Kenwood TS-440 repair

This page describes some repair tasks related to the Kenwood TS-440 transceiver. You will be guided through the repair proces step-by-step followed by descriptions and pictures. Pictures on this page can be enlarged by clicking on it.

I recommend reading TS-440S Repair FAQ by Mike Freeman KC8QNO. Another big help is the TS-440 Yahoo group: TS-440. 

Look for the obvious

Troubleshooting a transceiver is a complicated task. There are thousands of components, and dozens of wires, connectors, and cables.

The obvious: Is the circuit getting power?

In the case of dots-in-display, the main suspect is VCO1. It contains 4 oscillators. If an oscillator stops, the dot-in-display will show. Check the active oscillator for about 7 volts DC on the collector. R102, R107, R112 or R118 can be used as measuring points. 

In the case of transceiver-2, the oscillator Q21 is active when the TS-440 dial shows 29.999 MHz to 22.000 MHz. I measured 7 V on the collector of  Q21 when the dial showed 29, 28, 27 or 26 MHz. But the oscillator died when the dial showed 25, 24, 23 or 22 MHz. Why? Because there was no power on the collector of Q21 (the collector measured 0 volt). The error was traced down to a faulty Q44. Replacing that transistor solved the problem.

It was obvious that oscillator Q21 died when no power was supplied!

Checking the Band Pass Filter on the RF-unit

The Band Pass Filter on the RF-unit can have leaky diodes, and this will cause dots-in-display. I discovered this problem when I checked Transceiver-2.
The TS-440 display showed 14.000 MHz and I measured the following values:
Only L52 should be high, and L11 should be at 0 volt. Replacing the two diodes (D4, D5) associated with L11 solved the problem.

Rebuilding VCO1 on the RF-unit

                VCO1 on the RF-unit (TS-440 Service Manual p. 75)

VCO1 filled with goop.
Fig. 1. VCO1 filled with glue.

The problem 
Kenwood TS-440 is designed as a mobile transceiver. To withstand mobile use, some of the circuit boards are partly covered with glue ("goop") in order to prevent components from vibrating and causing frequency instability.

As time goes by, the glue becomes slightly conductive and can cause leak currents flowing between components. This is very bad in VCO circuits, where the varicap diode is extremely sensitive to any voltage change.   

If a TS-440 has conductive glue in the VCO circuits, it may sound distorted when receiving a signal and transmitted SSB sounds strange.

Conductive glue was discovered when I measured 1-2 V DC between ground and the brown glue in VCO1. I used a voltmeter with 11 Mohm input impedance. All conductive glue must be removed.  


Goop halfway removed.
Fig. 2. Glue halfway removed.

Removing the glue

Make notes on how cables etc. are connected to the RF-unit, or take a close-up picture. Now disconnect the cables, remove the screws, and extract the RF-unit. 

The glue is removed using hand tools. Use a stanley knife for cutting out pieces of glue and for cleaning (scraping) the component side of the circuit board. A screwdriver is handy for lifting components when unsoldering. A long-nosed plier is used for removing components and pieces of goop. This work requires several hours of intensive labour. I recommend you take regular breaks!

Some people say that components dirty with glue can be removed, cleaned, and reused. I could not do that. The glue removal proces will probably destroy many components. Furthermore, it is easier to clean the circuit board when everything on it has been removed.  

Do not destroy the 2SC2668(Y) transistors. Unsolder them, clean them carefully, and reuse them. I offer this advice because it was difficult for me to obtain new transistors. I've later discovered, that 2SC2668(Y) is available on Ebay. 

Eucalyptus oil bottled

Alternative method of removing the glue

Update September 2021. VK5NNT Neville describes his method of removing the glue: "What I have discovered is that I put pure eucalyptus oil on the glue and it softened it to a jelly condition. I used a small plastic syringe to put drops of eucalyptus oil on the glue. It acts fairly quickly and does not require a lot of oil; just put some on an area and test to see how long it takes to soften. I was then able to remove the glue easily with a paper cutting scalpel and tweezers and used a small suction nozzle connected to my vacuum cleaner and suck up the jellied glue. I did not remove the components".

I have not tried this method, but I would encourage using it. Removing the glue and not the components will greatly simplify the repair work and speed up the task!

When the glue has been removed from VCO1, go to "VCO1 Adjustment" and follow the description. 

When the glue has been removed from VCO5, go to "VCO5 Adjustment" and follow the description.

VCO1 cleaned.
Fig. 3. VCO 1 fully cleaned.
Rebuilding VCO1

I ordered new components from East Coast Transistor in New York, USA. The rebuild work should be divided into steps. Build one VCO and test it. Then continue with the next one. 

The original varicap ITT310TE (D44, D46, D48, D50) can be substituted by BB139. The original switching diode MA858 (D45, D47, D49, D51) can be substituted by BA244.

For C175, C182, C188, and C195 the factory value of 5 pF may be too low. A low value can make the oscillator stop. Increasing the value may be necessary to keep the VCO running. Here are the values from my rebuild of transceiver-2 (varicaps were BB139):

Q21 (67 - 75 MHz)Q22 (59.5 - 67 MHz)Q23 (52.5 - 59.5 MHz)Q24 (45 - 52.5 MHz)
C174 = 33 pF
C175 = 6.8 pF
C182 = (5 + 2.2) pFC187 = 47 pF
C188 = 10 pF
C194 = 82 pF
C195 = 10 pF

WB7DFV Dave used NTE614 as varicap replacement. He wrote: "It turns out that the NTE614 was a good replacement. I could not get the oscillator with T21 to work and had to reduce the 39 pF cap to a 22 pF cap. Now the radio works perfectly."

KA5ZWY Larry told me that he had problems with T22. However, when he applied pressure to the PCB near T22, the problem disappeared. Resoldering T22 and surrounding components will solve that problem. 

Fig. 4. Hooking up wires before testing VCO1.
Hooking up wires 
It is a good idea (but not mandatory) to test the four oscillators in VCO1 before the RF-board is put back in the chassis. This test requires some wiring to emulate the voltages within the transceiver. 

An external PSU supplies 7.7 V DC at 60 mA.

One end of L101 and L97 is desoldered from the PCB, and this end is connected to 7.7 V DC. The other end feeds the circuit.

The 10 nF capacitor is important. It decouples RF from the circuits in VCO1 and from the potmeter wires. Once I forgot this capacitor, the active oscillator's frequency jumped up and down!

The 10 kohm potmeter creates the VCO voltage. 

VCO1 finished
Fig. 5. VCO1 rebuild finished.
Testing VCO1
Figure 5 shows the rebuilt VCO1. Some of the new components are soldered on the copper side of the PCB (C171, C178, C184, C190). There are 2 empty holes near Q21, Q22, and Q23 where no components are fitted. 

An oscillator is activated by applying 7.7 V to its collector via the 100 ohm resistor (R102, R107, R112, or R118). One side of the resistor (the side connected to Q17-Q20) is desoldered from the PCB, and 7.7 V is applied to this end.

The voltage range for testing each VCO should be 2.0 V to 6 V. The reason for testing at 2.0 V is that this is the default control voltage. If the VCO cannot oscillate at 2.0 V, the PLL unlocks causing dots-in-display. When the PLL is locked, the VCO voltage should be within the range 2.5 - 6 V.

A good connection point for a frequency counter is "PLL VCO" which is at the edge of the PCB. I added a 22 pF capacitor between "PLL VCO" and the counter. The frequency range of each oscillator is printed in the circuit diagram.

VCO1 adjustment 
Mount the RF-unit and reconnect all cables. Follow the Service Manual p. 92 item 13. Please note the typo at 3rd step. The correct value is 14.4999 MHz, not 14.9999 MHz.

Main dialTP10

29.9999 MHz2.49 V2.47 VAdjust T21 to get this reading
22.0000 MHz5.94 V6.06 V
21.9999 MHz2.51 V2.45 V
Adjust T22 to get this reading
14.5000 MHz5.86 V5.94 V
14.4999 MHz2.50 V2.48 V
Adjust T23 to get this reading
7.5000 MHz5.93 V6.21 V
7.4999 MHz2.38 V2.49 V
Adjust T24 to get this reading
30 kHz6.62 V6.13 V

TP10 is the top of R138 (PLL-unit) and it can be accessed from the side.

Rebuilding VCO5 on the PLL-unit

VCO5 on the PLL-unit Kenwood TS-440
VCO5 on  the PLL-unit (Service Manual p.72)
VCO5 before goop removal
Fig. 6. VCO5 filled with glue.

The problem 
Some of the components in VCO5 are covered with glue in order to prevent components from vibrating and causing frequency instability.

As time goes by, the glue becomes slightly conductive and can cause leak currents flowing between components. This is very bad in VCO circuits, where the varicap diode is extremely sensitive to any voltage change.   

I measured about 300 mV DC between ground and the glue in VCO5 using a voltmeter with 11 Mohm input impedance. This voltage level may seem low, but it is poison to a voltage controlled circuit!   

VCO5 cleaned.
Fig. 7. VCO5 fully cleaned.

Removing the glue 
Make notes on how cables etc. are connected to the PLL-unit, or take a close-up picture. Disconnect the cables, remove the screws, and extract the unit. 

Remove the lid of the shielded box. The glue in VCO5 can be removed using hand tools. Use a stanley knife for cutting out pieces of glue and for cleaning (scraping) the component side of the circuit board. A screwdriver is handy for lifting components when unsoldering. A long-nosed plier is good for removing components and pieces of glue.

Eventually I used propyl alcohol on a cotton bud to clean the area.             

VCO5 rebuilt finished
Fig. 8. VCO5 rebuild finished.
Rebuilding and testing VCO5 
Figure 8 shows the rebuilt VCO5. C186 and C192 were fitted on the trace side of the PCB.

Q36 on the circuit diagram is 2SK192A, but this FET is not available any more. I substituted it with BF256B, which works just fine. I used 2SC3200 instead of Q34 and Q35.

I divided the work into steps. I first build the amplifier Q34 and Q35, and tested the DC levels. Then I continued with the oscillator part Q36 and Q37. I checked the DC levels, and measured the output with a diode probe. Finally, I connected a frequency counter to R154, applied 5.0 V DC to TP 11, and adjusted T20 until the counter showed 36.22 MHz. 

Please note that the Service Manual has swapped the DC readings of Q36's gate and source. The gate should read 0.0 V and the source should read 1.0 V. The drain level is correct (7.4 V).  
VCO5 adjustment
Fit the PLL-unit into the chassis and reconnect all cables. Follow the Service Manual p.91 item 3. Connect your voltmeter to TP11 (top of L40) and turn the T20 slug slowly until 5.0 V DC is measured. This adjustment should be carried out only when the PLL is locked (that includes VCO1). Finally, fit the lid on the VCO5 shield box.

Help - there are still dots-in-display...

After having rebuilt VCO1 and VCO5 in transceiver-1, I still had dots-in-display. It happened at any frequency, and often right after I had touched wires or cables on the RF-unit. At other times, the TS-440 behaved normally. What to do now?

Step 1: Check the band select circuit (RF-unit).
Service Manual page 23. My advice is to follow the frequency values in table 13 and check B0-B3 at connector 15. Check also the control signals to ATU, filter unit, and band-pass filters.

Step 2 : Check diodes in the band pass filters (RF-unit).
The diodes in the band pass filters on the RF-unit were checked by following the procedure described in TS-440S Repair FAQ. 

Step 3: Force dots-in-display.
The idea is to poke cables and wires, and slightly move components until the error becomes permanent. As I was working on this in the RF-unit, I discovered that one of the wires in the flat-cable between VCO1 and Q41 was unstable. When I touched the wire, dots-in-display turned up. Touching the wire again made the error go away.

Loose wire (RF-unit)
Fig. 9. The loose wire on the RF-unit.

The "problem wire" carried DC to the buffer circuit in VCO1. When the wire became loose, the output of VCO1 disappeared completely (!) and caused the dots-in-display error. The solution was simple: I soldered a replacement wire under the PCB. The display problem was solved. I have not seen dots-in-display on my TS-440 since then.

Excessive hiss from the loudspeaker

When the AF control is turned fully down (fully CCW), a rather loud hiss is heard from the loudspeaker. There was something wrong with the AF amplifier circuit on the IF-unit. The electrolytic capacitors around IC7 have probably dried out.

Step 1:
Replace C173 (IC7 feed-back). No change, still excessive hiss.

Step 2:
Replace C170 (IC7 input), C172 (IC7 decoupling), and C175 (IC7 output).

The problem is apparently solved, no hiss anymore, at least until the next day, then the hiss came back! I dismounted the 4-hole molex header at J2 which connects the AF potmeter. The loudspeaker became silent immediately. My conclusion: The hiss comes into the AF amplifier IC7 via J2. As I remounted the molex header onto J2, the sound suddenly became right. I discovered that the ground pin at J2 had a bad connection. 

Step 3: J2 has to be cleaned. I applied Contact 60 (cleaner agent) to J2 in order to re-establish the ground connection. Problem solved.

If the AF potmeter has no ground connection (as it was in my case), it will be connected in series with the input of IC7. The noise can never be fully turned down. When the ground connection of the potmeter is good, the pot is connected in parallel with IC7's input. The potmeter can now completely silence the loudspeaker when the AF pot is fully CCW.

Hint: By tilting the IF-unit 135 degrees, you can access the solder side as seen in this picture.

Written by OZ1BXM Lars Petersen October 2009. Latest revision 08-September-2021.