Lebowski
10 MW
I've had it up to 98000 erpm, don't know how high Arlo went...Tek said:
Would you like a very small Lebowski brain board?
- Thread starter zombiess
- Start date
Futterama
1 kW
I don't understand the meaning of that sentence. What power supply issues are there when using isolated supplies?zombiess said:
Liniear Technology has a 150V input synchronous step-down regulator, but it's only rated 100mA:
http://www.linear.com/product/LTC3639
They also have a 400V input device, but it requires external switches and a whole bunch of other supporting components:
http://www.linear.com/product/LT3752
Lebowski
10 MW
When i was thinking about using this 150V 100mA device, I was considering a 2 stage approach. Use this one to step down to around 60V, the 100mA means there's about 6W available from 60V. Use a second more low voltage 500mA device to make 5V out of 60, this is about 2.5W. Another one can make 15V at 200mA, taking 3W.
If you make the small power pcb like this you have the option of leaving out the 150V device if your battery is 48V...
If you make the small power pcb like this you have the option of leaving out the 150V device if your battery is 48V...
Futterama
1 kW
Lebowski, that seems like an OK idea. I'm using the LTC3630A for 5V and 15V, one device for each voltage, this gives me 500mA at each voltage. Also, I need 12V around 1A to drive cooling fans, the LTC3630A supports parallel connection, but I chose to use the LTC3637 which is much similar to the LTC3630A but with more powerful internal switches.
Both the LTC3630A and the LTC3637 can operate at 76V max input.
I made some spreadsheets for calculations, and also got a spreadsheet from a Linear Tech rep. My calculated efficiency is above 90% so with maybe 70V after the first stage, you have 7W, and after the second stage, you have 6.3W total. The Lebowski chip will use around 1W, right? 200mA at 5V. That leaves a good 5.3W for the gate drivers (with or without isolated supplies) and other 5V stuff like current sensors ect.
Both the LTC3630A and the LTC3637 can operate at 76V max input.
I made some spreadsheets for calculations, and also got a spreadsheet from a Linear Tech rep. My calculated efficiency is above 90% so with maybe 70V after the first stage, you have 7W, and after the second stage, you have 6.3W total. The Lebowski chip will use around 1W, right? 200mA at 5V. That leaves a good 5.3W for the gate drivers (with or without isolated supplies) and other 5V stuff like current sensors ect.
Lebowski
10 MW
Pruple jeep is busy with something like that, AFAIK his end goal is a controller box combined with a salvaged Honda Hybrid power stage...Tek said:98000 erpm? Oh, I like where this is headed...I like it a lot!
Is there anything stopping us from having a 2 box system? Like one control box, and one power stage that can be any size we desire?
heathyoung
100 kW
One of these and a Mitsubishi PM300CLA060 would be killer 
300A 600V IGBT 3 Phase module - integrated gate drive, short circuit protection, complete with desat detection. Bolts not solder for connectors... And a completely different SOAR curve to mosfets (you can use all its capability).
Very difficult to kill these little buggers too. You usually melt off a terminal first.
300A 600V IGBT 3 Phase module - integrated gate drive, short circuit protection, complete with desat detection. Bolts not solder for connectors... And a completely different SOAR curve to mosfets (you can use all its capability).
Very difficult to kill these little buggers too. You usually melt off a terminal first.
IanFiTheDwarf
10 W
That’s what I was planning on using, or the one other manufacturers 300A 600V three phase IPM’s.
All you need is four DC-DC converters and four opto isolators between the brain board and the IPM and you are there. And three current sensors.
These IPM’s will take the full 300A and more for a short time but you will need to keep supply voltage to under 450V not 600 and preferably under 400V for full short circuit protection.
Or if you want real extreme power http://theelectrostore.com/shopsite_sc/store/html/Skiip-603GD172-3DUW-Semikron-New.html
All you need is four DC-DC converters and four opto isolators between the brain board and the IPM and you are there. And three current sensors.
These IPM’s will take the full 300A and more for a short time but you will need to keep supply voltage to under 450V not 600 and preferably under 400V for full short circuit protection.
Or if you want real extreme power http://theelectrostore.com/shopsite_sc/store/html/Skiip-603GD172-3DUW-Semikron-New.html
liveforphysics
100 TW
Those big IGBT modules in practice have never done even 50% of the rated current continuously in my own experience. Kinda a lot of cost and weight in return for ho-hum performance.
You will notice the Model S used TO-247 packages because they required both much higher performance (1800A phase current peaks) and wanted it to fit in a much smaller package.
I'm not a hater of big solid state modules, I think they are awesome when done right. The IGBT's they are packaging that are what don't impress me.
MOSFET's are good enough now that IGBT's days are numbered IMHO. Eating the Vf drop as heat becomes impractical and silly at some point.
You will notice the Model S used TO-247 packages because they required both much higher performance (1800A phase current peaks) and wanted it to fit in a much smaller package.
I'm not a hater of big solid state modules, I think they are awesome when done right. The IGBT's they are packaging that are what don't impress me.
MOSFET's are good enough now that IGBT's days are numbered IMHO. Eating the Vf drop as heat becomes impractical and silly at some point.
zombiess
10 MW
liveforphysics said:
MOSFETs tend to switch really fast and have fast recovery. These are all good things, except if you are trying to meet emi spec. Yes you can slow down the switch times but how does one slow down the diode?
There will be a place for both devices for quite a while, the lines just get more blurred and the designer needs to choose the best device for their application. Very few of us on here have to worry about snappy diodes, but other industries do... *Cough* aerospace, medical.
heathyoung
100 kW
I've regularly pulled 300A out of those modules - the biggest appeal is the SOAR curves of IGBT's vs that of MOSFETS.
IGBT's can be used at their maximum voltage and current ratings concurrently. MOSFETS can't.
IGBT's can be used at their maximum voltage and current ratings concurrently. MOSFETS can't.
Lebowski
10 MW
Lebowski said:
Looks like there is a new one out, the LTC3638 which will go upto 140V in and delivers 250mA:
http://www.linear.com/product/LTC3638
Futterama
1 kW
Not quite new, but have been well hidden in their parametric searches. Is 250mA any good then?
Lebowski
10 MW
Futterama said:
Maybe then you need only 2 converters for a high voltage system, this one to make 15V and a 3630 to make 5V from 15V
starting to consider to make a new PCB, one with SMD components and only 3 gate drivers, for 12 FETs max.
With 4110's it would make a nice compact ~80A phase peak (60A rms) controller for bikes. Thinking about something
that'll fit in a 17cm x 11cm x 5cm box. Controller would need to fit a 8x10cm PCB (power supply, controller IC, current sensors
and gate drivers) with the low-inductance output stage in the other half of the box.
Futterama
1 kW
Holy cow, Lebowski, are you going the SMD route now? Congratulations!
The MS16 package the Linear devices come in, has real tiny pins, but I managed to solder it on a homemade PCB without soldermask, so it should be easy to solder it on a professional made PCB. The only tricky part is the thermal pad underneath, but I guess a big via could be used to solder through. What I did was add flux to the thermal pad and then add a bit of solder to the PCB area and keep the PCB solder hot while sticking the IC on the hot solder. The only thing that can go wrong is adding too much solder so it will squeeze out and short the pins, but it worked fine for me in my first try 8)
The MS16 package the Linear devices come in, has real tiny pins, but I managed to solder it on a homemade PCB without soldermask, so it should be easy to solder it on a professional made PCB. The only tricky part is the thermal pad underneath, but I guess a big via could be used to solder through. What I did was add flux to the thermal pad and then add a bit of solder to the PCB area and keep the PCB solder hot while sticking the IC on the hot solder. The only thing that can go wrong is adding too much solder so it will squeeze out and short the pins, but it worked fine for me in my first try 8)
Lebowski
10 MW
I'm too old for this nonsenceFutterama said:Holy cow, Lebowski, are you going the SMD route now? Congratulations!
The MS16 package the Linear devices come in, has real tiny pins, but I managed to solder it on a homemade PCB without soldermask, so it should be easy to solder it on a professional made PCB. The only tricky part is the thermal pad underneath, but I guess a big via could be used to solder through. What I did was add flux to the thermal pad and then add a bit of solder to the PCB area and keep the PCB solder hot while sticking the IC on the hot solder. The only thing that can go wrong is adding too much solder so it will squeeze out and short the pins, but it worked fine for me in my first try 8)
Futterama
1 kW
Then send me the board and SMD components and I'll solder it for you and send it back. Seriously. I would do that for you.
Lebowski
10 MW
thanksFutterama said:
but I have a friend here who's also into SMD's, so it should be ok...zombiess
10 MW
Lebowski said:I'm too old for this nonsenceFutterama said:Holy cow, Lebowski, are you going the SMD route now? Congratulations!
The MS16 package the Linear devices come in, has real tiny pins, but I managed to solder it on a homemade PCB without soldermask, so it should be easy to solder it on a professional made PCB. The only tricky part is the thermal pad underneath, but I guess a big via could be used to solder through. What I did was add flux to the thermal pad and then add a bit of solder to the PCB area and keep the PCB solder hot while sticking the IC on the hot solder. The only thing that can go wrong is adding too much solder so it will squeeze out and short the pins, but it worked fine for me in my first try 8)
Lebowski, it is worth it to give smd a shot. I have been living in the stone age for too long myself. I feared SMD work, but with $200 in soldering gear you pretty much have everything you need.
Now that I have a few SMD projects under my belt it is by far my preferred method of construction. I started out with 1206 size components and found it was super easy to solder and get good results. Then I purchased a cheap station that has a decent hot air setup for $200 and smd became even easier so I dropped down to 0803 size. I am now confident enough with hand soldering 0803, I know I can do 0603 with a bit more patience. I really like smd design for anything that needs a good ground plane, such as a gate driver.
I had +25 years of working with through hole so smd appeared more difficult than it is. My only training was 1-2 hours of you tube videos to learn techniques. Drag soldering any SOIC package is easier and cleaner than the same through hole version. Hard to believe until you try it with a flux pen and a flat or chisel tip. It really is as easy as you see the teachers demonstrating in the video.
bowlofsalad
100 kW
In a discussion with Kingfish, he briefly mentioned Lebowski's controller design. The discussion was about variable regenerative braking. I think that is a potentially extremely useful concept, though one I have never used before. Anyway, he mentioned that Lebowski's controller design has variable regenerative braking. So I am very interested in this project, be it through simply buying a completed board or buying a kit for assembling it myself.
I have very little PCB soldering experience, and I had never heard of SMD(surface mount device) until reading this thread. However, I am certainly not shy to try my hand at soldering SMD. In fact, I'd really like the opportunity.
https://www.youtube.com/watch?v=ihoX7x0RBz8 This video gives a decent illustration of through hole and comparing it to SMD.
I think I like the idea of having the mosfets separate from the rest of the controller. I ponder the idea of methods keeping the mosfets extremely cool but that often leads me to involving the entire controller. I don't know if anything else in the controller gets warm at all, but I am sure that the mosfets get pretty warm.
I have very little PCB soldering experience, and I had never heard of SMD(surface mount device) until reading this thread. However, I am certainly not shy to try my hand at soldering SMD. In fact, I'd really like the opportunity.
https://www.youtube.com/watch?v=ihoX7x0RBz8 This video gives a decent illustration of through hole and comparing it to SMD.
I think I like the idea of having the mosfets separate from the rest of the controller. I ponder the idea of methods keeping the mosfets extremely cool but that often leads me to involving the entire controller. I don't know if anything else in the controller gets warm at all, but I am sure that the mosfets get pretty warm.
zombiess
10 MW
bowlofsalad said:
Keeping MOSFETs cool is actually not very hard. The hard part is designing an optimal layout for the gate driver and power stage while still allowing good cooling. I like to lay the MOSFETs flat. This allows for the best use of surface area for passive cooling and heat distribution while allowing me to obey the gate drive/power stage best practices. If I need to increase power density, then water cooling could become an option, but that comes at the cost of increased complexity, cost and most likely more overall weight, but the gain is lower volume.
bowlofsalad
100 kW
zombiess said:
What I was thinking was making a mosfet box out of pre-made heat sinks, at least using a few heatsinks as a few of the walls, and then filling the box with mineral oil. There are pros and cons to everything, but I prefer the idea of filling the box with mineral oil to pumping water around. Using connectors stops fluid from wicking, you place the connector like you would a door to a room.
Similar threads
