SONY TA-5550 VFET amp - need help

spaghetti

Active Member
I bought recently this amplifier sold as not working, but with clicking relay according to the seller. I did not test it. Right when it arrived I removed the precious VFET transistors, checked them and they're all still OK, so I put them aside.

I started replacing the killer diodes, then while I was at it I recapped the unit (expect for the 2 main big filter caps... there's some small ripple current but I think it's not an urgent issue) and replaced the trimpots with new ones.

Now I am testing the amp (no VFET installed yet) injecting the usual 1Khz sine wave and checking the results at the speakers terminal. with the scope. I see that the signal looks good as long as the volume is quite low, let's say up to 2 of 10, then it clips quickly and I get a distorted signal also on the other channel.

I checked all the transistors, diodes and resistors on both boards, replaced a few resistors that were out of spec, but found nothing serious.

I was wondering if it makes sense testing the amp without the VFET installed, or if the strange behavior is caused by their absence... but of course I don't feel like testing it with the VFET installed unless I'm confident that everything is OK.

Any suggestion/help is highly appreciated, thanks

Andrea
 
You're trying to drive a speaker without the output devices installed? You may be getting something through the driver emitter resistors, but it's not surprising that it gets distorted as you try to increase the output.
 
Now I am testing the amp (no VFET installed yet) injecting the usual 1Khz sine wave and checking the results at the speakers terminal. with the scope. I see that the signal looks good as long as the volume is quite low, let's say up to 2 of 10, then it clips quickly and I get a distorted signal also on the other channel.
not sure how you are getting anything at the speaker terminals with no output drivers installed:idea:
 
@rottalpha the protection relay clicks so the signal can pass.

I think this is quite normal, it's a behavior I've experienced with may other amps (you get signal even when the outputs are out... of course power is very limited)
 
I was looking at other Sony VFET amps schematics in order to find some hints/inspiration right now... it's rather similar to TAN-8550 and 8650. Rather different than the TA-4650 and TA-5650 that I'm familiar with.

And by the way it's TAN-5550, I'm sorry I keep writing TA forgetting the extra N.
 
I'm writing a few notes after some measurements I took today.

One thing I think it's worth noticing is that D101, represented as zener diode on the schematics, is still a VD-1221 killer diode. I don't know if the zener symbol is an error or if the diode can work also that way.

  • positive rails voltage to the class B amp is a bit off, it's +80V instead of 82. Q401 gets quite hotter than its counterpart for the negative rails, Q403
  • rails voltages to the VFET drains is +-55V instead of -+53... but at least the voltages are symmetrical here
  • rails voltages to class A amp are +62V -66V instead of +-60V... these don't look very good to me.

Trimpots RT101 and RT201 have a significant effect on many transistor voltages. They also have strong effects on the voltages at the gates of power outputs... I don't know if this is normal. Also, these voltages are not much consistent in time, and differ also after turning the amp on/off. This somehow reminds me of the TA-4650/5650 behavior without the diode modification, but I'm not aware of this modification also for the TAN-5550.

Since the manual does not provide any voltage information for the VFET gates, I'm not even sure what voltages are supposed to be correct.
 
One of the things to check is to have proper bias adjustment for the vfets; check, with no vfets installed, the gate-source pin voltages, and verify the min/max voltage you get varying the bias pots RT501 and RT601. The TA-5650 has a 0-28V range, to give you an idea. Keep in mind that, since the vfets are depletion mode devices, if Vgate is 0 the vfet will basically be a short (1 to 2 Ohm of D-S resistance out of circuit, they will get destroyed with the amount of current provided by the supply rail voltages across such a low resistance); therefore, to install the vfets you need to adjust the pots to have max voltage at the gate. If you have small Vgs ranges, or different ranges for different vfets, there are issues with the bias circuit.

RT101 and RT201 are the pots that make sure the two transistors in the differential pairs Q101 and 201 are working symmetrically varying ultimately the DC that can reach the speakers when you have the vfets installed. If they are not adjusted properly, the imbalance will affect all the following stages, as they are all DC coupled. It's always good to make sure that all the transistors in the differential pairs are well matched. Any DC from the previous stages impacts the bias point of all the successive amplification stages.
 
@MattiaM thank you for the suggestions. VFETs are arranged differently on TA-4650/5650, S and D are swapped in the TAN-5550, this increases my sense of insecurity regarding the voltages I get from this amp... I have nothing to compare them to.

I can adjust the bias for Vgs between 0~25V... but I can't keep symmetry (achieved by regulating RT101/201... turning the amp off and then after a while on again the values vary,for example P channels may get 12V, while N channel may get 35V) and the voltages are jumpy, meaning that while I look at the multimeter I see the value change between ~20V and 30V, with more time spent in the middle of the range.

In this thread about the TA-4650 @RealWorld suggested me to put 68 ohm resistors in the B+/B- voltage going to the VFETs in order to protect them from destruction. I wonder if the same approach (I still have the resistors around) may save the transistors also in this amp.
 
Yes, they’re swapped, but notice that on the tan5550 you have the 2sk60 connected to the positive rail and the 2sj18 connected to the negative rail, whereas it’s the opposite in the ta-5650. In terms of potentials, you have the drain at a higher potential than the source for the 2sk60 and the source at a higher potential than the drain for the 2sj18 in both amps.

Have you tried checking with the oscilloscope if you have any oscillation at the gate pins? (Without any input signal, and with the vfets out)
 
Last edited:
I took a few pictures of what I get at collectors of Q602 and Q603 (the same shape appears later at the VFET gates) with a 1khz signal at very low volume. You can see the sine wave on the strange shape. The other channel at collectors of Q502-503 has a different shape, still not flat.
q602-q603.jpg
The second picture is what I see at collectors of Q604-Q605, still while injecting a sine wave in the amp (no wave visible here).
q604-q605.jpg
The last picture is what I see at the gate of Q612 (no sine injected now) which is the same shape present at Q602/603. That thing does not look good to me, but I don't understand where it comes from.
q612.jpg
 
That looks like the 50Hz (20ms period) from your mains picked up by the scope. What I was looking for is whether you get a voltage swing (something like 0V-25V/30V) at the gates of the vfets due to oscillation, without any signal sent through, and with the signal inputs shorted to ground to reduce noise (if you have some spare shorting plugs for the rca phono inputs of some other amp, those are perfect). Given that the TA-5650 and 4650 (not sure about the 5550) tend to do that kind of oscillation with the vfets out, it's worth checking. The voltages to me seem OK, the only value a bit off is that -66V (and its transistor running hotter than the one for the opposite rail); that might be just the age of the amplifier and the lack of load.

Try adjusting RT101 and 201 to get the closest you can get to -59V at the bases of both Q105 and Q106 (and 205 and 306 for the other channel). Once you do that, short the RCA signal inputs of the amplifier, and check the type of voltage swings/oscillations (if any)you can see with the oscilloscope at the gates of the vfets, and what the bias adjustment pots do.
 
The second picture is your ripple from the +/- 82V rails after the voltage drop due to the collector resistors. The first and third are different because of capacitance etc. as they’re taken further away from the rails.
 
I added the short plugs and there's no significant difference at the VFET gates. Anyway if we were looking for big voltage swings then I guess we can conclude that there's no oscillation.

RT101 and 201 have little effect on the voltages at the base of these transistors... Q105 range is -66-68V, Q106 range is a bit smaller at -66-67V, so no way to get closer to -59V (I guess all this is because there is that nasty -66V rails voltage).
 
I added the short plugs and there's no significant difference at the VFET gates. Anyway if we were looking for big voltage swings then I guess we can conclude that there's no oscillation.
RT101 and 201 have little effect on the voltages at the base of these transistors... Q105 range is -66-68V, Q106 range is a bit smaller at -66-67V, so no way to get closer to -59V (I guess all this is because there is that nasty -66V rails voltage).

I guess a couple of volts of adjustment is all you can expect there, as they're meant to basically provide adjustment for small matching tolerances in the differential pairs; since you can't get the -59V, I would still set them in a way that the two bases of Q105 and Q106 have voltages as close as possible.

I can adjust the bias for Vgs between 0~25V... but I can't keep symmetry (achieved by regulating RT101/201... turning the amp off and then after a while on again the values vary,for example P channels may get 12V, while N channel may get 35V) and the voltages are jumpy, meaning that while I look at the multimeter I see the value change between ~20V and 30V, with more time spent in the middle of the range.

That is some sort of drifting behaviour; not as in this thread (bias adjustment reduced to 1 to 8V, source voltage dropping below the gate voltage and bias polarity reversing, together with a 580 kHz throughout the circuit), as you don't have bias polarity reversing (which is good) and, apart from drifting voltages, you seem to have control over the correct range of the bias settings.

I pass you one of the tests that Joshvi's (one of the vfet gurus here on the forum) advised me to try on a TA-5650:
"if it is oscillating the gate voltage will swing between 0-28, that is why it is showing less voltage when measuring with multi meter. Please check the gate voltage / wave form with a CRO. Connect 560 ohm 1 W resistor between speaker terminal and put 1kpf capacitor between gate and source of vfets, check whether oscillation is stopped or not.If stopped vary the bias pot and check the gate voltage . if every thing is okay, you set the bias pot @ position maximum gate voltage. Then remove the 1kpf gate source capacitors and 560 ohm resistors you connected. Put back the Vfets, adjust the bias as per service manual, enjoy music". Maybe worth trying this out on the 5550 as well?
 
@MattiaM thank you for your help. I won't be home for a few days, so I can't continue my troubleshooting.

Meanwhile I was thinking about replacing the main caps as well, sooner or later this is something that must be done.
Replacement caps are hard to source as these amps require screw terminal caps, and I last time I checked on mouser there was no model available with the correct spacing for the TA-4650. I guess for this amp it won't be different. But yesterday I found a picture of some adapters that seem to solve this problem:

adapter.jpg

so, anybody knows if these can be bought from mouser or similar ecommerce?
 
@MattiaM Replacement caps are hard to source as these amps require screw terminal caps, and I last time I checked on mouser there was no model available with the correct spacing for the TA-4650. I guess for this amp it won't be different.

FYI, the IEC standard that defines standard capacitor sizes changed somewhere early 90's, and the former 40mm standard diameter was no longer a standard, it became 35mm and 50mm.
The same for the terminal distance. But, now it's only 1 mm wider for most Nippon Chemi-Con / Nichicon / Kemet caps, so the usual trick is to widen the existing holes with just a little, and voila it fits.
Then, use some 3mm (1/8") thick foam tape (as used for draft stopping for doors and windows) and stick that around the bottom of the 35mm caps, and then put it back in the original black plastic chassis clamp.
Works like a charm.

I would not recommend to use additional clamps and wires between those caps and the PCB: keep it close, short and tight.
 
@Oilmaster Thank you for the detailed explanation. There are not many choices right now on Mouser, I think this one is cheap and should be good: https://www.mouser.com/ProductDetail/KEMET/ALS80A103DB063?qs=sGAEpiMZZMtZ1n0r9vR22YqjN7vGN3ZS4rCbF3CuDojs1M7ZALPypg==

Meanwhile I was able to fix the DC unbalance on the rails voltages to the class A amp. I now get -59V on Q105/Q106 and Q205/206 bases. The symmetry between B+ and B- is amazing.
The issue was R409, which measured 165K ohm instead of the expected 150K. I think I somehow overlooked it when I measured the first time.

The issue with the B+/B- rails voltage to class B amplifier remains: B+ is about 80V, while B- is 82. After the issue with R409 replaced all the resistors on the PSU board, so it should not be a resistor. Wonder what is wrong this time.

Anyway good progress so far, the bias at the gates of the VFET sockets is more steady and easier to adjust. I guess that when the +-82V issue is fixed then everything will be hunky-dory.
 
Last edited:
Those Kemet are very fine caps for these amps; tried and confirmed.
Just verify if the height suits your application; there is a PCB support in the 4650/5650 models that fixes the height of the PCB with respect to the chassis.
Of course some play may help as long as the PCB is not stressed.
 
The issue with the B+/B- rails voltage to class B amplifier remains: B+ is about 80V, while B- is 82. After the issue with R409 replaced all the resistors on the PSU board, so it should not be a resistor

From what I understand from the circuit, the +/-82V come from a very simple half wave rectification through diodes D301 and D306 from the secondary (violet and yellow) winding of the transformer. The two 1000uF 35V caps provide smoothing and they're rated only 35V because they're connected to the +/- 53V DC rails. If Q401 runs hotter than Q403, it means there's more current going through it, Vce is lower, and since Ve is regulated and stable at 60V it pulls down Vc from 82V to 80V. My guess is a leaky transistor between Q401, Q402 and Q405?
However, a 2V difference wouldn't really concern me too much, some voltage variations might be expected. Keep also in mind that the circuit is not loaded.
 
Back
Top Bottom