After nearly a year the 5650 finally is up and running. For those who haven't seen my previous posts i picked up a TA-5650 and a ST-4950 tuner from a garage sale last year for a bargain. Since then i have had no end of trouble getting the 5650 back into good health. It had a bunch of issues no one had really encountered before and has been nothing but a headache for me. A lot of threads on v-fets have also recently been destroyed thanks to photo buckets stunt, there are many photos i should have taken but i get carried away working and forget the camera quiet often. I will still try to make this as photo heavy as i can.
So to begin with here's the Amplifier and Tuner the day i found them complete with the manuals for both of them, a little dirty with the odd scratch but in fairly good condition.
[/url]
First up was to disassemble the unit, give it a good clean and check the health of the v-fets. Unfortunately on removal of the v-fet's one of the 8 failed a basic diode check on a meter despite the amplifier still working. On one meter it test fine and on my second meter it read faulty so i don't think it was completely blown but on edge. Fortunately i located a replacement 2sk60 of the same rank from fellow forum member @joshvi who shipped it over to me nice and quickly to Aus all the way from India, i got very lucky!. I'd like to thank him as none of this would have been possible without that one v-fet!. In the meantime i proceeded to recap the unit with Elna Silmic II's and apply the necessary mods mentioned in the Sony service bulletin.
Depending on the serial number of the particular TA-5650 it will have none, some or all of these mods applied. This bulletin is applicable to both the TA-4650 and TA-5650.
The main amplifier board removed and the unit opened up ready for cleaning and work to begin. Note the shrinking casing on the capacitors on the amplifier board due to high heat, these were definitely ready for replacement yet surprisingly still test within spec.
One of the most important things to do to any TA-4650/5650 is to replace the varactor diodes responsible for bias compensation with temperature changes within the amplifier. These commonly fail causing erratic bias and often results in failure of the v-fets. This particular unit could not hold a stable bias at all and would shoot all over the place as much as 180mv then down to 0mv. The suicide diodes as they are often referred to can be replaced with a pair of 1n4148 zener diodes wired in series. I chose to couple them in my unit to the transistors with thermal paste and a little bit of heatshrink over the top. The suicide diodes can be identified as a little black jelly bean with a blue dot on them. While replacing them i noticed the left channel had various replacement resistors and some burned, this was the channel with the faulty v-fet, it defiantly had been worked on before and i suspect all v-fets on that channel replaced as they all come out on the left channel with zero thermal paste on them??
Suicide diodes replaced with 1n4148 in series
With the suicide diodes out the way its time to get back on with the re cap. Along with all the electrolytic caps all the ceramic and mylar caps on the main amplifier board were replaced with wima/vishay equivalents. There was nothing wrong with the originals and these parts are generally reliable. However i have had a string of amplifiers from this era in the last couple of years that have had one or two failed film caps where they are close to a heat source. In all cases it has been silver mica caps in early Sansui units. It can be a massive pain to diagnose when they go belly up so i replaced them for the sake of it, with the rarity of v-fets I’m not chancing it on this one.
For the main filter caps i used 10,000uf 63c units by kemet, these were some of the closer sized options i could find that were readily available in Australia. Double sided foam tape was used around them to make them round enough to fit in the holder. Some mild filing was needed on the connectors to accommodate the slightly wider lead spacing of the replacements as well.
A few shots with all the new caps installed. Bourne's multi turn sealed trim pots were also installed for bias and power supply adjustments, this is pretty much a given practice for any old receiver. Sony uses a quality trim pot from factory however their role is too critical to skimp on imo.
Next up the speaker protection relay was replaced, an omron MY4-02 24VDC is the correct replacement.
It has been found by several owners of TA-5650's that the voltage doubler responsible for +-75 supply to the source pins of the v-fets is prone to becoming unbalanced when there is uneven loading on the supply. Aged driver transistors, resistors tolerances etc all can make the supply loading uneven. With the vfets installed they source enough current to keep the supply mostly stable however it is recommended to install an extra diode in series then tie the centre pin to ground in order to solve this issue entirely. This mod was originally for the TA-4650 but applies to the 5650 and has been done by many with success. I've drawn the correct placement for the extra diode on the circuit diagram, chose to use the ground pin of the power lamp rather than pin 5 as shown on the diagram, both connect to the same place it was just easier this way.
The power light in this unit was blown, i decided to convert it to led when replacing it. I replaced the original 820ohm resistor with a 3.3kohm resistor and diode in series. Because this is not full wave rectification the led will pulse at half the mains frequency. This is unacceptable for a back light for example and can be quiet annoying to look at. In this case though it is barely noticeable unless you look really hard so it does the job quite nicely.
Another threat to v-fets is the loss of source voltage. Earlier TA-5650's use a plug to connect to the amplifier board. In later models the connections are wire wrapped and soldered. While the plug is useful and makes removing the board a piece of cake i recommend wire wrapping and soldering all the connections. On even earlier models the bridge wire that runs from one side of the board to the other is also socketed with plugs.
Now with the bulk of the work done its time to verify that the amplifier can generate proper bias and wont destroy the v-fets once they are put back in. I've marked on the photo the correct voltages you should be seeing with the v-fets removed and the amplifier powered on. Note when adjusting the bias trimmers you should see a range of adjustment of approximately 8-28v. You want to set for the highest voltage possible. Higher voltage= lowest bias setting.
Note DC offset was riding a little high on the left channel. There is no adjustment on the TA-5650 for offset, you must replace the input differentials with a matched pair to lower it. Replace the 2sa705's for Q302-303 and Q352-353 with pairs matched for Hfe. BC556b works nicely.
This is where the headaches with my unit started to begin. It was discovered that by connecting the pre amp jumpers on the back the source voltage would drop below the gate voltage and bias polarity would reverse, the range of adjustment would drop to around 1-8v and a 580khz oscillation would occur throughout the circuit. Shorting the amplifier inputs to ground would cause the exact same issue. After tearing my hair out checking and double checking every component in the amplifier i was lost for a cause. A forum member over at AK was dealing with the exact same issue as me, we ended up concluding that the lack of v-fets/possible worn semiconductors throughout the amplifier board was causing the circuit to become unstable. He ended up taking the punt and installing the v-fets on his unit, the issue went away instantly.
I wasn't quiet game yet and spent some time sourcing nos transistors for as many as i could and using substitutes only where i couldn't get originals, the only thing this did was waste money and time. Not all was a waste though, the number of burned components on the amplifier board when i got it suggest a number of transistors while still functioning would have been stressed so i can relax knowing they are all new now and likely to perform correctly for a long time.
By this point i was completely out of ideas and decided to take the risk and install the v-fets. Before doing so i decided to install install 68ohm 5w resistors in line with the +- 45v supply to the amplifier board to prevent the v-fets being able to conduct destructive amounts of current in case of a fault. Others on various forums had used halogen lamps to do this job but i didn't like the idea as a filament lamp is relatively low resistance when cold. The time it would take for the filament to warm up and raise resistance could possibly allow the v-fets to blow up before it has a chance to limit current. I nervously flipped the switch monitoring source and gate voltages closely and all was good, circuit was now maintaining stable and correct voltages.
Never took a photo with the resistors as shown but install as marked on the photo.
By this point i was getting excited for v-fet goodness however that was soon ended. I found that if i turned the treble control to the max and the volume to about 10 o'clock with no load connected or signal the unit would start oscillating and the protection resistors would get hot and both source and gate voltages would drop very low due to the voltage drop across the resistors. Annoyed but glad nothing was damaged back out came all the v-fets and more searching for faults. After hours and hours of searching and no luck i put the unit up on the shelf and left it there for some months.
Something i did not own that was becoming increasingly important was an oscilloscope, i bought one and wish i had have done so years ago. Having a look at a 1khz signal through the scope all looked good until i increased the treble, the resulting sine wave was a mess and would get worse every click of treble i increased it by until it would oscillate so bad the sine wave would fill the whole screen with a solid block of colour.
Didn't take a photo but looked pretty much like this
After another round of fault finding and no luck i concluded that with the treble set to the maximum the resulting gain of high frequency circuit noise (hiss) would place enough current demand that the resulting voltage drop across the resistors would cause the circuit to become unstable. Armed with the oscilloscope to watch the outputs i removed the resistors and fired it up incrementally increasing the treble until i was at maximum. The result? a clean sine wave, steady bias of 50mv and v-fet magic out of the speakers. So we now know you can't use a variac on a v-fet, testing them with the v-fets removed is hit and miss and protection resistors causes the circuit to become unstable, how helpful!
I don't feel the need for a thread for the ST-4950 tuner but some work was done to it as well. Not too much went on with this one, It was in a sad shape, the numbers on the glass screen had all drifted around the place. I gave the unit a recap a clean and took to sorting out the numbering for the stations. They appear to be painted on and the back light shines up the side of the glass and reflects off the white paint. I used a fine jeweller’s screwdriver and pushed them around the place until they lined up as well as i could get them. They wanted to crack and stick back to the glass so i could only get so far but the result is much better. I have no idea what caused them to become unstuck in the first place, have never seen anything like it before.
The back lights also needed some work. The tuner uses regular 12v festoon lights with a blue plastic film wrapped over the edge of the glass at each end to produce a green back light. Years of heat from the bulbs had burned the plastic and while not noticeable in the photo half of the display was the yellow colour of the bulbs. I replaced the bulbs with some white leds and the result was a dark blue back light, i liked the look so i decided to keep it. I forgot to take photos but a rectifier cap and resistor was installed on the underside of the unit in order to stop the back lights flickering and the stereo indicator was replaced with a red led.
last up a shot of both units up and running back to their full glory! This is the first v-fet I've listened to and it has an interesting sound, it’s a very mellow sounding amp yet every detail is there. Not really like anything I've listened to before.
So to begin with here's the Amplifier and Tuner the day i found them complete with the manuals for both of them, a little dirty with the odd scratch but in fairly good condition.
First up was to disassemble the unit, give it a good clean and check the health of the v-fets. Unfortunately on removal of the v-fet's one of the 8 failed a basic diode check on a meter despite the amplifier still working. On one meter it test fine and on my second meter it read faulty so i don't think it was completely blown but on edge. Fortunately i located a replacement 2sk60 of the same rank from fellow forum member @joshvi who shipped it over to me nice and quickly to Aus all the way from India, i got very lucky!. I'd like to thank him as none of this would have been possible without that one v-fet!. In the meantime i proceeded to recap the unit with Elna Silmic II's and apply the necessary mods mentioned in the Sony service bulletin.
Depending on the serial number of the particular TA-5650 it will have none, some or all of these mods applied. This bulletin is applicable to both the TA-4650 and TA-5650.
The main amplifier board removed and the unit opened up ready for cleaning and work to begin. Note the shrinking casing on the capacitors on the amplifier board due to high heat, these were definitely ready for replacement yet surprisingly still test within spec.
One of the most important things to do to any TA-4650/5650 is to replace the varactor diodes responsible for bias compensation with temperature changes within the amplifier. These commonly fail causing erratic bias and often results in failure of the v-fets. This particular unit could not hold a stable bias at all and would shoot all over the place as much as 180mv then down to 0mv. The suicide diodes as they are often referred to can be replaced with a pair of 1n4148 zener diodes wired in series. I chose to couple them in my unit to the transistors with thermal paste and a little bit of heatshrink over the top. The suicide diodes can be identified as a little black jelly bean with a blue dot on them. While replacing them i noticed the left channel had various replacement resistors and some burned, this was the channel with the faulty v-fet, it defiantly had been worked on before and i suspect all v-fets on that channel replaced as they all come out on the left channel with zero thermal paste on them??
Suicide diodes replaced with 1n4148 in series
With the suicide diodes out the way its time to get back on with the re cap. Along with all the electrolytic caps all the ceramic and mylar caps on the main amplifier board were replaced with wima/vishay equivalents. There was nothing wrong with the originals and these parts are generally reliable. However i have had a string of amplifiers from this era in the last couple of years that have had one or two failed film caps where they are close to a heat source. In all cases it has been silver mica caps in early Sansui units. It can be a massive pain to diagnose when they go belly up so i replaced them for the sake of it, with the rarity of v-fets I’m not chancing it on this one.
For the main filter caps i used 10,000uf 63c units by kemet, these were some of the closer sized options i could find that were readily available in Australia. Double sided foam tape was used around them to make them round enough to fit in the holder. Some mild filing was needed on the connectors to accommodate the slightly wider lead spacing of the replacements as well.
A few shots with all the new caps installed. Bourne's multi turn sealed trim pots were also installed for bias and power supply adjustments, this is pretty much a given practice for any old receiver. Sony uses a quality trim pot from factory however their role is too critical to skimp on imo.
Next up the speaker protection relay was replaced, an omron MY4-02 24VDC is the correct replacement.
It has been found by several owners of TA-5650's that the voltage doubler responsible for +-75 supply to the source pins of the v-fets is prone to becoming unbalanced when there is uneven loading on the supply. Aged driver transistors, resistors tolerances etc all can make the supply loading uneven. With the vfets installed they source enough current to keep the supply mostly stable however it is recommended to install an extra diode in series then tie the centre pin to ground in order to solve this issue entirely. This mod was originally for the TA-4650 but applies to the 5650 and has been done by many with success. I've drawn the correct placement for the extra diode on the circuit diagram, chose to use the ground pin of the power lamp rather than pin 5 as shown on the diagram, both connect to the same place it was just easier this way.
The power light in this unit was blown, i decided to convert it to led when replacing it. I replaced the original 820ohm resistor with a 3.3kohm resistor and diode in series. Because this is not full wave rectification the led will pulse at half the mains frequency. This is unacceptable for a back light for example and can be quiet annoying to look at. In this case though it is barely noticeable unless you look really hard so it does the job quite nicely.
Another threat to v-fets is the loss of source voltage. Earlier TA-5650's use a plug to connect to the amplifier board. In later models the connections are wire wrapped and soldered. While the plug is useful and makes removing the board a piece of cake i recommend wire wrapping and soldering all the connections. On even earlier models the bridge wire that runs from one side of the board to the other is also socketed with plugs.
Now with the bulk of the work done its time to verify that the amplifier can generate proper bias and wont destroy the v-fets once they are put back in. I've marked on the photo the correct voltages you should be seeing with the v-fets removed and the amplifier powered on. Note when adjusting the bias trimmers you should see a range of adjustment of approximately 8-28v. You want to set for the highest voltage possible. Higher voltage= lowest bias setting.
Note DC offset was riding a little high on the left channel. There is no adjustment on the TA-5650 for offset, you must replace the input differentials with a matched pair to lower it. Replace the 2sa705's for Q302-303 and Q352-353 with pairs matched for Hfe. BC556b works nicely.
This is where the headaches with my unit started to begin. It was discovered that by connecting the pre amp jumpers on the back the source voltage would drop below the gate voltage and bias polarity would reverse, the range of adjustment would drop to around 1-8v and a 580khz oscillation would occur throughout the circuit. Shorting the amplifier inputs to ground would cause the exact same issue. After tearing my hair out checking and double checking every component in the amplifier i was lost for a cause. A forum member over at AK was dealing with the exact same issue as me, we ended up concluding that the lack of v-fets/possible worn semiconductors throughout the amplifier board was causing the circuit to become unstable. He ended up taking the punt and installing the v-fets on his unit, the issue went away instantly.
I wasn't quiet game yet and spent some time sourcing nos transistors for as many as i could and using substitutes only where i couldn't get originals, the only thing this did was waste money and time. Not all was a waste though, the number of burned components on the amplifier board when i got it suggest a number of transistors while still functioning would have been stressed so i can relax knowing they are all new now and likely to perform correctly for a long time.
By this point i was completely out of ideas and decided to take the risk and install the v-fets. Before doing so i decided to install install 68ohm 5w resistors in line with the +- 45v supply to the amplifier board to prevent the v-fets being able to conduct destructive amounts of current in case of a fault. Others on various forums had used halogen lamps to do this job but i didn't like the idea as a filament lamp is relatively low resistance when cold. The time it would take for the filament to warm up and raise resistance could possibly allow the v-fets to blow up before it has a chance to limit current. I nervously flipped the switch monitoring source and gate voltages closely and all was good, circuit was now maintaining stable and correct voltages.
Never took a photo with the resistors as shown but install as marked on the photo.
By this point i was getting excited for v-fet goodness however that was soon ended. I found that if i turned the treble control to the max and the volume to about 10 o'clock with no load connected or signal the unit would start oscillating and the protection resistors would get hot and both source and gate voltages would drop very low due to the voltage drop across the resistors. Annoyed but glad nothing was damaged back out came all the v-fets and more searching for faults. After hours and hours of searching and no luck i put the unit up on the shelf and left it there for some months.
Something i did not own that was becoming increasingly important was an oscilloscope, i bought one and wish i had have done so years ago. Having a look at a 1khz signal through the scope all looked good until i increased the treble, the resulting sine wave was a mess and would get worse every click of treble i increased it by until it would oscillate so bad the sine wave would fill the whole screen with a solid block of colour.
Didn't take a photo but looked pretty much like this
After another round of fault finding and no luck i concluded that with the treble set to the maximum the resulting gain of high frequency circuit noise (hiss) would place enough current demand that the resulting voltage drop across the resistors would cause the circuit to become unstable. Armed with the oscilloscope to watch the outputs i removed the resistors and fired it up incrementally increasing the treble until i was at maximum. The result? a clean sine wave, steady bias of 50mv and v-fet magic out of the speakers. So we now know you can't use a variac on a v-fet, testing them with the v-fets removed is hit and miss and protection resistors causes the circuit to become unstable, how helpful!
I don't feel the need for a thread for the ST-4950 tuner but some work was done to it as well. Not too much went on with this one, It was in a sad shape, the numbers on the glass screen had all drifted around the place. I gave the unit a recap a clean and took to sorting out the numbering for the stations. They appear to be painted on and the back light shines up the side of the glass and reflects off the white paint. I used a fine jeweller’s screwdriver and pushed them around the place until they lined up as well as i could get them. They wanted to crack and stick back to the glass so i could only get so far but the result is much better. I have no idea what caused them to become unstuck in the first place, have never seen anything like it before.
The back lights also needed some work. The tuner uses regular 12v festoon lights with a blue plastic film wrapped over the edge of the glass at each end to produce a green back light. Years of heat from the bulbs had burned the plastic and while not noticeable in the photo half of the display was the yellow colour of the bulbs. I replaced the bulbs with some white leds and the result was a dark blue back light, i liked the look so i decided to keep it. I forgot to take photos but a rectifier cap and resistor was installed on the underside of the unit in order to stop the back lights flickering and the stereo indicator was replaced with a red led.
last up a shot of both units up and running back to their full glory! This is the first v-fet I've listened to and it has an interesting sound, it’s a very mellow sounding amp yet every detail is there. Not really like anything I've listened to before.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.