Would something with the dimensions of the Tesla QC adapter work? It has the QC socket, but you would have to replace its Tesla plug with a CHAdeMO plug. It might also have room for a circuit board. Most likely not the cheapest approach.garygid said:
Mini-QC Rapid-Charger (RC) Project for LEAF QC Port
- Thread starter garygid
- Start date
Help Support My Nissan Leaf Forum:
garygid
Well-known member
tb...,
I had understood that the Tesla QC adapter could be ordered
for about $1000 by current owners, or obtained for $500 if
ordered along with a car. I wonder if one could order a
spare in case theirs got stolen or vandalized?
If so, I would encourage someone to do order a spare
or backup adapter, along ith their car.
Hopefully somebody with their opposing digit in the Tesla pie
will soon be able to verify the availability of this Adapter.
It will be interesting, and possibly even informative to see
one, and have a chance to observe it in operation.
Presumably, this Tesla adapter must simulate being both
a QC vehicle, and being a Tesla Supercharger.
A QC Logging adapter would allow such an investigation
to be done easily, at least on the QC end of the adapter.
Note that it would also be an interesting experiment to
monitor the electrical activity at the SC end of the adapter,
which could be facilitated with a SC-to-SC Logging Adapter.
It is even more interesting to monitor the Tesla SC
connection if it is supposed to be very close to the
proposed SAE L3 DC standard. Anyone with more
non-proprietary insight on that supposition?
Cheers, Gary
I had understood that the Tesla QC adapter could be ordered
for about $1000 by current owners, or obtained for $500 if
ordered along with a car. I wonder if one could order a
spare in case theirs got stolen or vandalized?
If so, I would encourage someone to do order a spare
or backup adapter, along ith their car.
Hopefully somebody with their opposing digit in the Tesla pie
will soon be able to verify the availability of this Adapter.
It will be interesting, and possibly even informative to see
one, and have a chance to observe it in operation.
Presumably, this Tesla adapter must simulate being both
a QC vehicle, and being a Tesla Supercharger.
A QC Logging adapter would allow such an investigation
to be done easily, at least on the QC end of the adapter.
Note that it would also be an interesting experiment to
monitor the electrical activity at the SC end of the adapter,
which could be facilitated with a SC-to-SC Logging Adapter.
It is even more interesting to monitor the Tesla SC
connection if it is supposed to be very close to the
proposed SAE L3 DC standard. Anyone with more
non-proprietary insight on that supposition?
Cheers, Gary
valerun
Well-known member
All - a bit of update on isolation stage.
PCBs for the initial batch of these are in - see the photo. Yes, this is ~30 of them ;-) Turned out ordering 30 is about the same price as ordering 15. Go figure. These are heavy duty, 4oz copper, 0.1" thick boards so they don't flex and pass a lot of current. We will expect to clock these at 25kW peak power.
On the board (from left to right on the horizontal one):
1. Isolation stage power board
2. Controlled PFC stage control board (with 1 current and 2 voltage channels, 20A IGBT drivers, hardware PFC on board, etc)
3. Precharge & sensor board (2 current & 2 voltage channels - fully isolated)
Will be assembling & testing a couple of units based on these in the next few days.
Thanks,
Valery
PCBs for the initial batch of these are in - see the photo. Yes, this is ~30 of them ;-) Turned out ordering 30 is about the same price as ordering 15. Go figure. These are heavy duty, 4oz copper, 0.1" thick boards so they don't flex and pass a lot of current. We will expect to clock these at 25kW peak power.
On the board (from left to right on the horizontal one):
1. Isolation stage power board
2. Controlled PFC stage control board (with 1 current and 2 voltage channels, 20A IGBT drivers, hardware PFC on board, etc)
3. Precharge & sensor board (2 current & 2 voltage channels - fully isolated)
Will be assembling & testing a couple of units based on these in the next few days.
Thanks,
Valery
Attachments
jclemens
Well-known member
valerun said:
What would happen if we put rectified 240VAC into this?
otherwise, is your PFC stage in your 12kw charger modular enough to include in this new one? can it do at least 17kw?
As for an update from me, I know it has been slow. But I have been making progress.
The female end of the "15 inch extension cord" is essentially done, what is holding things up is the complete re-design of the main body of the male end. This my version 2.0 and will include gland nuts as strain relief and a pistol grip.
valerun
Well-known member
jclemens said:
very good questions.
1. Yes you can put rectified 240VAC into this - or rectified 3-phase (subject to voltage limits on components, of course). The transformer would have to be wound a bit differently as your would need higher transfer ratio (1:1.6 instead of 1:1.2) as you are feeding ~330VDC peak into the stage. The main disadvantage will be much higher peak current for the same power output - with corresponding increase in heat losses. These would be most pronounced in the transformer as winding losses are quadratic to current. Since the transformer happens to be the most thermally stressed component at this point, this will result in lower max power. Still - you will probably be able to push 12-13kW through a transformer designed for 20kW. Another disadvantage (at least for now) will be lack of output current control as the isolation stage is NOT controlled at this point. The current WILL be naturally limited by the nature of the transfer function of the iso stage (higher current means longer reset of the leakage inductance which means longer deadtime when no energy transfer takes place which means transfer ratio falls as current rises therefore naturally limiting max current). However, getting that limit right will likely require precise turns ratio tuned for a particular battery spec (e.g., transformer for a Leaf would be different from transformer for RAV4, etc).
2. Better approach is to use a controlled PFC stage. In our 12kW units, PFC stage is integrated with buck output stage and has a fixed output voltage. Our new 20kW controlled PFC stage is different and can vary the voltage output across the required range. We will be processing 5 initial orders for 'beta' versions of these next week (2 spots still open). One could take one of those and pair it up with an isolation stage to get an isolated 20kW charger. As you remember from one of the previous posts, we will be taking 3 initial orders for 'beta' versions of the isolation stage next week, as well.
So in a nutshell, starting next week you should be able to get an isolated 20kW charger hardware for ~$4,000. Or, at ~$2,000, you would be able to 'hack' things together with a rectifier directly feeding the isolation stage. In this case, you might need to tweak transformer's turn ratios / wound your own transformer.
Thanks,
V
valerun
Well-known member
hi Guys - a quick but exciting update.
In the last few days, we have tried a number of different core options for the main transformer. Lots of samples FedEx'ed from China, at least 20lbs of copper, 5 rolls of Kapton tape, etc. ;-)
Our latest transformer design just completed a 30-minute 15kW run at 330V output with steady-state transformer temperature at just 20C above ambient. The core is still very far from saturation at 330V so there's a pretty good chance that we can push 20kW at 420-440V output through this same setup.
Next up is packaging this into our 10x10x8" standard box and wiring up the new control board.
Gary - any update on the QC side?
V
PS. a couple of pics for you - the new transformer and the assembled power PCB for the packaged version of the isolation stage.
In the last few days, we have tried a number of different core options for the main transformer. Lots of samples FedEx'ed from China, at least 20lbs of copper, 5 rolls of Kapton tape, etc. ;-)
Our latest transformer design just completed a 30-minute 15kW run at 330V output with steady-state transformer temperature at just 20C above ambient. The core is still very far from saturation at 330V so there's a pretty good chance that we can push 20kW at 420-440V output through this same setup.
Next up is packaging this into our 10x10x8" standard box and wiring up the new control board.
Gary - any update on the QC side?
V
PS. a couple of pics for you - the new transformer and the assembled power PCB for the packaged version of the isolation stage.
Attachments
philipscoggins
Well-known member
Wow!
You guys are actually doing it!
(Cheering from the sidelines)
You guys are actually doing it!
(Cheering from the sidelines)
valerun
Well-known member
philipscoggins said:Wow!
You guys are actually doing it!
(Cheering from the sidelines)
;-) yes we can! ;-)
some more test data for the iso stage. Format: input V, input C -> output V, output C -> eff %:
Vin Cin Vout Cout Eff Pin, kW Pout, kW
317 28 323 26 95% 8.9 8.4
331 52.5 329 50 95% 17.4 16.5
333 57 332 54 94% 19.0 17.9
336 61 333 57 93% 20.5 19.0
You can see the transfer ratio dropping as the power increases - from 1.02 to 0.99. The drop is much smaller than with previous transformer due to much smaller leakage inductance.
We think we can gain another 0.5-1% in efficiency by doing a couple of things to switching transitions to take advantage of smaller leakage inductance
Note that the above results mean that at the Leaf's higher voltage levels, 60A output shown above will mean ~25kW!
We were able to run continuous at 16kW output and could probably be continuous at 19kW but we don't yet have proper overheating protection set up on this bench unit so don't want to push it. Will see what we can do once the system is packaged into the box with defined airflow and all sensing properly wired.
Thanks,
Valery
garygid
Well-known member
Valery,
Many thanks for keeping us posted.
It will be very interesting to see what a complete
(AC to regulated DC) unit will be able to do.
Cheers, Gary
Many thanks for keeping us posted.
It will be very interesting to see what a complete
(AC to regulated DC) unit will be able to do.
Cheers, Gary
valerun
Well-known member
garygid said:
Gary - just multiply the above efficiency numbers by 97% to get AC input ->DC output efficiency. You will get 90% at 20kW. I believe we will be able to get 91-92% with the tweaks I mentioned.
So any word on when will might be able to try QC control code on this?
Valery.
garygid
Well-known member
When will a complete unit be available for indoor testing?
We would try around 12 kW first, or about 30 amps max, right?
Exhausting 1 to 2 kW with air ventilation will be a good space heater?
What ambient air temps have you used for testing the cooling?
Are you controlling with the Arduino Due yet?
Are you logging data to SD card or a PC yet?
Are the control circuits and boards for the whole unit available yet?
At this point, we are only controlling the final stage PWM.
About 0 to 5000 out of 6000, which is 14 kHz on the Due.
Will your isolated unit need additional control?
Whenever you have compiled and run the early Due version
that I sent to you, let me know and I will send another version.
I can well imagine that you are very busy.
Cheers, Gary
We would try around 12 kW first, or about 30 amps max, right?
Exhausting 1 to 2 kW with air ventilation will be a good space heater?
What ambient air temps have you used for testing the cooling?
Are you controlling with the Arduino Due yet?
Are you logging data to SD card or a PC yet?
Are the control circuits and boards for the whole unit available yet?
At this point, we are only controlling the final stage PWM.
About 0 to 5000 out of 6000, which is 14 kHz on the Due.
Will your isolated unit need additional control?
Whenever you have compiled and run the early Due version
that I sent to you, let me know and I will send another version.
I can well imagine that you are very busy.
Cheers, Gary
valerun
Well-known member
garygid said:
Thanks Gary.
What we will have next week (fingers crossed) is a system that can do 20kW continuous into 400-450V Leaf pack. Yes, probably a good practice to try at lower power first but hardware will be able to handle up to 20kW. It will have thermal protection so you won't be able to blow it up.
The tests we reported on above were done with our regular control boards (Pro Mini based). The control and sensor boards are ready and being integrated now to form a complete system. What it will be capable of doing next week is taking serial input commands and respond to them very quickly. The serial comms part is already there in the latest firmware for our 12kW units so it will be easy to port and it has been tested already. The 'very quickly' part is what we are working on now and will be based on proper PID loop control.
This brings me to a fundamental proposal of how we work together. I suggest that we separate the work into 2 modules - power hardware & software and control hardware & software. We will take care of the former, you - of the latter. The API in between will be serial based. We will guarantee response time. Otherwise it will be very hard to have 2 groups working on the same thing - controlling these power stages is not trivial and isolated control is substantially harder than non-isolated. I suggest we deal with all the complexities on our end and give you a simple interface to command what to do.
If you agree to this concept, we need to work out the command set that would be sufficient. My initial list is:
1. Max voltage - ramp to max voltage of 450V or so and stay there
2. Zero voltage - ramp down to 0v and stay there
3. X Amps - ramp and maintain X A output current
This approach will get us the working system in the shortest time. We will eventually integrate both stage on a single Due board.
Thoughts?
Thanks,
Valery.
garygid
Well-known member
Sounds like a good start.
Perhaps add these commands:
4. Stop (get to 0 volts and 0 amps ASAP)
5. Sleep (power down, essentially Off)
6. Wake up (power up, essentially On)
7. Request Status (temperature, etc.)
8. Set and Get the Max DC Voltage
9. Set and Get the Max DC Current
Command Timing:
Accept commands at least 10 per second,
perhaps more, at almost any time?
Are you suggesting any synchronizing,
or just asynchronous communication?
Under normal circumstances, for the
2011 Leaf, the current-request commands
arrive 10 times per second, and usually
rise no faster than 20 amps per second.
Or, so we have observed.
The fall to zero current can be abrupt.
Feedback:
Some response to each command, at least
accepted or not?
Provide at least the output Volts and Amp,
perhaps as a response to each command?
Really, we would want to log those values
at least 100 times per second, but we need
to send the values to the car at least 10
times per second.
By Serial, do you mean RS232 serial?
Or, CAN message communication?
During the voltage ramp test, measure
current for any unexpected flow, as part
of a safety ground fault check.
Perhaps provide the ground to HV+ voltage
so we can log the HV symmetry. If the car
objects, we will want to have enough data
logged to be able to determine what went wrong.
Perhaps include the duty cycle being used
by the final regulator stage, again, to help
with the debugging?
The first amps-request might need to be
100 ma, or less, to get the output voltage
up to the battery voltage.
The output will, most likely, need a diode
on the output so that no reverse current
will flow when the battery is connected
to the power supply.
Good to hear that you are getting a unit
together for doing some real testing.
Are you getting a QC cable and plug
assembled to use for the testing?
The QC controller will need a clean 12v
power supply, and possibly a 5v supply.
Tomorrow I will try to get these ideas
and plans reviewed by others.
Cherrs, Gary
Perhaps add these commands:
4. Stop (get to 0 volts and 0 amps ASAP)
5. Sleep (power down, essentially Off)
6. Wake up (power up, essentially On)
7. Request Status (temperature, etc.)
8. Set and Get the Max DC Voltage
9. Set and Get the Max DC Current
Command Timing:
Accept commands at least 10 per second,
perhaps more, at almost any time?
Are you suggesting any synchronizing,
or just asynchronous communication?
Under normal circumstances, for the
2011 Leaf, the current-request commands
arrive 10 times per second, and usually
rise no faster than 20 amps per second.
Or, so we have observed.
The fall to zero current can be abrupt.
Feedback:
Some response to each command, at least
accepted or not?
Provide at least the output Volts and Amp,
perhaps as a response to each command?
Really, we would want to log those values
at least 100 times per second, but we need
to send the values to the car at least 10
times per second.
By Serial, do you mean RS232 serial?
Or, CAN message communication?
During the voltage ramp test, measure
current for any unexpected flow, as part
of a safety ground fault check.
Perhaps provide the ground to HV+ voltage
so we can log the HV symmetry. If the car
objects, we will want to have enough data
logged to be able to determine what went wrong.
Perhaps include the duty cycle being used
by the final regulator stage, again, to help
with the debugging?
The first amps-request might need to be
100 ma, or less, to get the output voltage
up to the battery voltage.
The output will, most likely, need a diode
on the output so that no reverse current
will flow when the battery is connected
to the power supply.
Good to hear that you are getting a unit
together for doing some real testing.
Are you getting a QC cable and plug
assembled to use for the testing?
The QC controller will need a clean 12v
power supply, and possibly a 5v supply.
Tomorrow I will try to get these ideas
and plans reviewed by others.
Cherrs, Gary
jclemens
Well-known member
garygid said:
I am still making progress on the plug redesign. The 3 holes at the back are for cable glands that will act as strain relief for 2 power cables an 1 data cable.
valerun
Well-known member
Sounds great Gary. All doable. RS232, yes. Will work on getting this API implemented next week. I think this modularization will help us minimize 'time-to-market' ;-)
If anyone has a good source for QC cable+connector assembly, we could use some (Joel is helping us get one but I think we could use more than one here).
Thanks,
Valery
If anyone has a good source for QC cable+connector assembly, we could use some (Joel is helping us get one but I think we could use more than one here).
Thanks,
Valery
garygid
Well-known member
Joel,
Really nice pistol grip.
You do good work.
However, the cables going out the back, with
a heavy cord (or male connector, a real QC plug,
and its heavy cable) hanging from it might be
an overly strong downward twisting force
on the front end of this Jolomo2 plug?
Might it be better if the cables came out the
butt end of the grip?
Just a thought.
Really nice pistol grip.
You do good work.
However, the cables going out the back, with
a heavy cord (or male connector, a real QC plug,
and its heavy cable) hanging from it might be
an overly strong downward twisting force
on the front end of this Jolomo2 plug?
Might it be better if the cables came out the
butt end of the grip?
Just a thought.
valerun
Well-known member
another quick update - played a bit with switching timing & adding back some leakage inductance. Turns out the latest version of the transformer we wound was too perfect - almost no leakage inductance. As we depend on leakage inductance to ensure zero-voltage-switching, this actually hurt us in efficiency. Also, as I mentioned in my previous post, the lower leakage inductance, the stiffer the supply - which makes regulation harder and current ripple higher.
So we ended up adding a small 5uH inductor in series with transformer's primary. Now we are back to 94% efficiency in isolation stage - all the way to ~17kW output. At 3kW output, we measured 97% but we really need some more precise equipment to measure efficiency at these levels. Absolute best in class designs I have seen in academic papers reach 99%, albeit at the cost of some ridiculous number of parallel parts and serious hit in power density. In fact, there are so many FETs in paralel that they didn't even use heatsinks on them - the total surface of FET backplates was sufficient to dissipate heat they generate!
With that 5uH inductor, we also made the stage much softer, which reduced the 120Hz current ripple to perfectly manageable 20-30% (vs. 50-70% before). This also contributes to higher efficiency as the peak currents are now lower.
Finally, a couple of words about the front-end stage. As I mentioned before, this is a controlled 25kW PFC stage built on a 600A 600V IGBT. In all the tests above, the heating of that stage is pretty much imperceptible (<10C over ambient). The inductor temp is ~30 over ambient. 97% efficiency at 20kW. So the full-system efficiency is ~91%.
Next:
1. get a few more cores (same as we used for the latest design - on order already) and wind transformer versions with deliberately higher leakage inductance (e.g., primary on one half of the core, secondaries on the other half, etc.)
2. Break out secondary diodes onto a separate PCB. When trying to assemble the current PCB-based design, it became obvious that otherwise disassembly would be pretty hard. Separate diode PCB would also allow us to try different configurations of secondary diodes, as well. I think that we could gain another 0.3-0.5% efficiency at 20kW by using 4 diodes instead of 2, etc.
3. Finish packaging into a 10x10x8
4. Hook up current / voltage sensors from secondary side back into PFC stage controls (PCB is ready, using same sensing approach as our 12kW units so everything has been thoroighly proven - just need to connect)
5. Write code to process additional commands discussed above
6. Rewrite the control loop to use proper PID loop control (same as our DC motor controllers so, again, has been tested before)
V
So we ended up adding a small 5uH inductor in series with transformer's primary. Now we are back to 94% efficiency in isolation stage - all the way to ~17kW output. At 3kW output, we measured 97% but we really need some more precise equipment to measure efficiency at these levels. Absolute best in class designs I have seen in academic papers reach 99%, albeit at the cost of some ridiculous number of parallel parts and serious hit in power density. In fact, there are so many FETs in paralel that they didn't even use heatsinks on them - the total surface of FET backplates was sufficient to dissipate heat they generate!
With that 5uH inductor, we also made the stage much softer, which reduced the 120Hz current ripple to perfectly manageable 20-30% (vs. 50-70% before). This also contributes to higher efficiency as the peak currents are now lower.
Finally, a couple of words about the front-end stage. As I mentioned before, this is a controlled 25kW PFC stage built on a 600A 600V IGBT. In all the tests above, the heating of that stage is pretty much imperceptible (<10C over ambient). The inductor temp is ~30 over ambient. 97% efficiency at 20kW. So the full-system efficiency is ~91%.
Next:
1. get a few more cores (same as we used for the latest design - on order already) and wind transformer versions with deliberately higher leakage inductance (e.g., primary on one half of the core, secondaries on the other half, etc.)
2. Break out secondary diodes onto a separate PCB. When trying to assemble the current PCB-based design, it became obvious that otherwise disassembly would be pretty hard. Separate diode PCB would also allow us to try different configurations of secondary diodes, as well. I think that we could gain another 0.3-0.5% efficiency at 20kW by using 4 diodes instead of 2, etc.
3. Finish packaging into a 10x10x8
4. Hook up current / voltage sensors from secondary side back into PFC stage controls (PCB is ready, using same sensing approach as our 12kW units so everything has been thoroighly proven - just need to connect)
5. Write code to process additional commands discussed above
6. Rewrite the control loop to use proper PID loop control (same as our DC motor controllers so, again, has been tested before)
V
valerun
Well-known member
BTW, Gary - I chatted with George Betak last Sat and we thought it would be awesome to bring this project to Nissan hackathon mid-Jan. I think this would be the star of the show. What do you say?
valerun
Well-known member
jclemens said:garygid said:
I am still making progress on the plug redesign. The 3 holes at the back are for cable glands that will act as strain relief for 2 power cables an 1 data cable.
awesome! BTW Joel we are interested in selling your Delux Hook/Hooded Dummy Inlet as option for our JuiceBox EVSE. Perhaps moving to injection molded version? Please email me.
garygid
Well-known member
The commands:
Query Max Amps, and Max Volts.
Set MaxOut Amps and MaxOut Volts
Read AmpsOut and VoltsOut.
Set VoltsRamp rate.
Perhaps ramp to specified voltage,
using the specified ramp rate.
What is the output voltage and current noise
and ripple from the final stage while charging?
Later, Gary
Query Max Amps, and Max Volts.
Set MaxOut Amps and MaxOut Volts
Read AmpsOut and VoltsOut.
Set VoltsRamp rate.
Perhaps ramp to specified voltage,
using the specified ramp rate.
What is the output voltage and current noise
and ripple from the final stage while charging?
Later, Gary
