Re: Question about schottky
I have been told to use a saftey margin like 4-6 times Vac but I would also like to see some good guidelines for dimensioning, both on voltage and current ratings for individual diodes in a bridge.
/
lumanauw said:
I have been told to use a saftey margin like 4-6 times Vac but I would also like to see some good guidelines for dimensioning, both on voltage and current ratings for individual diodes in a bridge.
/
i doubt, it was also powered off and on again within few second for hundreds or thousand times for testing firmware versions during development:
http://home.tu-clausthal.de/~tpa/relais/index.html
This weekend I tried using an AM radio as a detector of
the broadband noise that a rectifier generates. I could
not reliably determine what was rectifier noise from the
background racket on my initial attempt.
I'm going to try again with a Topaz Ultra-Isolation
transformer at the end of a long(!) extension cord
in my back yard. Also, I'm going to see if running an
amplifier with a swept signal will modulate the noise
and make it more identifiable amidst the racket of
modern civilization.
A rf spectrum analzyer is what we really need, but I
don't have access to one. I do have a HP 100 MHz
'scope, and might even attempt to rig up some kind
of shielded enclosure. It'll be a while before I've got
something set up, if I attempt it at all.
I'll be trying convention, Schottky, and HexFRED rectifiers,
and as I can get them, some of the newer soft recovery
types from Fairchild and On-Semi. I also plan to use
snubbers, both RC and small values of series R for low
current supplies. If I have to, I'll try to obtain some
RF absorbent material and see if that'll sponge up noise,
but it'll have to wait until I can afford some.
I >can< see some transformer ringing caused by
rectifier switching; a tuned RC snubber on the
secondary will easily damp it, and likely reduces the
RF noise as well. I just can't yet measure the latter.
Measure, measure, measure, I say; and then listen.
I hate doing things blind.
the broadband noise that a rectifier generates. I could
not reliably determine what was rectifier noise from the
background racket on my initial attempt.
I'm going to try again with a Topaz Ultra-Isolation
transformer at the end of a long(!) extension cord
in my back yard. Also, I'm going to see if running an
amplifier with a swept signal will modulate the noise
and make it more identifiable amidst the racket of
modern civilization.
A rf spectrum analzyer is what we really need, but I
don't have access to one. I do have a HP 100 MHz
'scope, and might even attempt to rig up some kind
of shielded enclosure. It'll be a while before I've got
something set up, if I attempt it at all.
I'll be trying convention, Schottky, and HexFRED rectifiers,
and as I can get them, some of the newer soft recovery
types from Fairchild and On-Semi. I also plan to use
snubbers, both RC and small values of series R for low
current supplies. If I have to, I'll try to obtain some
RF absorbent material and see if that'll sponge up noise,
but it'll have to wait until I can afford some.
I >can< see some transformer ringing caused by
rectifier switching; a tuned RC snubber on the
secondary will easily damp it, and likely reduces the
RF noise as well. I just can't yet measure the latter.
Measure, measure, measure, I say; and then listen.
I hate doing things blind.
Getting a night sleep after the experiments makes me wonder
if I measured the right thing. John certainly has a point when
he says we should measure the current, since it is the reverse
recovery current of the diodes that supposedly causes
the problem. However, apart from direct radiation from the
rectifying circuitry into the equipment powered by it, it is
after all the supply voltage that matters, which is
why I choose to measure the load voltage rather than the
diode or charge current. Another objection might be how
much the HP filter affects the readings I get. It is after all
a derivation circuit, and maybe the undershoot pulse I see
in the voltage is still too wide to be something that has passed
reasonably unaltered throught the filter and it is rather some
artifact due to the derivation character of the filter? I did one
follow upp measurement and the pulse roughly had the
values ta = 325-350us and tb = 75 - 100us, but already here
it is becoming difficult to measure reliably with the equipment
I have. (ta is the width of the negative slope of the pulse,
measured from the zero crossing to its negative peak and
tb is the width of the positive (recovery) slope):
I would also like to point out once more that the main point of
my exercise was to see if I could measure something at all
without a lot of fancy lab equipment that I don't have access
to. Given what John has said on earlier occassion I might
perhaps not even have tried it unless till had managed to
capture ringing on his scope. I also agree with Damon, I too
hate to do things blind. There is, of course, the Hagerman
article on calculating snubbers, but it is still nice if one can
also measure the effect.
Till,
I basically agree with your comment on the types of answers
one gets, but you summed it up nicely. It is not quite that bad
though, one often gets useful answers too. I think all who
commented on my experiment here had something useful to say.
if I measured the right thing. John certainly has a point when
he says we should measure the current, since it is the reverse
recovery current of the diodes that supposedly causes
the problem. However, apart from direct radiation from the
rectifying circuitry into the equipment powered by it, it is
after all the supply voltage that matters, which is
why I choose to measure the load voltage rather than the
diode or charge current. Another objection might be how
much the HP filter affects the readings I get. It is after all
a derivation circuit, and maybe the undershoot pulse I see
in the voltage is still too wide to be something that has passed
reasonably unaltered throught the filter and it is rather some
artifact due to the derivation character of the filter? I did one
follow upp measurement and the pulse roughly had the
values ta = 325-350us and tb = 75 - 100us, but already here
it is becoming difficult to measure reliably with the equipment
I have. (ta is the width of the negative slope of the pulse,
measured from the zero crossing to its negative peak and
tb is the width of the positive (recovery) slope):
I would also like to point out once more that the main point of
my exercise was to see if I could measure something at all
without a lot of fancy lab equipment that I don't have access
to. Given what John has said on earlier occassion I might
perhaps not even have tried it unless till had managed to
capture ringing on his scope. I also agree with Damon, I too
hate to do things blind. There is, of course, the Hagerman
article on calculating snubbers, but it is still nice if one can
also measure the effect.
Till,
I basically agree with your comment on the types of answers
one gets, but you summed it up nicely. It is not quite that bad
though, one often gets useful answers too. I think all who
commented on my experiment here had something useful to say.
Houston we have a problem!!
Given that I couldn't measure any ringing, I've concentrated on
the undershoot in the load voltage as measured through the
HP filter. I have tried four different types of diodes and
the results make me wonder if the measurements make
any sense at all, since the differences are so small.
Anyway, this is what I got (all figures very approximate):
ampl.________ta________ tb________diode type
-----------------------------------------------------------------------------
-2.3mV___325-350us___75-100us___KPA204G (bridge)
-1.1mV_______325us___75-100us___BYW29 (ultra fast)
-2.0mV_______350us___75-100us___10ETF02 (ultra fast soft rec.)
-2.6mV_______350us_______50us___MBR745 (Schottky)
These results makes me question the whole experiment.
Maybe I am rather measuring artifacts from the HP filter, which
is after all a derivation circuit?
Given that I couldn't measure any ringing, I've concentrated on
the undershoot in the load voltage as measured through the
HP filter. I have tried four different types of diodes and
the results make me wonder if the measurements make
any sense at all, since the differences are so small.
Anyway, this is what I got (all figures very approximate):
ampl.________ta________ tb________diode type
-----------------------------------------------------------------------------
-2.3mV___325-350us___75-100us___KPA204G (bridge)
-1.1mV_______325us___75-100us___BYW29 (ultra fast)
-2.0mV_______350us___75-100us___10ETF02 (ultra fast soft rec.)
-2.6mV_______350us_______50us___MBR745 (Schottky)
These results makes me question the whole experiment.
Maybe I am rather measuring artifacts from the HP filter, which
is after all a derivation circuit?
what stands pwm for? in german it is pulsweitenmodulation what would be pulse wideness modulation or so ?
We use "width".
I have an e-mail......(and can't share it.....sinbin threat) from a place in the EU in the PWM business. They claim it is analog. Good enough for me.
I know that there are some companies that make a PCM front-end, as opposed to good ol' analog input. I believe that Mueta does not advise PCM inputs.......check their website.
Jocko
I have an e-mail......(and can't share it.....sinbin threat) from a place in the EU in the PWM business. They claim it is analog. Good enough for me.
I know that there are some companies that make a PCM front-end, as opposed to good ol' analog input. I believe that Mueta does not advise PCM inputs.......check their website.
Jocko
Nelson Pass said:
made my day ...
So - anyone willing to state "the" 3 basic rules for diod demensioning ?
/
TNT said:
Do you mean current rating of bridge diodes? Are there any
particular rules you have seen that you are referring to?
Otherwise, this is a topic that has been discussed a couple
of times, but I can't remember it leading any further than
speculations and guesses. Usually you have an average
current rating and a surge current rating, and usually also
a surge current derating depending on the number of pulses.
Note however that the surge current often derates to the
average current when the number of pulses increase.
From this it is not clear, to me at least, how to properly
choose diodes. A straightforward approach is to say that
if we assume a constant load, the charge of the capacitors
will only change with the ripple and thus stay constant on
the average. That means that average charge current
equals average load current. Hence, choosing diodes with
the same rating as the load current should be sufficient.
It is not that simple however. The charge current occur in
peaks and are much higher during these peaks than the
load current is. Hence, even if you can (almost) calculate
the average power dissipation through the diodes using
the load current, you do get a higher instantaneous power
dissipation during the pulses, which will probably be long
enough to force you to derate the diodes. The higher charge
current also increases the voltage drop slightly which also
increases the power dissipation (which is why I wrote almost
above). I have no golden rule for how to do this, but wish
I had. Good old overdimensioning seems the way to go. To
be on the safe side, choose diodes that can handle the peak
charge currents as average current, although that is most
certainly to overdo it. Some diodes, like the ultra fast soft
recovery ones Elfa is now selling have a diagram for how much
peak current they can take depending on the conduction angle,
which should make a reasonable calculation possible in this case.
Hi,
You should choose the diodes so they can at least withstand 3 times the supply RMS voltage so the diode can withstand the peak voltages.
For example:
Diode Maximum Current Maximum Reverse Voltage
1N4001 1A 50V
1N4002 1A 100V
1N4007 1A 1000V
1N5401 3A 100V
1N5408 3A 1000V
Cheers,😉
You should choose the diodes so they can at least withstand 3 times the supply RMS voltage so the diode can withstand the peak voltages.
For example:
Diode Maximum Current Maximum Reverse Voltage
1N4001 1A 50V
1N4002 1A 100V
1N4007 1A 1000V
1N5401 3A 100V
1N5408 3A 1000V
Cheers,😉
Are the MUR860's a soft-recovery type?
I'm under the impression that they're fast-recovery but not soft-recovery, yet they seem to be the diode of choice for building chip amps at the moment.
If so, how does their preference in sonic terms correlate with the presence of hard reverse-voltage spikes?
TIA,
KT
I'm under the impression that they're fast-recovery but not soft-recovery, yet they seem to be the diode of choice for building chip amps at the moment.
If so, how does their preference in sonic terms correlate with the presence of hard reverse-voltage spikes?
TIA,
KT
Sorry,
Just looked up the MUR860 datasheet.
They ARE specified as "Ultrafast with Soft Recovery"
Thanks,
KT
Just looked up the MUR860 datasheet.
They ARE specified as "Ultrafast with Soft Recovery"
Thanks,
KT
I have a batch of those rectifiers and they are miserable. They have a pronounced snap-off characteristic which will ruin your sound. Stay away!
Christer !
Thanks for the effort - I would not have asked if I know and due to lack of knowledge I can not add any info.
I'm shure the inhabitants of this forum could come up with some rules. And it would save bandwith (and diodes) in the long run.
/
Thanks for the effort - I would not have asked if I know and due to lack of knowledge I can not add any info.
I'm shure the inhabitants of this forum could come up with some rules. And it would save bandwith (and diodes) in the long run.
/
RFI: Via rails or via air?
I don't think my measurements made any of us much wiser, but
at least it raised a question that is still hanging in the air: is
the presumed interference effect du to RFI on the supply rails
or due to RFI emitted from the PSU and radiated into the powered
equipment? In the first case, there must be measurable RFI
on the supply voltage (my measurements do not disprove that).
In the other case, it is not necessarily so, it is just sufficient that
there is RFI on the charge currents between transformer and
capacitors. John seems to hint at the latter by advising to focus
on measuring currents. I also found in my earlier Spice
experiments that it seemed easier to get ringing on the charge
currents than on the supply voltage. Both were possible, but
I had to "provoke" the circuit with a much higher winding
inductance to get any mentionable effect on the supply voltage.
Are there any opinions on this? It seems to be a question just
as important as the choice of diodes. If it is the question of
radiated RFI, then it would make a lot of sense to have a
separate enclosure for the PSU, for instance.
I don't think my measurements made any of us much wiser, but
at least it raised a question that is still hanging in the air: is
the presumed interference effect du to RFI on the supply rails
or due to RFI emitted from the PSU and radiated into the powered
equipment? In the first case, there must be measurable RFI
on the supply voltage (my measurements do not disprove that).
In the other case, it is not necessarily so, it is just sufficient that
there is RFI on the charge currents between transformer and
capacitors. John seems to hint at the latter by advising to focus
on measuring currents. I also found in my earlier Spice
experiments that it seemed easier to get ringing on the charge
currents than on the supply voltage. Both were possible, but
I had to "provoke" the circuit with a much higher winding
inductance to get any mentionable effect on the supply voltage.
Are there any opinions on this? It seems to be a question just
as important as the choice of diodes. If it is the question of
radiated RFI, then it would make a lot of sense to have a
separate enclosure for the PSU, for instance.
I have a probe that consists of a resistor and several turns of very fine wire wrapped in a coil. When I wave it around inside a piece of equipment, I can find where RF signals are being radiated into the ether. I think, if I can pick up such a signal with my probe, surely other parts of the instrument can also pick it up. Therefore I say, anything that can be proven to be radiating strongly should just be fixed. If it can't be fixed, then it should definitely be sheilded from the rest of the equipment.
- Status
- Not open for further replies.