Fullscreen HTML5Med Res (??MB)HTML4

TERMINALS POWER F1 8A +17V 0V 3x 2200 µF 25V 1k Q2 IRF1405 DC SOCKET REG1 7812 +17V D G K S A GND 220nF LOW ESR +12V OUT IN S4 2.2k 100 µF A 10k λ LED3 K ZD1 15V 1W +12V POWER B 1 0 0k C Q6 BC337 E +12V 100 µF 4.7k 470Ω 100k 100k LEVEL VR6 10k 8 5 6 4.7k 7 IC1b 100k 47k 1 220k VR2 10k 4 MAX SET IC2a 3 MIN SET 4.7k 160Hz TRIANGLE GENERATOR 5 RUN BRAKE 470Ω VR5 250k A 1 µF IC1: LM358 IC3: LM393 ZD2 8.2 V IC2: LM324 IC4: 4013B 12 14 K IC2d TRACK VOLTAGE LOCKOUT A 220k 100Ω D3 1N4148 UNDER-VOLTAGE LOCK-OUT 10 LI'L PULSER MK2 (REVISED) S1 VR4 1M 10k 470k SC 10k 10k K 2013 S2 7 IC2b +12V TP GND 10k VR1 10k 6 22nF 1 SPEED TP1 ERROR AMP 4.7k IC1a 2 100k 10nF 100k 3 VR3 10k INERTIA 2 10nF 9 +1 .09 V CUT x 13 A +12V D2 1N4148 4 IC2c K 8 11 Fig.1: this circuit diagram shows the required changes to the original Li’l Pulser Mk.2 circuit in orange. Note that Q6 and its associated base divider resistors are only added as a ‘belts and braces’ measure if you are building the revised PCB (see text & Fig.3). 1N4148 A K Li’l Pulser Mk2: fixing the switch-off lurch By NICHOLAS VINEN Our new Li’l Pulser Model Train Controller, described in the July 2013 issue, has been very popular but a design flaw has become apparent. At switch-off, any locomotive(s) on the track can suddenly lurch forward, even if they are stationary at the time. This is regardless of the position of the speed control knob and the brake switch. Here’s the cure. 68  Silicon Chip siliconchip.com.au +12V 100 µF +17V +12V LED1 TRACK λ 2.2k D6 FR607 K λ 1 µF MMC C B 47 µF K A D 10Ω G 7 IC3b 6 A A E Q1 IRF1405 A D5 1N4004 10k K 2 x 0.1Ω 5W (R1,R2) D4 1N4148 B BC337 S D7 1N4148 4 D1 1N4004 The Li’l Pulser Mk.2 Train Controller was originally described in the C to fix the switch-off “glitch”. July 2013 issue. Follow this article Q3 Q5 BC337 E 8 K RLY1b 10k 5 RELAY1 TRACK TERMINALS RLY1a A 47k K +12V 100nF 2.2k 10k +12V 1k + 100 µF Q4 BC327 – PIEZO SIREN E +1 .09 V TO PIN 9 OF IC2c 2 1 IC3a 3 3 10k 100nF REV 1M C 4.7k OVERCURRENT CURRENT 1N4004, FR607 ZD1 K I F YOU’VE BUILT the Li’l Pulser Mk.2, then you’ll want to fix the switch-off flaw. The magnitude of the effect varies, depending on how the unit is switched off (via its front panel controls or the mains power supply), what type of supply is being used, the types of locos involved and so on. It can range from a minor issue to one serious enough to cause derailment. While this can be solved by taking the locos off the track or disconnecting the controller from the track before switching it off, that’s inconvenient. So we set about figuring out why this was happening and how to fix it. The cause Take a look now at the circuit of A K FWD VR7 1k D 14 Vdd Q S 1 K 8 IC4a CLK R REVERSE λ LED2 Q 2 9 6 4 11 S3 D S Q IC4b CLK Q Vss R 10 7 13 12 TP2 LEDS siliconchip.com.au 5 1 µF B A A BC327, BC337 B K A E G C Fig.1. This shows the relevant sections of the original circuit published in the July 2013 issue but with a number of changes shown in orange. Ignore the changes for the moment while we discuss the problem and how it occurs. Comparator IC3b generates the PWM waveform to drive Mosfet Q1, which switches the supply voltage to the tracks, controlling how much power the locos receive. This works by comparing a 160Hz triangle waveform to a control voltage, with the control voltage indicating the desired loco speed; the higher the control voltage, the higher the output PWM duty cycle and thus the higher the motor speed. The control voltage is low-pass filtered by an RC network, to prevent it 7812 IRF1405 D D GND IN S GND OUT from changing too rapidly and also to simulate train inertia. The amount of filtering applied depends on whether or not the inertia switch is on and the position of inertia control pot VR4 but regardless, there is always some filtering of this signal. When the unit is switched off and its power supply capacitors discharge, the power supply to the op amp generating the triangle signal collapses and so the triangle signal’s voltage drops rapidly. But this filtering of the control signal causes the control voltage to drop much more slowly. In other words, the 47µF capacitor at pin 5 of comparator IC3b remains charged for some time after power is removed. This means that at switch-off, the January 2014  69 DC INPUT TERMINALS TERMINALS TO TRACK LED3 POWER 2.2k 4004 4004 2.2k 100nF 10k 47k S3 A FOR/REV 47 µF LL IC4 4013B 10nF 1 S2 S1 BC337 TP1 100k LEVEL 4.7k INERTIA VR4 1M INERTIA 10nF TRACK TRACK Q3 100k 100k 10k 470k 100k 1 RUN/BRAKE VR1 10k TP GND 220k 10k SPEED 10k 100k 4.7k 100 µF 250kΩ VR5 STOP 4148 100Ω x ADD WIRE VR2 CUT BOTTOM LAYER TRACK 470Ω 10k K D1 D5 VR6 10k D3 MODEL TRAIN CONTROLLER 1 µF IC1 LM358 ZD2 8.2V 10k 4148 D2 IIC2 C2 LM324 10k MIN. A IC3 LM393 1k 10k MAX. 470Ω 1k 100 µF 220k 22nF 1 VR3 S4 POWER C 2013 1 µF MMC 100nF TP2 220nF REG1 7812 NIART LED O M RELL ORT N O C 0 910 76 131 0190 131 1M 4.7k 1 µF 10k 1 VR7 BC337 Q5 BC327 2.2k R2 100 µF 4.7k Q4 OVERCURRENT R1 COM NC PIEZO LOW ESR 100 µF 4148 + 2200 µF 25V 47k 4148 + F1 D6 D7 D4 NO 10k LOW ESR RELAY1 4.7k 2200 µF 25V LOW ESR 0.1 Ω 5W 2200 µF 25V 10Ω + 0.1 Ω 5W ZD1 1k 8A + Q1 2x IRF1405 FR607 Q2 DC IN 0V 15V 1W DC IN +17V LED2 REV LED1 TRACK Fig.2: here’s how to make the changes to the original PCB to eliminate the switch-off lurch. The changes are all indicated in red and are easy to do (see text). control voltage rises relative to the triangle waveform (by dint of the triangle voltage dropping) and so the PWM duty cycle increases until it reaches 100%. It stays at 100% until the power supply has collapsed to the point where there is no longer sufficient voltage to turn the Mosfet on, at around 3-4V. This can take a significant fraction of a second. During that time, the full input supply voltage, typically around 17V, is applied across the tracks. Hence the sudden jerk from the locomotive(s). This can happen regardless as to whether the Li’l Pulser’s power switch (S4) is thrown or its power supply is turned off at the mains outlet but it tends to be worse when switched off via S4. That’s because if the supply is switched off at the wall, the Li’l Pulser’s input capacitors remain in parallel with its output capacitors and so the supply voltage drops more 70  Silicon Chip slowly. Depending on that amount of capacitance, the supply voltage may drop slowly enough that the control voltage drops as fast or faster, preventing any output pulses. The solution We have taken a two-pronged approach to solving this. The first set of modifications pretty much eliminates the jerking and can be easily applied to existing PCBs. We have also produced a revised PCB which incorporates these changes and will supply this to new constructors. The revised board also incorporates a few extra components which provide further protection against a switch-off pulse when power is switched via S4, which as described above, tends to be the worst case. These circuit changes are shown in orange on Fig.1, as noted above. First, we have taken the power-up reset circuit, based around op amp IC2c and converted it into an under-voltage lock-out, which still also performs the original reset function although by a different means. Originally, an RC filter from the 12V rail, connected to pin 10 of IC2c, provided a time delay. This was compared against a reference voltage at pin 9, which was derived from the outputs of the min/max speed buffer op amps IC2a & IC2b. This is the same reference voltage used by op amp IC2d (at pin 12) to time the switch-over of the reversing relay. In operation, some time after poweron, reset is asserted and Mosfet Q1 is held off until the capacitor at pin 10 charges to a higher voltage than the reference. The reset is then released and normal operation begins. For the new circuit, we drastically reduced that capacitor value from 10µF to 10nF, effectively eliminating the time delay. Instead, 8.2V zener diode ZD2 plus the voltage divider formed by the 470kΩ resistor and an additional 220kΩ resistor prevent the reset from being released until the power supply voltage has risen past about 11V. This takes some time (for the supply capacitors to charge, etc) so despite the much smaller capacitor value, there is still a reset delay at start-up. This 11V threshold must be reasonably accurate; it has to be below the minimum supply voltage, or else reset will not be released at power-up. At the same time, it can’t be too far below the supply voltage as we want reset to occur shortly after switch-off, before any unwanted output pulses can be delivered to the tracks. To this end, we have changed the reset reference voltage from one which varies depending on the positions of VR2 and VR3 to a fixed 1.09V (nominal) derived from an existing divider across the 12V rail (10kΩ/1kΩ). Pin 10 must rise above this voltage in order for the reset to be released and since the 470kΩ and 220kΩ resistors form a roughly 2:1 divider, that sets the threshold at 8.2V + 1.09V x (470kΩ + 220kΩ) ÷ 220kΩ = 11.6V. In practice, at the low current it is being operated, ZD2 will be at the lower end of its breakdown voltage range, so the actual threshold will tend to be closer to 11V. The minimum output of REG1 is 11.4V but also consider that the 1.09V reference is derived from the supply voltage and so the threshold siliconchip.com.au DC INPUT TERMINALS Extra Parts For PCB Modifications TERMINALS TO TRACK Making the changes We made these changes to our prototype and it no longer causes any noticeable motor pulse at switch-off. Fig.2 shows what is required. Start by removing the 470kΩ resistor to the right of IC3, the 10kΩ resistor directly below it and the 10µF electrolytic capacitor to the left of S1. Since it’s a double-sided board, it has plated through-holes so the easiest way to remove the resistors is to clip their leads off close to the body, then pull the stubs out with pliers while heating the solder joints. The holes can then be cleared with a solder sucker. The electro can be rocked out while heating the pads and gently pushing on the body and its mounting holes cleared of solder too. Next, cut the track to pin 9 of IC2, on the underside of the board (shown in Fig.2 with a red ‘x’). Fit a fresh 470kΩ resistor and ZD2 to the pads originally used for the 470kΩ resistor, siliconchip.com.au VR1 10k TP GND LED3 POWER 10nF Q3 2.2k 2.2k 100nF 47k 10k IC4 4013B 10k 10nF 1 4.7k INERTIA VR4 1M BC337 TP1 100k LEVEL 4.7k Q6 4004 4004 100k BC337 S3 S2 S1 47 µF LL A FOR/REV 10k 250kΩ VR5 STOP INERTIA 10k TRACK TRACK D6 FR607 100k 100k 10k 1 RUN/BRAKE 10k 220k 470Ω SPEED K D1 D5 4.7k 100 µF 100k 470k 4148 4148 IC2 LM324 10k 10k MIN. VR2 S4 A 220k D3 D2 VR3 100 µF 10k 10k IC3 LM393 1k 1 µF MMC 1M 470Ω 1k 2.2k POWER C 2013 TP2 220nF REG1 7812 1 µF VR6 10k 1 MAX. 10k 4.7k 1 µF 100nF VR7 OVERCURRENT 100 µF 4.7k BC337 MODEL TRAIN CONTROLLER NC 22nF 1 NO 100k Q5 BC327 RELAY1 COM IC1 LM358 Q4 100 µF R2 ZD2 8.2V PIEZO LOW ESR R1 100Ω 2200 µF 25V 47k 4148 + 0.1 Ω 5W 10Ω D7 0.1 Ω 5W ZD1 LOW ESR + F1 15V 1W + 2200 µF 25V LOW ESR 0 910 7 13 4 drops somewhat if the supply is on the low side. Now since the 10nF capacitor only provides a very short delay (with a time constant of 10ms) and with a threshold of about 11V, once the unit is switched off, the 12V supply doesn’t have to drop by much before it enters the reset state which forces Q1 to stay off while the supply voltage decays to zero. + 2200 µF 25V Q1 2x IRF1405 D4 Additional parts 1 BC337 NPN transistor (Q6) 1 8.2V zener diode (ZD2) 1 10nF MKT capacitor 1 220kΩ 0.25W resistor 1 100kΩ 0.25W resistor 1 10kΩ 0.25W resistor 1 100Ω 0.25W resistor Deleted part 1 10µF electrolytic capacitor Q2 DC IN 0V 1k 8A Parts List Changes For Revised PCB DC IN +17V 4148 1 8.2V zener diode (ZD2) 1 10nF MKT capacitor 1 470kΩ 0.25W resistor 1 220kΩ 0.25W resistor 1 100Ω 0.25W resistor 1 short length light duty hook-up wire LED2 REV LED1 TRACK Fig.3: follow this parts layout diagram if you’re building a new unit using the revised PCB (09107134). This version also adds transistor Q6 and two associated resistors. with the cathode of ZD2 to the top of the board and ‘air wire’ them together. The 100Ω resistor and 10nF capacitor can be fitted as usual, with the added 220kΩ resistor wired across the new capacitor under the board. Finally, run a short length of insulated wire (eg, Bell wire or Kynar) under the board, from the now-isolated pin 9 of IC2 to the top-most pad of VR7. When you reassemble and test the unit, you should find that it now operates as before but without the switchoff pulse from the motor(s). If the unit fails to operate, check the voltages at pins 9 & 10 of IC2. Pin 10 should have a slightly higher voltage when power is applied; it’s unlikely that it won’t but if not, you may need to change ZD2 to the next lowest voltage (eg, 7.5V). New PCB & further changes To make it easier for new constructors, we can now supply a revised PCB for the Li’l Pulser, incorporating all the modifications. This new PCB is coded 09107134 and is available via the SILICON CHIP Online Shop. Fig.3 shows the overlay diagram for the revised PCB. You will need to refer to the original assembly notes in the July 2013 issue when building it. In addition to the above changes, we have also added NPN transistor Q6 and two more resistors so that as soon as S4 is switched to the off position, Q6 turns on and rapidly discharges the 47µF control voltage filter capacitor, so there is no possibility of an output pulse regardless of how quickly the under-voltage lock-out circuit kicks in. This is a bit of a ‘belts and braces’ approach, ie, it isn’t totally necessary but it provides some extra cheap insurance against any sort of output pulse being delivered to the tracks. With these changes the unit will now behave itself at switch-off but SC otherwise operate identically. January 2014  71