I have been working on an economical way to replace ~40 8ft F-Tube lighting fixtures for the past decade or so...with each advancement in LEDs I re-figure, recalculate and procrastinate.....The most recent cycle involves the 10W chips nominally 12V / 900mA with peak efficiency ~ 10.0V ~ 600mA.....Real World testing suggests 12 - 18 chips provide equal or greater lighting effects than a single 8ft 2-tube fixture (LEDs @ 72W to 108W when driven @ 600mA......compared to the base line ~160W of the F-Tubes).....So the problem is now reduced to driving 12 to 18 "10W" (actual power ~ 6W each) LED chips * ~40 fixtures.....Obviously this is a non-trivial consideration....A single 650mA 190Vmax driver is available for ~$60.....another possible solution is a 36V 600mA driver for ~$3.50ea....obviously this would require 4 to 6 of the drivers netting $14 to $21 per fixture, while the installation would be a bit more involved this route offers the potential of reduced long-term maintenance by decreasing the cost per failed driver.......There are a myriad of other potential solutions using readily available drivers.....But.....
At the end of the day,it would be ideal ito have ~160V to ~180V regulated @ 600mA.....
And this has led me down a dark and potentially dangerous path....As luck would have it rectified line voltage is nominally ~170V....assuming 17 LED chips in series @ a nominal forward voltage of 10.0V each @ a "natural forward current" of 550mA it is very tempting to forgo any driver at all.....I know, it is an insane idea! But is it really? The rated voltage of each LED is nominally 12V so a circuit to "clamp" the voltage below 204V would be important to protect the LEDs from surges.....beyond that, actual testing of the LEDs demonstrates they operate continuously @ 11.0V @ ~900mA (9.9W) if properly cooled....this would imply that the series could withstand ~187V indefinitely....Further testing indicates that @ 153V the LEDs operate @ ~310mA (2.82W) at roughly half the brightness they have @ the nominal 170.0V, so the series would natively still be "functional" @ +/-10% ( 170Vdc - 17Vdc = 153V 170Vdc + 17Vdc = 187Vdc) without any dire implications....
Because the LEDs are intentionally being driven @ only ~65% of their rated current (Rated Current = 950mA....Peak efficiency//Target Current = ~600mA) they are relatively immune to destructive run away from thermal drift and normal grid line variations.....Which only leaves "surge protection" for voltages over 204V.....a traditional inductor/capacitor post rectifier filter is certainly worth the effort, but installing a "whole house surge protector" on the mains and then routing the power through a smaller surge protector should ensure that transient spikes are significantly muted before reaching the rectifier and its filters.....
The existing F-Lights are on five separate breaker-circuits and wired with 6-10 lights per circuit, so my thought is to build a rectifier for each existing circuit with fairly robust filtering and then add second order filters at each fixture in the circuit followed by a 1A Circuit Breaker or fuse and a 180V (Standoff Voltage) TVS as the final line of defense (which should trip the breaker/blow the fuse if a voltage spike makes it past the primary surge suppressors, secondary and tertiary supply filters) While all of this may sound like a lot, the actual cost of materials for a one-time short run for the ~40 fixtures is actually almost negligible....
The Elephant in the room is the complete lack of galvanic isolation.....but this is obviously already true for the 120V mains run to the existing lights....and with a 1A circuit breaker/fuse at each fixture the potential hazard is lower than that of the existing fixtures.....
A dark and dangerous path.....I would love some thoughts on why I shouldn't proceed to build/test a prototype....Intuitively the lack of galvanic isolation worries me, AND the idea of operating LEDs w/o at least passive current limiting is troublesome.....but the math would SEEM to suggest the circuit (or lack there of) will operate nominally under a fairly large range of conditions....And I can't see any way a "real driver" would increase safety or efficiency.....
Thanks in Advance,
Fish
At the end of the day,it would be ideal ito have ~160V to ~180V regulated @ 600mA.....
And this has led me down a dark and potentially dangerous path....As luck would have it rectified line voltage is nominally ~170V....assuming 17 LED chips in series @ a nominal forward voltage of 10.0V each @ a "natural forward current" of 550mA it is very tempting to forgo any driver at all.....I know, it is an insane idea! But is it really? The rated voltage of each LED is nominally 12V so a circuit to "clamp" the voltage below 204V would be important to protect the LEDs from surges.....beyond that, actual testing of the LEDs demonstrates they operate continuously @ 11.0V @ ~900mA (9.9W) if properly cooled....this would imply that the series could withstand ~187V indefinitely....Further testing indicates that @ 153V the LEDs operate @ ~310mA (2.82W) at roughly half the brightness they have @ the nominal 170.0V, so the series would natively still be "functional" @ +/-10% ( 170Vdc - 17Vdc = 153V 170Vdc + 17Vdc = 187Vdc) without any dire implications....
Because the LEDs are intentionally being driven @ only ~65% of their rated current (Rated Current = 950mA....Peak efficiency//Target Current = ~600mA) they are relatively immune to destructive run away from thermal drift and normal grid line variations.....Which only leaves "surge protection" for voltages over 204V.....a traditional inductor/capacitor post rectifier filter is certainly worth the effort, but installing a "whole house surge protector" on the mains and then routing the power through a smaller surge protector should ensure that transient spikes are significantly muted before reaching the rectifier and its filters.....
The existing F-Lights are on five separate breaker-circuits and wired with 6-10 lights per circuit, so my thought is to build a rectifier for each existing circuit with fairly robust filtering and then add second order filters at each fixture in the circuit followed by a 1A Circuit Breaker or fuse and a 180V (Standoff Voltage) TVS as the final line of defense (which should trip the breaker/blow the fuse if a voltage spike makes it past the primary surge suppressors, secondary and tertiary supply filters) While all of this may sound like a lot, the actual cost of materials for a one-time short run for the ~40 fixtures is actually almost negligible....
The Elephant in the room is the complete lack of galvanic isolation.....but this is obviously already true for the 120V mains run to the existing lights....and with a 1A circuit breaker/fuse at each fixture the potential hazard is lower than that of the existing fixtures.....
A dark and dangerous path.....I would love some thoughts on why I shouldn't proceed to build/test a prototype....Intuitively the lack of galvanic isolation worries me, AND the idea of operating LEDs w/o at least passive current limiting is troublesome.....but the math would SEEM to suggest the circuit (or lack there of) will operate nominally under a fairly large range of conditions....And I can't see any way a "real driver" would increase safety or efficiency.....
Thanks in Advance,
Fish
Probably saved me some $$$. PoCo came out and took me off the grid while I wired in a new one. 
