Post by thread:Need another set of eyes for MOSFET capacitive swi

In the process of building a DC-DC converter for my car MP3 player I've come across the need for -5 and -12 volt power. While I am aware of that there are a number of single chip solutions, especially from Maxim, I'm attempting to attack a capacitive switcher for reasons of understandability and parts availablity. In short I have no idea where to get parts for or how to wind flyback transformers. Also I wanted to practice/learn some discrete design skills. I understand capacitive switchers. An example can be found here: http://ourworld.compuserve.com/homepages/Bill_Bowden/page10.htm#5volts.gif That uses a 555 to generate the drive the charge pump. Now the problem is that the current available from the circuit depends on several factors: 1) The current available from the 555. 2) The size of the caps. 3) The discharge rate of the reservoir cap. So what I'm trying to figure out is how to get 0.5 A at both voltages from the circuit. For the sake of argument I imagined the circuit with a fixed load that drew 0.5A at all times. For -5V that would be a 10 ohm load. The input voltage is a nominal 12V with a 1.4V drop due to the diodes. Lastly to ensure regulation we'd like to have the reservior cap never go above -7 volts. Now the discharge rate depends on the RC constant (specifically Vt = V0*e(-t/RC). Through experimentation I figured out that when t/RC is 0.4 that the cap discharges to 2/3's of its original value, which is near the ratio from -10.6 to -7 that I'm looking for. So plugging in: 0.4=t/10C (R=10) 0.4*10C=t 4C=t So the discharge time and hence the required charge frequency is directly related to the size of the reservoir cap. Some common values results in easily achievable frequencies. Even the 100 uC shown in the circuit only requires a 2500 Hz frequency to maintain its charge. Now all I have to do is charge it fast enough. I decided to use a pair of logic level Power MOSFETs to charge the flying cap. The IRFZ44 with its 0.028 ohm Ron resistance will completely charge a 100 uC flying cap in 14 uS which is more than fast enough. The only thing I was trying to figure out is how to put the MOSFETs in a totem pole configuration. After researching CMOS configurations I finally think I understand that the purpose of the P-channel MOSFET is so that the complementary MOSFETS can be driven by the same input. But since the characteristics of the N-channel is so much better I want to use a pair of N-channels. So I plan to use this config: +12 | D A-->G S | |-----| | ------ | D B-->G S | | gnd with A and B having to be complemetary. Whichever point A or B that's 5V will conduct with the other off. Right? Now on to current limiting resistors. Since the gate capacitance is 1900 pf there will need to be current limiting resistors on A and B presuming that I'm driving them with a PIC. I saw this in the previous MOSFET thread here. Also will I require current limiting resistors between the MOSFET's and +12 and GND? Since the Ron is so low, it's possible to draw more than 50A (rated max current) through the MOSFET. Or do I not need to worry about it because the regulator at the far end cannot draw that much current? One last thing. To prevent both MOSFET's from conducting linearly together I plan to program the pic with brief delays where both MOSFET's are turned off. So instead of a AB pattern of 01 10 01 10 using 01 00 10 00 01 00 with the 00 periods much shorter than the power on periods. BTW this is where the PIC comes into play, using it to drive the MOSFETs in the above pattern. That's why I didn't lable the thread as OT. One last question. Can anyone explain the difference between using a Zener and a 7905 regulator? That's all I have so far. Just wanted to see if I doofed up anything really badly in the design before I put this together. Thanks, BAJ

On Wed, Dec 29, 1999 at 06:15:20PM -0500, Byron A Jeff wrote: > So I plan to use this config: > > +12 > | > D > A-->G > S > | > |-----| |------ > | > D > B-->G > S > | > | > gnd To activate MOSFET A you would need the gate to be 5V higher than the source. This is easy with MOSFET B because the source is at ground. But MOSFET A would only trigger if the voltage at source is at or below ground potential, right? > One last question. Can anyone explain the difference between using a Zener and > a 7905 regulator? As sole regulation? Zener would try to sink any excess current. 7905 would instead partially close output in order to avoid exceeding volts. I still can't figure out why a 7805 can't be used upside down. -- James Cameron spam_OUTquozlTakeThisOuTus.netrek.org http://quozl.us.netrek.org/

{Quote hidden}> > On Wed, Dec 29, 1999 at 06:15:20PM -0500, Byron A Jeff wrote: > > So I plan to use this config: > > > > +12 > > | > > D > > A-->G > > S > > | > > |-----| |------ > > | > > D > > B-->G > > S > > | > > | > > gnd > > To activate MOSFET A you would need the gate to be 5V higher than the > source. This is easy with MOSFET B because the source is at ground. > But MOSFET A would only trigger if the voltage at source is at or below > ground potential, right? Normally true except I'm using IRFZ44 logic level MOSFETs. They have internal circuitry so that they are fully on at 5V. > > > One last question. Can anyone explain the difference between using a Zener and > > a 7905 regulator? > > As sole regulation? Zener would try to sink any excess current. 7905 > would instead partially close output in order to avoid exceeding volts. So I guess I'll use a 7905. BAJ

Byron, my man. you are doing it the hard way. The charge pump chips I have used are suited for low currents. I know you are going to reinvent for more current, but why? 6+2.5 watts is not a place to start breaking ground on leading edge inductorless concepts. National simple switcher (and linear tech and others, for that matter) show an easy to build Buck-Boost that goes from +12 to -12 with a relatively easy cookbook circuit. And even though a flyback would be a fabulous solution, yes, you need to have them wound for you ($$$). The apps on the simple switchers show easy to get toroid inductors. They list Schott, Pulse, and Renco. I quickly crossed a 68uH 2A inductor to a Miller 5707 in Digikey. It is oversize, but what the hell, you are only building one. I see this app under LM2576, near the end. It shows a 700mA output, +12 to -12 converter. Build two of those, and you are in business! Remember, when you build it and it has a massive amount of ripple, to go back to the app note and study their recommended current path connection diagram. Following layout practices like they say will take a horendously noisy switcher and make it very manageable. The dead giveaway is when you ground the scope, then put the probe tip on the same ground point, and you get lots of nasty noise. If so, it is time to take an asprin and warm up your iron. Make the layout paths correct, and recheck. And with the home made cap converter, watch the RMS current through the reservoir capacitors. You must study the ESR and max ripple ratings. If you go over them, the cap heats up inside and becomes damaged over time, like about when you have your babe in the car flying down the highway on a sunny spring day and the stereo system goes (BBBBZZZZEEEEAAAAACCCCKKKKKK POP). Byron A Jeff wrote: > In the process of building a DC-DC converter for my car MP3 player I've > come across the need for -5 and -12 volt power. While I am aware of that > there are a number of single chip solutions, especially from Maxim, I'm > attempting to attack a capacitive switcher for reasons of understandability > and parts availablity. In short I have no idea where to get parts for or > how to wind flyback transformers. Also I wanted to practice/learn some > discrete design skills. PS real engineers don't use Maxim.

> > Byron, my man. you are doing it the hard way. I know. I know. But every time I start to look at inductors I get the willies. Every time I read a switcher datasheet, it seems to come up with an inductor component that's completely unavailable. The cross referencing you list below isn't a cut and dried activity. Also I'm trying to limit myself to stuff I can get in town here in Atlanta. Fortunately we have one of the few RadioShack.com stores which carries everything in their commercial catalog, including inductors. So I'm listening. {Quote hidden}> > The charge pump chips I have used are suited for low currents. I know you are > going to reinvent for more current, but why? 6+2.5 watts is not a place to > start breaking ground on leading edge inductorless concepts. > > National simple switcher (and linear tech and others, for that matter) show an easy > to build Buck-Boost that goes from +12 to -12 with a relatively easy cookbook > circuit. And even though a flyback would be a fabulous solution, yes, you need to > have them wound for you ($$$). The apps on the simple switchers show easy to get > toroid inductors. They list Schott, Pulse, and Renco. I quickly crossed a 68uH 2A > inductor to a Miller 5707 in Digikey. It is oversize, but what the hell, you are > only building one. I see this app under LM2576, near the end. It shows a 700mA > output, +12 to -12 converter. Build two of those, and you are in business! I see. I read that data sheet. Every other example show's an input voltage that's higher than the regulated voltage. > > Remember, when you build it and it has a massive amount of ripple, to go back to > the app note and study their recommended current path connection diagram. > Following layout practices like they say will take a horendously noisy switcher and > make it very manageable. The dead giveaway is when you ground the scope, then put > the probe tip on the same ground point, and you get lots of nasty noise. If so, it > is time to take an asprin and warm up your iron. Make the layout paths correct, > and recheck. I see that I can get the parts intown. Will take a crack at it. BAJ

As someone else noted - a Natsemi "simple switcher" or similar is very very easy to get going and probably better suited to this application. However - the FET drive totem-pole you have shown is OK BUT the gate of the top FET MUST be driven to eg 5 volts above its Source when it is turned on. As the source will be at +12 volts at this point then the gate must be driven to 17 volts (or more). For this you need a "high side" voltage source. A capacitor pump of sorts is one way of getting this :-) This is one reason to use the P channel upper FET - in that case you pull the gate down towards ground to turn the FET on so no high side drive voltage is needed. There are many app notes around showing high side supplies. One basic form works like this. - Connect capacitor Cpump side A to junction of 2 FETS. - Connect a diode, anode to +12V and cathode to Cpump side B. - Connect second diode, anode to Cpump side B, cathode to Vhigherstill. - Connect a reservoir capacitor from Vhigherstill to ground. When the FET pair output is low Cpump will charge to +12v less a diode drop. When the top FET turns on it will pull the FET junction towards +12V so the top of Cpump will be pulled to ABOVE +12v. Second diode will conduct raising Vhigherstill to about 2 x +12v. This is a low power diode pump being used to provide drive voltage for a high power diode pump :-). The first time its a bit flaky as the drive voltage is being made as the upper FET is turned on (bootstrapping) but once the top rail is at about 2 x +12v it's away. BUT - use a simple switcher :-) PS -Re someone's comment on Maxim. Real engineers will often find a cheaper solution than a Maxim part but anyone who will NEVER use Maxim is not a real engineer :-). Maxim make some very useful products. In small quantity use they are often very hard to beat. In volume use they compete with everyone else in the market and are sometimes the right solution, sometimes not. Russell McMahon _____________________________ >From other worlds - http://www.easttimor.com http://www.sudan.com What can one man* do? Help the hungry at no cost to yourself! at http://www.thehungersite.com/ (* - or woman, child or internet enabled intelligent entity :-)) {Original Message removed}