Revisiting my USB-powered active load: providing the MOSFET gate a good voltage.

[technix]

So the idea of using a logic-level power MOSFET and super low sense resistors got swept into the junk bin for being too unreliable (and I have to burn a power TO-247 MOSFET and my finger to find out) so I am back to the drawing boards in the power MOSFET gate drive. The sense resistors are back at 10x 10R in parallel (1R total) and the entire module is still USB bus powered and headless, operated completely through an app on the computer.

So here is back to the how to drive the power MOSFET problem. The maximum current rating is now 2.5A thanks to the sense resistor, and this means smaller TO-220 power devices like the IRL540 can be used. Also, I am thinking driving the MOSFET with a higher gate voltage.

So how good is ICL7660 as a voltage doubler? Can I generate the 8-10V using ICL7660, and control the power MOSFET using that? ICL7660 will power the TLC2272 op amp, ADR02B voltage reference (used by both on-chip ADC and DAC for setting voltage) and the MOSFET gate. The TLC2272 will have two uses: comparing the sense resistor voltage to the setting voltage, and double the first voltage output by two. (Maybe I can optimize away one op amp?)

[SeanB]

Use the 7660 as a voltage quadrupler, which will give you at least a usable 15V supply, easy to stabilise with either a 15V zener diode ( as it will be current limited to around 3mA in any case) to drive the Mosfets and the CMOS opamps. You probably will be better off using another 7660 to drive the reference separately, just to have a low noise supply rail for it. 8V will be enough, and low current draw will make the ripple lower.

[technix]

Use the 7660 as a voltage quadrupler, which will give you at least a usable 15V supply, easy to stabilise with either a 15V zener diode ( as it will be current limited to around 3mA in any case) to drive the Mosfets and the CMOS opamps. You probably will be better off using another 7660 to drive the reference separately, just to have a low noise supply rail for it. 8V will be enough, and low current draw will make the ripple lower.

It seemed to me that quadrupling the voltage needs two 7660. Cascade them like two stages of separate voltage doubler? Do I need to clock synchronize them or a reservoir cap is all it needs?

Or should I just go with MC34063? Keep in mind that USB bus power is limited.

[SeanB]

No need for a clock, and you can use the first doubler to drive the second and the reference, though you will want some more electrolytic capacitors for decoupling. Will be a roughly constant 50mA of power draw in any case, though you will probably want to use lower value caps on the second doubler to increase the ESR of this stage.

[technix]

Here is the analog board, all caps ceramic (even lower ESR.) 10µ caps are 0805 size, load resistors are through-hole ones, and every else R and C 0603.

[SeanB]

Make C2,C6 10uF as well, you will need some more capacitance on that 15V rail and the USB supply. Saves buying 2 different values, and the voltage degradation means you probably want 2 parallel units in those locations as well. Always easier to leave them off than to add later.

You can also simply remove IC2 (note you did not connect ground pin 3 in any case or show a supply) and drive the pump capacitor from IC1 and connect D2 to D3.  Will give 12V minimum to drive the opamps.

[technix]

Make C2,C6 10uF as well, you will need some more capacitance on that 15V rail and the USB supply. Saves buying 2 different values, and the voltage degradation means you probably want 2 parallel units in those locations as well. Always easier to leave them off than to add later.

You can also simply remove IC2 (note you did not connect ground pin 3 in any case or show a supply) and drive the pump capacitor from IC1 and connect D2 to D3.  Will give 12V minimum to drive the opamps.

Can you show me the circuit without IC2? Also the digital board will have lots of 100n/220n there so I am already buying two values. BTW this fits in a single-sided board.

Brain fart regarding pin 3 of IC2. Also the four diodes are actually Schottky (Eagle does not came with the appropriate footprint/symbol combo and I am not bothered with creating one either.)

[eneuro]

I need turn ON/OFF mosfets from 3.3V-4.1V using MPU, so I need VGS 12V-15V, but It looks like @EEVBLOG patent shown in Dickson Voltage Doubler with more than one stage looks much simplier and can be quite compact since we need those caps for voltage double/tripple/ anyway, so I do not know, probably will stick with this, but unsure how to renew licence 



I like much better this discrete stright forward way to step up voltage shown in Dave vblog  than using 7660

This simple circuit simulation shows that after a few input pin switches I've nice 15V from 4.1 (I've assumed 0.3Vf, which is probably possible using eg. BAT85 I've used in other low Vf drop projects before) 


BTW: If Dave provided his own Bitcoin number, like I've did, I could from time to time send some $es for good beer, etc, thanks to  tricky things shown in his vblog 

[Jorpy]

BTW: If Dave provided his own Bitcoin number, like I've did, I could from time to time send some $es for good beer, etc, thanks to  tricky things shown in his vblog 

It's on the donations page: 1KgAGM3BPP6tCAYaVX4trELx3vUTmg4C4x 

[technix]

I need turn ON/OFF mosfets from 3.3V-4.1V using MPU, so I need VGS 12V-15V, but It looks like @EEVBLOG patent shown in Dickson Voltage Doubler with more than one stage looks much simplier and can be quite compact since we need those caps for voltage double/tripple/ anyway, so I do not know, probably will stick with this, but unsure how to renew licence 



I like much better this discrete stright forward way to step up voltage shown in Dave vblog  than using 7660

This simple circuit simulation shows that after a few input pin switches I've nice 15V from 4.1 (I've assumed 0.3Vf, which is probably possible using eg. BAT85 I've used in other low Vf drop projects before) 


BTW: If Dave provided his own Bitcoin number, like I've did, I could from time to time send some $es for good beer, etc, thanks to  tricky things shown in his vblog 

Thanks for your info and I will modify accordingly - my circuit is using SS14, the SMD version of 1N5319 Schottky barrier rectifier, in fact.

[NiHaoMike]

If you have a spare PWM on the microcontroller, you could use it to drive a small open loop boost converter, with a zener diode to limit the output voltage.

[technix]

I need turn ON/OFF mosfets from 3.3V-4.1V using MPU, so I need VGS 12V-15V, but It looks like @EEVBLOG patent shown in Dickson Voltage Doubler with more than one stage looks much simplier and can be quite compact since we need those caps for voltage double/tripple/ anyway, so I do not know, probably will stick with this, but unsure how to renew licence 



I like much better this discrete stright forward way to step up voltage shown in Dave vblog  than using 7660

This simple circuit simulation shows that after a few input pin switches I've nice 15V from 4.1 (I've assumed 0.3Vf, which is probably possible using eg. BAT85 I've used in other low Vf drop projects before) 


BTW: If Dave provided his own Bitcoin number, like I've did, I could from time to time send some $es for good beer, etc, thanks to  tricky things shown in his vblog 

Breadboarded this with 1uF polymer caps (I don't have nonpolarized 10uF caps yet) a string of 1N5819s and an ICL7660 in DIP package, got 12V when driving an idling TLC2272 (all pins except power floating). 15V without load.

[technix]

Here is the analog board, single sided through-hole construction. (This kind of construction is cheap, and "star grounding" is replaced with a big continuous ground plane)

The two smaller resistors are polyfuses.

C1 through C7 are footprint for 1µF polymer caps although I may end up using 10µF through-hole MLCC

If I want to also tap off the input voltage without loading it, how can I do that? This is required to allow me to have a constant resistance mode by keeping the sense resistor voltage a constant ratio of the input voltage.

[eneuro]

Breadboarded this with 1uF polymer caps a string of 1N5819s and an ICL7660 in DIP package

Do you have idea what and how switching frequency is managed in ICL7660?
I wonder also what about limiting switching current (eg. MPU pin)-I've rdawn this 330R by by mistake on VCC, but my intention was limit pin current somehow to lets say <20mA, so I've did another simulation with 330R on MPU pin, as well as I will try adjust (optimise) switching frequency based on caapcitance used and current limit resistor.
Quick calculation of RC constant for eg. 330R and 1uF, gives 330us for RC, so lets use 3RC time to delay (half period) toggling this PWM MPU pin and we'll see what happends.
In hand made simulation it looked like, when we wait anougth time to charge/discharge througth this 330R resistor than I've got decent results even with small load (100k resistor on mosfets gate to ensure OFF state when gate driver circuit not powered yet, etc).
What I mean-I looked like we have to adjust somehow this switching fequency to allow capacitor fully charge/discharge and inserting calculated delay based on RC constant is for the moment best idea, but having another spare MPU pin probably it could be done easier-just but enabling pin change interrupts, however in this case not sure what logic voltage levels will be for low/high (need to verify datasheet depending on MPU used and VCC), those threshold logic voltage levels will be above zero and less than VCC.
It is interesting how ICL7660 manages this switching-it has build in oscilator set to predefined frequency, or comparator to watch when this first stage capacitor is close to fully charged/discharged? 
Anyway, ICL7660 probably cheaper than MPU so it might me interesting option too, I knew details how its swithing frequency is managed 

[technix]

Breadboarded this with 1uF polymer caps a string of 1N5819s and an ICL7660 in DIP package

Do you have idea what and how switching frequency is managed in ICL7660?
I wonder also what about limiting switching current (eg. MPU pin)-I've rdawn this 330R by by mistake on VCC, but my intention was limit pin current somehow to lets say <20mA, so I've did another simulation with 330R on MPU pin, as well as I will try adjust (optimise) switching frequency based on caapcitance used and current limit resistor.
Quick calculation of RC constant for eg. 330R and 1uF, gives 330us for RC, so lets use 3RC time to delay (half period) toggling this PWM MPU pin and we'll see what happends.
In hand made simulation it looked like, when we wait anougth time to charge/discharge througth this 330R resistor than I've got decent results even with small load (100k resistor on mosfets gate to ensure OFF state when gate driver circuit not powered yet, etc).
What I mean-I looked like we have to adjust somehow this switching fequency to allow capacitor fully charge/discharge and inserting calculated delay based on RC constant is for the moment best idea, but having another spare MPU pin probably it could be done easier-just but enabling pin change interrupts, however in this case not sure what logic voltage levels will be for low/high (need to verify datasheet depending on MPU used and VCC), those threshold logic voltage levels will be above zero and less than VCC.
It is interesting how ICL7660 manages this switching-it has build in oscilator set to predefined frequency, or comparator to watch when this first stage capacitor is close to fully charged/discharged? 
Anyway, ICL7660 probably cheaper than MPU so it might me interesting option too, I knew details how its swithing frequency is managed 

ICL7660 have a predefined 10kHz switching frequency (measured using a scope) or it can be slowed down using a cap on a pin.

[eneuro]

OK since we know what this frequency is in the case of 7660, than proper resistor and caps can be used, because of when I'd like to use for example 1nF caps like in this another sim, this switching frequency had to be different to fit into charge/discharge rates, but still with 15Vz I was able in by hand simulation keep mosfet VGS (after 27R gate resistor) at desired voltage, but I had to click PWM switch really fast   



There i some  advantage of using MPU since we can input to doftware resistor and caps values and let MPU calculate swithing frequency-no need to mess with additional caps to adjust this frequency etc 

[technix]

OK since we know what this frequency is in the case of 7660, than proper resistor and caps can be used, because of when I'd like to use for example 1nF caps like in this another sim, this switching frequency had to be different to fit into charge/discharge rates, but still with 15Vz I was able in by hand simulation keep mosfet VGS (after 27R gate resistor) at desired voltage, but I had to click PWM switch really fast   



There i some  advantage of using MPU since we can input to doftware resistor and caps values and let MPU calculate swithing frequency-no need to mess with additional caps to adjust this frequency etc 

This is a lot of stages so I settled down with 6 Schottky diodes and 7 10uF caps. There is no 330R resistor though.

[technix]

No longer bother tackling the charge pump, I replaced it with an MC34063 circuit. Still single-sided through-hole only.

[eneuro]

No longer bother tackling the charge pump,

It will depend on project requirements and in application like mine where I want on the fly controll mosfet VGS to provide something like soft start by adjusting its RDSOn by slowly rising VGS such one pin PWM controll when used in 1nF caps in simulations seams to work fine, especially while as I've shown this thing will drive mosfet gate directly, so no another power losses, but 100k resistor on mosfet gate, with lets say a few nF capacitance 100k is still able to turn off this mosfet without any additional gate driver (load less than 100W) within 1ms-2ms and power consumption at 15VGS is 0.1mA at 100k gate resistor, so from my point of view in my project it is still a winner and MC34063 is not what I'm looking for, but for you maybe it will provide more curent needed to power another ICs needed.
I do not need any additional ICs and I've 0.1mA (100uA) load on mosfet gate, so this tricky charge pump should be fine