regulated switched LED regulator designs (experimental)

the thing ::

how to set it up in practice . . . involves a number of tests ::

1. Detach the mosfet and the feedback Mod line -- from simulation and from build
   Regulate the frequency and duty to match those of the simulation
2. Re-attach the mosfet , ( /!\ /!\ /!\ ) set some 20Ω or greater in series with the 6LF22 to limit the current through LED-s --and-- set electrical fuse or limitting resistor in series with the LED-s --or-- double the number of LED-s ( /!\ /!\ /!\ ) - - - - - - - - - we are going to verify the transformer ::
   do the same mods on the simulation and test for the differences . . .
   -- if the real build performs worse you need to recalculate and rebuild the transformer accounting for resistance not present in ↑this↑ circuit
   -- if the build performs better then the primary side is "over-driven" the timing of the switching wave-forms has to be re-evaluated and adjusted
3. Attach the Mod line --and-- ( /!\ /!\ /!\ ) set the regulation to be greater than in simulation (!! dont remove electrical fuse or extra LED-s !!) ( /!\ /!\ /!\ ) ,
   -- regulate it bit below the desired LED current (in case the regulation failure -- the regulator has to be re-designed --or-- replaced with better known one)
4. if everything worked out so far the additional tests are required such as ::
• test for bit higher than simulated input voltage (the Cl batteries may have up to 1.8V per cell e.g. 10.8V for fresh 6LF22)
• test for battery contact strobe !!!
  -- here everything can fail -- use fuses and perhaps build a protective controller that can fast switch off the circuit under test
• test for the lower than in simulation input voltage !!!
  -- may result in oscillations failure and the MOS-FET being "constantly ON" (attempting to remain in conduction mode) . . .
• etc. ...

---

Basic simple ::

add VCO , Mod2 divider , the 2 to 4 decoder and a current sense / +integrator -- and it's basically done . ..

---

[Eop]

Making working/alternate TC4405 model

TEST

CLM

Better than nothing OR the Microchip's model

---

[Eop]

Comparing Constant Current Circuits (to sort/verify the LED-s with)

▼Fast/Simple/Robust/Trivial Op-Amp + BJT Setup▼
→ apx. 16mA ±30µA(0.2%)

[fig.1]

▼Complex/Trivial Op-Amp + BJT Setup▼
 → apx. 18mA ±24µA(0.1%)

[fig.2]

▼Simple/Experimental BJT Setup ( !? the Best ?! )▼
→ apx. 18mA ±60nA(0.0003%) 

[fig.3]

▼Full/Trivial BJT + N-jFet Setup ( also better than any op-amp variant ??? )▼
→ apx. 18mA ±300nA(0.002%) 

[fig.4]

▼Complex/Experimental Op-Amp + MOS-Fet Setup▼
(likely worse than it's BJT alternate -- the 2-nd ↑↑ fig. )
→ apx. 18mA ±120µA(0.7%) 

[fig.5]

▼Trying to get it right here▼
-- it seems that the op-amps should be fed from the separate supplies or otherwise the PSRR won't recover the precision reached at reference voltages
→ apx. 18mA ±48µA(0.3%) 

[fig.6]

((the work is in progress ...))

▼If you are stupid then knowing the right Op-Amp-s could save your day▼
(the node "SE" had the greatest error although the node "a0" has the least ??? -- about LT1012A)
→ apx. 18mA ±10µA(0.06%)

[fig.7]

-------- Update 2017.12.24 ::

▼a conditional opposite to "If you are stupid then knowing the right Op-Amp-s could save your day"▼

→ apx. 18mA ±7µA(0.04%)

 [fig.8.1]

a "Russian LM308" **

 [fig.8.2]

** it's nearest LT's substitute (so far)

---

[Eop]

3-transistor "precision" LDO

Left-Below is the ® , Below-Right is over current-(/SC-) proof advance of it (as usual -- not tested as a real thing !)

---

[Eop]

Depletion mode MOSFET - macro model - some essencial tests

about BSS139 -- R_DSON

10MHz - random drive (i donno any specifics about these devices)

---

[Eop]

Verifying the MOSFETs´ RDS(ON)

. . . being used to BJT-s i always thought the R_DSON being some illustrative variable that has nothing to do with reality - amazingly enough the tests i performed verified it being NOT a theoretical figure !

used the MOSFETs (IRFZ44N and BUK455-60A) unmounted from cordless drills´ speed adjustment circuits

the IRFZ44N existed in Spice DB but for the BUK455-60A i used a model compilation method described in the [LT Spice Tutorial 6] website (which actually resulted in better match with reality this time ??)

 as for R_DS(ON)

as and for directly measured I_D, U_D, U_G

ah! - also modified in ↑that↑ spice tutorial the files RDSon test jig and Switching Time Test Jig coz they were totally impractical to use for given purpose (fine tuning your MOS-Fet models) respectively to Cust MosFETs - 01.asc and Cust MosFETs - 02.asc (note that you have partially pre-calculate and modify accordingly the component parameter values to get your response right)

---

Update!

PS! Since the power supply (Mod.: SMP-100KB, Rev.: A3, Date Code: 8945, +12Vdc/3.75A - output) was not very stable (or depended on it´s ~AC input ???) the source data can be trusted from 1 to 3 digits (depending on conduction state of the FET and stability of the PSU) and contains extra to that the unknown systematic +other errors of the multi-meters -- this was not intended to be a precision measurement anyway - but for setting the apx. (static) device margins (near +25°C).

also there were used voluntary statistical alterations to guess better the values required to form a "believable" (?accepatable) output for R_DS .


[Eop]

Random Voltage Regulator Tests

Random keywords : stability, efficiency, ripple, startup time, slew rate, deviation :::

---

---

(Added 2015-01-10) LTSpice Src ...

---

[Eof]