| tiroth |
Stupid question:
In (say) a 35A bridge, is each of the individual diodes rated for 35A continuous? |
|
|
| Christer |
Don't worry, stupid questions sometimes turn out to be quite
intelligent. Actually, if you just look at a datasheet you won't
find the answer clearly stated. However, as long as we connect
the bridge as intended, ie. to rectify AC, there will always be
exactly two diodes conducting (or none) and the same current
must flow through both of them. It is thus quite obvious that the
current rating for the bridge is the same as the current rating
for each diode, so there is no need to distinguish these ratings.
Then, the rating given is usually the average forward continous
current. I would think it is always the average, but I am not
sure of this, so better check the datasheet. |
|
|
| peranders |
| quote: | Originally posted by tiroth
Stupid question:
In (say) a 35A bridge, is each of the individual diodes rated for 35A continuous? |
Normally you have to read the datasheets but sometimes it can be hard to convert data into real life.
To be honest, I'm not totally sure what "Maximum average forward rectified current" means.
I think it is 63% of the peak current. It means 38 A rms on the AC side, assumed resistive load (not an power amp).
Peak current is normally 5-20 times the continious value.
Your exact question: I guess each diode can take at least half the rated current. |
|
|
| Christer |
| quote: | Originally posted by peranders
Normally you have to read the datasheets but sometimes it can be hard to convert data into real life.
To be honest, I'm not totally sure what "Maximum average forward rectified current" means.
I think it is 63% of the peak current. It means 38 A rms on the AC side, assumed resistive load (not an power amp).
Your exact question: I guess each diode can take at least half the rated current. |
Yes you are probably right, I got it wrong. Since it is the average
rectified current and each pair of diodes conducts only every
second half cycle it is more likely that each diode can take half
the rated current.
Peak current is normally 5-20 times the continious value.
Don't you mean peak surge current here? |
|
|
| tiroth |
Thanks for the help folks. I was guessing it might be 1/2 of the rated "maximum average" forward current, but after reading a few manufacturer's datasheets I was more confused than when I started.
I wanted to figure out what rating of diodes was necessary to make an equivilent discrete bridge, so I will try using paralleled 8A Schottkeys.
Thanks! |
|
|
| Christer |
Beware that diodes might not share the current equally when
paralleled. I don't know if there is any good way around this.
One might consider small resistors in series with each diode,
but that is hardly a good approach. Maybe you can find
something useful in this application note
http://www.st.com/stonline/books/pdf/docs/3602.pdf
I found it just a few days ago and haven't had time to read it
myself yet. |
|
|
| tiroth |
That is a pretty useful white paper! However I am not entirely sure how to apply it to the kind of situation commonly seen in charging a large C bank, i.e. very narrow conduction angle.
I guess I will select for lowest possible delta Vf and pray. ^_^ Really these are intentially overspec'd for most designs so I don't anticipate problems. |
|
|
| peranders |
tiroth, may I ask what you are giong to with these diodes?
If you want many amperes, check MURxxxx diodes from ON. |
|
|
| Christer |
The diode specs. are given for sine waves, and I suppose the
main limitation is power dissipation, not current, so I would
think they can take quite a lot more current when we have a
narrow conduction angle. We can easily compute the power
dissipation in a diode under the specified conditions. The power
dissipation under the narrow conduction angle might be more
difficult to compute, but to play it safe (I hope) one might
simplify and use the peak current as a constant current drawn
during this whole time period. Then compensate for this
higher current by finding the appropriate voltage drop and
compute this overestimation of the power dissipation. If it is
considerably below the power dissipation under spec'd
conditions and the peak current is not extreme, I would think
this is safe. How many actually do take the narrow conduction
angle into account? My guess would be that most don't or
use som rule of thumb because it seems too difficult.
* Disclaimer: As noted, I don't know for sure if this is safe
and reasonable, it is just my intuition and assesment, for what
it is worth. * |
|
|
| janneman |
I think it is not as bad as you think.
35A average current is just that, 35 amps average. (Wait, don't shoot me yet). So say if you build a ps to power a class A amp with 3A standing current, that is the average curent taken from the supply (x 2 for stereo). Your 35 amps bridge will have no trouble whatsoever with this. The limiting factor will most probably be the max repetitive peak current, normally specified for 50 or 60 Hz. That depends on the filter electrolytics as already noted above. Stay reasonabel here and you'll be fine.
Paralleling diodes is quite hard to do without wasting precious power and is totally unnecessary.
Jan Didden |
|
|
| tiroth |
Well, I already have a goodly quantity of TO-220 8A dual Schottkey-type rectifiers, so that is where the paralleling came in. I think I paid something like US$0.05 per piece surplus which is quite a lot less than new!
Unfortunately I have been unable to locate a datasheet so I was going to parallel as a safety measure. |
|
|
| tiroth |
Suimasen.
I just dug them out and they are actually single-junction packages, GI 837A. Couldn't find that part anywhere--OEM?
(Tried Questlink, Google, Chipdocs) |
|
|
| janneman |
| quote: | Originally posted by tiroth
Well, I already have a goodly quantity of TO-220 8A dual Schottkey-type rectifiers, so that is where the paralleling came in. I think I paid something like US$0.05 per piece surplus which is quite a lot less than new!
Unfortunately I have been unable to locate a datasheet so I was going to parallel as a safety measure. |
Well, unless you are planning to power an industrial welder or have fractional Farad capacitors in your supply, a bridge with your 8 amps diodes should do fine!
Jan Didden |
|
|
| peranders |
| quote: | Originally posted by tiroth
Suimasen.
I just dug them out and they are actually single-junction packages, GI 837A. Couldn't find that part anywhere--OEM?
(Tried Questlink, Google, Chipdocs) |
I tried Google.
http://pubpages.unh.edu/~aperkins/pdf/Misc/crossref.pdf
Maybe it can be at some help. GI = General Instruments if I'm not wrong? |
|
|
| peranders |
| quote: | Originally posted by janneman
Well, unless you are planning to power an industrial welder or have fractional Farad capacitors in your supply, a bridge with your 8 amps diodes should do fine! |
8 A is not very much in a stereo amp >100W out |
|
|
| janneman |
| quote: | Originally posted by peranders
8 A is not very much in a stereo amp >100W out |
Why not?
Jan Didden |
|
|
| peranders |
| quote: | Originally posted by janneman
Why not?
|
100 watts @ 4 ohm x 2 (stereo) = 10 A rms = 5 A from each rial => approx 10-15 A peak (due to smoothing caps) all the time. I think the "headroom" for the diodes is a little small.
It's maybe OK for normal music but for party use?
I think 15-25 A bridge is good choice for medium amps. Why save money on such important part which is rather cheap? |
|
|
| janneman |
| quote: | Originally posted by peranders
100 watts @ 4 ohm x 2 (stereo) = 10 A rms = 5 A from each rial => approx 10-15 A peak (due to smoothing caps) all the time. I think the "headroom" for the diodes is a little small.
It's maybe OK for normal music but for party use?
I think 15-25 A bridge is good choice for medium amps. Why save money on such important part which is rather cheap? |
5A from each rail for continuous 100W @ 4 Ohms? Must be some party! (Must be very special music also).
Even then, there is a 60% safety margin. And we all know that the real average is closer to 10 or 20W, which is just a couple of amps. So you get closer to 400% safety margin.
And the peak value is probably higher than you say, but that is no problem. repetitive peaks of 5 or 10 times the average spec is not an uncommon spec for these diodes.
And yes, a 15-15 amp is a good choice, but in this case the guy has the diodes laying around. So, you ask the wrong question: "Why save money on such important part which is rather cheap?" The real question is: "Should I spend more money without benefit because I already have what I need?" Answer: No.
Jan Didden |
|
|
| dhaen |
Heck, why worry on the detail?
I just work out the current, then choose a rectifier of 2 to 4 times the rating. They're cheap. It's not as if we are designing for mass production where every cent counts.
Oversize rectifiers also drop slightly less volts.
Schottkys are the coolest:cool: |
|
|
| tiroth |
I didn't realize this issue would cause such debate, but it has definitely been educational! I think I am assured that my safety margin is reasonable even without paralleling. Dissipation should not be high. I think the only worry is exceeding the peak half-cycle draw at turn-on, and there are ways to address that.
Thanks again! |
|
|
| till |
| a question here: why are schottky the coolest? suitable "cool?" schottkys would be FYPF1010DN or ON MBRF20100CT ? what about fairchild "stealth" super/hyper /giga - whatever - fastorsoftrecovery - ISL9K3060 or N218R1560G2 ? |
|
|
| peranders |
| quote: | Originally posted by janneman
5A from each rail for continuous 100W @ 4 Ohms? Must be some party! (Must be very special music also). |
If you don't want to fry the amp with a test signal it should withstand full power for at least 5-10 minutes but if it's OK to have a black hole after some unexpected input, that's fine by me.
Compare NAD the cheapest, this amp will burn if you drive it at party levels. How funny is that? |
|
|
| tiroth |
| quote: | Originally posted by till
a question here: why are schottky the coolest? |
I'll go out on a limb and see how right (wrong) I am.
Schottky diodes as a majority carrier device should be closer to ideal in terms of their recovery characteristic. Fast/soft recovery conventional PN are just trying to get closer to this standard. However not all Schottky are equal, especially in terms of reverse leakage current.
http://www.ixys.com/t070299c.pdf |
|
|
| dhaen |
| quote: | Originally posted by till
a question here: why are schottky the coolest?
|
Sorry guys, it was my little joke;)
'Cos of my age; coolest means "the lowest temperature" to me:rolleyes:
Schottky's may well be better for other reasons too:) |
|
|
| peranders |
| quote: | Originally posted by dhaen
Sorry guys, it was my little joke;)
'Cos of my age; coolest means "the lowest temperature" to me:rolleyes:
|
Are sure? It's often particulary the temperature which detemines the max current. I think the schottky diodes will run equally hot as silicone diodes. |
|
|
| dhaen |
| quote: | | Are sure? It's often particulary the temperature which detemines the max current. I think the schottky diodes will run equally hot as silicone diodes. |
They generally have a lower voltage drop.:yes: |
|
|
| janneman |
| quote: | Originally posted by dhaen
Heck, why worry on the detail?
I just work out the current, then choose a rectifier of 2 to 4 times the rating. They're cheap. It's not as if we are designing for mass production where every cent counts.
Oversize rectifiers also drop slightly less volts.
Schottkys are the coolest:cool: |
Sure, no problem. Your (or anybody else's) call anyway. I was just trying to answer a specific question as good and as precise as I could.
Jan Didden |
|
|
| Christer |
it seems your question wasn't stupid at all, but rather one of
those intelligent questions that nobody has thought about
asking before. Or perhaps we are all just a bunch of stupid
guys arguing about the answer? :) |
|
|
| tiroth |
| quote: | Originally posted by dhaen
They generally have a lower voltage drop.:yes: |
Here is another wrench in the works: I thought up until recently the high voltage ones also tended to have high reverse current which could lead to high dissipation. Arg!
Fortunately that problem seems to be going away. |
|
|
| janneman |
| quote: | Originally posted by tiroth
I'll go out on a limb and see how right (wrong) I am.
Schottky diodes as a majority carrier device should be closer to ideal in terms of their recovery characteristic. Fast/soft recovery conventional PN are just trying to get closer to this standard. However not all Schottky are equal, especially in terms of reverse leakage current.
http://www.ixys.com/t070299c.pdf |
Yes, and they also have less forward drop than conventional Si diodes, that why they were popular (maybe still are) in low output voltage smps. So, for the same current, they dissipate less and thus are cooler (lower temp...).
Jan Didden |
|
|
| jwb |
This matter of diode specification seems fraught with peril. In my next amp I'm using 5A average forward current, but with a 324000µF CRCRC power filter. The conduction angle with such a filter is difficult to determine, but I know it is roughly 11°-15°. With this duty cycle (3-4%) the peak repetitive forward current will be about 80A!
If you consider a device like IRF MBR4045WT, you see it has an average forward current of 40A and peak repetitive forward current of 80A. However this latter spec is for a 50% duty cycle, I think. With a 3% duty cycle I think I should be okay??? The max. surge current for this device is over 1000A.
Anyone have advice on a similar high-current flat-pack Schottky? I chose Schottky over HEXFRED and similar diodes because the low forward voltage of the Schottky allows me to have higher voltage rails with the same transformer. |
|
|
| peranders |
| As long as we talk about rectified currents it's a little bit hard to determine how big the diodes should be. One important factor is temperature of the bridge. It should not be too hot and if the smoothing package is bigger than "normal" maybe it's a good idea to have a little larger diodes due to those peak currents. |
|
|
| jwb |
Yeah, i'm planning to mount these flat-pack diodes on a 16"x12"x.125" aluminum flat sheet. Should be plenty of heatsink for the job.
:hot: :hot: :hot: :hot: |
|
|
| tiroth |
| Due to the dissipation law for diodes I don't think thermal overload will be a big concern, even with the narrow conduction angle. As long as they are rated for the peak current I would guess it is fine. |
|
|
| sully |
A bridge is typically current rated using the following:
1. Maximum average forward rectified output current, with the ratings based either on a case temperature if heatsinked, or ambient temperature, if it is supported by the leads.
2. Surge current. This is the non repetitive current pulse the unit will survive..usually rated as one 8.3 mSec pulse, but can be rated two pulses.
3. Maximum reverse leakage and Vbr.
4. Case temperature.
Each diode within will see half the average current..a 10 amp bridge will have each diode seeing 5 amps average. A ten amp bridge can be built with 5 amp diodes, but the surge ratings may be too low.. the surge rating for larger die is better, and the vf will be reduced.
Schottky's have much lower VF for the current, but depending on the barrier metal, may leak like a sieve....Shottky's can dissipate as much power in the reverse direction as forward, depending on the application.
Using 4 8 amp diodes for an 8 amp bridge is perfectly fine..But if the cap bank is really oversized, you will be better off with larger diodes just for the surge that the caps will have at turn on. Skimping on this will result in reliability issues quickly (it blow'd up).
For 60 or 50 cycle use, all diodes in production today do not dissipate any significant power during turn off, so the recovery speed of the die is not important for derating. Typical stuff will be 2 microsecond.
Some manu's also provide a fusing rating for the device..the 6 ampere GI device I'm looking at has a fusing rating of 127 amp squared seconds.
PN die will be rated to 150 C typical junction...shottky's 100 C or 125 C depending on manu.
BTW, GI is general instrument...I was an app. eng there for a while.
Cheers, John |
|
|
| tiroth |
John,
Then maybe you can tell me why those GI data sheets are so impossible to locate. (Just kidding! ;)) |
|
|
| jwb |
Great info John. So would you say that the current ratings for a diode are based mostly on junction/barrier temperature? The Schottky I'm pondering is rated to 150°C operating, and I estimate the Rθjs at 2.5K°/W. So do I merely need to analyze the heat transfer to keep the peak junction temp within the ratings, or is there another limit to peak repetitive current?
As for the surge rating, I'm dodging that problem by charging the filter initially through a resistor. |
|
|
| sully |
| quote: | Originally posted by tiroth
John,
Then maybe you can tell me why those GI data sheets are so impossible to locate. (Just kidding! ;)) |
Because I took them all..
Do you need any info?
| quote: | Originally posted by jwb
Great info John. So would you say that the current ratings for a diode are based mostly on junction/barrier temperature? The Schottky I'm pondering is rated to 150°C operating, and I estimate the Rθjs at 2.5K°/W. So do I merely need to analyze the heat transfer to keep the peak junction temp within the ratings, or is there another limit to peak repetitive current?
As for the surge rating, I'm dodging that problem by charging the filter initially through a resistor. |
The current ratings are derived by the junction temperature..If you look at the forward current derating curve of any diode, you will see it go to zero current (usually straight line) at some temperature..Most manu's will do so at 125 C or 150 C. Below the break temp, anywhere from 40C to 100C, it is constant current . So a 3 amp diode will be three amps up to the break, say 40 C...then they draw a straight line to the highest temp and zero current. The intent of that graph is to depict a constant junction temperature, although it is rather crude. VF lot variation will change that curve, and the manu's make that line conservative.
That graph is actually just an approximation..the reason is the current/voltage curve, which give a non linear power/current relation. But you won't get in trouble if you follow it..
Using theta Jc calcs with heat transfer will work fine.
The GI page I'm looking at also has a graph of peak forward current vs number of cycles at 60 hz..a 3 amp unit is about 150 amps 1 cycle, to about 100 cycles at 27 amps. Note that that is a half sine wave, as per a jedec method. Your diodes look to be rated for other than 60 hz operation, hence the 50% number. If you have transient thetaJc graph of the diode, you can estimate the junction rise max during each cycle.
In a supply, the average will be the only concern. The junction temp fluctuations during each cycle will be only in the degrees range, regardless of the amount of capacitance it is charging..
Try to stay as low in temp as possible..lifetime doubles every 10C drop in junction...and shottky's have some major leakage hot, and you may have to calculate the power loss as a result of the leakage current during the reverse half cycle.
If you find surge is a problem, then you could go with a bigger die size, with less heat sinking required.
Cheers, John |
|
|
| sully |
I looked up your devices..note that the reverse leakage at 100 C is 50 milliamps..
At 45 volts, that is 2.25 watts dissipation. If you run hotter, that will really skyrocket. They are capable of IR runaway if you don't consider that power dissipation. If your running 10 amps, that is about 6 watts forward power, 2.25 reverse..about 1/3 of the expected dissipation added.
The reverse leakage runaway used to be a major issue with schottky's and gold doped fast recovery diodes. But I haven't built or tested a small diode in 11 years. My experience of late is with 6000 amp 2.5 inch diameter diodes..
Cheers, John |
|
|
|
