Re: A Bridge Replacement A Good Idea?

Pooh Bear wrote: 
John Popelish wrote:


Pooh Bear wrote:

John Popelish wrote: 

(snip)

How long will that current last if this thing is plugged in at the
peak voltage.  It is charging 11 uF.  Hint:  It is no where a half
cycle.



I was just looking at that. I had assumed a larger value for the storage caps.

I'm still not keen on exceeding Ifsm though. They don't call it absolute max for nothing ! 

But that "absolute max current" includes the additional spec that it
is allowed to take place over 8.3 ms.  That isn't going to come close
to happening with this circuit.


It's true I hadn't initially considered the very low value of reservoir cap.

;-) It is very unusual. By the way, 34 amperes will charge 11 uF to 340 volts in 125 us.

By the way, a cold 1 amp slow blow
fuse typically has almost 1 ohm of resistance.  More as it gets hot.
That additional ohm lowers the peak current to 340/11=30.9 amps.


(snip)

I appreciate that you are a conservative guy.  But I think you should
back your conservatism up with actual reasons, not just hunches, when
you are giving advice.


I'll back it up with experience. Been bitten too often ( actually not that often - but with a boring
predictability ) when designing close to the edge. False economy. It's simply not worth it. I've been
pressured into designing in parts I wasn't entirely happy with too. Thankfully it was the client who
insisted ! They can pay for the repairs.

As long as you understand exactly how close to that edge you are, and are not just playing a hunch, I agree.

What do you think of the peak 10,000 or so watts dumped into that 5
watt resistor during the same start up current pulse we have been
discussing?  Do you think that might exceed its absolute maximum rating?



Funnily enough ( I assume you mean the 10R ) that just crossed my mind. A while back I had a small
SMPS design that used a 3W 4R7 inrush limiter R. I toyed with using a power metal film R as combined R
and fuse ( since it fails flame free ). Worked fine here but the Chinese sub-contractor couldn't find
a part with equivalent durability. They had to fit wire-wound to avoid it 'blowing', so it's back to a
fuse plus R ( I'm actually using a 'Surge Gard' this time ).

It's all about thermal mass basically. A wirewound will survive the shock. 

Prove it.  ;-)

I think that part is a lot closer to destruction that 1 amp diodes
are, but obviously survives, or the author would have had to pick
something else.  I suspect the author blew up some 1 amp diodes in
early experiments with bigger front end capacitors and learned more
from the lesson than there was in it.


On 120V supplies with the same 10R it would likely be fine with 1N4007s though. 

The peak heating (not the peak current, but I*T, which is roughly what
heats diodes) is about the same, because set up for 120 volts, it
charges 22 uF on the first half cycle, instead of 11 uF.

I don't see the long term heating issue being a problem actually.

I was talking about that first half cycle, not long term. I think we both agree that the circuit will work fine, long term with 1 amp diodes. Only the inrush has been under "serious" discussion.

The sad thing is that I think this sort of discussion is a fine way to spend between 1 and 3 A.M. on a Sunday morning, when the missus is out of town. But it is beginning to cut into tomorrows (well, later today's) motorcycle riding.

Never underestimated the damage you can do to diodes with pulses. I used to think you could
use a 1N914/4148 safely as a catch diode on typical small pcb relays until I saw several
failures. I fit 1N400Xs there now.


Good for you.  By the way, do you remember the coil current on the
diodes that failed?


Not offhand and it's a long time back. They would likely have been 12V or 24V coils though. Not
the 'sensitive type'.

This kind of thing actually....

http://uk.farnell.com/jsp/endecaSearch/partDetail.jsp?SKU=103069&N=401

12V 285R - 42mA 

I want to see you blow up a 1N4148 with 42 mA, interleaved by 12 volts
reverse, of any duration or number of cycles you are willing to wait
around for.  It ain't going to happen before we get struck by
lightning or swallowed up by an earthquake.  I have seen literally
thousands in such service with no unexpected failure rate.



Seen it several times with only a couple of units in service. It *may* have been a similar 24V relay
but with obviously lower operating current.

Consider the waveform shape and time and consider the diode's intrinsic R and then the heating effect.
It *has* to be transient thermal failure following a fatigue period.

How regularly ( time interval ) were your relays being switched btw ? That might explain something.

Every few minutes, round the clock. One application was a thermocouple data logger that cycled hundreds of relays to select thermocouples to be connected to the amplifier. No diode failures.

I've seen similar problems in another configuration too. Just last week did a similar swap drom 4148s
to 4004s.

It's *all* about high peak current. 

1N4148 diodes are rated for 200 mA, average rectified current at 25C:
http://www.fairchildsemi.com/ds/1N/1N4148.pdf
That implies peaks higher than that.

42 mA peak (relay coil discharge currents are roughly triangles that ramp to zero in some milliseconds) isn't going to hurt them with thermal cycles at any repetition rate that a relay will be operated at.

I designed quite a few circuits that pulse such diodes with hundreds
of mA for 10s of microsecond durations and adequately low duty cycles,
also with good reliability.

I might be wrong, but that is what I understand to be the case.  If
you witnessed a failure under those conditions and didn't investigate
enough to claim your Nobel Prize, you missed a great opportunity.

If this failure took place on a breadboard with test clips being moved
around and parts reused till the leads broke off, that is another
matter, entirely.  I've blown up lots of stuff, that way.



No. This was a fully functioning unit in service. It happened more than once. Part substituted - no
more failures.


Go for that prize, man!
.

Relevant Pages

  • Re: 4 ohms/8 ohms?
    ... >>>he seemed to be basing his calculations on, ... >>>of 'peak' and 'peak to peak'. ... >> half cycles, but the full cycles - it wouldn't make any sense otherwise. ... hey, good stuff John, that had the makings of a right mother of a flamewar, ...
    (alt.guitar.bass)
  • Re: 4 ohms/8 ohms?
    ... >>he seemed to be basing his calculations on, ... >>of 'peak' and 'peak to peak'. ... > half cycles, but the full cycles - it wouldn't make any sense otherwise. ... but NOT for a musician who isn't likely to ...
    (alt.guitar.bass)
  • Re: Cycles on SW Trains
    ... South West Trains is better than that, but most people don't read the rules closely enough to realise. ... "In the area bounded by Hook, Alton, Guildford, Reading and Dorking, cycles may not join or alight from services due to arrive in London between 07:15 and 10:00 inclusive, and departing London between 16:45 and 19:00 inclusive. ... In addition to the restrictions above, cycles are not permitted on morning peak services out of London in the area bounded by Clapham Junction, Feltham and Strawberry Hill, and on services timed to leave Clapham Junction between 07:45 and 09:00 inclusive. ...
    (uk.rec.cycling)
  • Re: 4 ohms/8 ohms?
    ... drive home the point to Nigel that rms values represent the absolute value of a full wave cycle rather than just the positive or negative 180 degrees he seemed to be basing his calculations on, thus my inclusion of 'peak' and 'peak to peak'. ... just half cycles, but the full cycles - it wouldn't make any sense otherwise. ... but NOT for a musician who isn't likely to have electrical knowledge. ...
    (alt.guitar.bass)
  • Re: Just way too nervous the Australians
    ... Always felt SRT didn't have the greatest support when he was at his ... peak. ... Gilly played in a great team and his failures were covered to a ...
    (rec.sport.cricket)
Quantcast