Spot the Design Error

bitrex wrote in news:TKwrF.135842$O snipped-for-privacy@fx39.iad:

General Instrument had a Set top satellite receiver that would, only after between 20 and 300 hours of operation, start exhibiting noise in the audio. The audio chip was a custom 68 pin PLCC design by GI done by Motorola.

Inspection of the chip revealed that Motorola had left out a mask layet in the chip, and it would, after some time, build up a charge and start spitting noise into the amp loop. Absolutely verified. New chip, no noise. Let it run, and it would begin at some point. Chips with the mask layer had no issue. Motorola was to pay, as nearly 350k were already integrated into units, and some had even made it to customer installations. We split the re-work and did

180,000 of them in San Diego, and Puerto Rico did the rest where they were actually built.

We figured all told, it was going to cost Motorola over $10M for the goof. Their answer was easy. They bought General Instrument uplink encoder division.

Training 7 solder assembly folks how to reflow and pull off a 68 pin PLCC without ripping up 1996 tech circuit traces on a commercial board not meant for rework or rough service was not easy.

They just cannot get that you cannot refow some and pull up and then hope the rest melt too.

It was all done with MetCal heads sized for the chip. Still.

We did not have hot air rework stations.

Took 'em a while to find that problem's source.

I don't understand the input ground connection. How does input shield-wire "E" get to ground?

--
 Thanks, 
    - Win

--------------------

** OK - especially for Win, cos he is *such* a nice guy.

1. From point E on the PCB ( mounted on a large heatsink ) it travels via a co-axial cable to an Alps, 21 step attenuator on the front panel.

2. From the same attenuator it also travels to the rear panel and meets up with a two pairs of metal XLRs ( male and female) plus two 1/4 inch jack sockets.

The *jack sockets* are firmly grounded to the aluminium metalwork.

The centre point of the two of 22,000uF, 100V filter caps also goes to chassis.

FYI:

Though the input is unbalanced, it is supply hum free and RF proof.

Note the use of two AC fuses, main and spare ;-)

Despite having an 800VA toroidal transformer and NO soft start system, it has a mild inrush surge.

Now, what is wrong with the two thermal reed switches ?

Surely a man of your experience can figure out that at easy one.

.... Phil

D112/D113 seem upside down. This way they decrease the current limit when there is less voltage over the output transistors, the reverse of a SOA protection.

There is something very strange about the EUROPE (220V/240V) and AUSTRALIA (240V) transformer connectionss in the bottom of the diagram. Looks like a drawing error. Must have been been noticed on testing...

Also, a connection is missing from anode D308 to D306/C301.

More to be found, probably...

Arie de Muijnck

Correction --> they _increase_ the current limit when the voltage is lower. Should be OK, unless wrongly dimensioned.

This feels like an episode of Batman where the riddler has them in an imminent death situation and starts asking Batman riddles to get out of their predicament.

snipped-for-privacy@columbus.rr.com wrote in news: snipped-for-privacy@googlegroups.com:

Then...

That was a Holy Dumb Post, Robin.

You mean where either of the THERMAL READ Switches do nothing but turn on LED 'PR'? They don't actually appear to cut any power to the amp if it warms up a bit much...

The power switch on the General Model version appears to be missing the point of having a second 220 winding in the primary transformer. Except for the Australian version where it appears to be wired correctly.

John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup) 
                      John's Jukes Ltd. 
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3 
          (604)872-5757 (Pinballs, Jukes, Video Games) 
                      www.flippers.com 
        "Old pinballers never die, they just flip out."

Hmm, I can see two things which look like errors:

- capacitor C123 seem way too small, bigger (say 1u) would give better transient response

- filetring of voltage for VAS and input stage is with respect to negative rail, so it couples variation from negative rail. Correct way is to filter with respect to ground (would cost few cents for higher voltage capacitors)

The second rises THD, but it probably still is within specs, so I would not call it "misbehave terribly".

Schematics is not clear about feedback takeof point. As drawn it suggest that it may by suboptimal, leading to increased THD.

--
                              Waldek Hebisch

Looks to me that if one side swings close to a rail, and current limits, it increases the current limit on the opposite side, namely the one with a lot of voltage drop across it.

Luckily, music is AC coupled so doesn't have really long duration swings.

That is really an ugly variation on the repeated-a-million times standard audio power amp.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I am suspicious of VR101 - if the wiper lifts off the track then TR115 and TR110 will stop conducting and each half of the output stage will rail?

piglet

---------------------

------------------------------

-------------------------------------------

** OK - the first cupie doll goes to John Robertson !!

The two thermal reed switches, bolted to L & R channel heatsinks and wired to a PCB then a tiny red LED on the front panel, labelled "Thermal" - do ab solutely nothing !!

s/r.18074.html

The Japanese designer did just as required, fitted an over-temp indicator a nd left it at that.

In practice, the indicator only comes on when output devices are sizzling h ot- about 120C on their cases. If nothing is done in the next few minutes t he channel fails with massive damage requiring a complete re-build. Sometim es a serious PCB fire breaks out too.

I reckon you are supposed to have a little man, sitting watching that red l ed all the time and able to take immediate action if it ever lights !!!

The amp PCBs use about a dozen fusible resistors, mostly small WW types in crucial spots with mA ratings printed on them, but still they fail to stop the fibreglass board catching alight.

The nickname I invented for the model is the " 22 oh boom " !

----------------------------------------------------------

** That is a fascinating issue, I have never been able to get to the bottom of it. Yamaha's schem shows versions with the power tranny's primary dead shorted and power fed to the centre tap. Cannot possibly work as drawn.

My best guess is there are really two toroidal transformers in the aluminiu m can that hides them, primaries in series wired out of phase to each other . Doing this reduces the magnitude of the inrush surge so no soft start is required.

FYI:

The general model is factory wired to suit the market while the others have a slide switch inside to select 220/240.

..... Phil

---------------

** The misbehaviour happens in normal use, it has no connection with any faults developing.

Affected uses had no choice but to get a different amplifier.

..... Phil

----------------------------------

.... LOL

Thanks Phil. Ok how about C112 being a bit small to bypass a noisy zener D105?

piglet

What's right with it? The two LED circuit is incomplete and meaningless, in the drawing, worse than the hidden input ground connection. Are you saying that SL119B, or whatever the p/n is, is two reed-relays? If so, where's the rest of it? If not, what's going on, nothing?

--
 Thanks, 
    - Win

Phil Allison wrote in news:7b392552-82df-40ae- snipped-for-privacy@googlegroups.com:

You have succeeded in building a better bat trap.

Another thing I am not sure about is the way the aux -ve rail is ac bypassed not to ground but instead to the -80v rail bobbing up and down with signal - could that invite motorboating. Like make C116 a higher voltage part with anode returned to ground not -80V.

piglet

-------------------------------------------

** That is what VI limiters in a power amp do, vary the available current as the output voltage swings about in reverse proportion to the Vce appearing on upper and lower side BJTs at any moment.

The aim is to not allow excursions beyond the maker's SOA curve and hence device failure.

** So says a music hater.

** Another unsupported, blown out his fat arse opinion John gets to keep.

Yawnnnnnn ....

...... Phil

** Says it all, really.