Mini-QC Rapid-Charger (RC) Project for LEAF QC Port

[klapauzius]

Yes, communication over standard wired network would be nice. Wifi can be unreliable at times (maybe just for the Blink units....).

 

[garygid]

If someone has working interface and measurement
circuit schematics to share, that would speed us along
towards having a common proto board, or boards.

I would be happy to not re-invent things that others
already have tested, or to evaluate, and possibly
test good suggestions. Specific part suggestions for
the Due 3.3v environment, and well thought out circuits
with tested component values would be most helpful.

Perhaps a second proto board for the QC interface and CAN bus,
with the lower proto board used for controlling/measuring the
power supply, with all the chips that attach the high voltage
on a separate high-voltage interface board, "remote" from
the Due stack.

To be useful, inrush current needs to be limited.
To be evaluated, logging must be included.

 

[klapauzius]

Not sure how precise the temperature has to be. Would a digital 1-wire themormeter like the DS18b20 do?
I have a working circuit for AC current measurement using a current transformer.
Together with the voltage you can have real-time current and power factor. Using a multiplexer, this just uses one ADC input.
It eats some clock cycles though, as you have to read a few samples of the AC current/voltage.

 

[valerun]

If someone has working interface and measurement
circuit schematics to share, that would speed us along
towards having a common proto board, or boards.

I would be happy to not re-invent things that others
already have tested, or to evaluate, and possibly
test good suggestions. Specific part suggestions for
the Due 3.3v environment, and well thought out circuits
with tested component values would be most helpful.

Perhaps a second proto board for the QC interface and CAN bus,
with the lower proto board used for controlling/measuring the
power supply, with all the chips that attach the high voltage
on a separate high-voltage interface board, "remote" from
the Due stack.

To be useful, inrush current needs to be limited.
To be evaluated, logging must be included.

I will get the schematics up later this week. All interfaces are there.

Current transformers are fine for AC but output current is mostly DC. Having different parts to measure different currents means BOM is longer and it bites you on total parts costs in the end as you can't order in volume. Standardization on universal sub-components is key. If we get the board designed that can truly be used as a universal power control board, we can manufacture it in 100s for less per unit that one would spend on components alone to build one by themselves.

 

[garygid]

Valery,
Splendid, it is very kind of you to share with us.

I look forward to understanding your circuit suggestions.

Safely masuring a higher voltage that could go negative,
like HV+ to ground, seems to require some attention.

Have you already designed or even made a custom
proto/control shield board for the Due?

Maybe we should be working from that, instead of trying
to build on these limited-space proto boards?

Your suggestions are welcome, thanks (spaciba??)

 

[garygid]

If one already has an Analog to Digital multiplex capability,
the TMP36 - Temperature Sensor, $1.50 from Sparkfun, might
integrate better.

However if you already have the one-wire communication
in hardware, then the digital temperature one-wire sensor
(about $5) might be good, but is it slow converting, at
about 750 ms per sample?

 

[garygid]

Any suggestions for handling inrush current?

The data sheet for the PFC chip (used in the power supply
kit that we bought) seems to indicate that it can handle
limiting the inrush current. However, it appears that it
does not do that, at least in the circuit supplied

Is this chip actually have that capability?

If so, any ideas on why it is not working?

Thanks, Gary

 

[klapauzius]

If one already has an Analog to Digital multiplex capability,
the TMP36 - Temperature Sensor, $1.50 from Sparkfun, might
integrate better.

However if you already have the one-wire communication
in hardware, then the digital temperature one-wire sensor
(about $5) might be good, but is it slow converting, at
about 750 ms per sample?

750 ms seems to be the max conversion time for the 1 wire sensor, so it is slow. Do we need faster conversion?
Another concern would be how to attach any of these to a heatsink properly.

 

[klapauzius]

Current transformers are fine for AC but output current is mostly DC. Having different parts to measure different currents means BOM is longer and it bites you on total parts costs in the end as you can't order in volume. Standardization on universal sub-components is key. If we get the board designed that can truly be used as a universal power control board, we can manufacture it in 100s for less per unit that one would spend on components alone to build one by themselves.

Yes, I agree.
I think e.g. AC input current measurements could always be added later, if e.g. the board that hosts the arduino controller would allow "add-ons". I dont think we are going to use all of the arduinos ADC pins in the "basic" design.

The beauty of an open source/open design powersupply is that it would allow such add-ons for later.

 

[garygid]

Some might be getting close to having a 440v (or more) power supply
that could produce somewhere between something like 10 amps
and around 30 amps.

So, what should we check before plugging it into an expensive car?

First, normal safety precautions and operation, like
Is there suitable wiring and breaker in place.
Can it be plugged in without blowing the breaker.
Is there an Off button?

Assuming you will be able to get a suitable QC type plug,
and pins for it, and are planning to have suitable wiring:

1. The initial handshake (4 control lines) is not difficult and
has pretty much been disclosed in public literature.
Read and understand this literature, and try to
understand what is missing or not fully specified.
The following is an example of available info.

The interface:
http://chademo.com/05_interface.html" onclick="window.open(this.href);return false;
The protocol:
http://chademo.com/05_protocol.html" onclick="window.open(this.href);return false;

You will observe that one does not get very far
before CAN communication is needed.

However, you can begin testing your "charger".

If you have your CAN Logging working, and you are
also logging pseudo-CAN messages that contain the
power supply vitals, you can run some tests on your
"Rapid-Charger" to see if it is ready for "prime time".

The pseudo message would have the power supply's
output voltage (before the output diode), its current
(not much at this point unless you have a dummy load),
its vital temperature, and its pulse width modulation
percentage index.

Using 2 bytes for the voltage, and possibly two for the
modulation index, one for the temperature, and one
for the current, you have 2 remaining, which we will
use later as we start to monitor the interface signal
states. I will suggest a specific format later.

However, you can choose the format to start logging
your power supply's basic operation. Later, we will
add another pseudo-message as we come to want it.

Now, pretending that you were connected to the car, it
would tell you its voltage requirement, and you would
agree, or pack up and go home. After you tell the car
how much current you can supply, and other negotiations
are comleted, the car will indicate that you should
ramp up the voltage, hold briefly, and ramp back down.
You can observe this voltage on the display of most
real QC machines. Yes, the timing is not specified, but
you can get a feeling for what is acceptable to the car
by watching a few real QC sessions.

So, test your candidate charger by having it do the voltage
display, and log the results. When successful, wait a few
seconds and have your "charger" ramp the current up,
into a dummy load, and hold it constant for a while, then
push the Stop button to terminate the current flow
as quickly as possible, logging the entire test.

If you can, add monitoring the input current to the
log data, and run the test again, logging from the initial
moment of plugging in. Using my CAN-Do on the PC,
and possibly my Sketch that supports some of this,
but will support all of it as we prgress,
graph the data and show us the results.

 

[davewill]

Yes, communication over standard wired network would be nice. Wifi can be unreliable at times (maybe just for the Blink units....).

If people want WiFi or Ethernet over power line, they can buy the appropriate adapter and plug it into the Ethernet.

 

[valerun]

Valery,
Splendid, it is very kind of you to share with us.

I look forward to understanding your circuit suggestions.

Safely masuring a higher voltage that could go negative,
like HV+ to ground, seems to require some attention.

Have you already designed or even made a custom
proto/control shield board for the Due?

Maybe we should be working from that, instead of trying
to build on these limited-space proto boards?

Your suggestions are welcome, thanks (spaciba??)

Thanks Gary -

to measure higher voltage going both ways we have a simple divider going into the A7520 isolated op amp chip. It handles -0.2-+0.2V range on its inputs so we design the divider mapping into that range. Works great. We have tried at least 4 different solutions before standardizing on this as the most robust, cheap, and universal. We tried: hall sensors loaded through resistors (too much dissipation), 1:1 isolator chips (too expensive, bad noise immunity), other op amp chips (inconsistent noise immunity, sub-optimal input ranges, slow).

In one of the posts above I have shared the pictures of the actual board that we built. It does not have ALL the MVS attributes I mentioned but it has ~90%. We are making a second revision now and I would like to include mods to cover rest of MVS attributes. Hence call for input. We can include some proto space on that board, as well - let me know if you think that would be a good addition.

Re PFC inrush - yes, the chip has ramp control but that applies only to how fast the voltage is raised from the input peak voltage to the target PFC output. The chip cannot control the rise from zero to peak input value. That would have to be done with either inrush resistors (our approach - a poor man's precharge circuit) or dedicated pre-charge circuit that would also command 'chip enable' signal when ready. In 25kW units, we will have to go with the proper precharge unit (already designed, built, and tested) as there are no inrush resistors capable of handling 100A current levels. For 12kW, I suggest to stick with inrush resistors for now.

V

 

[garygid]

If I knew what was on your board, I might possibly be able to
suggest additions. For that purpose, a schematic would be
very helpful.

Do you have a CAN transceiver with probably 4 resistors,
a 120, two 5k, and one 10k, or is the QC interface stuff
going on another board?

 

[garygid]

The A7520 looks good, but only 5% accuracy.
The A7510 is better at 3%, but will still need calibration.

Looks like it would be good for current measurement,
across a shunt, assuming the input sensitivity is sufficient
for something like a 50 millivolt shunt.

Nte: Some folks have managed to use the EEPROM
in the other uP on the Due board.

Does your control board for the Due use the pins used
by the ColdTears display Shield for the Due?

If so, maybe we could re-assign the conflicting pins?

 

[valerun]

The A7520 looks good, but only 5% accuracy.
The A7510 is better at 3%, but will still need calibration.

Looks like it would be good for current measurement,
across a shunt, assuming the input sensitivity is sufficient
for something like a 50 millivolt shunt.

Nte: Some folks have managed to use the EEPROM
in the other uP on the Due board.

Does your control board for the Due use the pins used
by the ColdTears display Shield for the Due?

If so, maybe we could re-assign the conflicting pins?

good question re pins. can you pm me your email so I can send you some files?

re shunts - I don't like those things - they are noisy and hot. I suggest sticking with Halls. Again, that's what all OEMs use.

re 5% accuracy - yes, for total precision, you need to calibrate. The thing is, nothing is perfect and you would need to calibrate your sensitive circuits anyway. Resistors have spread, AREF on the MCU has biases, etc. Again, we went through that journey, as well, and decided that per-unit calibration is the most economical and predictable way to deal with all these in one step. Connect some constant voltage to the output, see what the unit tells you and adjust per what you see on your trusted multimeter. With what is going to be a $2000-3000 device, per-unit manual calibration 'cost' is trivial IMO.

 

[valerun]

If I knew what was on your board, I might possibly be able to
suggest additions. For that purpose, a schematic would be
very helpful.

Do you have a CAN transceiver with probably 4 resistors,
a 120, two 5k, and one 10k, or is the QC interface stuff
going on another board?

yes, CANs are on board. Right now, just tranceivers with terminating 120R, selectable by jumpers. What are 5k & 10k for?

 

[surfingslovak]

While reviewing Uchen's product lineup earlier tonight, I couldn't help noticing that they apparently carry a CHAdeMO-compatible plug now. Perhaps it's worth looking at? If I recall correctly, Valery mentioned that he already was communicating with them, but I'm not sure if everyone was aware.


Click to open

 

[Ingineer]

That's not CHAdeMO, it's some strange Chinese standard.

-Phil

 

[surfingslovak]

That's not CHAdeMO, it's some strange Chinese standard.

Ah, thank you, Phil! Strangely, they do specifically list CHAdeMO, but use the same odd plug photo there too. Hmm. The German EV drivers I interacted with on goingelectric.de mentioned that the Chinese level 2 plug looks like a clone of the European Type 2 (Mennekes). It's not an identical copy, but it does resemble the original. They thought that this was fairly typical of the approach observed in China lately. Perhaps they followed the same pattern with their level 3 plugs?


Click to open

 

[garygid]

In about 25 countries, not China, a Chademo plug-in end resembles this:
http://en.wikipedia.org/wiki/File:CHAdeMO_Plug_VacavilleDavisStDC2.jpg" onclick="window.open(this.href);return false;

Our Jolomo plug is a bit different overall, but seems to plug into
the LEAF's QC socket, and at least work for careful, dry, indoor
Rapid-Charging.