Author Topic: Diode size  (Read 993 times)

0 Members and 1 Guest are viewing this topic.

edy252

  • Jr. Member
  • **
  • Posts: 54
Diode size
« on: July 08, 2004, 11:37:37 AM »
hi all.....


i'm intending to use individual diodes (6 diodes) to rectify the 3-phase output of my genny.....i expect a maximum of 55 amps (@ 12 volts).......how much amps should the diodes be able to take....in other words, do the whole 55 amps pass through one diode at a certain time, or is it divided among the others too? so should the diodes be able to handle 55 amps or less?


thx in advance

« Last Edit: July 08, 2004, 11:37:37 AM by (unknown) »

finnsawyer

  • Hero Member
  • *****
  • Posts: 1565
Re: Diode size
« Reply #1 on: July 08, 2004, 08:33:53 PM »
I assume you're using the wye connection in which case the diodes conduct in pairs.  Each diode when its conducting must carry 55 amps.  However, each diode conducts only one third of the time.  So the average diode current is 18.3 amps.  The diode heats up as though that was the current.  Consequently 20 amp diodes should be sufficient as they are designed to handle that amount of heating full time.
« Last Edit: July 08, 2004, 08:33:53 PM by (unknown) »

Ungrounded Lightning Rod

  • SuperHero Member
  • ******
  • Posts: 2866
Re: Diode size
« Reply #2 on: July 08, 2004, 11:43:25 PM »
Each diode when its conducting must carry 55 amps.  However, each diode conducts only one third of the time.  So the average diode current is 18.3 amps.  The diode heats up as though that was the current.  Consequently 20 amp diodes should be sufficient as they are designed to handle that amount of heating full time.


Though diode conduction is non-linear (that's the POINT, after all), it's not a step function.  Thus the dissipation goes up faster than the current.  So three times the current for one third the time is NOT the same dissipation.


It's not quite as bad as a resistor, which dissipates I squared R watts.  With a resistor you'd dissipate THREE times as much power in this situation.  (1*1*1 vs 3*3*(1/3)).  But it's pretty close.  And even within the conduction period it's not a step function:  Assuming the bulk of your load is resistive and your waveshape is sine the peak current would be 1.414 times the average current, which makes it more than a factor of 1.414 worse.  And if you're charging batteries the peak is sharper than a sine wave, making it worse yet.


I'd be conservative and buy diodes with a rating of at LEAST twice the average current, and preferably a lot more.


High-current diodes aren't all THAT expensive.  Especially compared with an alternator damaged by overcurrent, or by overspeed due to unloaded blades in a storm.

« Last Edit: July 08, 2004, 11:43:25 PM by (unknown) »

TomW

  • Super Hero Member Plus
  • *******
  • Posts: 5130
  • Country: us
Re: Diode size
« Reply #3 on: July 09, 2004, 06:43:56 AM »
rod;


I must agree on the over sizing of diodes.


Even my TDM mills have 100 amp Schottky blocking diodes on them and will never exceed 15 amps output.


The cost difference is so small that it just makes sense to over size the diodes by a factor of 3 and to heat sink well with good air flow. Always good design practice to not scrimp on the power rating of critical components. Replacing a diode is no big deal [unless it is on the tower] but, as you pointed out, a diode failure could have a domino effect and cause a cascade failure of other components or devices.


Cheers.


TomW

« Last Edit: July 09, 2004, 06:43:56 AM by (unknown) »

finnsawyer

  • Hero Member
  • *****
  • Posts: 1565
Re: Diode size
« Reply #4 on: July 09, 2004, 08:59:38 AM »
The diode junction acts like an imperfect switch.  In the reverse mode it will have a large voltage but negligible current.  In the forward mode the voltage varies logarithmically with current.  It will increase as the current goes from 20 amps to 55 amps, but not much.  Maybe from .6 volts to .7.  So the power dissipated depends essentially on the current.  If the current flows for only a third of the time one can use a diode with a lower current rating as I suggested.  Maybe he already has these diodes and can try them rather than spending money on the larger ones.  For an experimenter spending a few dollars more for a more robust design probably doesn't matter.  If one were going to mass produce these things it would.  A million units at $2.00 more per unit adds up.  That's why we have engineers and why they build prototypes of their designs.


If you've actually found diodes that behave the way you say, I am inclined to think that they have an excessively large internal resistance drop.  A poor design?  Still in this world all sort of things are possible.  I think the answer I gave was correct as far as the diode junction is concerned and was in the spirit of his question.  He can use this as a guide for his own work.  

« Last Edit: July 09, 2004, 08:59:38 AM by (unknown) »

Victor

  • Jr. Member
  • **
  • Posts: 93
Re: Diode size
« Reply #5 on: July 09, 2004, 10:56:00 AM »
No one has mentioned that the ratings  are dependent on a juction temperature of 25 deg. C . That Jt is probably a rarity in most home brew projects, another reason to be conservative.


Victor

« Last Edit: July 09, 2004, 10:56:00 AM by (unknown) »

(unknown)

  • Guest
Re: Diode size
« Reply #6 on: July 20, 2004, 05:49:57 PM »
Amen. If I may suggest a simple, CHEAP, way to get any and all the diodes you might want or need...

(Drum roll please)

Old computer power supply(ies)

Yep, they have pretty darned good Shottky 'fast response' diodes in them, in pretty robust packages. (And the great part is, they are usually already mounted to a pretty good heat sink) They often come in a package which looks at first like a large transistor, but will have marking like this:

->|<-

Representing that there are TWO diodes in the package, with one common connection. In the example I gave, which is the most common arraingment, the common connection is the cathode, leaving the anodes seperate. These diode pairs will be very, VERY closely matched... Better than you could do with a DMM or DVM. I have seen them with up to as much as 40 amps rating PER LEG! Typical is around 20 or 30 amps per packege, (pair) and then the next-up size, also common, about the same PER DIODE. 10 to 15 amps per diode also common. I can provide some links to sites which will help you find out what you have. (There are also 'discrete', or single, diodes in there too)

With such diodes redily available, no reason to skimp on the amps ratings. Simple to double up if needed, solder up another pair. Heck, they'er free!

Any computer power supply will have several of these, and if you start looking, it won't be long before you have people offering YOU old and/or dead power supplies, and whole computers too, for that matter. (Be careful, the leftover 'junk' can be hard to get rid of in some places) I salvage LOTS of old and not-so-old power supplies, and I use everything but the squeal. Especially the fans. And large electro caps. And bridge rectifiers, and so on and so on. Anyone, even in the most out-of-the-way places can come by an old computer or two. (There's always eBay... Shipping might cost more than the danged thing is worth, though!)

Point is, a very simple test with a DMM, or even a simple homemade go/nogo tester will sort out which diodes are good, and which way is anode or cathode. No reason to pass up free or almost free parts, and put them back to work, rather than in the landfill. (If interested, email for plans for a very simple, cheap, easy go/nogo tester made with a flashlight 'D' cell and a bulb from same)

Saftey note: Capacitors in these power supplies can and DO hold a LEATHAL charge for some time after unpluging... Discharge caps first THING! If you don't know how... Email me, and I'll teach you how.


Junkman


Therealjunkman@Yahoo.com

« Last Edit: July 20, 2004, 05:49:57 PM by (unknown) »