[EVDL List]

Hi All,

Is there a good analog EE in the house? I finished converting a
Porsche 914 to an electric vehicle a few weeks ago and the first
thing to die was the DC-DC converter that transfers charge between
the main 144V pack to the 12V accessory battery. I took the unit
apart and found a push-pull switching power supply with shorted out
FETs. After replacing the FETs, the unit shorted again under a 20
amp current load. I replaced the FETs again and captured several
waveforms of the internal nodes.

The schematic and datasheets for the CCPower unit are here:

http://914ev.blogspot.com/2007/09/initial-ccpower-schematic.html

The oscilloscope waveforms are here:

http://914ev.blogspot.com/2007/09/debugging-dc-dc-waveforms.html

The main 144V battery pack can range anywhere from 120V to 170V
based on charging and regenerative braking. Again, the unit dies and
shorts the FETs when pulling a significant current out the 12V side of
about 20amps.

I'm still commuting in the vehicle, but I need to charge the 12V accessory
battery every night along with the main pack (argh!). Any suggestions for
how to fix this would be most welcome. Thanks for your time!

Cheers,
Tim
http://914ev.blogspot.com/






____________________________________________________________________________________
Sick sense of humor? Visit Yahoo! TV's
Comedy with an Edge to see what's on, when.
http://tv.yahoo.com/collections/222
_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev

 

[EVDL List]

From: Tim Kutscha
> the first thing to die was the DC-DC converter...
> The main 144V battery pack can range anywhere from 120V to 170V...
> found a push-pull switching power supply with shorted out FETs...
> After replacing the FETs, the unit shorted again under a 20 amp
> current load.
>
> The schematic and datasheets for the CCPower unit are here:
> http://914ev.blogspot.com/2007/09/initial-ccpower-schematic.html
>
> The oscilloscope waveforms are here:
> http://914ev.blogspot.com/2007/09/debugging-dc-dc-waveforms.html

Good detective work! The schematic and waveforms help a lot!

The schematic strikes me as a "textbook" circuit, designed by an inexperienced engineer. It is missing about half the parts needed to work reliably. Not enough filtering, not enough bypassing, no inrush limiting; just plain not enough safeguards to protect against damage. No overvoltage, undervoltage, or overtemperature protection. The overcurrent sensing circuit looks like it's intended to provide gross shutdown, not smooth current limiting. No output fuse in case the battery is connected backwards.

Your understanding of how this circuit works is a little off. The duty cycle of the MOSFETs controls voltage; not current. The SG3525 is a PWM controller that adjusts the average input voltage to the transformer primary, just like the PWM controller for a series motor. The transformer's fixed turns ratio steps this down to the desired output voltage.

For example, if the transformer has a 10:1 ratio, and you want 12v out, then you need 120v in. If your pack voltage is 160v, then the PWM needs to run at 120v/160v = 75%. It actually splits this equally between the two half-cycles; each MOSFET is on 37.5% of the time.

Increasing the load current doesn't change the duty cycle; it just increases the current in the diodes, transistors, and transformer windings. The only increase in duty cycle comes from the slight voltage drop caused by the winding resistance and transistor and diode drops.

PS: I think there is a missing diode in the output; its cathode goes to the cathodes of U1a and U1b, and its anode goes to the center tap of the transformer secondary. This is the freewheel diode for the output inductor, L2. Leaving it out provides no path for inductive current when both MOSFETs are off, leading to lots of ringing and voltage spikes. L1 should also be a fairly substantial inductor, so the current through it doesn't fall to zero during MOSFET off-time.

This circuit is likely to die if the input voltage gets low enough so the MOSFETs try to be on 50% of the time. The SG3525 has crossover protection, but apparently not enough from your waveforms. The 600ns non-overlap time is too short, and the gate drivers are too weak (R11 and R12 are too large).

Hope this helps!


--
"Excellence does not require perfection." -- Henry James
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart-at-earthlink.net

_______________________________________________
For subscription options, see
http://lists.sjsu.edu/mailman/listinfo/ev