[FIXED] Yamaha A-S700 intermittent power-on failure (and eventually full failure)

[Zapper]

Far out!

OK, here are a few ideas. You get what you pay for, so don't sue me ;-)

To adjust the DC offset, we want to inject a +/- adjustable current into the front end. The top of resistor R119 is a good spot, because a voltage developed there will alter the bias on Q103 through resistor R109. A simple way is to connect a potentiometer between the +/-15V supplies in the preamp, and then have a 100k resistor to limit the current to the top of R119. This should produce a swing of ~+/-100mV on the output. Reduce the 100k resistor if you want more, increase it if you want less. Excuse my lame MS Paint sketches.

A slightly more elaborate approach adds RC filters on either side of the potentiometer, which divides the voltage by 1/3, so the output resistor goes to 33k:

For the output bias, Q117 forms the bias circuit known as a Vbe multiplier. The collector to emitter voltage of Q117 will equal its base to emitter voltage Vbe times a divider ratio determined by R129, R133, R175, and R131: Vce = (R129 + R133 || R175 + R131)/(R133||R175+R131) * Vbe. The Vce of Q117 sets the bias across the output devices because it determines the voltage difference between the bases of Q119C and Q119A. We can adjust this bias by tampering with this resistor divider. I propose removing R133 and inserting a 1K trimpot in its place, as shown. The original divider has a ratio of 3.67. The modified circuit has a variable ratio from 3.20 to 4.29.

The failure mode of a potentiometer is the wiper going open, so if that happens we want to reduce bias and suffer more distortion rather than increasing bias and risking thermal failure. If the wiper goes open the ratio goes to its smallest value and reduces the bias on the output, so it fails safely. Likewise the failure mode of the offset circuits is you get no offset control and resort to factory settings.

 

[Doodski]

OK, here are a few ideas. You get what you pay for, so don't sue me ;-)

To adjust the DC offset, we want to inject a +/- adjustable current into the front end. The top of resistor R119 is a good spot, because a voltage developed there will alter the bias on Q103 through resistor R109. A simple way is to connect a potentiometer between the +/-15V supplies in the preamp, and then have a 100k resistor to limit the current to the top of R119. This should produce a swing of ~+/-100mV on the output. Reduce the 100k resistor if you want more, increase it if you want less. Excuse my lame MS Paint sketches.

View attachment 355957

A slightly more elaborate approach adds RC filters on either side of the potentiometer, which divides the voltage by 1/3, so the output resistor goes to 33k:
View attachment 355958

For the output bias, Q117 forms the bias circuit known as a Vbe multiplier. The collector to emitter voltage of Q117 will equal its base to emitter voltage Vbe times a divider ratio determined by R129, R133, R175, and R131: Vce = (R129 + R133 || R175 + R131)/(R133||R175+R131) * Vbe. The Vce of Q117 sets the bias across the output devices because it determines the voltage difference between the bases of Q119C and Q119A. We can adjust this bias by tampering with this resistor divider. I propose removing R133 and inserting a 1K trimpot in its place, as shown. The original divider has a ratio of 3.67. The modified circuit has a variable ratio from 3.20 to 4.29.
View attachment 355959
The failure mode of a potentiometer is the wiper going open, so if that happens we want to reduce bias and suffer more distortion rather than increasing bias and risking thermal failure. If the wiper goes open the ratio goes to its smallest value and reduces the bias on the output, so it fails safely. Likewise the failure mode of the offset circuits is you get no offset control and resort to factory settings.

That stuff you formulated and calc'd @Zapper is what separates a working electronics engineer from a electronic technician like me. That's why I don't even attempt to try to pretend that I know something about the stuff you know. You explained it so well that it makes sense to me. This is the teachable moment stuff that I never received in the electronics semi-conductor devices section training at the Institute of Technology where I studied electronics. After the study when I was employed I did receive circuit analysis training from engineers that where employed by manufactures to instruct electronic technicians like me although it was not like you do it and was not about designing or MODing circuitry. What it was is to start at the beginning of a device and go through the entire device and get a decent understanding of the circuitry operation so that one can service a cassette player, DAT, CD, DVD, MD, PWM power supplies, the usual audio amplifier types and any car audio device going at the time including absurd and intriguing monster car amps up to about 3+ feet long. Thanks for taking the time and for providing your valuable professional experience and knowledge so that I can see how you operate and so that possibly @catch22 can MOD his Yamaha amp and then have maybe the only one inexistence with proper offset and bias calibration adjustments. What are you thinking @catch22? Are you interested in MODing your Yamaha amp and making it a very rare piece with bias and offset calibration adjustments?

 

[catch22]

OK, here are a few ideas. You get what you pay for, so don't sue me ;-)

This is the kind of stuff I was talking about when I said I truly regret I do not understand this science. Stunning.

My attempt to understand this tells me: once you inject something for example right of R109, it goes left through R109 and up and right to Q103 - ok, fair enough, but it also goes from bottom of C165 through it up and further. And at the same time also from the left of R109 it goes to C107 and to R119 and propagates EVERYWHERE , altering the initial designed voltage EVERYWHERE like waves in the sea... And the final result is not predictable (for me). This is our human fear of unknown and it is also so impressive to see what you can do if you have proper knowledge So yep agree with @Doodski this is coming from true professional experience.

What are you thinking @catch22? Are you interested in MODing your Yamaha amp and making it a very rare piece with bias and offset calibration adjustments?

I am all in, just give me the list of parts (I will go to RS this week to collect the 22uF/630V cap) for all these 33k R, a potentiomenter and caps for RC filter... As a PoC we can do it on a faulty slightly misadjusted(80mA currently) left channel and bring it down to 20mA. If all works, replicate the same for Right channel for future possibility of adjusting if it starts to creep over time from 20mA up.
Only if you accept my limited speed of implementation

 

[catch22]

I would also want to bring the L down to 20mA even before the manual adjustment. So these 2:

Desolder and measure? Or replace anyway? Measure would only show a reduced/increased capacity, but won't show if they are leaking?

 

[Doodski]

This is the kind of stuff I was talking about when I said I truly regret I do not understand this science. Stunning.

My attempt to understand this tells me: once you inject something for example right of R109, it goes left through R109 and up and right to Q103 - ok, fair enough, but it also goes from bottom of C165 through it and further. And at the same time also from the left of R109 it goes to C107 and to R119 and propagates EVERYWHERE , altering the initial designed voltage EVERYWHERE like waves in the sea... And the final result is not predictable (for me). This is our human fear of unknown and it is also so impressive to see what you can do if you have proper knowledge So yep agree with @Doodski this is coming from true professional experience.


I am all in, just give me the list of parts (I will go to RS this week to collect the 22uF/630V cap) for all these 33k R, a potentiomenter and caps for RC filter... As a PoC we can do it on a faulty slightly misadjusted(80mA currently) left channel and bring it down to 20mA. If all works, replicate the same for Right channel for future possibility of adjusting if it starts to creep over time from 20mA up.
Only if you accept my limited speed of implementation

That's very interesting that you are willing to take the initiative and do this MOD stuff. Keep in mind this is R&D at this point and not yet torture tested in your specific unit so it's not written in stone that you can effectively access the parts required without excess expense incurred. Let's wait till after the unit is fully 100% repaired and confirmation of repair(s) is completed and then address any MODs. That's how it's done.

Otherwise what you are seeing from @Zapper is the culmination of ~several years of intense multi discipline formal study, years and years of working knowledge decades of experience and a very nice knowledge donation to your cause in getting your Yamaha amp operational.

As per offset voltage if a calibration adjustment is in operation in pretty much all decent amplifiers the offset voltage is generally calibrated in mV and not tens of mVolts.

 

[Doodski]

I would also want to bring the L down to 20mA even before the manual adjustment. So these 2:
View attachment 355992
Desolder and measure? Or replace anyway? Measure would only show a reduced/increased capacity, but won't show if they are leaking?

Are those the caps that @Zapper outlined as being the amplifier circuitry DC voltage blocker capacitors that can cause the offset voltage at the speaker terminals? I'll let you check that for practice.

 

[Doodski]

Only if you accept my limited speed of implementation

Haha... That's funny and intriguing but otherwise I am setup permanently here both @ ASR and @home and not going anywhere unless I drop dead out of the blue...LoL.

 

[catch22]

That's very interesting that you are willing to take the initiative and do this MOD stuff. Keep in mind this is R&D at this point and not yet torture tested in your specific unit so it's not written in stone that you can effectively access the parts required without excess expense incurred. Let's wait till after the unit is fully 100% repaired and confirmation of repair(s) is completed and then address any MODs. That's how it's done.

Deal

Are those the caps that @Zapper outlined as being the amplifier circuitry DC voltage blocker capacitors that can cause the offset voltage at the speaker terminals? I'll let you check that for practice.

Yes, they are:

If either of these capacitors is leaky, it will cause increased output offsets.

There are many other potential sources of DC offset, but these caps are the first suspects.

Haha... That's funny and intriguing but otherwise I am setup permanently here both @ ASR and @home and not going anywhere unless I drop dead out of the blue...LoL.

True

 

[Doodski]

@catch22. Yes, please do order in the amplifier circuitry DC blocker capacitors and replace them when you replace the capacitor located @ the standby power supply circuitry. After that and BEFORE (Really really important.) powering the unit ON at a minimum the amplifier PCB must be removed (You will most likely already have the amp PCB out of the chassis for the DC blocker capacitors replacement anyway so you can hopefully kill multiple birds with one stone per say.) and very thoroughly inspected with your snazzy illuminated magnifier tool and remove all old solder from overheated and suspect solder joints that I am confident you might find and apply fresh new solder to those joints. For a beginner like yourself keep in mind to only unsolder one component lead leg at a time on one component at a time as you rework the PCB so that you don't have electronic components fall out from the PCB and then get all confused about where they came from when you find them loose on your workbench surface...LoL... I've been there and done that and it is pretty commonplace for some and causes lotsa extra work and confusion. The name of the game in electronics repair is being very clean, very organized, very thorough, not being invasive and not creating extra work. So as things progress I will dump tidbits of information into your mind and get you up to speed as best as think I can.

 

[catch22]

@catch22. Yes, please do order in the amplifier circuitry DC blocker capacitors and replace them when you replace the capacitor located @ the standby power supply circuitry. After that and BEFORE (Really really important.) powering the unit ON at a minimum the amplifier PCB must be removed (You will most likely already have the amp PCB out of the chassis for the DC blocker capacitors replacement anyway so you can hopefully kill multiple birds with one stone per say.) and very thoroughly inspected with your snazzy illuminated magnifier tool and remove all old solder from overheated and suspect solder joints that I am confident you might find and apply fresh new solder to those joints. For a beginner like yourself keep in mind to only unsolder one component lead leg at a time on one component at a time as you rework the PCB so that you don't have electronic components fall out from the PCB and then get all confused about where they came from when you find them loose on your workbench surface...LoL... I've been there and done that and it is pretty commonplace for some and causes lotsa extra work and confusion. The name of the game in electronics repair is being very clean, very organized, very thorough, not being invasive and not creating extra work. So as things progress I will dump tidbits of information into your mind and get you up to speed as best as think I can.

Repeating the sequence
Main problem:
1. Replace C254 (easily accessable from the side, no disassembly requireed)
2. Check if the main problem is gone

DC Offset:
3. Remove PCB MAIN(1) to be able to access C101 and C107 from the bottom of ghe PCB
4. Replace both caps
5. Use the opportunity, while the PCB is removed, to assess the condition of other joints in MAIN(1) and desolder/resolder suspect joints
6. Reassembly
7. Check DC Offset again

Right?

 

[Doodski]

Repeating the sequence
Main problem:
1. Replace C254 (easily accessable from the side, no disassembly requireed)
2. Check if the main problem is gone

DC Offset:
3. Remove PCB MAIN(1) to be able to access C101 and C107 from the bottom of ghe PCB
4. Replace both caps
5. Use the opportunity, while the PCB is removed, to assess the condition of other joints in MAIN(1) and desolder/resolder suspect joints
6. Reassembly
7. Check DC Offset again

Right?

Yes, perfectly stated in proper order. After all that occurs and the unit has been very thoroughly tested, confirmation of repair is done and using impacts sustained to the chassis with a screwdriver handle etc to check for intermittent connections then if you are still hot to trot on the MOD action that may proceed. So the amp might be on your workbench for weeks while the parts order(s) are in progress and whatever is happening at any given time. With repairs I never reassemble a unit completely when waiting for parts or other service stuff needs to be done because it just wears out the screws and causes scratches etc and is invasive technique.

 

[catch22]

Yes, perfectly stated in proper order. After all that occurs and the unit has been very thoroughly tested, confirmation of repair is done and using impacts sustained to the chassis with a screwdriver handle etc to check for intermittent connections then if you are still hot to trot on the MOD action that may proceed. So the amp might be on your workbench for weeks while the parts order(s) are in progress and whatever is happening at any given time. With repairs I never reassemble a unit completely when waiting for parts or other service stuff needs to be done because it just wears out the screws and causes scratches etc and is invasive technique.

have some guys waiting for the mod.
Used these to assemble NodeMCU with some pullups and screw terminals for home light automations

 

[catch22]

All these are 220u 50V as needed for C107
So many flavours...
https://ie.rs-online.com/web/p/aluminium-capacitors/7152852 UPW1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/2437292 UHE1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/5194346 UPS1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/7621755 UBT1H221MPD8 <- this one is High Temperature Range, For +125°C
https://ie.rs-online.com/web/p/aluminium-capacitors/5201012 UVR1H221MPD +85˚C, series VR(M) same as in my amp
All these are Nichicon just as mine, in fact there are much more others.
Do I really need to get the VR? they are only sold in packs of 25pcs whereas 1-3 are 5pcs or 10pcs
tolerance +-20% for all

The ones in my amp are Series VR(M) attached

 

[Doodski]

All below are 220u 50V as needed for C107
So many flavours...
https://ie.rs-online.com/web/p/aluminium-capacitors/7152852 UPW1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/2437292 UHE1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/5194346 UPS1H221MPD +105˚C
https://ie.rs-online.com/web/p/aluminium-capacitors/7621755 UBT1H221MPD8 <- this one is High Temperature Range, For +125°C
but not much difference between 1-3
tolerance +-20% for all
All 4 are Nichicon just as mine, in fact there are much more others.

The ones in my amp are Series VR(M) attached

I am going to leave the capacitor selection for peeps more expert in capacitor(s) selection than I am. I am accustomed to in-warranty service and servicing out-of-warranty gear that is the same name brand as the in-warranty service gear that I serviced. So as a in-warranty service depot electronic technician I had 100% access and supply of service manuals, addendums, MODs, supplemental service literature and would simply order the exact manufacturer's part number from the service manual and then receive the exact same part/capacitor as was found in the repair unit. So I never bothered to select random capacitors etc. I have little to no knowledge of all the various near daily new and improved capacitor technical details and of the huge assortment of capacitor differences and stuff like that. Somebody here @ ASR will read this and select a suitable or better replacement capacitor(s). I'm not going to profess to know everything and if I am lacking in specific areas I state that and don't pretend.

 

[catch22]

I am going to leave the capacitor selection for peeps more expert in capacitor(s) selection than I am. I am accustomed to in-warranty service and servicing out-of-warranty gear that is the same name brand as the in-warranty service gear that I serviced. So as a in-warranty service depot electronic technician I had 100% access and supply of service manuals, addendums, MODs, supplemental service literature and would simply order the exact manufacturer's part number from the service manual and then receive the exact same part/capacitor as was found in the repair unit. So I never bothered to select random capacitors etc. I have little to no knowledge of all the various near daily new and improved capacitor technical details and of the huge assortment of capacitor differences and stuff like that. Somebody here @ ASR will read this and select a suitable or better replacement capacitor(s). I'm not going to profess to know everything and if I am lacking in specific areas I state that and don't pretend.

Lucky you Understand.
I can google using Yamaha part number UR268220 and still find it, but then the price is €15 or so for 1 piece plus delivery Gut feeling it is not needed.
My feeling any of the above 5 (or at least 1,2,3,5) will do. If anyone here confirm, fine, if not I will pick up any.

 

[catch22]

One thing that puzzles me now.... I have a degree in maths, and I just cannot understand how on Earth with the tolerance of 20% any quality of tuning is possible?
If you have 100s of components and if each has tolerance 20% it is a total mess. How this works at all?

 

[Doodski]

One thing that puzzles me now.... I have a degree in maths, and I just cannot understand how on Earth with the tolerance of 20% any quality of tuning is possible?
If you have 100s of components and if each has tolerance 20% it is a total mess. How this works at all?

Yes, it's twisted reasoning and requires mental somersaults but it works. It's all about the proportions in the reactance formulas and frequency response formulas and how capacitive reactance works with the frequency response and what the actual duty/function of the specific capacitor is when in circuit. In this amp's situation you have an AC coupling capacitor for the standby power supply and in another case you have DC decoupling with AC conduction and in some caps in your amp audio section they are strictly for shunting bad electrons distortion stuff from the audio voltage amplification circuitry to the ground potential and in all 3 of those applications the capacitive reactance proportions and laws are the same but the end result is different. It's really cool stuff and I was very hooked when studying it in class. We have not even approached inductive reactance yet but the frequency response laws and the time constants are the same but the operation is different. To thoroughly study the capacitive reactance and inductive reactance requires maybe 2 months of full time study in class study for mortals including the labs to get a very good handle on what is happening. For a peep like you with a advanced maths degree it should be pretty fast to get the study completed and be effective. If you want formal literature about this try Thomas L. Floyd - Principles of Electricity at your local university or technical institute library. That will give you great diagrams, schematics, examples, show the maths work clearly and in short order bring you up to speed.

 

[Zapper]

My attempt to understand this tells me: once you inject something for example right of R109, it goes left through R109 and up and right to Q103 - ok, fair enough, but it also goes from bottom of C165 through it up and further.

But, we are injecting only DC current (ideally), and C165 blocks DC. So it will have no effect on Q105.

I said "ideally", because we never have pure DC, there is always some noise, and we don't want to inject noise. That's why I sketched the second version, with additional RC filters from the +/-15V supplies, just in case they have any appreciable noise. They shouldn't, and almost certainly don't, but that's an extra precaution. Also resistors generate a small amount of electrical noise all by themselves, but the low value of R119 will greatly attenuate any noise that the additional resistor network would add.

And at the same time also from the left of R109 it goes to C107 and to R119 and propagates EVERYWHERE , altering the initial designed voltage EVERYWHERE like waves in the sea... And the final result is not predictable (for me).

It's the same thing. The small DC offset voltage we create is blocked by C107, C167, C101, C103, C165. The only paths remaining are through R119 to a local ground node, and through R109 to the base of Q103. So it really is very simple! A lot of electronic analysis is figuring out how to simplify the problem.

This is our human fear of unknown and it is also so impressive to see what you can do if you have proper knowledge So yep agree with @Doodski this is coming from true professional experience.

The flip side of the fear of the unknown is the excitement of trying and learning new things. I commend you for your adventurous spirit in considering modifying your Yamaha!

 

[Zapper]

One thing that puzzles me now.... I have a degree in maths, and I just cannot understand how on Earth with the tolerance of 20% any quality of tuning is possible?
If you have 100s of components and if each has tolerance 20% it is a total mess. How this works at all?

The modern approach is to run thousands and thousands of simulations. Monte Carlo simulations are used to randomly vary the values of each component at each trial. Sensitivity analyses are also used, where one component is varied per trial and its effect on the circuit is evaluated. Then all the random sensitivities can be RSS'd (root sum squared) together to estimate the net effect. Analog design involves spending a lot of time finding cases where the circuit breaks and fixing it, hopefully without breaking it some other case. Analog design is squishy - you squeeze it in one place, it pops out in another.

In addition to component variations, temperature is a huge effect. The properties of silicon devices change rapidly with temperature, some linearly, some exponentially. Voltage variations on unregulated supplies (as often used in power amps) due to mains variations has to be analyzed too.

Circuit design techniques have been developed over the years to reduce the sensitivity to these effects. For example, the base to emitter voltage Vbe of a bipolar transistor decreases linearly with temperature. This makes it hard to bias a single transistor so it conducts the same current at all temperatures. This is circumvented by using a pair of well matching transistors, such as the differential pair Q103 and Q105. Q105 buffers the negative feedback from the amp output at applies it to the emitter of Q103, as they have a common emitter connection. If an increase in temperature causes the Vbe of Q103 to decrease, it does the same to Q105, and the currents that each conduct remains unchanged. In effect the change in Vbe of Q103 is countered by a matching change in Vbe from Q105. This completely solves the problem of the temperature dependence of Vbe in this circuit.

Q107 and Q109 have a similar relationship. Together they form a very useful circuit called a current mirror. Q109 is the input to the mirror and Q107 is the output; Q107 will output a current that is (nearly) identical to the input current that flows into Q109. Q109's base is tied to its collector, so it will develop the exact Vbe it needs to conduct the current applied to its collector. But the base of Q109 is tied to Q107, so Q107 has the same Vbe as Q109. Therefore, it must conduct the same current as Q109 (give or take a bit for matching). The advantage of this circuit is that it is very temperature insensitive, for the same reason as the diff pair. Vbe changes a lot over temperature, but it changes the same way for the two transistors, so it cancels.

Before we could run thousands of simulations easily, the designer or apps engineer had to figure out which components mattered the most, and change them by hand in a breadboard circuit to evaluate their effects. This is a picture of the late Jim Williams, a well known apps engineer for Linear Technologies (now part of Analog Devices), and his famously messy workbench, where he would do that type of evaluation.

 

[catch22]

Yes, it's twisted reasoning and requires mental somersaults but it works. It's all about the proportions in the reactance formulas and frequency response formulas and how capacitive reactance works with the frequency response and what the actual duty/function of the specific capacitor is when in circuit. In this amp's situation you have an AC coupling capacitor for the standby power supply and in another case you have DC decoupling with AC conduction and in some caps in your amp audio section they are strictly for shunting bad electrons distortion stuff from the audio voltage amplification circuitry to the ground potential and in all 3 of those applications the capacitive reactance proportions and laws are the same but the end result is different. It's really cool stuff and I was very hooked when studying it in class. We have not even approached inductive reactance yet but the frequency response laws and the time constants are the same but the operation is different. To thoroughly study the capacitive reactance and inductive reactance requires maybe 2 months of full time study in class study for mortals including the labs to get a very good handle on what is happening. For a peep like you with a advanced maths degree it should be pretty fast to get the study completed and be effective. If you want formal literature about this try Thomas L. Floyd - Principles of Electricity at your local university or technical institute library. That will give you great diagrams, schematics, examples, show the maths work clearly and in short order bring you up to speed.

I realise how important is 1) understanding the theory behind each type of component and
2) this is physical world, not just theoretical models, so physical properties and behaviour change with time, temperature, voltage, current flowing through, etc.
This and so much more are taken into account during design