5668

Here I test the current-source with 12-bit DACs from Microchip. At first I tested the I2C DAC MCP4725, but the result was not so good: 3 to 6 mV offset (expected 0...1 mV), also the linearity was bad. Therefore I tested it with a SPI-DAC MCP4822 (dual). He is better from the offset (1... 2mV) , also the linearity (1-2 %). I tested it with different insulator-ic's : ADUM1200 for I2C and ADUM1400 for SPI and with opto-couplers.

The  isolated interface for SPI (3-4 Pins) is more complex compared with I2C ( 2 Pins).

Also the price for SPI-insulator chip (ADUM140x about 10 €) is higher compared with I2C (ADUM125x about 3 €) , therefore I tested it with four optocoupler . I found a small cheap breadboard ( Optoinsulator HW-399) with optocoupler TLP281-4 and 4 transistors to turn threw the coupler inverted signal.
I used 3 Pins for the SPI-interface (only CLOCK, MOSI, CS used for the DAC) and one for changing the 
output polararity of the currentsource .
It functions, but only very low SPI frequency (15 kHz max).
The slewrate from normal LED-couplers is to small.
But it's ok for applications like transistor-curvetracer changing I-base every 20 s or so.

For higher frequencies  I have tested the currentsource with an ADUM1400, making a small PCB with the same dimensions and pins like the HW-399 board, using the NC-Pin for +5V power supply.
(the ADUMxxxx insulators use magnetic-transformer and need on both side +3-5V power )

The max. speed I tested is 12 MHz .

21-08-25 :
Now I used for the ADUM1400 a normal adapter board from SO16W to DIL ,
dimension  21.6 mm (h) and 30.5 mm(w) and DIL-Pins with distance 22.86 m (9x2.54mm) from Pollin. 
The Enable-Pin on the isolated side must be on High-Level (with a resistor from 5 V) . 
The adapter is better from the insulation and easier to solder.
For save space, I placed the MCP4822 under them. 
Also I changed the max. input-voltage of the OPamp from 4V  to 1 V with a 10k / 3.3k divider.

The result is good, small offset and accectable linearity.