LTC7801/LTC7810 as a dynamically controlled current source?

How practical is using the Ith pin as a control input to make these parts behave as a controllable current source (And how good is the LTSpice model for this slightly odd use case)? 

What I have is a requirement for a two quadrant current source/sink with 55V compliance voltage and capable of some 40A output with > 10KHz modulation bandwidth, supply is ~75V.

It occurs to me that if the behaviour of the ITh pin on these regulators is reasonably well specified then it may be entirely practical to build a polyphase current source by controlling the voltage on the ITh pin, while retaining the over voltage trip behaviour and all the usual protections. Both have graphs for Maximum Current Sense Threshold vs ITH Voltage which hints that this may be workable. Unfortunately there is no bode plot for Ith -> output duty cycle.

The end game is actually a controlled voltage, that again can slew quickly (so low output capacitance which makes a standard dual loop design problematic), to act as a slow drain modulator in an RF transmitter application. 

What I have in mind is a two quadrant linear voltage source (possibly an audio amp chip!) driving a current shunt, then the voltage across the shunt will be integrated and fed to the Ith pin on the switchmode chips so that the switching supply servos the current in the shunt to zero. Hopefully with a 6 or 8 phase design at 200KHz per phase this should get me the dynamics I need without needing the massive output caps that a classic voltage mode loop would require. 

Spice says yes, at least for the LTC7801 with a control voltage on Ith, but suffers a convergence failure when I add the current sense stuff (Not really a surprise), but I don't know how accurate the model of the Ith pin is when externally driven. I also distrust spice for this sort of thing, it's 'wont work' is in my experience far more reliable then its 'that will do'.

Anyone out there with experience of this sort of thing? 

The reason for wanting to go with a integrated switchmode controller for this is the desire to get the protection, UVLO, high side driver and output over voltage protection that wind up taking a mess of parts if done piecemeal. 

Regards, Dan.

    •  Analog Employees 
    on Apr 27, 2019 1:36 AM over 1 year ago

    I have some experience with slewing the output voltage somewhat quickly in a slow drain modulator application. You might try started with a standard voltage divider to set the maximum output voltage (meaning a voltage source and peak current mode controller like the LTC7801 providing current limiting.) Next, compensate the loop and limit the slew rate of the output voltage command signal (into the divider) to avoid excessive reverse current (and/or tripping over-voltage). The IC I was using actually had crude reverse inductor current, but the LTC7801 appears to be a crowbar over-voltage protection (no limit).

  • Thanks for that. The crowbar instead of reverse current limit is rather less then ideal for this use case, but drain modulators are a bit of an odd application anyway.

    The problem with a classic voltage mode loop is the amount of output cap you need, makes the slew rate limit rather lower then is ideal.  

    I have also wondered about the available bandwidth at the SS pin, some experiments are possibly indicated.

    Regards, Dan.