LIBRA-UL-R repair on the Phobos Gate Opener System
more than two years of operation, the controller stopped working of the
Phobos BT L system at the
end of 2016. I wrote the
company via the website multiple times,
but received no response. From inspection, it was clear that
R23 power resistor (above the bridge heatsink in the previous image)
was burned up. I could not tell the original value, so I had
to removing the board, I checked the power supply voltage at the white
terminals at the transformer, and measured 33Vac (labelled as 25V on
the sticker). Taking the board to my workbench, I was able to
power up the board with a 24Vac
transformer and could then measure voltages and trace the schematic.
of the solder side of the board. The main driver for the
is U20 in the middle of the photo (pin 1 is top-right).
only traced the Power/IO section of the board. This is the
section most likely to fail and besides, it would be nearly impossible
to find a replacement for U6, the main chip of the logic section.
Power Supply Section
Referring to the schematic,
we start at the bottom of the page for the power section.
Incoming AC power is first filtered by C1 and RV1 and then
applied to JP9 terminals #11+12 as the Aux output, and also D1 for full
wave rectification. This creates the B+
bus, which is used to power the motor. At an input power of
28Vac, B+ will measure about 26Vdc. D20 isolates B+ from the
bus, which is filtered by the C3 2200uF electrolytic capacitor.
Due to the filtering, this bus will measure higher, or around
in my case. After that, this bus powers the red LED (LD1),
28V zener preregulator formed by R30 and D24. This in turn
the 5V regulator.
Motor Control Section
will focus first focus on how Motor2 is powered. Current for
is switched by K2 and K3. When these relay coils are
the motor is shorted, causing the motor to act as a brake.
Throwing K2 (power from U20-15) will cause B+ application to
terminal #4 on . If then Q4 is also turned on (via
from U6), current will flow through the motor and use R4 as the current
sense shunt. Per the note in the Project Log below, the K2 is
the CLOSE function. The opposite occurs when K3 is thrown,
runs in the opposite direction. The circuitry inside the
and its limit switch is a guess, but I am pretty certain this is
correct. It causes the switch output to go to B+ when that
is closed, which pulls down the open collector output to U6.
an interesting and initially puzzling action occurs before the motor
runs. That is when the remote button is pushed. The
resulting action is a 30 msec pulse low on the U20-12 pin (Sensor
Check). Via Q22, R23 and D25, this applies C+ to both
of the motor and Q4 (neither K2+K3 are actuated yet).
Q4 is OFF, but in the middle of the 30 msec pulse, MTR_DISABLE goes low
for 10 msec, and that causes about half an amp of current to flow in
Q4. This results in about 50mV across R4 for 10 msec, which
routed to the op-amp at U21. My best guess is that this 30
pulse is a sensor check. It allows the U6 controller to tell if
the system is up against a limit switch, and if the current sensor is
working. As an experiment, I shorted the base-emitter
Q20, causing the lack of the 50mV pulse. The result was that
motor move was immediately aborted, and none of the relays
actuated. This matches what occurs with my original
with the blown R23 and Q22.
Q22 is ON, the instantaneous power on R23 is 17W. This
is rated for 1W per my inspection. It is unusual for these
resistors to fail shorted (they
overheat and open), so my guess is that Q22 failed shorted in
the original controller.
of Motor1 is similar,
except power to it is not applied until about 920mseconds after Motor2
runs. This delay allows proper phasing of the gates.
Flasher output is composed by diode ORing all four motor output
terminals with diodes D27 through D30, and applied to #9.
from this feed is returned through steering diodes D31+D32.
interface from the logic section to the Power/IO section is nine logic
Two of them are the MTR_DISABLE lines mentioned above.
remaining seven get
buffered by U20. Momentarily
jumpering the lowered numbered pins to 5V causes a relay to throw and a motor
action to occur. These pins and functions are:
Motor 2 at +25V (Measured at JP9-3 to JP9-4) / OPEN
Motor 2 at -25V / CLOSE
Motor 1 at +25V (Measured at JP9-6 to JP9-7) / OPEN
Motor 1 at -25V / CLOSE
Enable Sensor Check (active low)
Small relay K8 (probably area illumination output on JP9)
Small relay K4 (probably "Safe" output on JP9 to power
light beam devices)
One thing that you can do is jumper these to 5V one at a time
and see what power output
does not work. Note that these logic control lines into the ULN2003 all have a
capacitor to return allowing you to find them easily.
the technical manual, the max working current to the motor is 3.5 Amps,
and 1.5 Amps nominal. So a good dummy load is about 20 ohms.
the case of my board, repair was not possible so I purchased a new one
for $211 (link below). With the new (working) controller in
I could better trace the schematic and could now read the value of R23.
Logic Section The main processor is the H8/3672 from Hitachi. From the hardware manual:
This LSI is equipped with ROM, RAM, an 8-bit timer (TMR), a 16-bit timer, a watchdog timer
(WDT), two types of serial communication interfaces (SCIs), a 10-bit A/D converter, and I/O
ports as on-chip peripheral modules. This LSI is suitable for use as an embedded processor for
high-level control systems. Its on-chip ROM is flash memory that provides
flexibility as it can be reprogrammed in no time to cope with all situations from the early stages of
mass production to full-scale mass production. This is particularly applicable to application
devices with specifications that will most probably change.
search shows this chip is not available in retail distributors and
needs to be purchased in bulk from sources in Asia. The markings
on one of my chips is:
first line is probably the actual part number, and the second line the
generic number. The third line may be some kind of production or batch
code, and I am guessing the last four digits is the date code, which
would make the date of manufacture the 38th week of 2011.
The main chip
is connected to a four-digit numeric LCD display with 18 interface
pins. I have not been able to find much about it, but it looks
like a simple direct multiplexed interface
from the processor to the display
segments. With 7 segments, a decimal point each, and four digits,
the minimum number of pins needed is this sum, or 12. During a
move, the display appears to indicate the peak current level of the motor and
should be multiplied by 4.5x to get the current in Amps. The
value is held while the move is ongoing, even if the limit switch is
hit. The display then goes blank once the move is completed or
Repair Case #1 I
was sent a broken unit in August 2020 and was able to repair it.
It failed in the same way as ours and I could verify that Q22 was
shorted. This would have led to R23 burning open. I was
unable to tell much about Q22 and the markings are "AL W72".
Based on the resistors around the control pin, I am pretty sure
this was originally a MOSFET. However, if you calculate the Vgs,
it is a very high value, and I think this would have caused long term
stress. This is probably a vulnerability in the design. I
addressed that in the repair by changing the resistors around the
Damaged Q22 transistor.
decided to build an interface board to help test the controller, which
is shown below. It allows me to easily tell if the motor has
power and the polarity (red/green LEDs). It also allows me to
easily start and stop the motor as you can see in the video.
There are also equivalent buttons for the travel limit switches.
The Libra Tester board. Start/Stop button is on the right. Red / Green LEDs indicate polarity of the motor power and the switches on the left are the travel limit switches.
Video of repaired Libra.
Photo of new MOSFET installed onto the back of the board (middle right). Note the old FETs is the size of the ones in the bottom left. New one is much bigger and stronger.
Repair Case #2 Information will be posted in the future on this repair case.
the light of the Libra being phased out and replaced by the (much more
expensive) Thalia, I think it makes sense to repair these boards.
Send me an email if you want to have me do it.
Spring 2014 - BFT dual gate and controller installed
December 2016 - Gate controller fails and this blog started
January 2017 - Controller successfully replaced and
2018 - Got an email from Frank W that his single motor system only
would open and not CLOSE. With the help of this blog, he
it down to a bad K2 relay, and got his system working by swapping it
for his K6 relay (he only has a single gate). This proves
that K2 is for the CLOSE
2019. I have purchased numerous after-market remotes on Ebay
all seem to work well. We prefer the type with the sigle row
buttons. See link below.