diodes in series/parallel
Signification, Wed May 30 2018, 11:08AMA while back, in a coilgun design, I needed a high voltage/current diode. I heard that single diodes can be wired in series/parallel configurations. I also heard this is very bad--unless you had well matched diodes, preferably from the same batch/manufacturer. I saw an application where a professional engineer put over 100 1N4007's in series/parallel to get the required voltage/current. Something about that wild pattern just bothered me.
Then I had an idea: What better a match are the diodes inside a bridge rectifier? The questions...what if you took a hefty full wave bridge rectifier and just used the "+" and "-" pins as the output, treating it as a single diode? This would wire the four diodes in the package in a 2x2 configuration, theoretically giving twice the voltage AND current rating of each diode. Also, would it be best to jumper the two "AC" pins? Of course "+" and "-" labeled pins would have to be reversed in this case.
One more thing, this method should also apply for a three phase rectifier giving a similar 3x3 property with the same use of the + and - pins....don't know what the best thing would be to do with the three AC pins...except NOTHING. Is anyone familiar with this?
Re: diodes in series/parallel
Dr. Slack, Wed May 30 2018, 01:19PM
The reason we try to avoid putting diodes in series to increase the PIV is that the diode charge storage delays the off time. Imbalances between diodes mean they turn off and start blocking at different times, and the first diode that tries to block will tend to get fried. The workaround for this is to use twice as many diodes as are needed for the simple PIV calculation. The first few times they get full reverse voltage, the fastest few will die. Fortunately, silicon junction diodes fail (almost) invariably short-circuit, leaving the self-selected better-matched survivors as a working diode stack. Obviously a string of 2 diodes does not give any advantage over a single diode, when trying to use this rule.
As a alternative to 2x overdesign, we can bridge each diode with a fixed capacitor, large enough to swamp the charge storage, and equalise the rate of voltage rise. Obviously this requires extra components, and makes the diodes less efficient
The reason we try to avoid putting diodes in parallel to increase current handling is they they don't share current nicely, their voltage tempco is the wrong sign. Mounting them on the same heatsink helps, as does under-running them heavily, though this latter tends to negate the point of using several. A small resistor in series with each can enforce current sharing, though this requires extra components, and makes the diodes less efficient.
Dr. Slack, Wed May 30 2018, 01:19PM
The reason we try to avoid putting diodes in series to increase the PIV is that the diode charge storage delays the off time. Imbalances between diodes mean they turn off and start blocking at different times, and the first diode that tries to block will tend to get fried. The workaround for this is to use twice as many diodes as are needed for the simple PIV calculation. The first few times they get full reverse voltage, the fastest few will die. Fortunately, silicon junction diodes fail (almost) invariably short-circuit, leaving the self-selected better-matched survivors as a working diode stack. Obviously a string of 2 diodes does not give any advantage over a single diode, when trying to use this rule.
As a alternative to 2x overdesign, we can bridge each diode with a fixed capacitor, large enough to swamp the charge storage, and equalise the rate of voltage rise. Obviously this requires extra components, and makes the diodes less efficient
The reason we try to avoid putting diodes in parallel to increase current handling is they they don't share current nicely, their voltage tempco is the wrong sign. Mounting them on the same heatsink helps, as does under-running them heavily, though this latter tends to negate the point of using several. A small resistor in series with each can enforce current sharing, though this requires extra components, and makes the diodes less efficient.
Re: diodes in series/parallel
DerAlbi, Fri Jun 01 2018, 03:16PM
Welcome back ^_^
I would like to add that diodes dont immediately die if they break through.. it really depends on the current and thus power/energy dissipation during breakthrough. As with mosfet body diodes normal PN-diodes can have an avalanche energy rating, may it be official or not; in fact there is always some reverse current (=leakage) in blocking operation - so this alone is never a reason for a fault.
This implies that seeing string of 100 diodes is on its own neither good nor bad or interesting or weird. It really depends on the surrounding circuit - i would call it strange in a flyback, but maybe OK for a 50Hz rectifier. (because the reverse current is dominated by capacitance as Dr.Slack said, and thus the lower the frequency the better the odds for the diodes to live; Of course relying on "odds" is always a sign of bad engineering.. )
DerAlbi, Fri Jun 01 2018, 03:16PM
Welcome back ^_^
I would like to add that diodes dont immediately die if they break through.. it really depends on the current and thus power/energy dissipation during breakthrough. As with mosfet body diodes normal PN-diodes can have an avalanche energy rating, may it be official or not; in fact there is always some reverse current (=leakage) in blocking operation - so this alone is never a reason for a fault.
This implies that seeing string of 100 diodes is on its own neither good nor bad or interesting or weird. It really depends on the surrounding circuit - i would call it strange in a flyback, but maybe OK for a 50Hz rectifier. (because the reverse current is dominated by capacitance as Dr.Slack said, and thus the lower the frequency the better the odds for the diodes to live; Of course relying on "odds" is always a sign of bad engineering.. )
Re: diodes in series/parallel
Patrick, Sun Jun 03 2018, 02:18AM
I was thinking instead of capacitors i was told to use conventional resistors in parallel to divide the voltage evenly, since high speed impulses can kill the caps and even if they dont, the capactive reactance drops with increasing speed, though the latter may be advantageous.
a conventional 1N4007 or HER108 diode (both 1kv) and a normal resistor (500v) pair would probly be fine, as we dont make consumer electronics for sale here on the forum. im thinking 250k to 500k ohms, but not sure about excessive loading or non- rectified bleed through.
Patrick, Sun Jun 03 2018, 02:18AM
I was thinking instead of capacitors i was told to use conventional resistors in parallel to divide the voltage evenly, since high speed impulses can kill the caps and even if they dont, the capactive reactance drops with increasing speed, though the latter may be advantageous.
a conventional 1N4007 or HER108 diode (both 1kv) and a normal resistor (500v) pair would probly be fine, as we dont make consumer electronics for sale here on the forum. im thinking 250k to 500k ohms, but not sure about excessive loading or non- rectified bleed through.
Re: diodes in series/parallel
Signification, Sun Jun 03 2018, 03:52AM
HI GUYS! It's great to be back with y'all...
Thanks for all the help--fell free to keep it coming!
My home recently flooded and of course, you know where I kept most of my huge electronic / electrical 'stuff'...LOW! I attempted to dig out the sources of my first two inquires: The many series/parallel configuration of the 100+ array of 1N4007 diodes, and the bridge rectifier example using it as a single 2x2 diode. The latter was out of an industrial -dual magnatron microwave oven- It was made by a reputable manufacturer (something like GE). It was from a restaurant, ran on 120VAC! and in additional to using the "+"and "-" terminals as the output, had the two AC pins jumpered together together with a heavy gauge crimped jumper. I wonder...in a bridge rectifier, are there actually four diodes inside, or are they integrated in a single die? I may bust one (this one) open. I have seen many series / parallel components (especially capacitor arrays) connected with and without this -center- connection added--I assume it is related to failure modes and different types of protection. The former (...huge diode array...) was the work of that rip-off guy from "Information unlimited" I am sure you all know of whom I speak. It was part of a power supply from that old Hughes Q-Switched ruby laser rangefinder head. I had a couple of these (one complete--including the original PFN and colliminating telescope!) I finally got around to designing a power supply that worked and was tunable. The mechanical rotary Q-Switch was exciting, as it contained a prism on the platform mounted on the shaft of a custom motor that had a rotation rate of 36,000 (not 3600 RPM!). That's 600 rev's per second! When fired at a piece of copper, it would leave a small crater in a wiff of smoke. A great fun project! I tried to get some photos of the crimson beam in a smoke-filled area, but never could catch the thing.
I have a ruby rod, 6-1/8" x 3/8" (which was above water) with polished, coated, and mirrored ends, a matching flashlamp, and a custom solid aluminum elliptical reflector fitted for the rod and lamp. I am seriously hoping that the moisture didn't effect the fragile optics. A shrunken quarter will be among the targets!...maybe some discussions will be generated for a different category.
Signification, Sun Jun 03 2018, 03:52AM
HI GUYS! It's great to be back with y'all...
Thanks for all the help--fell free to keep it coming!
My home recently flooded and of course, you know where I kept most of my huge electronic / electrical 'stuff'...LOW! I attempted to dig out the sources of my first two inquires: The many series/parallel configuration of the 100+ array of 1N4007 diodes, and the bridge rectifier example using it as a single 2x2 diode. The latter was out of an industrial -dual magnatron microwave oven- It was made by a reputable manufacturer (something like GE). It was from a restaurant, ran on 120VAC! and in additional to using the "+"and "-" terminals as the output, had the two AC pins jumpered together together with a heavy gauge crimped jumper. I wonder...in a bridge rectifier, are there actually four diodes inside, or are they integrated in a single die? I may bust one (this one) open. I have seen many series / parallel components (especially capacitor arrays) connected with and without this -center- connection added--I assume it is related to failure modes and different types of protection. The former (...huge diode array...) was the work of that rip-off guy from "Information unlimited" I am sure you all know of whom I speak. It was part of a power supply from that old Hughes Q-Switched ruby laser rangefinder head. I had a couple of these (one complete--including the original PFN and colliminating telescope!) I finally got around to designing a power supply that worked and was tunable. The mechanical rotary Q-Switch was exciting, as it contained a prism on the platform mounted on the shaft of a custom motor that had a rotation rate of 36,000 (not 3600 RPM!). That's 600 rev's per second! When fired at a piece of copper, it would leave a small crater in a wiff of smoke. A great fun project! I tried to get some photos of the crimson beam in a smoke-filled area, but never could catch the thing.
I have a ruby rod, 6-1/8" x 3/8" (which was above water) with polished, coated, and mirrored ends, a matching flashlamp, and a custom solid aluminum elliptical reflector fitted for the rod and lamp. I am seriously hoping that the moisture didn't effect the fragile optics. A shrunken quarter will be among the targets!...maybe some discussions will be generated for a different category.
Re: diodes in series/parallel
Patrick, Sun Jun 03 2018, 06:34PM
yes ive been laid-off 3 times in 18 months. Now it looks like Ive got a job building 320 pound drones and a 23 million $ series of prototype drones for the USAF.
Patrick, Sun Jun 03 2018, 06:34PM
yes ive been laid-off 3 times in 18 months. Now it looks like Ive got a job building 320 pound drones and a 23 million $ series of prototype drones for the USAF.
Re: diodes in series/parallel
Plasma, Mon Jun 04 2018, 12:23AM
Patrick is replying to the second to last reply
the first six months don't comment on anyone work, weather it is off topic or not. The theory you gained is nothing compared to some one that has worked ten years in the narrow field, mainly more so if they have been in the company awhile. What some one tells you to do do it it's coming from the boss that pay you wage, indirect, you are trading your time and skills for money so make your time valuable, never say you think someone is useless, if they have been working their awhile, you're the one that is going to look bad.
Off topic I hope you have the mature nature to learn than retaliate
Plasma, Mon Jun 04 2018, 12:23AM
Patrick is replying to the second to last reply
the first six months don't comment on anyone work, weather it is off topic or not. The theory you gained is nothing compared to some one that has worked ten years in the narrow field, mainly more so if they have been in the company awhile. What some one tells you to do do it it's coming from the boss that pay you wage, indirect, you are trading your time and skills for money so make your time valuable, never say you think someone is useless, if they have been working their awhile, you're the one that is going to look bad.
Off topic I hope you have the mature nature to learn than retaliate
Re: diodes in series/parallel
Signification, Mon Jun 04 2018, 02:44AM
...agreed...and, no SOUL is useless!!
I dug up this schema last night:
Signification, Mon Jun 04 2018, 02:44AM
...agreed...and, no SOUL is useless!!
I dug up this schema last night:
Re: diodes in series/parallel
DerAlbi, Mon Jun 04 2018, 12:23PM
This schematic makes no sense to me.
There are 2 circuits, sharing the same ground.
Both circuits deliver rectified power.
The upper one uses resonance to create AC - when the SCR fires, the transformer oscillates with the capacitors. The voltage is transformed up and output on the BNC.
Then there is the lower circuit. It just rectifies high voltage and puts it also on the BNC. The one with the higher voltage dominates the output voltage.
It seems like the lower portion can deliver DC while the upper circuit is some kind of firing circuit like in a gas discharge lamp. I however dont get, why one strand of diodes needs a different voltage rating than the other.
Anyway, if the large diode-strand is reverse stressed it only happens due to the resonance from the upper circuit. Since this should follow a agreeable sine wave its like i said in my post.. in such a case the diodes will live since the capacitive current will be limited and breakthrough is fine.
I dont really agree with the decision to connect the individual diodes in parallel then put the parallel diodes in series. I would have put the series strands in parallel - which should improve current sharing. Not a good circuit.
..sry to be on topic.. (not sure why this was worth mentioning, getting a job is only interesting if you can keep it, which seems to be the bigger problem - if you solve THIS issue, then we throw a party )
DerAlbi, Mon Jun 04 2018, 12:23PM
This schematic makes no sense to me.
There are 2 circuits, sharing the same ground.
Both circuits deliver rectified power.
The upper one uses resonance to create AC - when the SCR fires, the transformer oscillates with the capacitors. The voltage is transformed up and output on the BNC.
Then there is the lower circuit. It just rectifies high voltage and puts it also on the BNC. The one with the higher voltage dominates the output voltage.
It seems like the lower portion can deliver DC while the upper circuit is some kind of firing circuit like in a gas discharge lamp. I however dont get, why one strand of diodes needs a different voltage rating than the other.
Anyway, if the large diode-strand is reverse stressed it only happens due to the resonance from the upper circuit. Since this should follow a agreeable sine wave its like i said in my post.. in such a case the diodes will live since the capacitive current will be limited and breakthrough is fine.
I dont really agree with the decision to connect the individual diodes in parallel then put the parallel diodes in series. I would have put the series strands in parallel - which should improve current sharing. Not a good circuit.
..sry to be on topic.. (not sure why this was worth mentioning, getting a job is only interesting if you can keep it, which seems to be the bigger problem - if you solve THIS issue, then we throw a party )
Re: diodes in series/parallel
hen918, Mon Jun 04 2018, 06:19PM
I think the hope is, there will be only a positive output pulse from T, so it will only have to block the 1200 V from the really nasty power side, the HV for triggering will be conducted by the diodes.
hen918, Mon Jun 04 2018, 06:19PM
DerAlbi wrote ...
This schematic makes no sense to me.
There are 2 circuits, sharing the same ground.
Both circuits deliver rectified power.
The upper one uses resonance to create AC - when the SCR fires, the transformer oscillates with the capacitors. The voltage is transformed up and output on the BNC.
Then there is the lower circuit. It just rectifies high voltage and puts it also on the BNC. The one with the higher voltage dominates the output voltage.
It seems like the lower portion can deliver DC while the upper circuit is some kind of firing circuit like in a gas discharge lamp. I however dont get, why one strand of diodes needs a different voltage rating than the other.
Anyway, if the large diode-strand is reverse stressed it only happens due to the resonance from the upper circuit. Since this should follow a agreeable sine wave its like i said in my post.. in such a case the diodes will live since the capacitive current will be limited and breakthrough is fine.
I dont really agree with the decision to connect the individual diodes in parallel then put the parallel diodes in series. I would have put the series strands in parallel - which should improve current sharing. Not a good circuit.
..sry to be on topic.. (not sure why this was worth mentioning, getting a job is only interesting if you can keep it, which seems to be the bigger problem - if you solve THIS issue, then we throw a party )
This schematic makes no sense to me.
There are 2 circuits, sharing the same ground.
Both circuits deliver rectified power.
The upper one uses resonance to create AC - when the SCR fires, the transformer oscillates with the capacitors. The voltage is transformed up and output on the BNC.
Then there is the lower circuit. It just rectifies high voltage and puts it also on the BNC. The one with the higher voltage dominates the output voltage.
It seems like the lower portion can deliver DC while the upper circuit is some kind of firing circuit like in a gas discharge lamp. I however dont get, why one strand of diodes needs a different voltage rating than the other.
Anyway, if the large diode-strand is reverse stressed it only happens due to the resonance from the upper circuit. Since this should follow a agreeable sine wave its like i said in my post.. in such a case the diodes will live since the capacitive current will be limited and breakthrough is fine.
I dont really agree with the decision to connect the individual diodes in parallel then put the parallel diodes in series. I would have put the series strands in parallel - which should improve current sharing. Not a good circuit.
..sry to be on topic.. (not sure why this was worth mentioning, getting a job is only interesting if you can keep it, which seems to be the bigger problem - if you solve THIS issue, then we throw a party )
I think the hope is, there will be only a positive output pulse from T, so it will only have to block the 1200 V from the really nasty power side, the HV for triggering will be conducted by the diodes.
Re: diodes in series/parallel
DerAlbi, Tue Jun 05 2018, 05:52AM
i disagree. Look at the bottom of the the transformer T. Its secondary is grounded @ J2. So there cannot be any higher voltage than the 3x D3 allow making all the diodes D2 unnecessary.. The only reason for the diodes D2 must be then that the output @ L2 may swing extremely low which i dont see happening.
My conclusion: the guy who made the circuit didnt know what he was doing or didnt draw the actual circuit. Discussing the content of the schematic is therefore leading nowhere. Just my opinion - based on not understanding the circuit.
Regarding the original question of paralleling series strands of diodes; prefer to parallel whole strands.. dont parallel individual diodes and put the "simulated bigger diodes" in series. This will flatten the V-I-curves and add all parasitic junction resistances instead of minimizing them which leads to better current sharing of the strands. At least this is my theory.
DerAlbi, Tue Jun 05 2018, 05:52AM
i disagree. Look at the bottom of the the transformer T. Its secondary is grounded @ J2. So there cannot be any higher voltage than the 3x D3 allow making all the diodes D2 unnecessary.. The only reason for the diodes D2 must be then that the output @ L2 may swing extremely low which i dont see happening.
My conclusion: the guy who made the circuit didnt know what he was doing or didnt draw the actual circuit. Discussing the content of the schematic is therefore leading nowhere. Just my opinion - based on not understanding the circuit.
Regarding the original question of paralleling series strands of diodes; prefer to parallel whole strands.. dont parallel individual diodes and put the "simulated bigger diodes" in series. This will flatten the V-I-curves and add all parasitic junction resistances instead of minimizing them which leads to better current sharing of the strands. At least this is my theory.
Re: diodes in series/parallel
Signification, Tue Jun 05 2018, 10:22AM
From way back; here is what I recall of the original power supply for this Hughes Q-Switched ruby laser rangefinder for the M60 tank ( I have NEVER seen the receiver...):
The spinning Q-switching 600Hz motor had a small magnet spinning with it. It would trigger a Hall effect sensor at J1. However, the orginal flashlamp was SERIES-TRIGGERED--no trigger transformer (T), and associated circuitry as you see here for pulse triggering--I think the guy used his own design. The original energy storage was in a single box (Sprague) with only two terminals "COM" and "COIL" It was refereed to as the -Pulse-forming-network- or "PFN". The best I could find on it was that it contained four 40uF caps and four coils. It was labeled (by hand): 160uF 1250V. There was no cap array. It looks like J3 connects to the lamp trigger wire. L1 is, no doubt, the pulse-forming inductor. The D2-D3 relationship is not clear to me.
The ruby was 3" x 1/4", and with the telescope mounted you could fire the laser in a large clear area and see a near-perfect crimson 'circle' on a very distant 'target' (at night of course). I cannot find the original block drawings for this laser, but I think it was all powered by 12v. There was a potted multi-voltage oscillator. Everything was light-weight...compared to the PFN which weighed only about 4-5lbs. I saw this PFN on ebay once.
Signification, Tue Jun 05 2018, 10:22AM
From way back; here is what I recall of the original power supply for this Hughes Q-Switched ruby laser rangefinder for the M60 tank ( I have NEVER seen the receiver...):
The spinning Q-switching 600Hz motor had a small magnet spinning with it. It would trigger a Hall effect sensor at J1. However, the orginal flashlamp was SERIES-TRIGGERED--no trigger transformer (T), and associated circuitry as you see here for pulse triggering--I think the guy used his own design. The original energy storage was in a single box (Sprague) with only two terminals "COM" and "COIL" It was refereed to as the -Pulse-forming-network- or "PFN". The best I could find on it was that it contained four 40uF caps and four coils. It was labeled (by hand): 160uF 1250V. There was no cap array. It looks like J3 connects to the lamp trigger wire. L1 is, no doubt, the pulse-forming inductor. The D2-D3 relationship is not clear to me.
The ruby was 3" x 1/4", and with the telescope mounted you could fire the laser in a large clear area and see a near-perfect crimson 'circle' on a very distant 'target' (at night of course). I cannot find the original block drawings for this laser, but I think it was all powered by 12v. There was a potted multi-voltage oscillator. Everything was light-weight...compared to the PFN which weighed only about 4-5lbs. I saw this PFN on ebay once.
Re: diodes in series/parallel
Flachzange, Sat Jun 23 2018, 10:24AM
I am replying to the original question whether standard diodes can be connected in series for higher voltages.
I have build 20 high voltage diodes with each 140x 1N4007 which should give a voltage rating of 140 kV. However i am using them only to 100 kV peak. And they are not intended for 50Hz but for 10kHz for a voltage multiplier circuit. I am using this configuration for 4 years now almost weekly. So far NO diodes have failed even though i have made some shorts at 400 kV (4 Stage voltage multiplier) and 700 kV (8 Stages) without any current limiting resistors!
However, all the diodes were from the same batch, but no extra capacitors or resistors are used to equalise the diodes.
Flachzange, Sat Jun 23 2018, 10:24AM
I am replying to the original question whether standard diodes can be connected in series for higher voltages.
I have build 20 high voltage diodes with each 140x 1N4007 which should give a voltage rating of 140 kV. However i am using them only to 100 kV peak. And they are not intended for 50Hz but for 10kHz for a voltage multiplier circuit. I am using this configuration for 4 years now almost weekly. So far NO diodes have failed even though i have made some shorts at 400 kV (4 Stage voltage multiplier) and 700 kV (8 Stages) without any current limiting resistors!
However, all the diodes were from the same batch, but no extra capacitors or resistors are used to equalise the diodes.
Print this page