Return to the thumbnail page Display/hide file information See previous file See next file

NE555 led driver simulation

NE555 led driver simulation

Click to view full size image

This is the schematic of boost ne555 led driver, with simulation.

IMG_2996.JPG IMG_3080.JPG 555_simulazione.jpg P5258675.JPG

Light Information

Light Information

Manufacturer:marcopete87
Electrical
Wattage:11W
Voltage:12V
Optical
Lumen Efficacy:~80%
Physical/Production
Dimensions:too big to be pratical
Application/Use:12V emergency and car light

File information

File information

Download: Download this File
Filename:555_simulazione.jpg
Album name:marcopete87 / Led light
Keywords:Gear
File Size:185 KB
Date added:May 31, 2012
Dimensions:1280 x 774 pixels
Displayed:374 times
URL:https://www.lighting-gallery.net/gallery/displayimage.php?pos=-65801
Favorites:Add to Favorites
Comments
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 13, 2012 at 11:05 AM Author: Medved
I see it is quite unstable in regulation: Too much gain without any frequency compensation.
Connect about 1uF capacitor between the collector and the base and ~10kOhm between that point towards the shunt resistor.
You may play with exact values in the simulator...

No more selfballasted c***

marcopete87
Sr. Member
****
Offline

Gender: Male
Posts: 266
View Gallery

View Profile Personal Message (Offline)
Oct 13, 2012 at 04:47 PM Author: marcopete87
thank for the advice!
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 14, 2012 at 12:26 AM Author: Medved
The quite large capacitor and resistor are a kind of "universal" fix, but the result is quite slow loop response, suitable only when all the involved variables are changing only slowly. If the thing is supplied from it's own battery with no other loads being connected/disconnected, it should be sufficient. But e.g. electrical environment of a running car would be way too harsh and most likely the LED's or other components would be overstressed before the loop could respond.

In the meantime I just did some raw calculation:
The DCDC is operating in Continuous Conduction Mode, what is very difficult to stabilize without involving the coil current into the feedback loop and/or without using extra large output capacitor (zero in the right hand plane of the AC response, fast, jerky response,...). The simple dominant pole I advised would have to be really low in frequency for the circuit to be stable (the 1uF/10kOhm should suffice with the components involved), but the consequence would be quite slow dynamic response on external changes.
To check this, replace the input DC voltage source by "vpulse" and let it generate alternate voltages between two levels (e.g. 10 and 15V, with rise time about 1us, pulse width about 1ms, period about 2ms), you would see, what the circuit does then.

You could ease yourself in the loop stability design, when you operate the controller in the discontinuous mode (DCM) only (so the coil current always goes to zero in each switching cycle before the transistor turn ON the next time). For that replacing the coil by only 10uH (and maybe lowering the frequency a bit) would do the job. The peak current would be higher, but the IRFZ44 have quite large margin there.
The trick is, the coil current is always reset to zero, so there is no converter "history" in the coil current (what is responsible for the lag in the loop response, so tendency to overreact later on with higher bandwidth).
With the converter operating in DCM you put ~47kOhm in series with the "1uF" capacitor and then you could reduce it's value to a few nF (try one change at a time in the simulator with the pulsed source, observe the result). Of course, play with the values a bit as well...


In books you could find an advice to simulate the open loop response in a small signal AC simulation and tune the loop frequency compensation there, but you would get stuck by obtaining the representative converter small signal model of the boost converter.
And even when you succeed with the model, the real-life circuit is highly nonlinear, so even when it may appear stable in a small signal model, it won't be in the transient.
Therefore I usually advice to think about the stuff in the "small signal perspective" first, but do not spent as much time with the modelling, usually it is not worth the effort. Do more extensive transient simulations instead and include marginal component values. In transients, look for aperiodic settling with no ringing after any disturbance you may invent (input voltage change, open/close of a switch parallel to some of the LED's, adding extra current pulse load between the LED top and ground,...). If the loop is stable with good margin, all these disturbances lead to aperiodic (RC-like) settling of all parameters (current pulse amplitudes, output voltage, pulse widths,...)

No more selfballasted c***

marcopete87
Sr. Member
****
Offline

Gender: Male
Posts: 266
View Gallery

View Profile Personal Message (Offline)
Oct 14, 2012 at 01:32 AM Author: marcopete87
thank you!
when i'll have some time to spend, i'll try to simulate this.
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 18, 2012 at 12:21 AM Author: Medved
Let me know the result...

No more selfballasted c***

marcopete87
Sr. Member
****
Offline

Gender: Male
Posts: 266
View Gallery

View Profile Personal Message (Offline)
Oct 18, 2012 at 12:43 AM Author: marcopete87
i'm simulating...

edit:
1µF cause more issues than it solves.
100nF make current more stable.
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 22, 2012 at 04:44 PM Author: Medved
I did some experiments:
Indeed the 1uF is quite tttoooooo ssslllooowww tttooo rrreeessspppooonnnssseee, but it was aimed to so slow, so the rest of the circuit would not form any significant inertia, so the circuit would be stable. Of course, it would be useful only for a "single battery - single lamp - nothing else connected" system, where no dynamic changes other than the slow battery discharge by this circuit by itself could be expected.

I didn't resist and made a few simulations as well, but I find a few inconsistencies:
- First the "1nF" as the timing capacitor is too low for 250kHz switching frequency as per your presented simulation plot (it yield about 500kHz, I had to change it to 2.2nF with the same resistors). However I used quite an idealized NE555 model, with very low simulated internal delays. But if that make the difference, the IC is operated way too fast, beyond it's limit for that purpose (it's behaviour won't be reasonably stable).
- The IRFZ44 have too high gate capacitance for such high frequency and only 200mA gate drive capability of the 555 would cause quite high transient (switching) losses. I've seen there ~10..20% losses purely on switching transients. That would require the operating frequency to be below 100kHz in order to bring the switching losses into reasonable figure. I haven't count for losses related to reverse recovery charge of the diode (I used Schottky, what have no reverse recovery charge, at least the one in my model; it would be Freq*Trr*Vout*Icoil; it would disappear, when the converter operate in DCM)

What worked well in the simulation (kept ~200..250kHz operating frequency) was "type III" compensation network:
Between C and B of the BC547: 22nF (dominant pole) with series 22kOhm resistor (1. compensation zero), all that bypassed by 150pF (low pass filter)
Between B of the BC547 and cathode of the rectifier diode: Series 1kOhm and 470pF to form a kind of pre-emphasis
Between the B of BC547 and the sense resistor would be the 10kOhm (mentioned earlier)

In order to suppress the temperature dependency of the output current, I would use a TLV431 instead of the BC547 ("Cathode" instead of the collector, "Anode" indtead of the emitter, "Feedback" instead of the base). This would keep exactly 1.25V as the drop over the sense resistor (so for 320mA it have to be changed to 3.9Ohm; the "10kOhm" resistor between the sense resistor and the FB of the TLV431 would have to be replaced by ~18kOhm in order to maintain the AC loop response)

No more selfballasted c***

marcopete87
Sr. Member
****
Offline

Gender: Male
Posts: 266
View Gallery

View Profile Personal Message (Offline)
Oct 23, 2012 at 12:11 AM Author: marcopete87
ok, but:
1) refer to my sheet http://marcopete87.altervista.org/files/progetti/DRIVER_555/led_555.ods
2) i think i mistook capacitor in calculator (actually i think i used 2.2nF)
3) do you have any affordable mosfet to suggest?
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 23, 2012 at 02:10 AM Author: Medved
1) I didn't managet to make the table calculating (M$ Office only open it without working equations)
2) That match my simulations well, it was about 200kHz (depend on battery voltage) with 4k7/1kOhm resistors
3) http://cz.farnell.com/on-semiconductor/ntd6414ant4g/mosfet-n-ch-w-diode-100v-32a-dpak/dp/1879965
http://cz.farnell.com/international-rectifier/irfsl3806pbf/mosfet-n-to-262/dp/1602235
The IRFZ44 is marginal, but still usable on the 250kHz (due to it's large case)

No more selfballasted c***

marcopete87
Sr. Member
****
Offline

Gender: Male
Posts: 266
View Gallery

View Profile Personal Message (Offline)
Oct 23, 2012 at 11:42 AM Author: marcopete87
i used openoffice italian version to write this spreadsheet.

to get working sheet, replace "ARROTONDA" with "ROUND" (or your language translation) in all formulas.

and there are some translations

"C consigliato": suggested output capacitor
GREEN->User data
CYAN->Components calculated values
GREY->values use only for information
RED->don't touch!
"1) Scegliere un induttore con corrente di saturazione >> ILPEAK" -> 1) select an inductor with saturation current >> ILPEAK
"2) Il valore di C è il minimo sindacabile, più lo aumenti, meglio è." 2) C value is the minimum value, more is better for ripple filtering.
consigliata->suggested
Resistenza definita dall'utente->user defined resistance
Resistenza massima dissipatore->heatsink max resistance

thank for the links!

p.s. select boost sheet!
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4307
View Gallery

View Profile Personal Message (Offline)
Oct 23, 2012 at 12:06 PM Author: Medved
The problem was there were no equations at all, the Excell put there numbers, so I could not check the equations.
When opened in Oo, it works well (better than if it would have been created in M$ excell, as the M$ would require me to switch to Italian, while the Oo translate the displayed function names automatically, so I see them in English...)

For DCDC's I usually use different approach (more hand calculations), but I'm going to check the equations, as at least the timing resistor calculation seems to be incorrect...


Added: Now I see the problem: The sheet make all the calculation only for the minimum Vin, while at first glance it give the impression the frequency would be on the nominal Vin...

Edit2: The delta(Il) is there with fixed relation to the average current, with quite a low value for the boost, so it yield quite large coil. This ratio would be better to make selectable. And as stabilizing the loop is way simpler in the discontinuous mode, it would be beneficial to take into account the whole supply range (if designed for DCM, the DCDC should stay in the DCM in the complete Vin range).
But it is true, than the output power capability calculation for the DCM would look in a different way and it would be steered by different equations (e.g. the Duty would differ)...

No more selfballasted c***

© 2005-2019 Lighting-Gallery.net | Powered by: Coppermine Photo Gallery