Hi, I've been following the App notes on the datasheet for the LTC 1923 TEC controller and have sun into some issues with its control. I have the LTC1923 driving a 12V H bridge with optocouplers (5V PWM -> 12V PWM) and have picked RT and CT values so the f_osc is well below the response time of the optocouplers.
I've also got the TEC voltage being divided from 12V max to 2.5V max so as to clamp the voltage to 12V. NTC thermistor is set up as seen in the datasheet, except i am not using an LTC2053 or LTC1658.
I am using a fixed 10k/10k divider as my reference voltage (noninverting input) and an ADA4523-1 as my instrumentation amplifier (+-15V rails) with unity gain. The output from the ADA4523-1 is fed to an OP27 with Rf/Rg=10 with 0.22727V on the non inverting input for a 2.5V DC output with the ADA4523-1 signal being added to that.
This result is fed to the CNTRL pin on the LTC1923. The I_LIM and V_SET are connected to the 5V reference I am using the bias the NTC and divider and the EAOUT is connected to FB with 10M resistor which is then connected to the same 5V reference with a 100K resistor as seen on page 1 on the datasheet. V_THRM is connected to the NTC as well.
As far as I can tell, this should work fine, but on startup, the controller ramps up current to the max set by RS and heats up the NTC until the ADA output is around 500mV rather than 0. It then shuts off all current until the ADA output drops to ~300mV and then maxes current again until output is again at ~500mV. It does this over again and I can't seem to pinpoint why. I've tried Using VRef as by biasing source (connecting FB and EAOUT to it as desribed above), but the controller just drops VRef to maintain a fixed ADA output at 333mV (which doesn't cool the NTC when heated externally.)
