CB-Fan Card Power Supplies for the ECL Circuits and Termination ----------------------------------------------------- Initial Rev. 30-Jan-2012 Current Rev. 10-May-2012 The following notes are about the -5V power supply for the ECL integrated circuits and the -2V power supply for the ECL termination resistors. The are 3 basic topologies for these power supplies: - Use an isolated DC/DC converters running from the +5V or +12V VME supply - Use a non-isolated DC/DC buck converters running from the -12V VME supply. - Use a non-isolated inverting DC/DC converter running from the +5V or +12V VME supply. An appropriate positive to negative inverting non-isolated POL converter appears to be: TI Power Trends PTN04050A The "A" suffix is important. I *believe* that: PTN04050AAZ is their current part for SMD assembly process PTN04050AAH is thier current part for THD assembly process They also have a current lower temperature SMD assembly process part. Vin range is positive 2.9V to positive 7.0V Vout range is negative 3.3V to negative 15V maximum output current is 1.0 Amp when the output voltage is between -3.3V and -6.0V for continuous operation the Inhibit pin, pin #3, should be left open circuited Vout adjust pin, pin #4, if left open the output will be -1.79V 37.4k --> -3.30V, 19.1k --> -4.50V, 15.4k --> -5.00V, 10.7k --> -6.00V, 1.91k --> -12.02V Operation of this converter with a -2V output has a couple of issues: - The current reference resistor setup and feedback resistor setup makes the adjustment for -2V output very sloopy and very dependent on the exact value of things. Need to modify the reference or feedback resistor networks for good operation at -2V output. - Without changing the inductor, the FET ON duty cycle is going to be rather low. That should be fine if we do not run too near full output current. Could swap inductors or chenge the switcher's frequency to make cleaner operation at -2V if necessary. Work with the Power Trends PTN04050A Inverting Non-Isolated Converter - The intent is to run these converters from +5 Volts which is available at a higher current level from the backplane than the -12V bus. - Need to check / understand how well this converter would run making a -2.0 Volt output. - The PTN04050A appears to use the Semitech SC4508A controller in a setup that varies the Set-Point Reference to the FB- pin 8 input to the controller to adjust the desired output voltage. - Operation with Vin = 5.035 Volts and a 10 Ohm Load (except as noted). R_Set V_Out V_RSet V_FB+_Pin_9 ------- ------- -------- ------------- open -1.942 +0.436 +0.430 100 K -2.546 +0.418 +0.412 60 K -2.926 +0.406 +0.401 40 K -3.377 +0.393 +0.388 20 K -4.58 +0.358 +0.354 10 K -6.43 +0.305 +0.300 5800 -8.30 +0.250 +0.245 3700 -9.97 +0.201 +0.196 For the 8.30V and 9.97V output tests R_Load was 20 Ohms. Recall that the actual on-chip reference is +0.500V so the 1.942V and 8.30V output test are most interesting to design the reference circuit. This implies that the PTN04050A has about 4.99K Ohm from its 0.500V reference to the R_Set pin and about 33.99K Ohm from its R_Set pin to ground. The ratio of these resistors should be about 0.1468 Assuming that the resistor from the +0.500V refenence to the feedback divider is 1 K Ohm as measured the the feedback input resistor in the divider must be about 34 or 35 K Ohm. A rational layout. - More PTN04050A test runs: -5V -2V Parameter Output Output --------- -------- -------- V_Out -5.001 V -2.001 V R_Load 10 Ohm 3.33 Ohm --> Iout 0.5001 A 0.6003 A --> W_Out 2.501 W 1.2012 W V_In +5.005 V +5.018 V I_In 0.61 A 0.33 A --> W_In 3.053 W 1.656 W --> Eff 81.9 % 72.5 % R_Set 16.66 K 840 K V@R_Set +0.346 V +0.434 V Gate_Low 1.87 us 1.12 us Gate_Hi 1.91 us 2.66 us --> % ON 49.5 % 29.6 % --> Freq 265 kHz 265 kHz - 9-May-12 Modify the feedback resistor network to make operation with a -2V output cleaner. Specifically pull off the 33.2k Ohm direct feedback resistor and replace it with a 14.99k Ohm feedback resistor. Note that both the +FB and -FB pins of the SC4508A controller (error amp input pins) must remain in the range -0.2V to +0.7V for the controller's outer loop to operate correctly. With the 14.99k DC feedback resistor, it required a 17.6k Vset resistor to make a -2.00V output. This was running from a +5.00V input and was providing about 0.59 A to the -2V load. The indicated input current (on the bench supply's meter) was 0.32A I could not measure the switcher's frequency or duty cycle because Jim has my scope. With the 14.99k FB and 17.6k Vset the inputs to the error amp should have been at about +0.345 Volts. ---------------------------------------------------------------- The following information comes from evaluating a negative buck converter for running the ECL from the -12V VME supply. I gave this up because: - I do not like to pull this much power from the one pin -12V VME supply. - I do not like to run a buck converter over the 6:1 range to make the -2V Vtt supply for the ECL. Look at the Power Trends PT79SR105 non-isolated DC/DC buck converter with negative voltage input and -5V output. This topology probably has the potential to be the most efficient. - It appears to be conventional negative voltage buck topology using a free-wheel catch diode instead of a second FET switch. - ST-Micro D12NF06 N channel FET switch 60V 12A 100 mOhm ON at 6A 10V has the normal built in reverse Source-Drain voltage zener diode looking at the top of the part with the Drain tab at the top, Gate is left, Source is right. - B240 Schottky Diode 40V 2A free-wheel V forward is 500 mV at 2A Cathode is marked by strip. Test run the PT79SR105: - Vin supply is -10.0V, Load is 10 Ohms, Vout measures -5.085V - Controlle Chip is an 8 pin SOIC 1. Measures about -1.0V loading with the scope probe pushes the output about 6mv more negative. 2. Measures about -9.0V loading with the scope probe pushes the output about 150 mV less negative. 3. Measures about -8.5 V 4. Measures about -8.1 V 5. Measures about -9.96 V 6. Measures about -10.00 V same as the Vin supply 7. Gate Drive: On for about 504 nsec Off for about 504 nsec amplitude about 10V 8. Appears to be at the circuits "ground" potential. - The circuit has a resistor between the Vin pin and the switches Source pin. This resistor is marked "R120". With the circuit running as described above the drop across this resistor is about 33 mV i.e. not inconsistent with this being a 0.12 Ohm resistor. - The controller chip may be a LM3489 or something kind of like it. Test runs the PT79SR105 on 31-Jan_12: Vin = 12.00V Vout = 5.00V Load = 10 Ohms Ton = 412 nsec Toff = 596 nsec Iin = 0.23A or 0.24A mostly 0.24A Rsrc = 27mV drop across the R120 FET Source Resistor if source resistor is 0.12 Ohms --> 0.225A source current 947 Ohm Pot to Gnd \ 4059 Ohm Pot to Vout / Res of just the Pot Must match internal divider 944 mV at Pin #1 0.24A x 12V = 2.88W input assume 0.5A x 5V = 2.5W output --> 86.8% eff Vin = 12.00V Vout = 2.00V Load = 10 Ohms Ton = 156 nsec Toff = 848 nsec Iin = about 60 mA Rsrc = 5mV drop across the R120 FET Source Resistor if source resistor is 0.12 Ohms --> 0.042A source current 0.06A x 12V = 0.72W input assume 0.2A x 2V = 0.4W output --> 55.5% eff Vin = 12.00V Vout = 2.00V Load = 2 Ohms Ton = 176 nsec Toff = 836 nsec Iin = 0.22A or 0.23A mostly 0.23A Rsrc = 26mV drop across the R120 FET Source Resistor if source resistor is 0.12 Ohms --> 0.217A source current 2514 Ohm Pot to Gnd \ / In parallel with the 2498 Ohm Pot to Vout / Res of just the Pot \ internal diviser makes 2v 947 mV at Pin #1 0.23A x 12V = 2.76W input assume 1.0A x 2V = 2.0W output --> 72.5% eff