You need a desoldering station for the electrolytic caps swap job. Chinese made budget tool is the ZD-915 (or 917) which costs around 70 to 150 €. They might have stuck-sluggish starting vacuum pumps, but with some manual help they start pumping - maybe just a motor lubrication problem or bad electronic design?
A hint on the through-hole desoldering work: use proper desoldering tip temperature, not too hot. Never use force when pulling out a component or you break PCB's through hole rivets, which have inside-layer connections! Do not forcefully push and twist or grind the desolder tip against a PCB pad, this will damage the PCB copper pads&traces. Instead, add a bit of new solder to form a good thermal bridge and once the solder joint melts, suck for some 2+ seconds and slightly wiggle the caps from the other side to make the leads clear away from the hot rivet walls, while the fast airflow through the rivet cools the cap's lead and it will not re-solder back on to rivet's walls. If the lead is removed but the hole is not open, add new solder and re-suck. Sometimes you need also to heat the hole from the other side with sharp tip soldering iron and then suck from the other side. After a new capacitor is soldered in, use some softish plastic tip tool to scratch off tiny splashed tin droplets which may cause short-circuits. Brush off all loose tin and resin from PCB. Inspect every solder joint's quality with a magnifying loop! It may be wise to overhaul just one card, or module at a time - test the MT8802 still works before resuming work on the next card. If at some stage it does not boot up any longer, find out what went wrong with the last card you just worked on.
Bag each unit's screws once you remove them and label the bags accordingly. At some stage you have a big pile of various sized screws around and it's time consuming to learn where they belong.
These instructions should apply to other MT8802 parallel models (A,B,C). Proceed with your own risk and be very careful.
Enclosure top and bottom panels open up by removing two M5 screws from center backside of top and bottom cover panels and removing all the green padded legs from all corners (each secured with 1 or 2 screws) and the folding the support bracket from underside. Slide back / lift away the top and bottom panels. If you work with main PSU or Converter unit A10, also remove the side cover panels by opening 8 + 2 +1 screws and then slide the side panel backwards.
Trimmer resistor oxidation prevention tip:
Spray some vaseline type contact grease on every gray trimmer you see on the PCBs and those blue trimmer resistors on the card edges, but: do not soak the RF units with contact grease, it may cause problems.
Dead 12 V STBY power suppy:
STBY power supply failed, fuse blown. TO-220 MOSFETs IRFBF20
The G to S in short circuit. The 4 mm x 5 mm full-wave rectifier also had to be replaced.
(Info: thanks to Tieguan Shao)
Also an other fault found: the broad-band high power MMIC was blown in some RF unit. (perhaps meaning the SG output MMIC discussed below)
This describes the main PSU SBP-172A service:
Unplug AC cord, then the 12V STBY PSU plugs, open 3 screws and remove the STBY PSU. Loosen 2+1 screws holding the main PSU on bottom plate, slide it, so the front side peeks out of the side frame's hole and then unplug the DC connectors (4 pcs) and AC input connector. Pull it out and remove all screws holding the main PSU aluminum covers, so you end up with exposed PCB underside visible. Replace all small electrolytic capacitors:
1 uF 50V *3 on the small vertical PCBs and 2.2 uF 4.7
uF, 22 uF and 47 uF all 50V electrolytic caps (C20, C11, C9,
C52...). Their ESRs are probably gone high (ranging from 7 ohms up
to over 10 kohm (22 uF totally toasted) if the PSU makes a hissing
noise and not just a very high pitched whine.
This describes the A131 SG LOCAL SG 1 and A10 CONVERTOR service:
The MT8802A showed spurious signals on the spectrum analyzer,
about every 500 kHz +- 5 MHz away starting at levels -60 dBc, and
also signal generator output was not clean.
The most probable reason for this would have been noise getting to YTO control causing FM signal component to be added to it and thus eventually polluting the mixer(s) output(s). The mechanism may though be more complex.
First you better remove the top cabinet panel and the right side handle panel, which exposes the A10 and A15 units. Also remove card stack topside bracket plate and the the slotted card stack top panel and note which M3 screws are long or short or countersunk
Unplug SMA connector from A10 unit's middle low side, remove screws from A33 unit from top of A10 if it gives you troubles when removing the A10 converter. Loosen the base holding screws (2) with a very long screw driver, open the hard line SMA connector on top (A33), lift the A10 convertor unit out gently, but watch the front KB module wires from not being torn as you lift. Unplug another SMA connector from the front side low end on the A10 unit. Unplug A33 unit's X7 multi pole connector too and put A33 aside.
Take detailed digicam photos once unit's shielding covers are opened, so you have good documents where the A131 SG LOCAL's YTO drive wires go on the PCB and all those feed-through caps wires to PCBs. Also mark up the wires and their positions, if possible on the PCB the best you can.
Both PCBs on the A10 converter unit have electrolytic capacitors, 12 pcs of 47 uF 50 V, one 22 uF 50V and some 3 pcs 100 uF 50V. To replace those you need to unsolder several feed-through's pins coming from the shielded enclosures and open some power transistor securing screws before the PCBs can be folded up to replace those electrolytic capacitors.
There are some caps not soldered through the holes, but just installed on their sides. If no other properly mounted through hole radial caps exist on those PCB's do not remove the PCB from the slot, but just unsolder the caps on the topside, or cut lead off the old cap and re-solder new one on the old leads. Not by the books, but saves a lot of time.
Signal generator output MMIC amp. rebuild:
The SG unit's output amplifier has a big rectangular golden HiTite MMIC amplifier chip HMMC-5104, which is mounted on the milled unit aluminium body. If you decide to remove the PCB, you might as well see if there is good thermal contact from the big MMIC to the aluminum - the little thermal past there is, should be evenly spread. In my unit there was almost no thermal contact and the reason was the small threaded holes on the aluminum slab edges, had swollen the slab - it was bent and far off from being flat. I filed it to be more flat and added a bit of thermal paste and remounted the MMIC and the PCB with those small screws. This hard-to-find and expensive HMMC-5104 MMIC failed 7 years later. It was replaced (rebuild) with cheap SNA-586 (or NGA-586) MMIC. Removed the HMMC-5104, cut brass sheet in same shape as the flange with 4 holes and 5 mm wide double-sided FR-4 PCB strip (line) that was soldered in the middle of the new flange and the strip line cut from middle to install new MMIC there. MMIC ground pins were connected with solder using 3 mm wide brass strips to the flange and the flange to unit PCB ground pads on both sides with 3 mm wide brass strips.
SNA-586 needs only about 5 V and 60 mA - there is the old
HMMC-5104's D3 supply that should have +8 V: use series resistor
(see SNA-586 datasheet for value, should be 47 ohms ½ watts from
+8 V) to feed the MMIC DC from the small RF toroid choke to
SNA-586 output side on the strip line. Also the input and output
strips are connected with 3 mm wide brass strips to unit's PCB.
All other HMMC's soldering pads with D1, D2, and G biases are left
unconnected. The new brass flange is secured with original 4
screws, but with compression washers. CPU type silver paste migh
be better heat conductor under the flange, but maybe it is not
necessary as the SNA-586's Ptot is only 0.3 watts, a lot less heat
There is no need for MMIC with +29 dBm output, or is
there? The AUX output connector is routed via relays
and step attenuator to SG output, so excessive power fed to this
connector, if the step attenuator is set at 0 dB attenuation (max.
SG output at AUX +7 dBm) or close to it, reverse RF may damage the
SG output stage MMIC. Using Main input/output connector, there is
20 dB extra attenuation from the power attenuator and the risk is
less. If the above is of concern and you have +29 dBm a MMIC (like
MAAMSS049, GVA-81+ etc.) it be better to opt for that MMIC, just
feed the DC it needs using info from datasheets. The front panel
has no warning on AUX out connector on maximum allowed RF
power level, the manual says -3 dBm (0.5 mW) and "Do
not input external signal to the output connector".
This output stage MMIC is driven with -8 dBm from the
two-stage MMIC driver, the output is +13 dB and SNA-586 can easily
provide that level and necessary gain and take the -8 dBm input up
to 3 GHz. Also ERA-5SM+ might work. The AGC system levels the
output to constant +13 dBm, that power is seen at AUX connector as
+7 dBm maximum. I noted there is slight lever rise with frequency,
up to +1.4 dB from nominal (at 50 MHz) on the S.G. unit's SMA
output connector. This migh be designed feature to compensate
losses on internal cabling, step attenuator and relays.
While working on the MMIC replacement, a thin strand of metal had
sneaked inside the 4 GHz DRO's (there is a hole on one corner!)
over the PCB resonator line, and it pulled the PLL off lock and
made frequency unstable. There was no need to retune the DRO, once
the metal stand was removed. The PLL locking range of the screw is
about 45 degrees.
The 4106 MHz cavity filter on the unit's edge may do with some
re-tweaking, use voltmeter at SG power amp's PCB's AGC voltage to
tune the slugs, but note that there could be saturation and the
last tweaks seem like they have no effect, but all the cavities do
Also at the same time, a nasty fault in the IDC card edge connector developed; 1B line's pins 1 and 2 got pushed inwards and intermittently lost contact with the backplane receptacle. This made the SG synthetizer to be dead or on stuck some odd frequency and did not tune, or if it did,
when securing the SG unit firmly with two screws - lost signal after the tuning. Check the IDC card connector pins are still OK! Not bent, not pushed back.
This describes the A15 REF service:
There are also electrolytic caps to swap in the 10 MHz crystal
oven unit (A15), installed vertically on the right rear corner.
Easy to un install: remove top bracket and lift up, open covers
and change the caps.
2rd IF: 6.8 MHz
DSP demodulator at 32 MHz sampling rate + IF power detector
RF output: AUX +7 dBm, or MAIN via -20 dB attenuator &
audio signal generator, output to A13 SG1
STBY power supply failed, fuse blown. TO-220 MOSFETs IRFBF20
The G to S in short circuit. The 4mmx5mm full-wave rectifier also had to be replaced.
(Info: thanks to Tieguan Shao)
This describes the CPU card rack (9+3 cards) service:
Please pay attention to avoid ESD damage to these PCBs - use wrist strap.
It may sound odd, but the more electrolytic capacitors are
replaced in any/all of the units and cards, the lesser the
spurious signals (as seen in spectrum analyzer with 5 MHz span)
These cards have some 240 pcs of low profile radial electrolytic capacitors, ranging from 0.15 uF to 220 uF from 16 to 63 Volts. If you decide to replace them, every electrolytic capacitor has to be aligned in right polarity! The PCB is marked but just make sure you look at right corresponding caps "+" sign. I did this all, you can do it too without any mistakes.
I measured the ESR at 10 kHz for the old caps once desoldered and there were several with ESR of over 200 ohms with just a fraction of original capacitance left. Most of them were those of sizes under 10 uF and the 47 uF 16 V caps. Comparing to new caps, about 75 % of all electrolytic capacitors in the MT8802A were clearly deteriorated with elevated ESR values in the range of 5 ohms to 20 ohms. It would be waste of time to even try to locate which are the ones gone totally bad, just replace them all! It does not cost that much, tough it takes, say a full working day or two, to complete.
Use capacitor can size heights (low profile) as listed which fit on the PCB (stacking of cards, or having aluminum enclosures limits the can height!) with +105 C high temperature handling types and preferably lowish ESR if available and you can afford them. Avoid using cheap +85 C electrolytics. 100 uF and 220 uF 25 V caps were of +85 C type and also high canned. Why not use here 105 C caps with low can as all others.
Mark the only interconnecting SMA cable's plug's position (it goes to A11), so you do no plug it back by mistake to adjacent card's void SMA connector.
Mark card edge raiser tabs with the corresponding A## number as marked in the frame, so you do not loose track of which card goes to what slot.
The three M00008A DEMOD RF, A11 LOCAL and MM00001 IF cards have cast Aluminum enclosures bolted on the PCB, some on both sides. These also include electrolytics to replace. Mark first every Aluminum cast box with numbers and also their alignment position, as some of them are unsymmetrical and some have RF absorbers in the other end only. When you open these cast boxes, look for loose and fractured carbon RF absorber pads. Replace the cracked or pulverized ones with black ESD foam and glue the loose, but still usable pads back on the enclosures top wall with super glue.
In one of the BASE BAND cards, there are two 10 uF caps directly soldered to SMD IC chip pins. Just solder new caps to old caps leads before cutting them out, instead of trying to desolder the leads from the chip and solder new caps to SMD chip's pins directly - too big risk to fail and no big gain.
Last: check the CPU card's Lithium cell's voltage, if under 3 V,
might be time now to replace it (carefully).Recommend soldering in
the new cell (with connection wires) before cutting out the old
Li-cell, so there is no loss of DC.
After every card and module is overhauled and checked, reassemble, but don't put back on the card stack top covers yet. Just power up. What happened to me, was the booting screen check list OK's were as before, but then the MT8802A beeped and the screen went blue (blank), key LEDs lit up and the keyboard appeared dead. The manual instructs show how to install programs from diskettes (if you have them?). The unit I have has PCI card and I think it's were the SW is. So I intuitively started the unit with Preset key down and this time the boot screen showed OK also on SRAM (normally "un checked") and while keeping Preset down (and/or hitting it), after a while, the screen flashed alive with the usual GSM setup screen. I do not know what caused the boot failure, replacing the Li-cell later caused no such problems.
uF V (recommend pcs to buy) (can height)
0.15 50 (2) replace with polystyrene or ceramic 150 nF 50V (6 mm)
1.5 50 (5) (6 mm)
2.2 35 (5) (6 mm)
2.2 63 (5) (6 mm)
4.7 35 (10) (6 mm)
4.7 50 (5) (6 mm)
10 16 (20) (6 mm)
10 25 (10) (6 mm)
10 35 (15) (6 mm)
22 16 (10) (6 mm)
22 35 (15) (6 mm)
22 50 (5) (6 mm)
47 16 (80) (6 mm)
47 25 (15) (12 mm)
47 50 (45) (10 mm)
100 16 (30) (8 mm)
100 25 (5) (12 mm, recommend 10 mm)
100 35 (20) (10 mm)
220 16 (5) (10 mm)
220 25 (5) (12 mm, recommend 10 mm)
220 35 (5) (10 mm)
total: little over 230 pcs to replace, 322 pcs to shop.
Note: the amounts per rating are based on a quick "trash bin count" - amount of each sort is rounded upwards to next 5 or 10 pcs. You have less than 80 pcs of leftover caps after job is done, but you should not have ran out of stocks either.
This describes the LCD and keyboard panel service:
After trying new CCFLs, taking out all diffuser foils between LCD
backlight and the panel itself, I found still the display to be
unevenly lit and too dark to read in normal room light
I can not say if this is partially just poor performance of the Panasonic LCD or is it just worn (solarized) over time by CCFL's UV rays, but the work described below, proved futile and lead to efforts in finding spare LCD panel. Unfortunately, the Sharp (LQ9D168K) from Helmut Singer Elektronik is not compatible (LVDS connector) with unit I have with Panasonic LTM08C015KA flat males, 15 pins and 10 pins. Sharp LQ9D168K comes with a 31 pin connector as service manual says - so far unresolved issue!!!.
The CCFL replacement process (no help in my case! ,Roy Singh suggest tearing off the EMC mesh on the face plate as it adsorbs some of the brighness - which though may cause RF leak through the display screen's hole):
CCFLs used in MT8802A: 190 mm long (glass-to-glass) and 3 mm diameter, 2 pcs.
Remove four screws and pull out the gray plastic frame around front panel. Unscrew the two DUT Interface's connector securing screws. Next unscrew the now-exposed M4 countersunk screws around the aluminum front panel's edges, all of them, except the 3 smaller countersunk ones (M3) located front just above LCD screen topside. Cut carefully the white plastic cable ties (underside) securing gray shielded AF input/output cables leading to front panel AF in & out BNC- connectors. Mark their socket positions on the bottom back plane and unplug them. The RF connectors are mounted on body frame and stay there. Gently pry out forward the face panel, unplug all flat cable connectors leading from the front panel unit to the main unit and slide out the AF cables, so you now can totally remove the face panel away and work on it on your work bench.
Remove dust from the face panel as you proceed disassembling. Unscrew the LCD shield aluminum plate and resume by unplugging both two CCFL (cold cathode fluorescent lamp) connectors from the now exposed HV inverter unit, mounted on left side of the LCD's back. Dismantle (by removing four M3 screws and four small screws holding CCFLs) the plastic LCD backlight illumination diffuser - polarizer stack along with the CCFL assys.
The Toshiba INVC193A inverter seems to have two aluminum SMD electrolytic capacitors (10 uF and 2.2 uF 50V). They eventually fail and might destroy the whole inverter - replace those now! The usual first symptom is they begin to run hot. Their ESR's had gone to 3-fold on the 10 uF and 20-fold on the 2.2 uF @ 10 kHz in one such unit to be repaired.
Very gently bent open the tabs holding the CCFLs assys' ends and remove the reflector & CCFL cable harness from the backlight diffuser. Inspect if the lamps have dark ends. They usually do and might be worn enough to be replaced (www.CCFLoutlet.com). Generally a typical symptom of worn out CCFL lamps is they become redish in hue and a bit dim. I have seen some that give very faint neon like light yellow only eventually.
Next also remove the Panasonic LCD unit very gently and if dusty, clean and store safely, until time to reassemble. Do not touch any optical suffices or damage the fine EMC screen mesh.
You now have left only the keyboard processor card and underside the green keyboard PCB. Pull out the dial wheel knob from outside (no locking mechanism, may help if you turn the axle screw from behind the encoder while holding and pulling the knob). Unscrew the five M4 screws holding the keyboard processor card and unplug the flat keyboard cable connector and lift the card away, clean off any dust from the encoder with canned air. Next, unscrew the countersunk M3 screws (many!) from the keyboard PCB and lift it out with the rubber keyboard mats. Remove the rubber keyboard mats, if dirty, soak in mild soap, but do not rub, or use any chemicals on the black carbon contact pads in the back side of the keys! Dry the rubber mats. Use abrasive fiber pen or other mildly abrasive material to clean the gold plated PCB keypad contact surfaces. Though gold plated, they seem to turn brownish yellow (oxidize) and this eventually causes bad working keyboard. Clean all dust away from PCB and install the keyboard rubber mats on the aluminum front face plate holes and drop the keyboard PCB over it. Make sure every gray rubber tab comes through the holes in the PCB, then fasten the PCB with those M3 screws. Check the keys feel loose and working before re-installing the keyboard processor card and dial knob. Re-attach flat cable connector from the keyboard PCB to keyboard processor unit.
Re-assemble the Panasonic LTM08C015KA LCD display unit and
the backlight stack with HV inverter and everything you have
removed or disconnected (also the AF shielded cable connectors
under the unit to the back plane), so that the front unit is
eventually mounted securely back on the main unit with all cable
connectors properly attached. Reconnect AC cable and start up the
MT8802A to see if the backlight system, LCD display and keyboard
keys all work properly now? If so, power down and finalize the
work by assembling back the plastic frame and top and bottom cover
(see no loose flat cable gets pinched!) panels with all
The INVC139A CCFL inverter later failed (blew the on PCB
fuse) soon after the SG output MMIC, the encapsulated version of
HMMC-5004 also failed. Found a used working DS-1305WK
CCFL inverter, unfortunately none of the connectors mated and
the PCB had to be installed horisontally with one screw (GND
foiled one) and a piece of insulation material to support free end
of PCB with some Pattex glue. The new inverter's own cables must
be connected to the PCB, on original INVC the pin 1 is GND (white)
and pin 5 is +12V (one of many grays in the MT8802A), that is all
you need from the old CCFL cable. The DS-1305WK side cables: CN1
pin 3 black = GND, CN1 pin 10 Red=+12V, CN1 pin 1 +5 V
(datasheet says BRT, but works more like disable=0V): use
6k8 and 4k7 resistors in series from +12 V to GND and take +5 V
from the middle tap. There are two pads to solder the resistors
chain ends to (GND and +12V pads). If you did not save the HV CCFL
side sockets, you need to remove the HV connectors on the new
inverter board and solder the CCFL cables directly on the HV
This describes the cooling fan replacement and air-flow reversal modification (may not appy fully on MT8801C):
The MT8802A unit uses two 80 mm DC fans to cool the cabinet (seems the MT8801C has 92 mm fans though, and thus the loose fan casing issue is not valid). I recommend you loosen the back panel (open M5 screws around and low side back panel D-connector area with M3 screws), unbolt the fan casings (2* 4 M3 screws) and mark the DC cables' plug positions under the back plane PCB, and dismantle the old fans from the metal casings (and wash the dirty casings and meshes). Buy a pair of new 12 V 80*80 mm quiet (typically 20 dBa) PC cooling fans and replace those old (noisy and worn) fans. The DC plugs won't fit, so you cut cables and solder / heat shrink tube the old PCB plugs to new fan's cables. The fans are originally sucking air out of the cabinet. This seems inefficient and specially the A131 SG LOCAL and A10 CONVERTOR modules run very hot and this all leads to premature drying of the electrolytic capacitors. So I installed the new fans to blow air in to the cabinet. The metal fan casings have huge 5 mm gaps around the fan frame. I filled those air leaking void gaps with foam (silicon would also do) to stop wasting good blower CFMs. The cards and RF modules now run much cooler and I presume this makes the unit to work much longer.
Reassemble the back panel, but make sure no loose wire gets pinched between the panel and chassis.
Replacing buffer IC on A01 CPU to make the unit boot again:
D11 (74 ACT541 SMD buffer IC) This Ic is located on the back side of A01 Main CPU board. [credits finding this fault goes to Artur, SP5TAV]
Any further repair tips or mods etc. for MT8802A would be
appreciated via e-mail and added here crediting the source.
Finally, short tips for level calibrations (for exact procedures:
see MT8801C operating manual found on web):
Allow the MT8802A to warm up long enough.
Analog TX power meter calibration: Main connector. Select
"RF Level/Power", set "Ref Level" to say +10 dBm, hit "Adjust
range", select sub menu bar 2 with arrow up key and hit "Power
Meter Zero Set" with no RF input source connected,
select with arrow up key the sub menu 1, hit "Calibration", connect a +10 dBm RF source to Main input ("RF Frequency" must be first set on the signal source frequency), hit "Manual Calibration"......."succesful", and its done, "failed" and you have something wrong,
like a blown RF power attenuator behind the Main N-connector.
Spectrum analyzer refercence level calibration: Main
connector. Hit arrow left key, hit "Coupled functions", make sure
RF input is diconnected from any signal source, hit "Cal" and its
done. Analog TX level meter and spectrum analyzer levels
be now showing very close to identical dBm values.
© I. Yrjölä, 2010... 2017