Is it reasonable to expect that the turn-on (not off -- _on_) time of a diode should be pretty much consistent with the inductance of the package, and not much else?
A customer just sent me some O-scope traces of a circuit where a voltage at a catch diode anode is (at least apparently) going several hundred volts above its cathode for a microsecond before settling out to the nominal anode voltage of the diode.
We're not sure if it's a measurement artifact because the coil has some
700A in it, or if it's a real event; I wouldn't be willing to even believe it's a real event except that I know that funny things can happen at high currents.
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Augh. So -- what to do? Are there diodes that are specifically rated for quick turn-on? If not, are there diodes (or diode types) that are noted for turning on rapidly? Can multiple diodes in parallel be used to speed up the turn-on (this seems like all sorts of wrong to me, by the way -- but I feel I have to check).
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
I don'g actually know why that is, but some diodes overshoot a lot. There's more info in one of Jim Williams's app notes,
formatting link
.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Tim Wescott wrote in news:T_OdnWGfG_V6BB7SnZ2dnUVZ snipped-for-privacy@web-ster.com:
(part)cure: never turn the laser completely off, just slightly below or above lasing, should mean a big decrease in capacity. Somebody have a forward voltage versus capacity plot?
Not a PN junction diode. They take a while to conduct, the phenom being called, strangely, "forward recovery."
We did a pulse generator that used power diodes in DSRD (drift step recovery) mode. We applied 48 volts in the forward direction, and waited about 100 ns for the current to ramp up to 50 amps. The rampup looked pretty linear. Then we applied -400 volts to get them to snap.
That can happen. You may need a high-voltage schottky, SiC or GaN or whatever, or a few silicon schottkies in series.
John
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
There is always the issue of current hogging with paralleling diodes. Given your posts in the past, I think you knew that but were in need of a cup of coffee prior to hitting the send button. ;-)
That was a big part of the "all sorts of wrong" that I was referring to. Presumably if one did it one would need current balancing resistors, and one might leave blood on the floor before one got things working.
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
I've never done diode testing myself, but I've seen it done with transmission line tricks, not a driver circuit.
The turn on time is a real pain in the ass in switcher design. Not only for efficiency issues, but the pin on the chip that touches the external catch diode rises like a bat out of hell until the diode begins conducting. That fast rise can lead to C*dv/dt current injection into nearby pins on the chip. So pin layout is critical. Some designs use internal "blanking" periods to avoid making any "decisions" by the internet circuitry while it is expected that the diode hasn't turned on yet.
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
Trr is the first thing to come to mind. Rule of thumb is the higher the reverse voltage the slower the diode is. The SIC diodes are very fast in any voltage.
Not really worthwhile for small apps, but big switchers can benefit greatly from a dV/dt snubber. A dV/dt snubber, of course, works even with junction diodes, because although the diode won't turn on instantly, it has hundreds of volts to go before it's crashing into rails or anything, so it's forced on, *very* quickly. There's a Unitrode appnote, I believe, which plots transistor, snubber and total losses vs. snubber size -- there's a minima where, when the snubber just about doubles the fall time, total losses are about 10% lower than the unsnubbed transistor alone, despite wasting power in the snubber resistor itself.
More snubbing reduces transistor switching losses further, but dramatically increases snubber losses. If you can arrange your circuit to use lossless snubbers, you can achieve even better gains of course.
Even just a wad of capacitance can help (without fancy R's or D's) -- I recently did a discrete (external switch) buck converter, 24V supply, which exhibited 5V undershoot on the switching node (and this with a schottky diode) -- just putting a dumb 1nF across the diode was enough to flatten out the C vs. V curve, swamping it to a nice smooth linear falling edge and dropping the overshoot down to 2V. The supply rail is dI/dt snubbed (L || R, just a few nH really) so it doesn't impact switching losses much -- when the MOSFET switches on, the supply dips by about half, absorbing the energy required to charge both snubber capacitance and diode junction capacitance.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
I once build an inverter prototype which (due to insufficient theory at the time) resonated quite excellently. It rang around 15MHz, generating some
80% overshoot. At 350VDC supply, that's over 600V peak, enough to toast a poor 600V MOSFET! The equivalent circuit was roughly 18nH (the inductance between power planes and MOSFET terminals) and 6nF (junction + snubber capacitance), so the peak current was quite large (~160A) relative to ratings (70A design load current).
In order to get those boards to work, I tried squashing the overshoot with diodes to the opposing rail. This hardly reduced the overshoot by half: by all accounts, a full 100V spike was appearing across the diode during this event. Even taking off ~10nH lead and bond inductance, that still leaves ~50V across the die, real voltage applied to silicon.
Though the average current was small, the peak current was still enough to toast 6A diodes, even 12A diodes -- finally, 30A diodes survived the abuse. What kind of failure mechanism do diodes succumb to under peak conditions -- is it an electromigration thing, perhaps? The failure is gradual and not obviously thermal: the circuit can run happy for minutes with no significant temp rise (~10C), then the diodes suddenly go shorted.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
"Tim Wescott" wrote in message
news:T_OdnWafG_WVER7SnZ2dnUVZ_gqdnZ2d@web-ster.com...
> Is it reasonable to expect that the turn-on (not off -- _on_) time of a
> diode should be pretty much consistent with the inductance of the
> package, and not much else?
>
> A customer just sent me some O-scope traces of a circuit where a voltage
> at a catch diode anode is (at least apparently) going several hundred
> volts above its cathode for a microsecond before settling out to the
> nominal anode voltage of the diode.
>
> We're not sure if it's a measurement artifact because the coil has some
> 700A in it, or if it's a real event; I wouldn't be willing to even
> believe it's a real event except that I know that funny things can happen
> at high currents.
>
> --
> My liberal friends think I'm a conservative kook.
> My conservative friends think I'm a liberal kook.
> Why am I not happy that they have found common ground?
>
> Tim Wescott, Communications, Control, Circuits & Software
> http://www.wescottdesign.com
It depends on the device. High voltage diodes have a forward recovery time (Tfr and Vfr) that results in voltage peaks that cannot be explained by package inductance.
I would think of tens of volts, but I don't know what happens with
700A and very high dI/dt.
--
Wim
PA3DJS
www.tetech.nl
Please remove abc first in case of PM
On Semi Rectifier Applications Handbook (HB-214/D)
formatting link
Handbook.pdf pg. 26
"In general, Schottky technology offers switching performance because Schottky is a majority carrier technology.
In fact, switching both forward and reverse directions are major drawbacks of p+n n+ technology.
For example, when a p+n+n+ rectifier is switched from reverse blocking to on=96conduction as shown in Figure 23, its forward voltage drop exceeds its steady=96state forward voltage drop. This phenomenon is called forward voltage overshoot during the turn=96on transport. This is due to the fact that during Ugh=96speed switching from the off=96state to the on=96state, current through the rectifier is limited by the maximum rate at which minority carriers are injected into the junction. A high voltage drop therefore develops across the diode for a short period of time until minority carriers diffuse into the junction and reduce the drift region resistance.
This process was explained in an earlier section as conductivity modulation. The time it takes for the diode to recover to within 10% of its steady=96state forward conduction voltage is called the forward recovery time."
--
Or, your equipment just wasn't showing you the real signal, like a
digital scope for example.
You would be surprised to see what is there using a spectrum analyzer.
Jamie
Unfortunately we only had 100MHz scopes, a travesty I know, but I looked at the noise spectrum with a >>100MHz spec and didn't see any substantial harmonics over 60MHz, and not just on the inverter, but on any of our boards actually; this includes monolithic switcher chips and the FPGA, but that really just says what everyone already knows -- or should know: ground planes keep stray fields down. With a vengance.
It's true that a given measurement could be pretty crappy. I get that a lot when breadboarding: a switching node might fire common-mode noise into whatever the board is resting on; as a result, squigglies show up in every other measurement you make, even unrelated low-level circuits. But this isn't the case.
Two supporting factors: one, my Tek 475 wouldn't have seen anything more (as confirmed by the spec results -- the band ends at ~60MHz), and might turn out to have a harder time visualizing the details, due to vertical saturation and low duty cycle as you zoom in. (I'm not afraid of DSOs, I'm familiar with their operation and limitations.) Two, the signal literally appears everywhere on the board in this layout: it appears on the switching node (or Vds of both transistors); across the supply (high side drain to low side source); and, with an inductive loop probe, it appears (in phase or otherwise) around every opening and terminal connection. Because of this, it can be very difficult to troubleshoot -- it isn't really associated with any component, because it's a whole-board property.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.