Author Topic: EEVblog #378 - Dumpster Diving Teardown Repair  (Read 29732 times)

0 Members and 1 Guest are viewing this topic.

Offline Pentium100

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: lt
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #25 on: October 31, 2012, 01:24:10 am »
I wonder how long this amp would run with no power and the AC detect disabled. It has enough capacitance in it! I modded my cheap computer speakers to have a 6800µF bulk cap because the transformer was dropping out on high peaks (470µF was the original.) Now they last 3-4 seconds with no power and ordinary music.   ::)

Probably not very long as high quality amps have larger idle currents (it used about 40W when the relay was forced on). My tube headphone amp has 470uF and 330uF caps on B+, when I switch off the power tubes cool down faster than the caps discharge (that's why I have the bleeder resistor and a neon light to indicate whether it's still dangerous).
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #26 on: October 31, 2012, 01:50:12 am »
Useful video, but his advice about removing the ground on the scope, or isolating the scope is dangerous. The hifi amplifier should always be the isolated part.

I also suspect it doesn't work quite as he says - the data pin should be high most of the time (it essentially equals Vcc when opto off), the clock latches this data in (from the AC), which turns off the FET (through not-Q.) When data falls low, it latches that in, which turns on the FET. It ensures the transformer only ever sees 0V or one AC cycle (full cycle due to half-wave rectifier), quite neat actually...

I have also always wondered if there's a good reason not to use that full wave rectifier with FET to switch low voltage, high current AC -- over using a triac, for example.
I'm also surprised they used a relatively expensive MOSFET as opposed to a relatively cheap triac. Also, I don't see any snubbers to handle the inductive spikes, so I have no idea how it can be reliable. Apart from that, it's actually an interesting way to "period skip" a linear power supply!

Nowadays, they would just use a small switching power supply based on a single chip flyback converter. And the main power supply would be a switching unit as well (similar to a PC power supply), with variable output voltage in the fancier units.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline wastrix

  • Newbie
  • Posts: 5
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #27 on: October 31, 2012, 02:20:49 am »
Yamaha Training video on their "green" power supplies.

Youtube has a video for everything!, what are odds?

Anyway, followup:
It was C405, a 22nF metalised polyester film cap, 630V.
It read low, about 500pF.
A common fault in these Yamaha's apparently, and a few Youtube commenters pointed it out.

Dave.
Yep, I've repaired quite a few of these Yamaha receivers, every single time it was that cap. The boss rarely bothered checking anything else, just replaced that cap as a matter of course and 99% of the time that fixed it.

Quick way to make up an $80 service charge!
 

Offline Pentium100

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: lt
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #28 on: October 31, 2012, 03:04:06 am »
I also suspect it doesn't work quite as he says - the data pin should be high most of the time (it essentially equals Vcc when opto off), the clock latches this data in (from the AC), which turns off the FET (through not-Q.) When data falls low, it latches that in, which turns on the FET. It ensures the transformer only ever sees 0V or one AC cycle (full cycle due to half-wave rectifier), quite neat actually...
When data is low (opto is on) the FET would be turned off. Opto is turned on when the output voltage is too high.

Nowadays, they would just use a small switching power supply based on a single chip flyback converter. And the main power supply would be a switching unit as well (similar to a PC power supply), with variable output voltage in the fancier units.
Switching power supplies have more noise on the output than linear power supplies. Cheaper receivers use them of course, but the higher quality ones probably still use linear supplies (I do not have a new receiver, so I cannot take it apart and look). On the other hand I still prefer hard power switches instead of the soft ones. When I switch off the power to my amp it consumes 0W.
 

Offline kyndal

  • Regular Contributor
  • *
  • Posts: 54
  • Country: ca
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #29 on: October 31, 2012, 04:18:11 am »
i fixed a similar AV unit a few years back..  although it was an ONKIO? brand or something rather..

but ya  turned out to to be a failed Film cap aswell.   Took FOR EVER!!.
(including switching the board with a working one)
then basically de-soldering every part and test out of circuit
Sigh.

i guess these days.. with youtube and what have you.. i could have saved myself a few gray hairs
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #30 on: October 31, 2012, 05:22:33 am »
Switching power supplies have more noise on the output than linear power supplies. Cheaper receivers use them of course, but the higher quality ones probably still use linear supplies (I do not have a new receiver, so I cannot take it apart and look). On the other hand I still prefer hard power switches instead of the soft ones. When I switch off the power to my amp it consumes 0W.
That noise is mostly irrelevant in modern digital equipment. The digital logic doesn't mind as long as it can reliably distinguish 1s and 0s, and it's a digital signal all the way to the power stage. Where the signal does become analog occurs at rather high power levels, where a little noise would have no effect. The whole thing operates just like a low voltage servo motor drive. (Which if you think about it, that's exactly what it is...) Some manufacturers like Samsung are going as far as putting the modulators, drivers, and power MOSFETs in the same chip.

Reducing the supply voltage at low volume settings reduces the losses in the power stage and also allows the entire dynamic range of the Delta Sigma modulators to be used at low volume.

If the only inputs have their own power (like USB and HDMI), they could use that to activate the relay and eliminate all standby power use while retaining automatic power on. The problem is that the legacy analog and S/PDIF inputs don't have power as such and cannot reliably power anything. It might be possible with analog but it would rely on working with a low impedance source with a significant voltage level. (They could use a transformer to boost the voltage to charge a cap, which is later switched out when main power is on.) It could possibly work with coaxial S/PDIF as well but optical S/PDIF would be basically impossible. A possible workaround might be to have a momentary mains button across the relay, to allow it to be switched on for legacy inputs.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline Pentium100

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: lt
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #31 on: October 31, 2012, 07:24:30 am »
That noise is mostly irrelevant in modern digital equipment. The digital logic doesn't mind as long as it can reliably distinguish 1s and 0s, and it's a digital signal all the way to the power stage. Where the signal does become analog occurs at rather high power levels, where a little noise would have no effect.
But the amp has analog inputs, which care about noise. Or maybe modern amps do not have them, I wouldn't know as I would not buy an amp that does not have analog inputs.
 

Online oPossum

  • Super Contributor
  • ***
  • Posts: 1415
  • Country: us
  • Very dangerous - may attack at any time
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #32 on: October 31, 2012, 07:26:53 am »
Shouldn't that be a X2 rated cap?
 

Online tom66

  • Super Contributor
  • ***
  • Posts: 6678
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #33 on: October 31, 2012, 08:33:41 am »
I wouldn't have thought higher ripple; otherwise, it would likely still turn on once in a while. That C405 is a capacitive supply for the d-type. It provides a ~8-10V or so supply. The 6.8V zener powers an optocoupler. If the optocoupler output goes out the secondary senses an error, even if the PSU keeps working.
 

Offline mzacharias

  • Frequent Contributor
  • **
  • Posts: 709
  • Country: us
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #34 on: October 31, 2012, 09:23:31 am »


Troubleshooting waveforms.
 

Offline Pentium100

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: lt
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #35 on: October 31, 2012, 09:57:19 am »
LOL, the switched outlets were rated for ~100W so they used a 2A fuse, but users kept plugging in high power devices and blowing the fuse, so they replaced it with 10A. Now the relay will probably melt before the fuse does (since it is probably rated for only 10A).
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #36 on: October 31, 2012, 10:47:05 am »
Shouldn't that be a X2 rated cap?

It was not fitted with an X or Y rated cap, nor was it specified. It was a bog standard 630V metal poly.

Dave.
 

Online oPossum

  • Super Contributor
  • ***
  • Posts: 1415
  • Country: us
  • Very dangerous - may attack at any time
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #37 on: October 31, 2012, 10:53:54 am »
I know it wasn't specified, but that type of power supply typically has a X2 cap and a power resistor for inrush current.

Seems like a flawed design that has an above average failure rate.
 

Online tom66

  • Super Contributor
  • ***
  • Posts: 6678
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #38 on: October 31, 2012, 11:37:07 am »
I think it should be an X2 cap, as a normal poly may burn under failure. The 2.2k might be a fusible resistor though -- so they might be allowed to use an ordinary poly.
 

Offline jaycee

  • Regular Contributor
  • *
  • Posts: 206
  • Country: gb
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #39 on: October 31, 2012, 11:40:05 am »
A guy with one of these appeared on DIYAudio a month ago actually ! Despite having never seen it before, my suspicion fell on C405 immediately. In that case, the guy had measured the primary voltage of the transformer and found it was low. I did have a WTF moment when I saw the flipflop being used, but after a minute or two I suspected they were using it to pulse feed the primary rather than running it directly from line voltage.

It looks like they should have specified an X2 rated part, but the bean counters decided to cut a few cents and made it an ordinary film cap. The problem is, this cap will see all sorts of line spikes, which will ruin it. If they had put a VDR across the mains input, it might have been OK. Bean counters again! An X2 rated dielectric would have had some self healing properties which would have allowed it to tolerate line spikes, and the unit would have been reliable in the field. It is just as well the film capacitor they chose fails open, or there could have been a fire hazard!

As someone else mentioned, C405 is functioning as a dropper due to reactance. This was quite common for small "off line" supplies before SMPS became so easy/cheap to produce. That produces the voltage that the flipflop runs on. Also they are using it to AC couple the line frequency in, and the thing overall functions a bit like a light dimmer. My guess is because the MOSFET wasn't firing often enough due to the low coupling capacitance, there was the correct voltage on the secondary but not enough current backing it. The micro probably ran OK, but as soon as it tries to switch on the relay, the voltage collapses, resetting the micro - so all seems dead.

To those who have said "nowadays this would be an SMPS", true... the standby supply likely would be. However, the main supply, it depends. Class D (PWM) amplifiers are now becoming common for receivers due to their high efficiency and low heat output, and these are perfectly happy with an SMPS supply. On the other hand, a lot of high quality audio is still done with Class AB amplification, and it takes a LOT of effort to get a clean enough output from an SMPS. So typically, if it's Class AB, it will use a good old linear power supply.
 

Online tom66

  • Super Contributor
  • ***
  • Posts: 6678
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #40 on: October 31, 2012, 11:44:23 am »
If I were designing this, and I had to keep the 1W standby, I'd use a 680k/1W resistor in place of that cap, and replace the electro with one a little bigger, maybe 47µF (good high reliability cap), to give it some "backup". A separate tap off the transformer would provide some extra power once the circuit starts. It would use about 0.2W more, because of the resistor power dissipation, but would be more reliable. This is similar to the design in switch mode power supplies, but using lower frequency pulsing.

The 22nF cap has an impedance of 144k at 50 Hz, it is clearly used as a capacitive dropper, but these circuits don't seem to be very reliable. Since it was still oscillating down to 500pF, that is impressive...
 

Offline jaycee

  • Regular Contributor
  • *
  • Posts: 206
  • Country: gb
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #41 on: October 31, 2012, 12:01:44 pm »
tom66, if you were designing this today you'd use one of those one-chip SMPS's with onboard MOSFET. A bridge, a 400V electrolytic cap, the IC, a transformer, a few passives and bingo :) Same sort of thing they bundle in wall warts nowadays.

The Yamaha is not that old though, so I guess they chose this circuit for noise reasons. Probably easier to pass regulatory compliance too.
 

Offline Salas

  • Frequent Contributor
  • **
  • Posts: 292
  • Country: gr
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #42 on: October 31, 2012, 12:53:41 pm »
jaycee, your explanation is right on the money. That MKT film should be punched inside like Swiss cheese by now. Takes X2 rated or 1000V polypropylene film and foil pulse caps, polypropylene is self healing also. C405 stole Dave's thunder in the land of down under. Bummer.
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #43 on: October 31, 2012, 01:23:52 pm »
But the amp has analog inputs, which care about noise. Or maybe modern amps do not have them, I wouldn't know as I would not buy an amp that does not have analog inputs.
The legacy analog inputs are converted to digital to drive the modulators. That part is quite small and easy to shield.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline Pentium100

  • Frequent Contributor
  • **
  • Posts: 258
  • Country: lt
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #44 on: October 31, 2012, 01:48:31 pm »
But the amp has analog inputs, which care about noise. Or maybe modern amps do not have them, I wouldn't know as I would not buy an amp that does not have analog inputs.
The legacy analog inputs are converted to digital to drive the modulators. That part is quite small and easy to shield.
So I made the right choice in buying an older amp that is all analog - no A->D->A conversion adding noise and distortion and the circuit diagram is easier to understand. So, no new amps that use class A/AB, but are not audiophile-grade (and cost)?
 

Online tom66

  • Super Contributor
  • ***
  • Posts: 6678
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #45 on: October 31, 2012, 01:58:06 pm »
Plenty of new amps use class-AB (my brother bought one just a year ago -- a Denon 7.1 system), it's just it's becoming less common because you can reach the same performance with class-D nowadays. The only benefit is either wanting a nice room heater, or for audiophiles who like "the sound" of class-AB.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #46 on: October 31, 2012, 04:09:33 pm »
I have a similar one with a bad Dolby processor, low audio on the main channels but ok on surround info. Not the amplifiers but somewhere in that jungle of switching. Similar looking bard but has a dolby chipset in it. Must look at the model and see if i can get a manual for it and bypass the logic.

Anther is an Onkyo with a bad amplifier, needs about 10 transistors changed ( bloody expensive FET's as well, so not likely to be done) and they are of course on the massive heatsink buried in the bottom of the chassis, with the entire thing built on top of it. Switches on and works, but gets very toasty very fast then does a thermal shutdown, with one distorted channel.
 

Offline Salas

  • Frequent Contributor
  • **
  • Posts: 292
  • Country: gr
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #47 on: October 31, 2012, 11:18:42 pm »
Shouldn't that be a X2 rated cap?

It was not fitted with an X or Y rated cap, nor was it specified. It was a bog standard 630V metal poly.

Dave.

They designed and tested at 110VAC then they negotiated a 630V MKT type in large quantity and in several values for their whole production range. The large voltage margin will cover X spec inability they rest assured, is my scenario. Did the final boards, and went to production. Then all the 230-240VAC countries importers started receiving units to repair. They simply did not test at 240VAC. Was late, no X2 or 1000V PP cap at that size and value. My wild scenario, but if those units are not committing  suicide like Lemmings in America too, it maybe holds some water. 8)
 

Offline Salas

  • Frequent Contributor
  • **
  • Posts: 292
  • Country: gr
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #48 on: November 01, 2012, 12:24:07 am »
Found the excellent jaycee reference thread too. Interesting on how they tackled it not long ago.

http://www.diyaudio.com/forums/solid-state/216317-yamaha-dsp-ax750se-dead-no-power.html
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: EEVblog #378 - Dumpster Diving Teardown Repair
« Reply #49 on: November 01, 2012, 02:53:13 am »
So I made the right choice in buying an older amp that is all analog - no A->D->A conversion adding noise and distortion and the circuit diagram is easier to understand. So, no new amps that use class A/AB, but are not audiophile-grade (and cost)?
If you're using only "legacy" analog sources, then perhaps. There's the option of using a "hybrid digital" amplifier that gives the great efficiency of a digital amplifier without having to (in the traditional sense) convert the signal to digital. But if you're starting from a digital signal, going with a "pure digital" amplifier gives the best audio quality. In theory, it's possible to build an amplifier that can be switched between hybrid digital and pure digital operation (reusing the power supply and power stage), but I have never seen such a design.

Something I really like about analog is that mixing two signals can be done as easily as using a bunch of resistors. Doing it on digital would take a FPGA or ASIC and dealing with slightly different clock frequencies is a surprisingly complicated matter.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf