yamaha m-65 diagnostics and thermal, a tale of two yam's

[sae-7125]

the first amp seemed somewhat straightforward and was a 120v model, no broken output terminals. amp number two originally had a smoked channel although no chernobyl nor 3 mile island event transpired (yet, night is still young) perhaps a windscale.

interesting info about the above metaphor (word origin):

Chernobyl - Academic Kids

academickids.com

Windscale fire - Wikipedia

en.wikipedia.org

the bulk of this diagnostic evaluation will be in regards to amp #2, international model with voltage selector, since amp #1 had minimal issues and went back to its owner after some basic updates. obviously two different part number mains transformers between the two amps.

first step involved the usual glue removal and associated splatter on nearby components/jumpers. updated c157/158 and r141/142 and then diode checked the heat sinked components. tr135/137 (2sa1106) had base to collector shorted and r149/151 (2.2r) measured in the meg even though physically/cosmetically looked oem/new.

slowly replacing various diodes and zeners helped stabilize the channel to the point of not stuck in protection mode. the amp has been chained to dim bulb and variac (100v mains) until the final gremlin has been dealt with (unstable channel bias). dc offset for each channel is 15mv or less (right +13, left -15). when left channel bias is set to minimum the emitter resistors measure 10mv front and 13mv rear across each pair as shown in the service manual, should be zero mv. activating the "class a" cooker mode shows a roller coaster response going down a few mv then up several mv and back down a few mv. the right channel responds normal, active = up and stable new bias set point (circuit controlled/regulated).

if the left bias is too high when powered on then protection mode stays active even though dc offset is stable/low. adjusting the new bias pot from minimum to almost middle position helps to reduce the bias enough for protection mode to turn off. the amp plays fine in terms of audio output albeit limited by dim bulb ampacity.

a few more original low level transistors to replace and some diodes, after that it will be LCR interrogation of fusible resistors if needed. black gate capacitors removed/updated with fresh components especially since some measure low and have esr creeping up.

the hunt for stability and optimal functionality continues...



.


reference numbers from the first amp (120v model). still on dim bulb (limited to 75w using two 40w bulbs in parallel). dc bias: ~4mv (dim bulb) = ~10mv (no bulb. full mains). i should have captured full mains numbers, would have been more useful to others...


first amp


amp #2, international voltage selector, faceless and playing music on dim bulb (variac 100v mains)


.


output devices and new pot (left channel)


left channel new components, photo technically upside down.


thermal scan (amp 2), potential circuit imbalance tr119 (schematic, 2sa1306) although pin voltages match schematic and right channel tr120.


top side of pcb.

 

[sae-7125]

after rebuilding most of the damaged channel there seems to be an odd dc bias quirk or sensitivity. with the channel pot turned down to minimum measuring across the left channel front pair of emitter resistors there is 5.5mv. when slowly increasing resistance the voltage goes down to around 0.1-0.3mv for an extended period of resistance adjustment and then jumps to 15mv and starts to climb which also increases mains wattage consumption, as one might expect. the low mv flat spot obviously highlights an ongoing issue not to mention the voltage at pot minimum.

activating class-a mode and watching the bias rise reveals a few more issues as well. is there a specific "bias set point" the class-a circuit is calibrated to achieve and what circuit components control the "set point", a resistor value that limits the rise to +10mv or +20mv above class-a/b operation?

there was a rather unique circuit sensitivity prior to replacing tr139/141/143 which involves touching a dmm test probe (meter connection optional) which would cause the bias to jump up and exercise the dim bulb. when touching the base of tr115 bulb gets bright (bias jumped to 20's), when touching collector of tr115 at random intervals bulb goes dim/normal idle wattage. after updating various diodes/zeners, fusible resistors, low level transistors and all electrolytics there isnt much left other than requisitioning an old priest and a young priest plus a hammer and stake plan "b".

when setting the left/repaired bias to 1.0mv (dim bulb) which equates to 2.1mv full mains (no bulb) and... setting right channel to 0.7mv (bulb) plus watching the class-a swing for both channels to level off at 13.7mv left and 14.2mv right (relatively balanced with dim bulb) the amp performs without issue. no distortion nor protection circuit surprises, wattage consumption isnt unusually high.


music exercise has no issues, amp is pulling mains wattage peaks of around 118w, things seem functional and healthy i.e. firing on all cylinders.


thermals look fairly balanced as well, one possible exception would be tr159 and some nearby components (hot spots directly below centered green square) in addition to tr119 (schematic pnp).


when at low volume and still sufficiently warm the oddball area is quite obvious, higher current/load in one part of left channel.


left channel bias still has issues and the thermal scan shows a load anomaly (tr119/125 electrically isolated from heatsink). the hunt for root (cause) october continues...

a few extra notes when the amp is connected to full mains (no bulb). at idle when activating class-a mode the left channel screams up to mid 60's mv and might be tempted to flirt with disaster beyond that (no, replacing blown outputs due to runaway dc bias is not fun). the right channel casually climbs to mid 40's mv which doesnt seem as severe especially when comparing the rate at which each channel increases bias.

with dim bulb left channel, carefully adjusted to 5.0mv, dc bias shows a behavior of decreasing to 0.2mv then rapid increase to a peak of 25.8mv and then settling down to 13.6mv. the voltage swings seem less capacitor/smooth ramp and more like chaotic jumping intervals. the right channel has a slow smooth linear ramp up and minimal overshoot when the new bias limit is reached. the odd couple indeed..

*note to self: thermal camera emissivity setting adjusted up to 0.80 due to potential thermal exaggeration of surfaces compared with 0.50 previous measurements. a "one setting fits all" solution for electronics might be ongoing.

 

[avionic]

is there a specific "bias set point" the class-a circuit is calibrated to

No

 

[sae-7125]

another odd circuit reaction quirk is when touching the top of c117 (human finger, not grounded) the idle 5.5mv at pot minimum decreases to 0.2mv, the electrolytic metal housing measures something like -1.2vdc and the charge dissipates slowly when probed.

using a can of freeze spray on d121 causes the stray dc bias to increase vs spraying tr115 which causes it to decrease. both components were updated earlier and continue to measure good.

i will have to see if amp #1 is still available for some measurement harvesting and post the relevant data, in particular would be class-a mode bias mv boost and full mains idle consumption. obviously monitor rate of bias upward swing and some extended performance testing to use as a baseline for comparison.

since amp #2 has shattered speaker-a binding posts i wonder if that is a telltale sign of how much class-a mode was exercised with the amp until one channel smoked. the obvious correlation being eventual runaway bias due to thermal stress. wouldnt be surprising since the mains txf heats up noticeably and holds it due to overall density (big blob of various metals) which happens to be parked next to the left channel heat sink. in terms of output terminal plastic damage the current/previous owner(s) used the "a" output most seems likely. the possibility of physical trauma to the predominantly plastic binding posts might have been a contributing factor as well.

 

[avionic]

the possibility of physical trauma to the predominantly plastic binding posts might have been a contributing factor as well.

100% possibilty of shipping damage... Class A had nothing to do with it.

 

[sae-7125]

amp #1 not available for measurement data collecting, so much for that idea.

it is odd that when the dc bias is set suitably low overall performance testing seemed quite capable and healthy, as shown above. afterward i found a 6v zener installed at d133 which should normally be 12v (hz12b2). when corrected now the bias is far worse and unable to set a low stable point.

after probing around with an LCR meter at 100khz resistance mode there are some distinct imbalances vs the right channel, which is being treated as "known good". measuring from non-wiper side of vr103 to ground also tr135/137 emitter to ground shows lower resistances vs right channel which would explain why tr119 (pnp) is noticeably hotter than all other heat sink mounted transistors. once again indicating there are circuit issues, as if that wasnt already obvious.

next step is interrogating high value resistors and film caps. the hunt continues.

i wonder why d139/140 (hz24-2) were positioned directly under the power filter caps (other than basic pcb layout/design)? zeners need to breath too, not to mention eventual replacement.

 

[sae-7125]

it would appear that yam's might have a stability issue depending on if an exact replacement is used (original vs alternative), in regards to replacing certain transistors.

bumped into this a few moments ago:



sourced from:

Power Amp Development

ESP - Power amplifier development over the years.

sound-au.com

although ax-490 is not m-65 it does shed some light on a few things.

 

[avionic]

 

[sae-7125]

a few more pieces to this rather odd puzzle:


in regards to the m-70 design, licensed tech from carver.

Yamaha M-70 Power Amplifier (Jan. 1984)

Yamaha M65 Bias adjustment

Adjusting bias on a Yamaha M65. Manual states adjust to 10mv which is no problem then adds that when press "class a" button bias will be over 20mv. When class a is pressed it's 80mv. I'm thinking this is too high. The unit gets very hot in only 5 minutes. It will play fine like this over an hour...

www.diyaudio.com

Audiogon Discussion Forum

I just read an advertisment on Ebay for a Cap Kit for my Yamaha M70 Amp It Warned that the Mylar Capacitor #607 on board C MUST be replaced because i...

forum.audiogon.com

*although for the m-70, still interesting info when slogging through a persnickety quagmire of an m-65.

in regards to future diagnostics "use the force luke" seems appropriate, although venturing closer to dmz charlie foxtrot is highly probably.

for the circuit resistance issue measured at tr135/137 i may end up ordering some nos sanken components (2sa1106b), unless fine tuning the circuit for the newer pnp's has a beneficial effect i.e. re-engineering that part of the circuit.

more component interrogations and speculation inbound..

 

[sae-7125]

after a few component adjustments and achieving stability, albeit at a lower bias setting, amp passes performance tests and class-a bias measured 62mv left and 67mv right (cooker mode). an interesting side note/observation is that the two output pnp's that shorted are directly across from the txf, think of it as a thermal sandwich which is to say nothing of the e-field radiating from the power magnetic component into the heat sink (might be no concern). a test sample of one instance is clearly no proof nor trend in terms of accelerated failure mode but it does paint an interesting picture.


happy little amplifier, although not shown in class-a mode. exercised up to 155w on the mains with no issues (watt meter has slow update rate).


final idle numbers, display active and speakers connected/active.


repairs would be easier if things like this didnt occur, perhaps its time to buy a lotto ticket..


fyi/hint: ksa1837 is a pnp part number, or silkscreen in this case vs contents that are npn. two npn's didnt make a pnp.


outside thermal in a/b mode. nice heat soaked txf. the temp shown (min/center/max) is lower vs actual since the camera is set to 6 inch distance and the full screen shot was around 18 inches away.


next up is an onkyo grand integra m-508, such a beauty

terrible photo, too much background bench lighting.