Project:HUGnet Power Supply Theory
From HUG Wiki
This page needs to be edited. It was copied verbatim from documentation on the original power supply. Mostly just the reference designators need to be changed, and a couple of places where the CPU on the controller board takes care of things that used to be done analog. --Prices 13:04, 4 January 2007 (CST)
The main function is to provide current limited power (about 12 volts) to the HUGnet line. When a endpoint suddenly drops current by more than 10ma the power supply boosts the voltage by 6V for 300uS.
After a pulse it ignores any current changes for the next 300uS, after which it will be ready to look for another current drop. The sensor will adjust to any rise in current within about 50uS, making that the new level from which to measure any current drop.
Contents |
Power Feed
Power comes from an external source such as a plug-in-the-wall 12volt dc adaptor. Positive power comes through D1 to the HUGnet. D1 is used with C2 to couple the voltage boost onto the line.
The negative power couples to the line through current sense resistor R3 and current limit pass transistor Q2.
When current reaches near 300 ma the voltage drop across R3 gets high enough to begin turning on Q1, reducing voltage on the gate of Q2, reducing the current that will pass through Q2 to the line.
Signal processing
Sense Amp
A 10mA change in line current causes an 18mV change in voltage across R3. This voltage change is coupled through C4 into op amp IC1D where it comes out at pin 14 as a positive going 250mV. Diodes D3 and D4 limit the output to +-600mV from the bias voltage no matter how large the current change was. They quickly adjust the voltage on C4 so that the amp quickly settles after a large change. D2 and IC1A form an active diode to settle the amp quickly and consistently after a current increase. Q3 acts as a switch shorting input to output. It allows no output and settles the amp almost instantly if there are any current changes for 300uS after the voltage boost pulse is removed. After that Q3 is turned off and the amp is ready to see new current changes.
Latch
When the voltage exceeds about 250mV it is high enough to trigger the latch IC1B. The output of IC1B goes high turning on the voltage boost.
Timer
When the latch goes high, R13 begins to charge C6. After 300uS the voltage on C6 is at 2/3 of the op-amp supply voltage, high enough to trigger IC1C bringing its output low. While it is low, diode D5 turns on Q3 shorting out the sense amp and diode D6 forces the latch back into the low state. Now R13 discharges C6, bringing the voltage down until it reaches 1/3 of the op-amp power supply and triggers the output high again. Now the sense amp is active again and the latch is re-armed. While the timer output is high, R11 and R12 and D7 prevent C6 from settling any lower than the 1/3 voltage. This keeps the timing consistent for the next cycle, whether it begins immediately or later. R8, R9 and R10 create the 1/3 and 2/3 voltage references used by the timer.
Voltage Booster
Normally the latch is low and Q4 is off, leaving Q5 off and voltage on the negative side of C2 about 6V below +12H. When the latch goes high, Q4 turns on, turning on Q5. The voltage on Q5 collector rises to +12H, pulling the emitter of Q7 up about 6V to about .6V below +12H. The rise in voltage at the negative side of C2 brings the positive side of C2 about 6 volts above +12H. D1 gets reversed biased, allowing the voltage to rise. When the latch goes low again, Q6 pulls C2 low again. Any charge that C2 had lost is quickly put back in when the voltage is low enough that D1 is again forward biased. The charging current is limited by R23.
Parts Description
Current Limit Circuit
R3 converts line current into a voltage. It determines the current limit point. At 1.8 ohms the current limit is about 300mA. Power dissipation at limit is about 180 mA. Higher resistance will reduce current at limit. Lower resistance will increase current at limit.
R2 smoothes response at limit, helping prevent loop oscillation. Higher value makes a smoother response and less tight regulation. Lower value tightens regulation and increases chance of oscillation.
Q1 pulls current from R1, reducing voltage on Q2 gate until current is reduced to limit level. Q1 base-emitter turn-on voltage determines limit threshold. Virtually any small signal npn transistor will work.
R1 pulls gate of Q2 on. Its value is not at all critical. A higher resistance will slow loop recovery time. A lower resistance will speed up recovery and affect loop stability. Q2 passes current to the HUGnet line. It increases resistance as it goes into current limit, dropping whatever voltage is necessary. At 16 volts and 300 mA it dissipates about 5 watts. Any N-channel power fet that can handle the current and wattage is likely to work.
ZD1 protects Q2 gate from excess voltage, including voltages capacitively coupled to the gate from the drain pin. The value of ZD1 must be high enough to assure that Q2 gate gets to high enough voltage to fully turn on Q2, even at low zener currents. The zener value must be low enough to stay below gate breakdown voltage.
LED3 Turns on when the current limit circuit begins to drop appreciable voltage. Virtually any visible LED will work.
R26 determines the brightness of LED3. At 15 volts it allows 15mA and dissipates 225mA. A lower value may allow too much current into LED3. A higher value will work fine at lower LED3 brightness.
LED2 is normally on. It reduces brightness as LED3 increases brightness. Virtually any visible LED will work.
R25 determines the brightness of LED2. At 15 volts it allows 15mA and dissipates 225mA. A lower value may allow too much current into LED2. A higher value will work fine at lower LED2 brightness.
Power Conditioning
ZD2 forms a shunt regulator. At 16 volts input it sees 19mA and dissipates about 125mW. A higher or lower zener voltage will change op-amp voltage accordingly. 6.8 volts is a minimum working voltage for lm324 in this circuit. LMC660 can work fine down to 5 volts.
C1 provides high transient currents to the op-amps if they need it. Value is very non critical.
R4,R5 provide current to the regulator and op-amps. Each one dissipates 100mW with a 16 volt source. With a 26 volt source they could each dissipate up to 400mW. A lower resistance would increase power dissipation. A higher value will reduce dissipation. It may be ok to reduce current to just what the op-amp needs when the source voltage is at about 8V.
R6,R7 set the bias voltage. This is set to about ¼ op-amp voltage so that the latch cannot be reset to low by any signal from the sense amp. For minimum noise, the ground from R7,C1,ZD2 should be routed directly to R3 as a single point.
Sense Amp
IC1 LMC660 is very well suited to the task. LM324 is a bit slow and does not go to the top rail, but the circuit values are optimized to be stable using the LM324. When the faster LMC660 is installed, the circuit behaves cleaner and more stable.
C1 couples the change in line current to the amp. A higher value will lower the low frequency response of the circuit. A lower value will raise the low frequency response. The value is not terribly critical.
R28 sets the gain of the sense amp in ratio with R16. We want a current drop of 10mA to cause a voltage at op-amp output to rise about ¼ volt, (about ½ diode limit voltage). Higher resistance lowers the gain and lowers the lower frequency pass. A lower value increases gain and increases lower frequency response.
R16 sets gain of sense amp in a ratio with R28. An increased value increases the gain and lowers the upper frequency response. A decreased value decreases gain and raises upper frequency response.
C5 limits the upper frequency response. A higher value lowers the upper frequency response. A lower value increases upper frequency response.
D3,D4 limit the output to plus or minus .6 volts max, measured from bias voltage. Any good low leakage current silicon diode will work.
Q3 shorts input to output reducing gain to zero when it is on. Any good pnp transistor should work.
R27 keeps the loop stable and sets how fast C4 settles to a new line current during times that Q3,D3,D4 are conducting. Higher resistance is more stable and slower settling time. Lower resistance is less likely to be stable and faster settling time.
D2 and op-amp form an active diode ( a diode with no voltage drop) used to quickly settle the sense amp any time the signal at sense amp goes negative. Any low leakage diode will work.
R17 keeps the loop stable. Lower value is less stable and settles faster. Higher value is more stable (especially with slower op-amps) and settles slower.
R18 is current limit for Q3. Value is non critical as long as current can turn on Q3 and does not get excessive.
D5 acts as a gate to only apply current to turn on Q3 when timer is low.
Latch
The latch op amp is biased at its negative input at about 1.5 volts, the same as the output of the sense amp. The latch output is normally low. R14 forms a voltage divider with R15, keeping the op amp + input lower than the – input by about 150 mV. When the sense amp goes positive by more than that voltage, the latch amp output rises. R14 and R15 provide positive feedback and the latch snaps high. The output was 1.5 volts below the bias point. Now it is about 5.3 volts above the bias. It would now take a 530mV negative signal from the signal amp to reset the latch. The signal amp can’t do that, so signals can’t reset the latch. Only a ground on D6 can do that.
R14 working with R15 sets the sensitivity of the latch to about 10 mA. A higher resistance value of R14 will increase sensitivity. A lower value will decrease sensitivity.
R15 working with R14 sets the sensitivity of the latch to about 10 mA. A higher resistance value of R15 will reduce sensitivity. A lower value will increase sensitivity.
Diode D6 acts as a gate that resets the latch. Any good signal diode will work.
Timer
Op amp C output is normally high. The latch being normally low keeps C6 discharged via R13, thus keeping the amp’s – input low, keeping the output high. R10 and R8 and R9 keep the + input at 2/3 supply voltage. When the latch goes high, R13 charges C6. Eventually it exceeds the 2/3 voltage and the timer output goes low. R9 pulls the +input to 1/3 supply voltage, assuring the output stays high. D5 and R18 turn on Q3 settling the signal amp. D6 forces the latch low. Now R13 starts discharging C6. When it gets below 1/3 supply voltage the timer output goes high again.
R11 and R12 form a bias of ½ supply voltage when the timer output is idle (high). D7 limits C6 to no lower than about 1/3 supply when idle. This keeps the timing consistent no matter how short a time between pulses. Any good silicon diode will work.
R11. A higher value will lower C6 idle voltage. Excessively high will start affecting timing consistency as pulse density changes. A lower value will increase C6 idle voltage, reducing boost pulse width.
R12. A higher value will increase C6 idle voltage, reducing boost pulse width. A lower value will decrease C6 idle voltage. Excessively low will start affecting timing consistency with pulse density changes.
R10 increase will lower the 1/3 and 2/3 bias voltage. Decrease will raise it. R8 decrease will lower the 1/3 and 2/3 bias voltage. Increase will raise it. R9 increase will lower the 2/3 and raise the 1/3 bias. Decrease will raise the 2/3 and lower the 1/3.
R13 increase will slow down timer. Decrease will speed up timer. C6 increase will slow down timer. Decrease will speed up timer.
Voltage boost
Q4 is a level translator. Any good npn will work
R20 sets the on current of Q4. Value not very critical. A higher value decreases current and slows pulse rise time.
Q5 inverts the translated pulse and, with C3, limits pulse rise and fall rates. Any good pnp will work.
R21 turns off Q5 and, with C3 limits the pulse fall rate. A higher value slows pulse fall time.
C3 higher value slows rise and fall time.
R22 pulldown current for emitter followers. A higher value lowers max downward drive on C2 and may slow pulse fall time.
ZD3 sets pulse amplitude to about 6V.
R23 limits current charging C2. An excessively higher value may not fully charge C2.
Q6 and Q7 drive lots of current to C2. Any good signal transistor with a gain of >100 and >500 mA ratings is ok.
C2 must be large enough to provide minimal voltage drop when driving 400mA for 400uS. A higher value will have less voltage drop.
R24 sets brightness for LED1. A higher value lowers brightness.
See Also
|

