DIY Electric Car Forums banner

New EV Charger Design - Modular

43547 Views 68 Replies 14 Participants Last post by  PStechPaul
After delving deeply into the EMW 12 kW DIY charger, I think it would be best to make a new design. My concept is to make modules, each of which can be used on 120 or 240 VAC single or three phase, or up to 300 VDC. The modules would be 1.2 kVA for 120 VAC, and 2.4 kVA for 240 VAC or 300 VDC. They will be capable of being connected in parallel to obtain higher power. I think these modules could be built for a parts cost of less than $150 each. ;)

The IGBT I show here is an ultra-fast 35A 600V device that is designed for switching applications up to 100 kHz, so I think the inductor and capacitor size and cost may be greatly reduced. And this part is only about $1.50. :)

Here is a "first shot" at this design. It has been done using Mentor Graphics PADS 2004 and most of the parts are fully characterized with part numbers and approximate cost, as well as PCB decals so that a board can be made directly from the schematic. A BOM in Excel format can also be produced easily with a VBA script.

Here is a PDF which is a little easier to read:

And the BOM (preliminary) showing total parts cost less than $150:

This version is not PFC and non-isolated. It also has only a single pushbutton for start/stop, does not have a BMS connection, and has no display. But it has a serial port which can be connected to a Bluetooth module for viewing and logging data, and for commands. I am using only a 14 pin PIC16F1825 but it will probably need a 28 pin processor to provide the additional I/O needed. I have an Arduino UNO and I will try to add the connections to match its pinouts. I might also see if I can adapt the EMW control board with its display and function keys.

I am putting together an order to Mouser for some of the parts I will need for this design, and I am also going to get an AVR Dragon which is a $53 emulator/debugger/programmer which is really the very minimum needed for development of any serious design with the Atmel series microcontrollers. I plan to use this for analyzing the EMW charger (when I get a complete unit or the boards and parts needed), and hopefully be able to provide recommended modifications to the hardware and firmware to improve the reliability and performance. I think it may require a complete new set of boards, but many of the expensive components may be able to be re-used.
See less See more
61 - 69 of 69 Posts
Mud on my face. Missed a few things.

EMW LM 211. Probably not hysteresis control. Ordinary PWM output to buck IGBT from Arduino via LM211 strobe input.
Comparator inputs seem to be overcurrent protection.

LLC. As I mentioned it might be good idea to use SiC in all cases. I can now say it's definitely a good idea. And a buck output stage is necessary. Only ordinary Si diodes in the LLC output bridge.

Although the DC gain is load indepedent at resonance, the Q factor is not. The extremely low reflected AC resistance sends the Q factor sky high (>2000). Making frequency setting close to resonance at high gain a major challenge.
The buck stage solves this issue (Q ~7 at max load). SiC doubles the range for frequency setting, since SiC has no high switching losses in the ZCS region below resonance.
I was going to mention that the EMW design seems to use the LM211 comparator only as an output current limiter, and does not measure peak inductor current for cycle-by-cycle switching control. I think this is a serious flaw but I don't plan to do anything about it in my initial retrofit. The current sensor is hard wired to the power PCB and I would like to be able to re-use it essentially as-is. But I might add a current sensor with one lead of the output inductor passing through it, for fast PWM shut-down as well as possibly implementing cycle-by-cycle control and waveform monitoring.

As for the LLC topology, I had to refresh my memory as to its details, and I see that it is a resonant circuit with two frequencies, determined by a tank capacitor and two inductors, one of which is contained in the output transformer. I found an app note that pretty well explains many topologies:

And here is a discussion of a digitally controlled LLC:
See less See more
The microchip paper is a great overview of SMPS topologies. calls its application note AN2644 an introduction to LLC. But it is in fact a very detailed explanation.

Including mosfet losses, ZCS-ZVS, graphs, step by step time analysis, and a very important conclusion:
ZCS on the secondary side is inherent to LLC in all states of operation.
No need for SiC there. :)
Digital control is an excellent option for low cost, high volume applications.
The content of the paper looks rather complex: models, bode plots, z-transformations.
And that's for a simplified case: a fixed load (LED TV, I believe).
Fortunately, for limited number DIY charger builds, it's even simpler.
There's no need for complex modelling (if it possible in the first place) or z-transformations (high Q factor makes feedback regulation very challenging).
Instead, just keep the operating frequency close to resonance for maximum (open loop, load independent) gain and let a dedicated HW controller handle everything else: dead time generation, overcurrent protection in three stages, overvoltage and undervoltage protection, soft start, HV bootstrap half bridge driver, and more. All in a $2 16 pin package.

A buck output stage regulates by hysteresis control. Great for DIY. Cycle by cycle accuracy and there's no need for analysis in the frequency domain.

For PFC, I can't see a way around frequency compensation. Anyone?
See less See more
I am ordering a couple Microchip devices:

and dsPIC33EP:

They appear to be virtually identical. I had thought to use them for a three phase VFD motor control but it looks like they can also be used for PFC and a buck converter suitable for a battery charger:

Has anyone used these or similar devices? Do you know what the difference is between the PIC24 and dsPIC33?
See less See more
I haven't used PICs yet, but I've taken a look at their range of chips in the past. I remember that the ds stands for digital signal.
The info on their website confirms that dsPIC is a range of digital signal processors. Here's the application note for a IPFC:

Complete with software, z-transformations and digital PI controllers.

I'd go for the dedicated Infineon ICE3PCS01G with all functions onboard in a single 14 pin package, including a digital voltage control loop.
It can be synchronised with a PWM channel of a MCU for interleaved operation.
Far less external HW and SW necessary and a quick turnaround.
This project was put on hold 4 years ago. It has not been built or tested except in simulation, and the only schematic (PDF) is the one linked above in a previous post. You are welcome to take it from there.
61 - 69 of 69 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.