Dear all,
For a while I have been thinking about using the "diode matrix rom" to get "binary to thermometer code" converter. And I got some (lovely) ideas to segment some 2 bit pieces in such a way that 6 bits or so unary segmented DACs might not be humanly impossible anymore. I will share these ideas soon.
But for now the thing is when I look at the specs of some fast switching small signal Schottky diodes (such as BAT54, SD103AW...) the reverse recovery time is almost always given for I_F = 10 mA and R_L = 100 Ohms.
I have reasons to employ some large series resistors. But the difficulty is, smaller I_F would yield faster recovery but larger resistor would slow down the "discharge" of the junction. So here is the question: Pick a diode from above, connect it in series with a 20 - 50 k resistor. And then:
1) When you forward bias this series combination with 5 volts, how quickly would you expect the diode to turn on?
2) When you switch the polarity, what could be a typical reverse recovery time?
No precision is needed here but I would like to get some ballpark figures like 10's of or 100's of ns or some micro seconds... Any ideas??
For a while I have been thinking about using the "diode matrix rom" to get "binary to thermometer code" converter. And I got some (lovely) ideas to segment some 2 bit pieces in such a way that 6 bits or so unary segmented DACs might not be humanly impossible anymore. I will share these ideas soon.
But for now the thing is when I look at the specs of some fast switching small signal Schottky diodes (such as BAT54, SD103AW...) the reverse recovery time is almost always given for I_F = 10 mA and R_L = 100 Ohms.
I have reasons to employ some large series resistors. But the difficulty is, smaller I_F would yield faster recovery but larger resistor would slow down the "discharge" of the junction. So here is the question: Pick a diode from above, connect it in series with a 20 - 50 k resistor. And then:
1) When you forward bias this series combination with 5 volts, how quickly would you expect the diode to turn on?
2) When you switch the polarity, what could be a typical reverse recovery time?
No precision is needed here but I would like to get some ballpark figures like 10's of or 100's of ns or some micro seconds... Any ideas??
Schottky's have no reverse recovery in the normal sense, they are majority carrier devices. They do have capacitance, which has the same effect of delaying recovery, so this is just a duplication of the information in the capacitance graph.
Gold doped fast signal diodes like the 1N4148 might be better as they have very little capacitance.
Gold doped fast signal diodes like the 1N4148 might be better as they have very little capacitance.
Last edited:
The reasonable thing to do is to build a little jig and measure the exact phenomenon you're interested in.
If you prefer SMD Schottky diodes, I suggest you include the LL101A among the candidates you test. Or if you prefer silicon thru-hole diodes, the 1N916 and the 1N4448 and the 1N4149 all have half the capacitance of the junkbox favorite 1N4148. Bob Cordell uses the 1N4149 throughout his power amp textbook.
_
If you prefer SMD Schottky diodes, I suggest you include the LL101A among the candidates you test. Or if you prefer silicon thru-hole diodes, the 1N916 and the 1N4448 and the 1N4149 all have half the capacitance of the junkbox favorite 1N4148. Bob Cordell uses the 1N4149 throughout his power amp textbook.
_
Attachments
Schottky diodes are majority carrier devices, so theoretically that have no recovery current transient. In practice, they have substantial non-linear junction capacitance, the charging of which will look like your usual minority carrier reverse recovery transient. The key here is to choose a Schottky device that is not too large for the job, so that the associated junction capacitance is minimized. Given that the forward current is small, a small signal Schottky device might be appropriate. Keep in mind also that Schottky diodes are leaky, especially at high temperature, so using one at room temperature and assuming that the leakage current is small because of that might result in a nasty surprise.
The forward recovery time for signal diodes should be extremely small, in the ns range.
Older power diodes could display a largish forward recovery time, but that was >40years ago. It was linked to the structure (the size of the "parent" -that's French, I don't know the English equivalent-). Nowadays, practically all diodes have a short forward recovery.
The reverse recovery depends on the amount of charge stored during the forward conduction, and basically it remains invariant if the forward current and the reverse extraction currents are scaled in the same proportion.
This certainly breaks down for very high or very small currents: at low currents, the junction capacitance will be dominant, including for schottky's which have practically no reverse recovery artifacts
Older power diodes could display a largish forward recovery time, but that was >40years ago. It was linked to the structure (the size of the "parent" -that's French, I don't know the English equivalent-). Nowadays, practically all diodes have a short forward recovery.
The reverse recovery depends on the amount of charge stored during the forward conduction, and basically it remains invariant if the forward current and the reverse extraction currents are scaled in the same proportion.
This certainly breaks down for very high or very small currents: at low currents, the junction capacitance will be dominant, including for schottky's which have practically no reverse recovery artifacts
Forward recovery is generally not specified for many diodes, so it is a crap shoot. 1N493X series "normal fast" rectifiers seem to be particular offenders for bad forward recovery, while the GE fast recovery diodes were nice, but expensive, in a glass bead package (too bad GE stopped making semiconductors). We actually went looking for a fast forward recovery diode for commutation service, and some Fairchild normal recovery parts turned out the best - go figure...
I use BAS70 Schottky diodes in a clamp circuit in a 27 Mbit/s level shifter, but those diodes are only supposed to restore the DC level rather than to switch for every bit.
There may be no good reasons to trust LTSPICE "models" of diodes, nor to trust LTSPICE simulations of diode models, but here they are anyway, for your amusement. "Worst case" scenario uses 55Kohm series resistor.
_
Attachments
I would not expect LT Spice to correctly model the dynamic capacitance of a small-signal Schottky diode - best to check it on the bench...
moved to 'Parts'.The BAT54 is used as a RF detector for 2.4 GHz microwave oven leakage detectors, so the switching time has to be << 1ns
Dear all,
The moderator rightfully moved my post to "parts" section. So I hope this message still finds you.
Many thanks for all the comments and simulations! Surely it would be best to measure it on the bench - when I hopefully retire and get time and a scope in a year or so.
For now let me share with you what I have in my mind below:
https://www.diyaudio.com/community/...thermometer-code-decoder.408621/#post-7586375
The moderator rightfully moved my post to "parts" section. So I hope this message still finds you.
Many thanks for all the comments and simulations! Surely it would be best to measure it on the bench - when I hopefully retire and get time and a scope in a year or so.
For now let me share with you what I have in my mind below:
https://www.diyaudio.com/community/...thermometer-code-decoder.408621/#post-7586375