Hi all,
I bought a cheap Sony FB920R amp from eBay as it was listed as not working so I thought I'd have a go at fixing it. I am very much an amateur at these sort of repairs but I am fine with the basics and no stranger to soldering and whatnot.
Once I got the amp I saw a couple of burnt resistors near one of the channel's MOSFETs and the amp stayed in protection mode when powered up. I immediately suspected the MOSFETs and confirmed with a multimeter that both of them were a dead short. I removed them all and replaced them with some close modern equivalents (IRFP340/IRFP9240), replaced the dead resistors, and also upped the rating of a couple of resistors in the bias circuit to allow for the higher voltage requirements of the new transistors, which then allowed me to set the bias for both channels at exactly 10mv, as per the service manual.
The amp now works and plays beautifully. Happy days.
However, here is my question. The DC offset on the left channel is about 15mv which is fine, but on the right channel (the one which was dead originally) it sits at around 69mv. This is the channel which had the failure.
Is this high enough to be cause for concern? I've read conflicting reports; some say anything under 100mv is no cause for concern, some say anything above 50 is not ideal. I am assuming something has perhaps been damaged or drifted out of spec during the failure of the channel. Any idea where I should start looking? I have read that the differential pairs are the most likely culprits of high offset. Can anyone point me in their direction for my amp?
The service manual can be seen here: https://www.manualslib.com/manual/1276748/Sony-Ta-Fb920r.html
Thank you in advance!
I bought a cheap Sony FB920R amp from eBay as it was listed as not working so I thought I'd have a go at fixing it. I am very much an amateur at these sort of repairs but I am fine with the basics and no stranger to soldering and whatnot.
Once I got the amp I saw a couple of burnt resistors near one of the channel's MOSFETs and the amp stayed in protection mode when powered up. I immediately suspected the MOSFETs and confirmed with a multimeter that both of them were a dead short. I removed them all and replaced them with some close modern equivalents (IRFP340/IRFP9240), replaced the dead resistors, and also upped the rating of a couple of resistors in the bias circuit to allow for the higher voltage requirements of the new transistors, which then allowed me to set the bias for both channels at exactly 10mv, as per the service manual.
The amp now works and plays beautifully. Happy days.
However, here is my question. The DC offset on the left channel is about 15mv which is fine, but on the right channel (the one which was dead originally) it sits at around 69mv. This is the channel which had the failure.
Is this high enough to be cause for concern? I've read conflicting reports; some say anything under 100mv is no cause for concern, some say anything above 50 is not ideal. I am assuming something has perhaps been damaged or drifted out of spec during the failure of the channel. Any idea where I should start looking? I have read that the differential pairs are the most likely culprits of high offset. Can anyone point me in their direction for my amp?
The service manual can be seen here: https://www.manualslib.com/manual/1276748/Sony-Ta-Fb920r.html
Thank you in advance!
The offset isn't an issue. If you use ohms law and work out the current flow in an 8 ohm load and the power dissipated in the load you will see its half of a quarter of nothing 😉
The original FET's are neither true Lateral FET's nor HEXFET's (which are what you have fitted). Provided there is no instability (which can in some cases show as an offset) there should be no issues. The output stage plays no part in the DC offset measured as long as it stable.
The bias generator should really be in thermal contact with the heatsink/FET's in the case of the IRFP devices as these have a positive tempco and that means the bias current will increase with temperature and in extreme cases can lead to thermal runaway.
The original FET's are neither true Lateral FET's nor HEXFET's (which are what you have fitted). Provided there is no instability (which can in some cases show as an offset) there should be no issues. The output stage plays no part in the DC offset measured as long as it stable.
The bias generator should really be in thermal contact with the heatsink/FET's in the case of the IRFP devices as these have a positive tempco and that means the bias current will increase with temperature and in extreme cases can lead to thermal runaway.
Thanks Mooly. I won't worry about the 69mv then! In terms of stability, both channels never really waver from these measurements even once the amp is nice and warm after playing for a while.
Could you explain your final paragraph please, as I'm not sure I follow what you mean and which bias generator components you are referring to. Thanks!
Could you explain your final paragraph please, as I'm not sure I follow what you mean and which bias generator components you are referring to. Thanks!
These are the bias generators.
Lateral FET's typically need no thermal compensation because they have a negative tempco over around 100ma current meaning current tends to reduce (for a fixed gate/source voltage) as temperature rises. The HEXFET's are the opposite.
Look in 99% of amps with either HEXFET or ordinary transistor outputs and you will see the bias generator fixed to the heatsink.
https://en.wikipedia.org/wiki/Rubber_diode
Lateral FET's typically need no thermal compensation because they have a negative tempco over around 100ma current meaning current tends to reduce (for a fixed gate/source voltage) as temperature rises. The HEXFET's are the opposite.
Look in 99% of amps with either HEXFET or ordinary transistor outputs and you will see the bias generator fixed to the heatsink.
https://en.wikipedia.org/wiki/Rubber_diode
Thank you. Those two transistors don't sit anywhere near the main heatsinks - would you advise fitting some sort of heatsink to them?
They don't need cooling, the one in the channel with the replacement FET's should heat and cool alongside the FET's it controls so as to try and compensate for changes in bias current that occur woth the change in temperature.
It would be worth seeing how stable the bias really is.
If you set both channels correctly, then loosely put the cover on and then play it loud so it gets hot and then recheck the bias it should ideally be unchanged. The bias should be the same cold, warm and hot. How close it reaches that ideal comes down to the circuit design but the HEXFETS's throw the positive tempco into the mix.
Look at most amps and you see the two output transistors on a heatsink and a third smaller one usually sat in the middle. That third one is the bias generator.
It would be worth seeing how stable the bias really is.
If you set both channels correctly, then loosely put the cover on and then play it loud so it gets hot and then recheck the bias it should ideally be unchanged. The bias should be the same cold, warm and hot. How close it reaches that ideal comes down to the circuit design but the HEXFETS's throw the positive tempco into the mix.
Look at most amps and you see the two output transistors on a heatsink and a third smaller one usually sat in the middle. That third one is the bias generator.
The M-fets IRFP340/9240 have a Vth of about 4VDC while the 2SK1529/2SJ200 have a Vth of about 2VDC, which means that the driver transistors Q509/Q510 on the repaired channel dissipate twice as much vs the channel with the original Toshiba M-fets. To correct this, you need to increase the resistance value of R519 to 680 Ohms. Important, set RV500 to the MIN position (counterclockwise) before turning on the amplifier.
That suggests 'leave it be' 🙂
Good thinking. The higher gate/source turn on voltage doubles the current in the 330 ohm. Its a very valid point.
What would be the consequences of not changing these two relevant resistors? Potential premature failure of the driver transistors?
The value of R519 is quite significant because it simultaneously defines the dissipation on the drivers Q509/Q510 (read: quiescent current through the driver stage) as well as the quiescent current through the output stage Q511/Q512 (in symbiosis with the Q508 bias transistor), but equally importantly it defines at what speed the driver stage will charge and discharge the very reactive and nonlinear capacities of the output M-fets.
With the value R519=330 Ohm the quiescent current of the driver stage is approx. 24mA which means a constant dissipation of approx. 1W on each of the drivers that do not have any heat sinks on them (as far as I can see). If you look at the datasheet for Q509/Q510 you will see that the max. allowed dissipation in free air is 1W5 (or 10W with an infinitely large heat sink), which in your case means that your driver transistors are getting pretty hot - I assume you can't keep your index finger on their heatsink tab for more than a few seconds). Therefore, either reduce the current to the original approx. 12mA or put appropriate heat sinks on the driver transistors. Keep in mind that these are very good-excellent driver transistors, and I assume almost impossible to obtain today, so choose your poison very carefully.
In addition to the above, as Moly already emphasized, it is very important to ensure correct temperature compensation of the output stage (output stage bias) through the largest possible temperature range, from completely cold to fully warmup amplifier. To successfully do this, Q508 should be mounted either on the output stage heat sink or directly on the heatsink tab of the output M-fets, with 3 twisted (or flat wires) soldered to the Q508 pins with a thin layer of thermal paste between the surfaces. I leave it to your imagination and suggestions from the forum brothers to figure out how to do it simply.