7
\$\begingroup\$

So I'm getting ready to build a power supply, and it calls for a bridge rectifier. However, I can't find any information about HOW rectifiers are rated, except for a note that (and I don't have it in front of me) reverse voltage should be three times normal to protect against spikes...or some such.

My input voltage is 110VAC, with a transformer converting it to 18VDC/1A.

Is the rectifier voltage the MAXIMUM it can take? or something else? What about amperage? Same deal? or, in my instance, do I need to have a 1A rectifier?

I have a choice between a 35V/10A, 50V/2A and 60V/4A rectifier, and I'm leaning toward the 60V/4A, guided ONLY by the fact that it's the only one that meets the 18V*3 note that I read.

Will I encounter a blue smoke demon? or is this a good way of thinking about it? How do I interpret rectifier ratings against a schematic when there's no specification beyond "bridge rectifier" in the parts list?

Crappy camera phone image. If it's illegible let me know, I can upload a scan in a couple hours.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ This PS will provide variable power up to 24V for breadboard used to develop and test synthesizer components, most ov which require between 15V-24V. Most of the power conditioning happens in the module itself, so I'm not terribly concerned about ultlra-ultra stable power coming out of the PS noted here. If that helps guide some clarifications on the answers.... \$\endgroup\$ – dwwilson66 Aug 8 '12 at 19:17
  • \$\begingroup\$ I've found good article about rectifier diode working ranges. I wasn't sure how it behaves but after reading it I've "seen" it better :) I hope it helped you too. \$\endgroup\$ – user83827 Aug 22 '15 at 16:34
6
\$\begingroup\$

I guess the \$\times\$3 rule of thumb refers to the transformer's secondary voltage, the AC voltage before rectifying. If your output is 18 V DC then the transformer will be a 14 V type. That 14 V is the RMS value, peak value is \$\sqrt{2}\$ times higher, or 20 V. Subtract 2 diode drops and you arrive at 18 V.

So applying the \$\times\$3 rule to the 14 V already gives you 100 % headroom, applying it to the 18 V gives you almost 200 % headroom, which is overly cautious IMO. Since 20 V is the highest voltage to expect all three will do fine.

The current is a different story.

enter image description here

If the smoothing capacitor is charged the input voltage will be too low for the greatest part of the cycle to make the diode conduct. Only for a short period you'll have a current that recharges the capacitor. So you can expect a large current peak there, several times the average of 1 A. A 1A diode like the 1N4001 can handle this current, though it's a good idea to have some headroom here as well. For a 1 A output a 2 A diode is a good idea, so also for current all three rectifiers will do.

\$\endgroup\$
  • \$\begingroup\$ Keep in mind that the 18V might actually be RMS, and so the peak would be about 25V. Given that there's a transformer in the way, I believe that the short margin to 36V is acceptable, since the transformer will effectively filter significant spikes. If the transformer output is used for any other purpose (such as driving a motor or relay) then you'd want a bigger margin to handle spikes, and the 3:1 rule the OP suggested is sufficient. \$\endgroup\$ – Adam Davis Aug 8 '12 at 16:33
  • \$\begingroup\$ Also, a better design would make sure that each diode could handle the peak to peak, so if one of them fails as short, the others don't cascade fail. So it wouldn't be a bad thing to spec something that can handle more than 50V reverse. But that's overkill for hobby applications. \$\endgroup\$ – Adam Davis Aug 8 '12 at 16:35
  • \$\begingroup\$ @Adam - OP states 18 V DC in his question. \$\endgroup\$ – stevenvh Aug 8 '12 at 16:36
  • \$\begingroup\$ Indeed he does. Given that he's asking this question, however, it's worth considering that the transformer specs provided by the OP are suspect, since a transformer doesn't have a DC output without additional circuitry. Perhaps it is a transformer designed for an 18V 1A DC supply, after rectification. Perhaps it's an 110V to 18V 1A transformer. Either way, for hobby application, any of the chosen bridge rectifiers would be sufficient. While I may seem to be splitting hairs, it's obvious this is new to him, so covering additional possibilities isn't a bad thing. \$\endgroup\$ – Adam Davis Aug 8 '12 at 16:52
  • 3
    \$\begingroup\$ @Adam - I know I'm going to let down many of my fans here :-), but I'm not clairvoyant. I have to work with what OP writes in his question. If he writes DC I assume he knows what that means. If he means AC he has to correct it. \$\endgroup\$ – stevenvh Aug 8 '12 at 17:14
5
\$\begingroup\$

The voltage rating for a typical bridge rectifier is the maximum recurrent peak reverse voltage. A 100V bridge rectifier contains diodes that will withstand peaks of 100V indefinitely.

Voltages that exceed this can lead to breakdown and failure, which is why it's common for 85-264VAC applications to use a 600V or 650V rated bridge, along with some surge-limiting circuitry. Your 3:1 derating guideline for voltage is pretty good, actually.

The current rating for a typical bridge rectifier is the maximum average forward output rectified current, typically measured at mains frequency. The limiting factor for the current is thermal - many bridges spec a certain current but say in the small print "with a particular heatsink, at 25C". There generally is also a surge current rating for things like initial inrush (when bulk caps are fully discharged) which is much higher than the average rating. It's good to have current margin to ensure thermal margin.

Exceeding the average current means that the device could get really hot and fail after a period of time, or it will work just fine if there's sufficient cooling. Your mileage may vary. Exceeding the surge current rating will most likely release the blue smoke of shame and possibly some debris :)

\$\endgroup\$
2
\$\begingroup\$

I think all three will do just fine and because of that I would go for the cheapest one fo the three. The ×3 is just a rule of thumb and with 18V you'll be fine with all rectifiers you list. The current they can handle is much better than what you need for your application.

\$\endgroup\$
  • \$\begingroup\$ Since I've got all three in a box of electronic goodies I picked up they're equally economical. I'll just pick the highest rated to give me the headroom. \$\endgroup\$ – dwwilson66 Aug 8 '12 at 18:53
2
\$\begingroup\$

I would roll my own in your position. The bridge rectifier is a simple circuit and the parts are cheap. The important thing is having diodes that will handle the current you're specing. Is the actual DC current draw going to be 1A? If so, I would go with 1N5400. If it's actually going to be significantly less than 1A most of the time, 1N4001s would probably work just fine. The 4001 is rated for 1A. I would not want to push one that hard. It's better to go with the higher current part if in doubt.

\$\endgroup\$
  • \$\begingroup\$ I've already got the rectifiers lying about, otherwise I woulda been soldering some leads together last night. :) \$\endgroup\$ – dwwilson66 Aug 8 '12 at 18:52
2
\$\begingroup\$

Update 1 THanks for photo of schematic. If you don't mind <20% efficiency, it might be ok. Transformer is 18Vac 4A = 72Watt max.

This full wave bridge in theory (with the right parts) can deliver 25.2(41% over AC) with no load but the 100Ω series gives low ripple but also huge loss.. eg 10V drop at 0.1A, so if you want 1V drop, consider what load you need.

for 72W @ 24V @3Adc 24dc/18Vac=1.33 using my chart 1.33 @ 72W >> equals 6Ω (not 100) Using Steven's photo you want the capacitor to hold up the voltage between pulses say 5~10% so if using 50Hz the pulses are 100Hz and instead of a time constant of 100mS, which would give you ~60% ripple you choose say 600ms so the load cap from T=RC, C=0.6sec/6Ω = 0.01Farad or 10,000 uF @30V or higher.

Now there happens to be 10 thousand capacitors that fit that description from $3 to $50 but many can't handle the ripple current. So using www.Digikey.com 's auto-filter we enter 10000 uF then select 30 or 35V then sort by Amp rating and for the cheapest that can handle 4A easily with margin. Great! that narrows it down to 11 parts (in stock) ranging from $3~$11. So I pick the most popular brand in reliable caps, Panasonic. Ripple Current 4.42A ESR (Equivalent Series Resistance) 50 mOhm, 10,000uF @ 25V 22mm x 45mm high. Now there is a huge inrush current, so we need to protect the diodes and the cap.

So we look for Inrush Current Limiters (ICL) from DK site again. and look for 3A continuous rating. Now I see only stock one of these and its only rated for 2A , so you need two. so buy 5.@$2.02 each.for spares.

The specs for the surge resistor are R @ 25°C =120 Ohm, R @ Current= 1.18 Ohm. Great! So it will have the soft filtering at low loads and thus low surge and very low ripple and drop to 1.18Ω at 2A or around 2Ω @ 1.5A with two in parallel thats 1Ω so the the 6Ω series R , that we calculated before, can drop to 5Ω. or so. Not too bad for a 78W 24V supply.. Now the 12V supply is going to get hot with 1A so then we consider.. maybe we should have split the supply to +/-12V so we dont lose so much in the 3-terminal regulator. But we need a current spec from your 1st. So for $6 to $15 incl spares plus shipping. the can ship same day on VISA.

That's how I would design around your transformer. But without knowing your application. especially the 100Ω pot ( poor regulation ) that's all I am going to do.

Oh yah the bridge, with the ICL resistor, you only need a 4A 50V bridge. with these specs $0.91 instock. Without, you need a bigger one.


Generally there are many Design Rules for safety of components and safety of public for consumer quality power supplies. Then for commercial and industrial a few more specifications. So for simplicity, when a consumer needs an unregulated DC bridged power supplied , they simply buy a wall adapter with the appropriate rating. These DC voltage ratings will be accurate when loaded with the rated current and then rise as the load is reduced to the peak voltage levels.

Since you wish to learn about how to specify components, there are only a few simple tradeoffs with choice you need to specify.

1) Budget cost of parts,

2) tolerance for ripple voltage,

3) expected load current and voltage

4) temperature rise of bridge.

5) The voltage drop on the bridge at expected load current - bridge power loss (=load current* voltage drop) - bridge temperature rise = Pwr * deg 'C/watt Junction to Ambient rise - ambient rise. If you enclose the project you need to consider having ventilation or consider the temperature rise inside the enclosure and the max temp. rating of the capacitors.

6) If you choose large uF with low ESR capacitors, you can reduce ripple significantly

7) If you add a series R after the bridge you can reduce the peak current and ripple voltage

8) If you add a series inductor after the bridge you can reduce ripple even more

9) If you need a precision regulator an adjustable LM317 or variation will work

10) If you want short circuit output protection (SCP) , a PTC can be used as a series resistor

11) if you need over-voltage protection (OVP) from flashover on transformer unless rated for 6kV (max passthru on power meter)

The bigger the bridge the lower the voltage drop and temperature rise. Typically 1V drop @1A = 1 Watt but if not heat sunk can get very hot, so choose lower drop Schottky bridge. But check the voltage Drop at your load current and temp rise coefficient.

Don't worry too much about voltage rating unless you live in Africa, India or some part of the world with wild power fluctuations. 50% margin should be adequate.

Here are some additional detail on items 7 not given in the other answers.

Courtesy of Dr Keith Billings "Switched Mode Power Supply Handbook" excellent book. enter image description here There are more curves for Ripple vs Cap value and R value.

There is a different curve for a half bridge if you choose that. Can you confirm output rating in Vac @Iac of transformer? It is not the same as DC. What accuracy do you need for Vdc output Ripple Vp-p? i.e. regulated or not?

"My input voltage is 110VAC, with a transformer converting it to 18VDC/1A"

\$\endgroup\$
  • \$\begingroup\$ Wow. A ton of information. That'll take a little while to digest. :) I've got no info about how accurate the 18v needs to be. The schematic in question is from a book, and text seems to indicate that the resistor and Caps AFTER the bridge will even everything out...There's a 100ohm 5W resistor, 2 - 5000uF and 1 - 1000uF caps before I get to a 100ohm linear taper pot which control the output from 0V to 24V. If it helps, I could post the schematic for a reasonability check? \$\endgroup\$ – dwwilson66 Aug 8 '12 at 18:47
  • \$\begingroup\$ Yes post it and also intended load \$\endgroup\$ – user11355 Aug 8 '12 at 19:06
  • \$\begingroup\$ You might be thinking about buying a used linear lab supply if that's what you need. But if you are planning PSU for powering LED's for photography,, just say so... or whatever you are planning to use this for. \$\endgroup\$ – user11355 Aug 8 '12 at 19:09
  • \$\begingroup\$ I'll post the schematic tonight. I've updated my OP with a comment on usage. \$\endgroup\$ – dwwilson66 Aug 8 '12 at 19:18
  • \$\begingroup\$ Oh, and I already have a lab supply; 0-30V, but I've never built a power supply before so I'm messing with that this month. :) \$\endgroup\$ – dwwilson66 Aug 8 '12 at 19:19
1
\$\begingroup\$

A picosecond of overvoltage is enough to fry a part. I've lost Mosfets to voltage peaks of that time scale. So yes you want that level of safety for grid-connected devices.

However your transformer being an inductor will block out any such peaks and you need not worry on the low-voltage side. It will put out a 30-40% larger voltage when unloaded though so you have to plan for at least twice the rating.

\$\endgroup\$

protected by Dave Tweed Aug 22 '15 at 17:59

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.