ph2t
Forum Replies Created
-
UPDATE: New IMPORTANT info about MOSFET switching in Kyosho/iWaver PCB’s
Recently, a customer of my Nelly v1.2 turbo sent me this email.
Quote:Dude,Update….i went out and got a 7.4V setup of a JK Lithium Cell. Basically i tried a hook up with 7.4V/150mah (max discharge of 8.4V) and 15C of burst. Basically the truck just went up on every single pull of the throttle. However the cell got heated too quick too fast…. Runtime about 2-3 minutes max!
So i went out and got a 7.4V/350mah setup, ran it parrallel with another identical battery to get 700mah and about 20C of burst on every pull. I hooked up the truck and it flew across my floor fronts wheels up all the way!!! Then all hell broke loose, your Nelly just burst into flames man. Damn!! I guess the current discharge was way too high..
When I read this I was like, holy crap! That’s like some serious current right there. As I thought about the setup he was using I started thinking that nelly should be alright with this.
Why did it blow?
So I started doing some testing. The only 7.4V setup I’m running at the moment is my lipo madforce. That’s running 7.4V/800mAh with a custom plasma dash based motor and the turbo is fine, no overheating, no blow ups….. The version of nelly I was running though was the deluxe version that uses MOSFET driver circuitry to help switch the mosfets on at the increased supply voltage of 7.4V.
So I thought I’d try a lipo setup just with a nelly v1.2, this version doesn’t have the mosfet driver circuitry, just a plain h-bridge. I hoooked a v1.2 up to a lipo’d iWaver running a plasma dash motor. A basic setup really with the iWaver PCB running now at 7.4V. What I found was that nelly v1.2 DID get hot, but still ran fine. So I took it outside and bashed it around for a while and all was OK. The motor did get hot, nelly got hot but the heat dissipated quickly and everything was OK.
Now, keep in mind that I’ve tested both v1.2 and v1.1 with a 6 cell nimh setup way back when I was first developing nelly and all is OK. You CAN run it with 6 cells no sweat.
I guess the key difference here is that lipo batts put out WAY more current than a set of 6 nimh’s/nicd’s. So, moving on from that, the email above shows that this customer run 2 set’s of lipo’s in PARALLEL, making the current going though the circuit incredibly high, hence nelly going up in flames.
This is where is starts to get interesting…..
Nelly deluxe barely gets warm on 7.4V 800mah lipo.
Nelly v1.2 gets hot but still perfoms fine on a 7.4V 800mah lipo.Why is this happening? Why is nelly v1.2 dissipating more power than the deluxe version when both turbos are using the same MOSFETS?????
My first thought was that the mosfets aren’t being turned on as well as they could be. This is an old, established fact in mosfet design. The higher the Gate “turn on” voltage, the lower the operating resistance of the MOSFET. I won’t go into the equations here, but when your resistance is lower, the dissipated power from the MOSFET becomes lower as well.
Shown below is a typical “On resistance Rds(on)” vs. “Gate drive Vgs” MOSFET graph. This graph is available on practically every MOSFET datasheet available from any vendor who manufactures MOSFETS.
It shows how as the voltage used to turn the MOSFET on increases, the resistance in the MOSFET decreases.
The red lines/dots indicate 3 different values for on resistance given the different Gate drive voltage. It’s pretty easy to see that as the voltage on the gate drive increases, the resistance in the MOSFET decreases….
So, I started to take some measurements and this is where I found the root of the problem. All the readings taken below are real values measured with my DMM. They were taken on a stock iWaver and Kyosho PCB’s that had the original MOSFETS removed and wires soldered in their place for access by my DMM.
A Stock PCB with 4 cells would only ever drive the (nchan) MOSFETS at 3V. A Stock PCB with lipo would also only drive the (nchan) MOSFETS at 3V!!!!! Given that the supply voltage has increased one would hopefully assume that the Gate drive votlage would of increased as well. Well it doesn’t!!! This is a BIG issue.
Consider it. The votage has now increased into the circuit, current therefore increases as well. Power dissipated must increase also. But the MOSFETS aren’t any more efficient than they were under the stock setup. This means that even more power will dissipate from the MOSFET pkg’s. Easily setting the stage for a blowup!
If you refer the the graph above, for 3V (in this case) the on resistance is 0.008 Ohm. Surely it can be better, even with the stock 4 cell setup, I’m loosing an extra 1.8V in turn on voltage that could be used to improve the MOSFETS on resistance down to only 0.002 Ohms!!! (see graph) That’s an increase in efficiency of 400%. See how such a small change can make an even greater
difference! Reading from the graph again. If I could turn the MOFSETS on at 7.4V I could get the resistiance down to (theoretically) 0.0016 Ohms. That’s an increase in efficiency of 500%!!!You see where I’m heading now? The higher gate voltage is clearly the key in being able to get the greatest peformance out of the MOSFETS.
So from this I started doing futher research on MOSFET DRIVERS. These components are available from several different vendors and appear to be very popular in the industry now, given by the shear amount of different models available across the market. MOSFET drivers are designed purely to drive MOSFET switches. They have no other function.
So I created a custom PCB of a nelly v1.2 and added a pair of MOSFET drivers that I got from Texas Instruments. Shown below is the test PCB WITHOUT the mosfets on it, just the MOSFET drivers.
Here’s a photo of the test setup. I used this to switch between a 4 cell stock setup and a lipo 7.4V 800mah setup. I took my measurements in circuit whilst the car was on.
See the (2nd half) of the table above for my results. As you can see now the (nchan) MOSFETS are being turned on at 5V, not 3V anymore. So I now have a clear performance gain. At 7.4V the (nchan) MOSFETS are now being turned on a 8V. A clear improvment!
I will finish the assembly of the custom turbo soon and start some real world testing and see if the theoretical improvements make an acutal PHYSICAL improvement. I believe it will though, just a matter of testing it.
So you can now see that for stock PCB’s and mosfet stacked PCB’s, YOU ARE LOSING A FAIR AMOUNT OF POSSIBLE PERFORMACE BY NOT DRIVING THE MOSFETS AT A HIGHER GATE VOLTAGE!
Needles to say that Nelly v1.3 will be including this MOSFET driver technology, :).
Sorry for such a huge rant but I wanted to show people in detail what is happening and hopefully my writing hasn’t been too much techo speak as I hope people can glean some further insight into what exactly is happening under the hood!
Cheers,
ph2t.
Edited by – ph2t on 26 September 2004 09:31:30
