Mosaic
Well-Known Member
Hi all,
I just spent a few hours chasing down an odd prob. in a battery charger circuit (500mA).
The setup:
Low side switched, 500mA, 12V battery charging, 36V supply.
A TL494 was pushing max DC of about 96% with a 10Khz PWM switching an IRFP3206 NFET to ground via a 5 ohm, 3W current sense source resistor.
I was using err amp # 1 sampling from the drain of the NFET to verify the presence or absence of a load battery being charged.
How the err amp worked:
I placed a 10K resistor across the common anode of the supply and the switched Drain of the NFET to provide a voltage reference (at the Drain) to use to detect a correctly connected battery. (Not reverse polarity or absent). Err Amp #1 compared the common anode vs the 10K pulled up Drain to detect the battery. The voltage divider feeding the err amp inputs ensured that the +ve input saw a little greater V ( than the -ve input) if no battery was present or the charger terminals were shorted. This disables the PWM and prevents reverse charging or short circuited charging.
The odd problem:
The polarity protection worked fine UNTIL a battery was disconnected. At which time the NFET seemed NOT to switch off during the dead time (4%). This was evidenced by the DRAIN seeming to pull to very near ground as a tiny current flowed thru the source resistor via the Err amp # 1 Volt ref divider (143k impedance). Naturally this near ground Vref kept the err amp from detecting the absence of a battery (should see V+ rail via the 10K pullup) and the err amp 'latched' . I could now put a battery in backwards or short the terminals and the 500mA would still flow. Shutting down and powering up restores expected operation until the battery is disconnected again.
Without expounding on all the figuring out.....the problem turned out to be.....large drain capacitance. The impedance of the 10K DRAIN pull up combined with the 10Khz PWM and 4% dead time combined to be NOT enough time to burn off the charge on the NFET DRAIN.
The solution is: Cut the PWM to 1Khz (increasing the absolute dead time) and cut the pull up resistor to about 3.3K thereby overcoming the residual NFET Drain charge and restoring proper function the the ERR amp # 1 inputs.
That's what u get from using an oversized NFET. I had this part used elsewhere in the circuit and I preferred to reuse it rather than spec a smaller unit as I may want to kit the project. Unique parts count should be minimized for simplicity and serviceability.
I just spent a few hours chasing down an odd prob. in a battery charger circuit (500mA).
The setup:
Low side switched, 500mA, 12V battery charging, 36V supply.
A TL494 was pushing max DC of about 96% with a 10Khz PWM switching an IRFP3206 NFET to ground via a 5 ohm, 3W current sense source resistor.
I was using err amp # 1 sampling from the drain of the NFET to verify the presence or absence of a load battery being charged.
How the err amp worked:
I placed a 10K resistor across the common anode of the supply and the switched Drain of the NFET to provide a voltage reference (at the Drain) to use to detect a correctly connected battery. (Not reverse polarity or absent). Err Amp #1 compared the common anode vs the 10K pulled up Drain to detect the battery. The voltage divider feeding the err amp inputs ensured that the +ve input saw a little greater V ( than the -ve input) if no battery was present or the charger terminals were shorted. This disables the PWM and prevents reverse charging or short circuited charging.
The odd problem:
The polarity protection worked fine UNTIL a battery was disconnected. At which time the NFET seemed NOT to switch off during the dead time (4%). This was evidenced by the DRAIN seeming to pull to very near ground as a tiny current flowed thru the source resistor via the Err amp # 1 Volt ref divider (143k impedance). Naturally this near ground Vref kept the err amp from detecting the absence of a battery (should see V+ rail via the 10K pullup) and the err amp 'latched' . I could now put a battery in backwards or short the terminals and the 500mA would still flow. Shutting down and powering up restores expected operation until the battery is disconnected again.
Without expounding on all the figuring out.....the problem turned out to be.....large drain capacitance. The impedance of the 10K DRAIN pull up combined with the 10Khz PWM and 4% dead time combined to be NOT enough time to burn off the charge on the NFET DRAIN.
The solution is: Cut the PWM to 1Khz (increasing the absolute dead time) and cut the pull up resistor to about 3.3K thereby overcoming the residual NFET Drain charge and restoring proper function the the ERR amp # 1 inputs.
That's what u get from using an oversized NFET. I had this part used elsewhere in the circuit and I preferred to reuse it rather than spec a smaller unit as I may want to kit the project. Unique parts count should be minimized for simplicity and serviceability.
