Hi everyone,
I am looking into building a high voltage charge pump doubler.
The idea is to convert 24 into 48V at 50Amp.
I am planing to use a PWM controller that will drive a capacitor in paralel with the imput to charge and in series to double the voltage at 10KHz. Never seen charge pumps so big, any advice? I am looking into protecting the mosfets against the big inrush current peak when the capacitor charges using some kind of current limiting to reduce the dutty cicles.
This is a uni project that I am doing. Using coils or transformers is not an option. It has to be a charge pump topology.
Thanks, Carlos
I am looking into building a high voltage charge pump doubler.
The idea is to convert 24 into 48V at 50Amp.
I am planing to use a PWM controller that will drive a capacitor in paralel with the imput to charge and in series to double the voltage at 10KHz. Never seen charge pumps so big, any advice? I am looking into protecting the mosfets against the big inrush current peak when the capacitor charges using some kind of current limiting to reduce the dutty cicles.
This is a uni project that I am doing. Using coils or transformers is not an option. It has to be a charge pump topology.
Thanks, Carlos
The high 10KHz will help you a lot. One way that is easy is to build a smaller voltage doubler that uses the conventional design with diodes. Then you connect a lot of these in parallel to get 50 amps.
In any case, I bet that you end up with many parallel components. I don't know how else to find a capacitor that could work. So the question is at what level do you implement parallelism? At the component or sub-circuit level?
I doubt you'll find either capacitors or FETs with the peak current capability to handle a charge pump of that size, and even if you could, the stray inductances in everything would most likely blow everything up. You'll have to run many circuits in parallel, and run them at staggered phases.
If it's for a university term project, you probably don't have to build the whole thing - instead, build two charge pump circuits capable of 48V at 1A, find a way to run them paralleled and out of phase to make 48V at 2A, and say "for the full design, we'll use 50 of these"
Any reason transformers aren't an option? Your best bet might be something like the attachment. Drive each side at near 50% duty cycle using a TL494 or SG3525, and use the biggest FETs you can find.
Also I'd read through the car audio forum on here, and look at some schematics for larger, 2KW+ amplifiers. Shouldn't be too hard to tweak a car audio power supply for 24V instead of 12V input.
If it's for a university term project, you probably don't have to build the whole thing - instead, build two charge pump circuits capable of 48V at 1A, find a way to run them paralleled and out of phase to make 48V at 2A, and say "for the full design, we'll use 50 of these"
Any reason transformers aren't an option? Your best bet might be something like the attachment. Drive each side at near 50% duty cycle using a TL494 or SG3525, and use the biggest FETs you can find.
Also I'd read through the car audio forum on here, and look at some schematics for larger, 2KW+ amplifiers. Shouldn't be too hard to tweak a car audio power supply for 24V instead of 12V input.
Attachments
It is difficult thing, but my idea would be: a CD4047 oscillating RC at 20KHz (pin13).
From pin 10 and 11, feed a IR2110, one per output. IR2110 drive a couple of hi power mosfet's (IRFP460 in example). Each driving a CP capacitor. At the other side of the cap, a couple of diodes, anodes to +24, and from cathodes, you'll get the 48V.
As the IR's and the IRF's are 180 out of phase, always is one supplying power to output, and ripple output is considerably lower. This is called iphase. Or, you can generate as phase as you want discretely or using a upc, and feeding IR2110 and IRFP460, one in each phase with the diodes. In such a way, reduces ripple and current per phase. An example, using a 4060 and a couple of gates, you can get 4 or 8 phases, and the same power stage. (4 rotated 90 deg, or 8 ph. rotated 45 deg, or n phases, or... 360 deg/n phases)
A bit more complex can be done using a synchronous rectifier in the other side of cap's, using more IR2110 and MOSFET's, but circuitry is more complex because the rectifier side is referred to +24 in place of gnd. This is my humble idea.
From pin 10 and 11, feed a IR2110, one per output. IR2110 drive a couple of hi power mosfet's (IRFP460 in example). Each driving a CP capacitor. At the other side of the cap, a couple of diodes, anodes to +24, and from cathodes, you'll get the 48V.
As the IR's and the IRF's are 180 out of phase, always is one supplying power to output, and ripple output is considerably lower. This is called iphase. Or, you can generate as phase as you want discretely or using a upc, and feeding IR2110 and IRFP460, one in each phase with the diodes. In such a way, reduces ripple and current per phase. An example, using a 4060 and a couple of gates, you can get 4 or 8 phases, and the same power stage. (4 rotated 90 deg, or 8 ph. rotated 45 deg, or n phases, or... 360 deg/n phases)
A bit more complex can be done using a synchronous rectifier in the other side of cap's, using more IR2110 and MOSFET's, but circuitry is more complex because the rectifier side is referred to +24 in place of gnd. This is my humble idea.
Thank you all for yout thoughts and ideas,
This is just a prototype.
We do not have coils for the power needed and I would like to do something never tried before. The Prius uses a Coil, for example. I have sucessefully reproduced the experiment using the AC Mains but could only get up to 2KW due to the low frequency and the low power factor.
The Real thing will run at 320V (250 to 390) and will double the output to 600V. This is for an Hybrid car.
The nominal output will be arround 20KW or 30Amps, half comes from the battery and half from the doubler (30V at 320V each)
I am going to use IGTBs rated at 600V and 100Amp Nominal.
The Power section will be two or three stages runing at 40% Dutty Cicle each, so they shold overlap themselves if using 3, if using only two the duty cicle will be about 80%. Capacitors will be Low ESR, probably 32V @ 1.000uF x 25
I just need to find a way to implement a current limiting device to allow the capacitors to charge slowly (1 or 2 seconds). Some kind of current limited PWM to 50Amps should do it.
A 0.001Ohm sense resistor would output 0.05V and dissipate 2.5W of power at 50Amps.
This is the basic system, but will have two or three groups as described before, to divide currents and reduce ripple.
For now I am planing to do a lower power version just to see if the system is Viable and if I have to do any considerable change into my design.
This is just a prototype.
We do not have coils for the power needed and I would like to do something never tried before. The Prius uses a Coil, for example. I have sucessefully reproduced the experiment using the AC Mains but could only get up to 2KW due to the low frequency and the low power factor.
The Real thing will run at 320V (250 to 390) and will double the output to 600V. This is for an Hybrid car.
The nominal output will be arround 20KW or 30Amps, half comes from the battery and half from the doubler (30V at 320V each)
I am going to use IGTBs rated at 600V and 100Amp Nominal.
The Power section will be two or three stages runing at 40% Dutty Cicle each, so they shold overlap themselves if using 3, if using only two the duty cicle will be about 80%. Capacitors will be Low ESR, probably 32V @ 1.000uF x 25
I just need to find a way to implement a current limiting device to allow the capacitors to charge slowly (1 or 2 seconds). Some kind of current limited PWM to 50Amps should do it.
A 0.001Ohm sense resistor would output 0.05V and dissipate 2.5W of power at 50Amps.
This is the basic system, but will have two or three groups as described before, to divide currents and reduce ripple.
For now I am planing to do a lower power version just to see if the system is Viable and if I have to do any considerable change into my design.
Each power stage is composed of 4 mosfets and a capacitor that is placed in series and paralel with the imput, with suitable driving logic for the gates.
When I mean paralel I mean several of this circuits feeding the output capacitor.
Lets say bank one is charging, bank two will be powering the output and so on. This would raise the dutty cicle from 40 to at least 80% with two phases.
Osvaldo,
Thats exactly what I am going to do, and using the IR2110. I am not sure about the 4047, I may opt to use something like an SG3525 or similar, but I would be limited to two outputs. Also some method of reducing frequency needs to be studied to save some power charging the gates at low outputs.
Thanks
The CA/CD/HCF/HEF4047 is a RC oscillator (pin 1, 2 and 3) with a direct oscillator pin (13) and two complemented 1/2 frequency (10 and 11). The other pins configure the IC as mono/bi/as/stable and go to fixed voltages. As all CMOS, power consumption is very low, and the opamps and analog circuit inside SG´s is unnecessary for you, charge pumps can't be PMW regulated nor stabilized.
Google for the datasheet, CD/HCF 4047 I had used very much and is very useful oscillator IC, RC can be set up to 1MHz!
Reducing frequency reduces wasted power in gate, but may cause interferences in the audio program, so don't go too low!
Good luck!
Hi Coloumb,
I am familiar with this IC, I have even thought about using a 4017 with something similar in case I needed further outputs, however I need PWM Regulation, otherwise the switches are going to blow in case of a short circuit if they dont first upon charging the caps. Whats your advice on this?
I am familiar with this IC, I have even thought about using a 4017 with something similar in case I needed further outputs, however I need PWM Regulation, otherwise the switches are going to blow in case of a short circuit if they dont first upon charging the caps. Whats your advice on this?
- Status
- Not open for further replies.