallelling disparate rectifiers is likely to fail due to current hogging,

have you got a clamp meter to probe for fault currents with?

only one thing: too much current.

--
For a good time: install ntp 

--- news://freenews.netfront.net/ - complaints: news@netfront.net ---

1. Check Capacitor polarity.
2. Connect any tungsten lamp between variac and rectifier. It will prevent future explosions.
3. Use only one rectifier. Google "diode parallel".
4. Do not forget discharge you capacitors after.

On Friday, December 27, 2013 8:49:47 AM UTC+2, Artem wrote: tungsten lamp

220 Volt!!!

I agree with Spehro. Perhaps one rectifier is bad or wrongly connected.

Also, regarding putting rectifiers in parallel:

Whilst it is not the cause of the present problem, it may be worth keeping in mind that the forward voltages of diodes tends to decrease when they get hotter. This means that whichever of the three rectifiers is hottest will tend to hog more than its fair share of the current, and may get hotter still, possibly leading to damage. The usual way to fix this problem is to mount the semiconductor devices in good thermal contact with each other, and then add individual resistors in series with each device so that if there were any current imbalance, it would tend to self-correct due to the voltage dropped in the resistance being large enough to swamp any thermal mismatch. These balancing resistors are most often seen when multiple bipolar transistors are used in parallel, in which case each transistor has the resistor in its emitter. In the case of the rectifiers, it might turn out to be the case that just using individual long wires for each rectifier, rather than joining them together with short thick wires, would be enough balancing resistance. It might even be the case that the rectifiers themselves have enough parasitic internal ohmic resistance to prevent thermal runaway. Anyway, checking the current sharing (e.g. with a hall effect clamp meter) under full load, especially after you artificially heat or cool one device, might be a test worth doing once you get the circuit working.

Chris

I'm not sure that capacitor is necessary. DC Choke will be more better solution.

I prefer use capacitor in series of primary coil for current stabilisation.

Why do you use multiple bridge rectifiers instead of four discrete sufficiently big rectifiers ?

Anyway, the specifications for most rectifier bridges seem to be quite "optimistic". In a single phase bridge rectifier two diodes connect all the time, causing two Vf voltage drops and the power dissipated P = 2 x Vf x I, thus at 35 A, one could expect about 70 W heat generation.

One data sheet for the KBPC3510 claimed Iav at Ta=55 C with 3/8" leads, which would suggest that the bridge was suspended in air without heatsink. Of course, this is ridiculous.

The derate curve makes more sense, in which the derate should start at

55 C _case_temperature, which is believable. With Tj(max)=150 C and Ta=20 C, the thermal resistance from junction to case is 1.4 C/W and from case to ambient air 0.5 C/W. To achieve such low heat sink thermal resistance, you would need huge fans to circulate cold air or have a constant supply of (ice)water :-). And as a reality check, who would design a product to operate at the maximum junction temperature anyway ?

Assuming some more realistic design parameters, say Tj=120C, Ta=40 C (inside an equipment) and Rth=3 C/W (from junction to air) a 27 W dissipation would be tolerated thus the load current should be kept below 15-20 A.

Perhaps a diode has been failed short in your previous experiment.

If you do not have sufficient balancing resistance (or inductance) in _each_ feed to the rectifiers, if one rectifier gets hotter (insufficient cooling in the middle unit etc.), the Vf drops, more current flows through this bridge, while the current in the other drops. Finally the hot diode overheats and breaks.

Use small separate resistors from the AC input point to each rectifier AC input to balance the current even if there is a temperature imbalance. Of course you cold use such resistors also between each bridge (+)-side and storage capacitors or if you have three storage capacitors, feed each with a dedicated bridge and combine the output from each capacitor to the final load.

Of course, balancing series resistor dissipate some power and cause voltage sag. One way to reduce the dissipation is to use series inductors from the main AC input to the individual AC inputs of the bridges. This is also gentler to the rectifiers and electric network, since it reduces the peak current during startup and during each mains cycle.

variac

the lamp only sees the difference.

a 12V 100W lamp might be a good start.

--
For a good time: install ntp 

--- news://freenews.netfront.net/ - complaints: news@netfront.net ---

I have not seen any specification for the actual thermal resistance from juction to ambient air without heat sink, but for packages of similar sizes, this is somewhere between 10-30 C/W, take 10 C/W, Tj=150, Ta=25 C, the maximum power dissipation would be 12.5 W. assuming Vf=0.9 V/diode, which would allow 7 A continuous current, For Rth(ja)= 30 C/W, we are down to 2.5-3.0 A.

Looking at data sheets from various manufacturers, the Vf=1.2 A at 35 A rather than 1.0 V that I had assumed in a previous post. The Rth(jc) is 2 C/W, while I had assumed 1.4 C/W based on the derate curve. Thus, even my calculations gave a too optimistic view of the situation.

What current? The pictured variac you referred to looks like less than 5 amps rating. What fuse does it have? It certainly doesn't look capable of the 35 amps that one KBPC3510 is rated at.

Did you test it under load, or unloaded?

Bear in mind that, with the rectifier-capacitor configuration you describe, the peak current drawn can be much higher than the average current, dependent on the impedance of the supply. Variacs have low resistance windings, and also, being autotransformers, the currents in each half, either side of the wiper are very unequal. Variacs are not really suitable for reservoir capacitor rectifier operation.

A 5 amp variac will only *deliver* 5 amps at whatever voltage. You can't use them for current step-up, I've seen melted windings where people have tried.

3*35 amps = 105 amps, which is a BIG variac, which would weigh hundreds of pounds, and cost thousands of bux. The only ones that big that I've seen were motor driven, three phase, and needed a crane.

Paralleling block bridge rectifiers is never a good idea, unless you can guarantee a close match on forward voltage and temperature coefficient thereof, and ensure they're all at the same temperature, under all conditions. Individual diodes can be paralleled with a suitable resistor in series with each.

If you can't find a suitable bridge, you'll just have to make one out of individual diodes. They're available up to hundreds of amps.

Check your rectifier connection (easy to get wrong), and your capacitor polarity.

--
"Design is the reverse of analysis" 
                   (R.D. Middlebrook)

Variac.html

Check that you didn't blow one of your bridges. If you shorted a rectifier in a bridge due to overload, then on the next half-cycle that bridge would become a short and something would give.

Diodes -- and by extension diode bridges -- do not share current well. That's because when a diode gets hot its forward voltage drop falls. With diodes in parallel, the diode with the lowest forward voltage drop steals all the current. Then the diode that's stealing all the current gets hotter, and it steals it even more. Then it fails -- and usually it fails short, not open.

When you parallel diodes (or bridges) you need to use ballast resistors in series with the diodes. You need to size the ballast resistors so that they have 100mV or so of drop at maximum current (and -- I'm not sure if the 100mV is enough).

So to use your parallel bridge scheme, you need those ballast resistors.

Or, as suggested, just find four individual diodes with sufficient current rating, make sure you're heat sinking them adequately, and make up your own bridge.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com

Oh -- and if you _do_ pursue the parallel bridges idea, then not only do you want them ballasted, you want them thermally coupled, as tightly as possible. That's so when one starts getting hot it warms the others, and so that one doesn't start out significantly warmer. Achieving that tight thermal coupling from the individual diodes one bridge to the individual diodes in another isn't trivial, which is another reason for just finding four herky diodes and building your own bridge.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com

Connecting power diodes in parallel or series is standard practice. It may or may not require resistors. Check with datasheet for particular diodes.

BTW, when paralleling FWR bridges, this is just one resistor per bridge.

Vladimir Vassilevsky DSP and Mixed Signal Designs

The variac in your link does not supply enough current to stress that KBPC3510 bridge you're using, not even close. Get real!

I know that RF power transistors are often built so that they're a bunch of small transistors with emitter loading resistors, all put in parallel on one chip.

If the diode manufacturer did that, or if they arranged their dopant levels to achieve that effect so that hot spots on the diode did not do the same current-stealing thing, then the overall effect may allow for paralleling.

I don't do much of that heavy-metal stuff, but the idea of diodes that can, by design, be paralleled makes sense.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com

Most RF power transistors I know of are a single transistor, with multiple emitters.

From memory, the humble 2N3866 has 9 emitters.

--
"Design is the reverse of analysis" 
                   (R.D. Middlebrook)

Most (RF) power transistors have multiple emitters with ballast resistors, and often multiple wirebonds (or welded strap clips, my favorite) to enhance balance and reduce inductance effects. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142   Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.