Yamaha A-520 clipping and distortion

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Dear forum members!

I have bought a defective Yamaha A-520 amp, and according to my measurements, it has a weird clipping problem, and it can be measured even at the RCA inputs at full volume, overloading my CD player's output. I have been trying to repair the amp, and all i could find was that if i remove the preamplifier stage PNP transistors after the input JFETs, the input is not distorted anymore. As soon as i put the transistors back, the distortion comes back again. The transistors seem to be good though, i have measured them with my multimeter.

Is there any of you who would be willing to assist me in repairing this amp?

Here is the service manual: YAMAHA A-520 SERVICE MANUAL Pdf Download.
 
Last edited:
there are some questions......
are both channels distorted (or only one)
signal distorted at the output of the preamp (or at the power amp)
what distortion (do you use a testsignal .....sinewave....)
have you measured voltages at some testpoints and compared these with
the schematic
and so on........
 
there are some questions......
are both channels distorted (or only one)
signal distorted at the output of the preamp (or at the power amp)
what distortion (do you use a testsignal .....sinewave....)
have you measured voltages at some testpoints and compared these with
the schematic
and so on........

Hi!

I am using a sinewave, it is a 1kHz signal played with a CD player. The left channel is working properly, this is the max signal i can get out from it to 8 Ohms without clipping, but it's more than enough.

BFoHFP9.png


The right channel looks like this (the volume pot is set to 20%).

uol57Vv.png
 
.........upper halfwave lost in the dark.........
good to hear that one channel is working.
i would begin to measure the dc-voltages (perhaps with a dmm, no input
signal)
in both channels and compare them - there must be somewhere
a mistake.

It seems like the JFET input stage is faulty, because the DC voltages are not OK. Also, the properly working channel uses the 2SK117-GR types, the faulty channel uses the 2SK117-BL types.
 
manual: ftp://77.172.185.158
gebr : schematic
ww : schematic

Hi! Thank you so much, i have the same serv manual. In fact i have repaired this amp today, D126 was 22Ohms, totally dead. I have replaced it and it works, but there is a 170mV DC offset on the RCh output, and the serv manual does not say anything about that. Got any ideas? Maybe the different JFETs are the problen?
 
If you have different J-FETs in the two channels then at least one pair of them was changed. Chances are they were not matched.
(I am talking about the TR103-105 pair and the TR104-106 pairs.)
J-FETs could have quite a bit of spreads in their parameters, so you really want to select matching pairs.
Take the J-FETs out and measure them.
You could start with the Idss measurement, the drain current when Vgs is zero, when the gate and source is shorted. Then you could measure the pinch-off voltage, the gate-source voltage when the current is decreased to zero.
Here is a tutorial on J-FETs: Junction Field Effect Transistor or JFET Tutorial
I am not sure how close you need to get with those parameters, but you would want the DC-offset of your amp less than 10mV. You could run some simulations to explore this.
You could also measure the FETs of the good channel for comparison.

Good luck, Peter
 
This amplifier is an example of minimalist overload protection.
- The upper VAS is a current source, about 3mA only. Total beta of output stage can drop to about >3000 at max. current, so >9A are allowed (unless something is shorted).
- The lower VAS can do about 90mA. Plenty of current.
- The system of the three diodes and the zener bypasses driver input to stage output, at over about +15A/-15A, not only covering short circuited speaker load, but also many types of internal short.
- In case of excessive current through the Vbe multiplier, as due to a transistor shorted to a rail, the zener shorts and removes drive from upper side of output stage, preventing destruction due to cross-conduction.
- The short circuit (and overload) detection has a threshold of about +/-7.5A (less when hot), a lag of about 1.2ms, a relay opening time of about 10ms (plus few dozen ms spark tail if load is inductive), recover of 3s (also power up relay close delay).
- The DC protection has thresholds of about -9V, +9V (assuming that 2SC2240 has a reverse beta of about 8), delay of about 700ms.

I'm still wondering how the zener managed to short. It was not with the 3mA of upper VAS current source. It was probably a genuine "component failure" due to mechanical and chemical aging, not electrical overload of any kind. That hypothesis would imply that the circuit design is good enough to expose "semiconductor manufacturer failure rate".
 
If you have different J-FETs in the two channels then at least one pair of them was changed. Chances are they were not matched.
(I am talking about the TR103-105 pair and the TR104-106 pairs.)
J-FETs could have quite a bit of spreads in their parameters, so you really want to select matching pairs.
Take the J-FETs out and measure them.
You could start with the Idss measurement, the drain current when Vgs is zero, when the gate and source is shorted. Then you could measure the pinch-off voltage, the gate-source voltage when the current is decreased to zero.
Here is a tutorial on J-FETs: Junction Field Effect Transistor or JFET Tutorial
I am not sure how close you need to get with those parameters, but you would want the DC-offset of your amp less than 10mV. You could run some simulations to explore this.
You could also measure the FETs of the good channel for comparison.

Good luck, Peter

Hi! Do i really need to have matched JFETs? I think i can replace some resistors with trimpots and adjust them, like in the Hiraga Le Monstre amplifier i have built. Do you think it's possible?
 
Hi! Do i really need to have matched JFETs? I think i can replace some resistors with trimpots and adjust them, like in the Hiraga Le Monstre amplifier i have built. Do you think it's possible?

You can re-engineer the input stage if you wish. Personally I think it would be simpler to find matched FETs.

If you re-engineer, putting a small pot between the sources of the long-tail pair, like in the Hiraga amp you mention, is not the best idea.
The parts of the pot acts as a source degeneration resistor for the FETs, further reducing their transconductance, that is smaller than that of a BJT to start with it. That will reduce the loop-gain of the PA, and that will "ruin" the various numbers of merit of the amp.
Not to mention that while this solution would give a DC balance, the differing resistances of the pot (the wiper would not be in the middle) would reduce the transconductance of the two FETs differently...
Do read up on this in the Doug Self audio amp book, or in his essay titled "Distortion in Power Amplifiers" that was published on his web-pages. (Those web pages are gone now, the extended version of that essay became his book, but probably you can still find it in the web archives. I saved it way back into a pdf file, I can send it to you if you require it.)
Again, I suggest you explore the effects of such a modification in simulation.

If you want to put some pot adjusting the DC output, I would suggest to use two Si diodes connected to the ground, fed from the supply rails through resistors. That would give you "stable" +/-0.6V, feed the trim-pots from it for each channel, and attach the wipers to the ends of the R161, R162 resistors that went to the ground. (I hope that is clear to you without a drawing!)
Long term reliability of trimmer pots are not too good, so as mentioned before, I would prefer finding matched FETs.

Good luck, Peter
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.