How does one do wide input voltage reduction from 50-150V?

[zombiess]

I am trying to figure out how to do a wide input voltage reduction stage for a FET driver setup. Something that can do 50-150V would be nice. Esitmated current requirements are 30mA total for the gate driver circuits at 15V. To get there I need to drop down to an easier to use voltage, probably 20-25V so a regulator can be utilized. DC-DC modules that do this cost too much. I was thinking of using a TL783 to drop to ~40V and then use another TL783 to drop to 15V for my supply. Not pretty or efficient, but it's cheap and I "think" it will get the work done, assuming I have my gate drive power requirements correct.

Gate drive current requirements are

Idrive = Qg * fsw - Average Current supplied to the gate driver
7.3mA = 454nC * 160000
Two gate drives = ~ 15mA

30mA = 2* 15mA to account for headroom an other parasitic losses in the circuit.

Suggestions, corrections? Ideas?

 

[Lebowski]

have a look here:
http://endless-sphere.com/forums/viewtopic.php?f=30&t=39820&p=582042#p582042

have a look at the low voltage power supply schematic. The code for the processor is given in the same post
it's the schematic and code for the 15v and 5v suppy on my bike

 

[zombiess]

Thanks Lebowski. Looks pretty straight forward. What kind of amperage does it supply? What is the switch freq? I have a bunch of 12f617 pics laying around here from a previous project. How high/low will this scale in voltage with appropriate voltage components? Do the FETs need heat sinking?

 

[Lebowski]

The output current is limited by either the power FET, the diode or the inductor. I use 1A inductors, which are the lowest rated
components and therefore limit the current. But in my case I use the circuit for powering the controller, not for a bunch of 60W 12V lamps
or something. I use the same gate driver IC and FET as for the motor, and a 200V diode, so it has a high voltage rating (150V with a 4115). Because of
the low output current (a few 100 mA) I don't use heatsinks.

The 12F 's function is very simple. It has an internal voltage reference of 0.6V The chip compares a resistive devided output voltage
signal with the 0.6, and when it's too low it turns the power transistor on for 10u-sec, followed by 20u-sec off. The 10-usec on charges the
inductor with a current, which is then dumped in the output capacitor raising the output voltage. The 20usec off is to prevent too much
rise in inductor current. After the 20usec off time the power transistor stays off, and the 12F waits until the voltage again drops below 0.6 to
give a new pulse. It really is that simple (we used to build the DCDC power supply inside a mobile phone this way, this is how the 4V battery
is turned into 1.1V for the processor). The source code is given in the linked post, so you can see how it works and make changes if you want.

At startup the 20usec wait time is longer (to prevent too much build up of current) but this is reduced to 20usec over a short period of time.
At startup the 12F first waits about half a second to let the bootstrap capacitor of the gate driver IC charge up via the zener diode voltage
generation construction. This because the first pulse to the high side NMOS FET must turn it on, else the power supply stays dead.

Last, I have the 12F give out a signal that the power is ready, this signal is tied to the reset of the controller IC. You don't want the controller
IC to already consume a lot of current while the 12F is still busy building up the supply. Power ready is turned on when the deadtime after
a 10usec pulse has dropped to the aforementioned 20usec.

 

[fechter]

Most AC switching power supplies will work in that voltage range. A small wall wart power supply is all you need. These are isolated too, and generally very inexpensive. A bit on the large size though.

 

[zombiess]

Lebowski, you have just taught me more about switching power supplies than I have ever known before

I do not do go with ASM, but I have looked through your code for the 12F617 and it looks pretty simple to figure out since I am familiar with PICs and registers, setting bits, clock cycles, etc. All my programming is done in BASIC which is then translated to machine code. I use Proton Plus which gives me low level access to the chips and fast code.

I just realize I do not have any 12F617's, I have it's little brother the 12F675; that's not going to work since it doesn't have a lot of the features needed. I'll be ordering a 12F617 in my next batch of parts.

Thanks for posting this as it's something I have been wondering about. Your explanation of operation really helps.

 

[Alan B]

They do make chips for this application:

http://www.linear.com/product/LTC3639

 

[zombiess]

They do make chips for this application:

http://www.linear.com/product/LTC3639

That is a nice chip and I missed it in my many hours of searching, but that chip only does up to 5v. I need around 20v.

*edit*
I just read the datasheet and saw it has an adjustable output, I spoke too soon. Thanks for the find Alan, I was searching all over yesterday for an IC just like this.

 

[zombiess]

I wonder if I can convert your code over to my setup so that I can utilize it easier in the future, and play around with it. Working with a single flat basic language file is much easier for me.

 

[Lebowski]

just watch out with the LTC3639, it only delivers 100mA. The controller IC for instance takes 200mA from 5V... What you can
do is use the 3639 to generate 50V (at 100mA -> so 5W) and then use a LTC3630 to step this down to 5V (at 500mA, so 2.5W)

 

[fechter]

They do make chips for this application:

http://www.linear.com/product/LTC3639

I haven't seen that one before. It looks pretty good. 100mA is enough for most controllers.

 

[dnmun]

mouser did not show it, here is the data sheet pdf:

http://cds.linear.com/docs/en/datasheet/3639fb.pdf