This is a proposed project to create a system based on KFLARE which would provide single-channel Laser/TEC driver boards controlled by a single central controller.

Required Changes

  • Float the current source so either anode or cathode of laser can be grounded
  • Upgrade the current source to realistically run at 4A. Using e.g a 3.3V power supply, the transistor would be dissipating about 3.5W.
  • Implement a floating photodiode readout
    • OPA365 is -5V only. Should find a +5/-5 replacement. Also, I(out) is only 65mA.
    • LMV641?
  • Possible transistor replacement: 2SA2222SG in TO-220 pkg

Design work Jan 2020


The DRV595 includes a PWM driver and seems overly complex. Best to just use an H-bridge to drive the TEC with a similar L-C filter on the output. For example the MC33886/7 which doesn't mention TEC in applications but seems otherwise suitable. Another more fancy option is IFX9201SG with an SPI control interface. Neither of these mentions a TEC application though.

Also see the ADN8830 ($10@100), LTC1923 ($30!) TEC controllers which do everything but needs external FETs or H-Bridge.

Best bet seems to be ADN8831 with two e.g. Si9801 half-bridge drivers.


Providing ~5A from a fixed supply is not very realistic. Looking at things like the LTM8064, which is a nice, constant-current step-down converter. They are $30/$20/$16 @1/100/1k. Another option is the LTM8024, which is not constant-current but could act as a programmable supply for a downstream driver.

Meeting Jan 2020

  • Plan to go ahead and fab 4-8 boards of essentially KFLARE design for testing
  • Begin design of upgraded version per notes below.

Discussions Nov 2019

Sylvain suggests that a 2-channel board would be OK but need to discuss changes above and mechanics / cooling.

A big issue with the old design was cooling of the driver circuit. A couple of thoughts come to mind. One is to optimize the power supply voltage for each laser (programmable switching regulator?) Another is to us a larger transistor package with a bigger heatsink.

Laser diode grounding note:


Handling a 4A LD with a simple linear constant-current driver is a challenge due to the dissipation in the power device. Thinking about alternatives, such as a switching regulator to provide a programmable bulk power source individually for each channel.

Existing System

  • Existing design uses an Atmel ATMega164A uC per laser in a 44-VQFN package. This device is efficiently used, with only about 4 spare I/O pins.
  • Communication is over a Modbus-like RS-485 serial shared bus. This works well since RS-485 has very good power supply rejection. See communications protocol spec document.

In the current system an additional ATMega64A uC was used as the system controller, which would manage up to 20 Laser/TEC modules and handle the interface with the host computer. The system controller also managed the power supply, fans and an additional white light source.

System Controller functions:

  • UART (RS-485) bus to Laser/TEC modules
  • UART or other interface to host computer
  • Fan control PWM output
  • Fan sense input
  • Interlock control logic inputs: TEMP_ALARM, AC_FAIL, FAN_FAIL, INTERLOCK_STATUS
  • Interlock control logic outputs: LASER_SHUTDOWN, LASER STATUS (per laser)
  • ~4 temperature sensors
  • I2C DAC for laser current setting (or maybe Laser/TEC control?)
  • White light source control (opt-isolated)
  • Laser enable outputs
  • LED indicators

Controller ideas

  • Zynq, e.g. Enclustra Mars ZX2 DIMM
    • Pros: Linux on-board, familiar, lots of GPIOs, complete control of peripherals
    • Cons: Analog inputs may be hard to use
  • Beaglebone:
    • Pros: Linux on-board, UARTs and analog inputs built-in, enough GPIOs
    • Cons: Unfamiliar, sounds like analog inputs are a real headache
  • Raspberry pi:
    • Pros: Linux on-board, widely used
    • Cons: Unfamiliar, share UART with serial console (?), analog inputs maybe a pain
  • FPGA Eval board:
    • Pros: Relatively easy to use (IPBus interface?)
    • Cons: Need to come up with an ADC solution

There exist some useful PMODs, such as Isolated RS-485 and 4 ch 12 bit ADC

Last modified 3 years ago Last modified on Jan 30, 2020, 3:27:12 PM