| jarthel |
situation:
I need a 100V 4A diodes to make a bridged diodes. But I only got 100V 2A ones.
would I get the same effect as a 100V 4A diode if I parallel 2x 100V 2A?
I was thinking along the lines of resistors. possibly incorrect thinking.
Thank you.
ps. no answer please of just to buy a 100V 4A. this thread's purpose is to gain some knowledge on the idea of paralleling diodes. :) |
|
|
| janneman |
The problem with your solution is that the two parallel diodes will have a slightly different forward voltage. The one with the lowest forward voltage will conduct more than the other, so you may end up with one diode carrying 80% of the total current and the other 20%.
One solution is to use small series resistors for each diodes, like 0.25 ohms, which will add (at nominal 2A per diode) some 0.5 volts to each forward voltage, equalising the current division better (but still not 50-50), the disadvantage is a small loss of voltage (across the resistors) and some extra heat.
Since you told us not to mention the obvious solution, I won't do that ;) .
Jan Didden |
|
|
| jarthel |
but if the diodes are of the same make/brand, why would it have different forward voltage?
ps. thank you for the explanation above :) |
|
|
| peranders |
| The advice here is: avoid paralleling diodes unless the are double diodes in the same package. Under those conditions the two diodes are at the same temperature and also rather equal. |
|
|
| sreten |
| quote: | Originally posted by janneman
The problem with your solution is that the two parallel diodes will have a slightly different forward voltage. The one with the lowest forward voltage will conduct more than the other, so you may end up with one diode carrying 80% of the total current and the other 20%.
One solution is to use small series resistors for each diodes, like 0.25 ohms, which will add (at nominal 2A per diode) some 0.5 volts to each forward voltage, equalising the current division better (but still not 50-50), the disadvantage is a small loss of voltage (across the resistors) and some extra heat.
Since you told us not to mention the obvious solution, I won't do that ;) .
Jan Didden |
Hi,
The voltage drop across rectifier diodes at peak current (~2V)
is higher than the forward voltage, the currents will equalise.
Due to the very high peak currents resistors in the bridge is not good.
The safest option is to parallel 3 diodes, why not ?
:)/sreten. |
|
|
| darkfenriz |
| quote: | Originally posted by sreten
why not ?
|
Highly positive thermal coefficient. |
|
|
| pinkmouse |
I could just say - "'cos they do", but I think you want a slightly more in depth explanation. :)
Every single component is made to a tolerance, and no two devices will ever be identical. If you look at some of the threads about matching transistors, you will see that you can narrow the selction down by using devices with the same date code, or from the same batch, but even so, that just reduces the amount of deviation, it doesn't eliminate it. As for it's importance, imagine a pencil, balanced on it's point. The slightest disturbance will cause the pencil to fall over, much as the slightest imbalance in the diodes will cause one to take more current and go bang |
|
|
| janneman |
| quote: | Originally posted by sreten
[snip]The voltage drop across rectifier diodes at peak current (~2V)
is higher than the forward voltage, the currents will equalise.
[snip]:)/sreten. |
The voltage drop IS the forward voltage. Whatever the currents, they will surely NOT be the same, unless you go through the trouble of carefull matching at the specified current. Even then, in actual use, the temp will be different and the current sharing will drift.
Jan Didden |
|
|
| jneutron |
It is very easy to parallel diodes. It requires:
1. VF matching at the highest current the die will see. They will have to be matched to within 1 millivolt, they will have to be tested at the same ambient temperature, within minutes of each. Drift in the test equipment, even at a 10 minute time scale, can throw the match off.
2. They have to be from the same diffusion run, and typically, from the same wafer. This guarantees that they will have the same VI curve, and that they will respond in the same manner to current over the range of temperatures you desire.
3. The diodes have to be in intimate thermal contact with each other. Any slight difference, and the hotter one will run away with the current. This stability of course, depends on the diode type, as some will have different bulk silicon resistivity depending on speed and process.
4. True paralleling requires absolutely no series resistance difference between the two, but this flies in the face of what is possible for diy'ers, as they cannot typically get same diffusion run or same wafer devices. Jan is correct in that a small series resistance will work.
5. Thermally, they have to have identical cooling parameters..It is possible to create an isothermal island to get around this to some extent, but that compromised device life.
It is possible to run a full range of tests to find diodes which can be paralleled, but that test is not very easy to do. I know, I've done it..in between dodging the velociraptors..:D
Cheers, John |
|
|
| Tweeker |
| Are you sure you need 4A IFAV? The diodes are each only on half the time. |
|
|
| janneman |
| quote: | Originally posted by Tweeker
Are you sure you need 4A IFAV? The diodes are each only on half the time. |
Even less, more like 10% of the time. Only if the sec AC is higher than whatever is on the reservoir cap. The current through these diodes will be high-level pulses, nothing like a continuous current. You *may* need 4A units if you have a huge reservoir cap because you need the repetitive puls rating of the diode.
Jan Didden |
|
|
| jarthel |
| I'll be using 56mF caps (cap input filter). |
|
|
| audioservice |
| The higher the current the more the diode's ohmic resistance affects its voltage drop. The conduction knee for a silicon rectifier diode such as a 1N4004 is around 0.5 volts. At the rated current the voltage drop will be around 1 volt. Hence, the diode's ohmic resistance acts to ballast the currents in the parallel diodes. At low currents it doesn't matter how well they ballast because they are operating below rating. The downside is that two parallel diodes having the combined current rating of a higher current diode will not have as great a combined **peak** current rating as the larger diode. The peak current rating is imporatant at power on when a high surge current occurs as the filter capacitor is charges up. It is best to parallel diodes from the same manufacturer. Even better, from the same batch. To be sure measure their voltage drops at rated current to "match" them. CHECK THE SPEC SHEETS! |
|
|
| TwoSpoons |
| Good grief! Just go buy some bigger diodes, and stop ****ing about. They're just not that expensive :clown: |
|
|
| janneman |
| quote: | Originally posted by jarthel
I'll be using 56mF caps (cap input filter). |
What is the repetitive pulse current rating of your diodes? The 2 or 4 amp continuous is really saying nothing in a rectifier application.
Jan Didden |
|
|
| Tweeker |
| Unless were using a choke. Edit: and were not. |
|
|
| Nordic |
| quote: | Originally posted by rpapps
Read post 1 |
That's like saying "I insist on breathing through my ****"
Sure I bet with some tinkering it may be possible... but compared to just going to the shop.... |
|
|
| Tony |
| quote: | | I need a 100V 4A diodes to make a bridged diodes. But I only got 100V 2A ones. |
if you are reffering to a full wave bridge, they i do not see why not, go ahead and use them. you will need 8 pieces instead of 4, i have done this lots of times.
but what is your total load btw, rail voltages? and power requirement of the load.
for all we know, those 2A diodes may be enough..:D |
|
|
| rpapps |
| quote: | [Originally posted by jarthel]
ps. no answer please of just to buy a 100V 4A. this thread's purpose is to gain some knowledge on the idea of paralleling diodes. |
| quote: | Originally posted by TwoSpoons
Good grief! Just go buy some bigger diodes, and stop ****ing about. They're just not that expensive :clown: |
| quote: | Originally posted by rpapps
Read post 1
|
| quote: | Originally posted by Nordic
That's like saying "I insist on breathing through my ****
|
Nordic, that's why I said read post 1.
If everybody bothered to read threads from the beginning instead of just leaping into the middle and posting an opinion, there would be a lot less confusion and misunderstanding.
Unlike a lot of people who post questions here, Jarthel wants to learn rather than be spoon fed so let's respect his wishes. |
|
|
| Nordic |
I suspect he just has to learn to wait till Monday like all of us at some point, when we realise we got the wrong components for a job on a weekend.
Sure, I understand being told, no it doesn't work, he would want an explanation.
And I think the one he got was that mass manufacturing is unable to produce identicaly matched diodes... which unless you load balance would have very asymetric thermal response... i.e. one will get alot warmer than the other...
I think as part of learning you would also think of planning for minimum component count and smallest PCB from a financial point of view.
I would say give the 2A diodes a shot... feel them for overheating after a few minutes of use... and decide from there wheter they are copeing or not... either way, its nearly Monday, and the shop will be open again. |
|
|
| EchoWars |
OK...another take on this, since he wants to learn something.
There's a right way to fix or build something, and a wrong way. Paralleling diodes is always a hack thing to do. If I open up an amp and someone has done this, I immediately know he's clueless.
There might be some gyrations you can do to get paralling to work, but compared to getting the diode you really ought to be using it's just wrong (especially when the next tech looks at it and wonders WTH you were thinking).
My $0.02...take it, leave it, ignore it. |
|
|
| sawreyrw |
EchoWars,
I know this guy wants to know if he can parallel diodes, but your response is right on. I've been trying to think of a 'good' way to say what you have said, but I couldn't come up with anything as good as your comments.
Rick |
|
|
| Tony |
| in the land of plenty such as the states, where everything is there for the picking, you thoughts are perfectly ok. |
|
|
| AndrewT |
Hi,
Nelson P. published a monster PSU schematic that used parallel diodes. With appropriate balancing.
Do you have the same opinion about his work? |
|
|
| EchoWars |
That's a silly baiting question and doesn't deserve an answer.
The original poster made it clear that he wants to learn something, so if you're hip to something that hasn't been discussed, the floor is yours. |
|
|
| AndrewT |
Hi,
there's two ways to read my comment.
First to recognise that paralleling of diodes has been done, done successfully and done by a reputable designer.
or
Secondly, it gives the reader a possible search clue to go and do some research.
Only a few readers would see it as antagonistic, but maybe I'm biased. However, is there a nicer way to say someone is being too generalistic in posting an opinion that cannot be supported by fact. There's a short phrase for this but I am not allowed to post it. |
|
|
| poobah |
Post 2 & 9 say it all... and well.
Paralleling diodes is common practice, and is quite often the only practical alternative.
In the audio world... it would be generally desirable to avoid the power losses and parts that Post 2 imply; simply because sufficiently sized diodes are readily available. As well, the complexity and constraints of Post 9's are also a hassle.
Somtimes people ask an audio question and get an engineering answer... nothing wrong with that at all. Engineering IS the parent science.
:rolleyes: |
|
|
| jarthel |
| thanks for the "on topic" responses :) |
|
|
|
