Intermittent fast failure or slow over heating?

[Mick Duprez]

Hi All,
Overthinking newbie here

I'm repairing a Yamaha DXR10 powered speaker for the second time with the same problem, it blew the thermal fuse and over cooked a 15k 2W resistor, I replaced the mosfet that feeds them (just in case) the first time but it's checks out ok this time. It has been running fine for a few months with regular gig work before failing again.
I powered it up using a current limiter (light bulb), it looked normal starting bright then settling but no lights on the dash to indicate it was running.

The 15k resistor turned into a 17.8k the first time, this one reads 12.3k and has visibly overheated again.
I've been over everything I can think of, pulled the main cap's and tested, done some in circuit testing of other diodes, fet's and transistors etc of and can't find anything obvious. I did however find the pot's on the DSP very gummy/dirty (a can of run and cola I'd be guessing ) and give them a good clean.
That's the only thing I can think of/find that may cause an intermittent short causing the large current flow causing the fuse to blow, could this be the case? If not I'm concerned it's a something making it draw just a little too much current causing a slow failure(?).

I'm about to replace these items again so I'm wondering what to look for/check when/if it starts running (I wish I had an infrared camera...).

I've attached the schematic, the items in question are F103, R116 and Q102 located at Grid O from grids 6 to 8.

thanks in advance,
Mick

[Tomorokoshi]

Let's start with some calculations to get a baseline:

Unloaded voltage at bridge rectifier D104:
1.4 * 120V = 168V

Fault current through R116 (assuming no loading down):
168V / 15000ohms = 11.2mA

Fault power dissipation in R116:
168V^2 / 15000ohms = 11.2mA^2 * 15000ohms = 1.88W

This is a worst case that would only happen in a fault condition.

Now trace the output of Q102. We find several connections:
1. It is connected to the output of the 7812 12V regulator through D110.
2. Zener diode clamping using D107 and D108 in series. It seems they are 6.2V, for a nominal voltage of 12.4V.

This hints at the next information to get:
1. The nominal operating voltage at the source of Q102.
2. If D110 is shorted or not.
3. The voltage across R116 in normal operation.

Back to the resistor, the part could very well be used beyond its ratings in elevated temperatures with continuous operation. The resistor is clearly being stressed. Is the resistor thermally connected to the thermal fuse? The thermal fuse seems to be rated to 115C. It might be as "simple" as replacing the 2W part with a 5W part. Details of the installation would help.

[Mick Duprez]

Thanks for your detailed info Tomorokoshi!

<snip>

Back to the resistor, the part could very well be used beyond its ratings in elevated temperatures with continuous operation. The resistor is clearly being stressed. Is the resistor thermally connected to the thermal fuse? The thermal fuse seems to be rated to 115C. It might be as "simple" as replacing the 2W part with a 5W part. Details of the installation would help.

By thermally connected do you mean via a heatsink or similar then no, they are apart from each other enough I'd think (~15mm).
The owner uses it mainly for a foldback or front of house on a stand etc, pretty well ventilated anyways.

I'll finish putting it back together and take some measurements, thanks again, cheers.

[Mick Duprez]

Let's start with some calculations to get a baseline:

Unloaded voltage at bridge rectifier D104:
1.4 * 120V = 168V

Missed this before but I'm on 240V:
at DC out at bridge rectifier I'm reading 328V (1.4 * 240 = 336, close enough)

Fault current through R116 (assuming no loading down):
168V / 15000ohms = 11.2mA

Fault power dissipation in R116:
168V^2 / 15000ohms = 11.2mA^2 * 15000ohms = 1.88W

This is a worst case that would only happen in a fault condition.

updated calculation:
328V / 15000 = 24.5mA
P = I*V = 24.5mA * 368V = 9.02 watts
does that sound right??

Now trace the output of Q102. We find several connections:
1. It is connected to the output of the 7812 12V regulator through D110.
2. Zener diode clamping using D107 and D108 in series. It seems they are 6.2V, for a nominal voltage of 12.4V.

This hints at the next information to get:
1. The nominal operating voltage at the source of Q102.
2. If D110 is shorted or not.
3. The voltage across R116 in normal operation.

1. Q102 voltage is 11.89, coming out of D110 is 11.6, going in is 15.58V
2. D110 is not short
3. 328V at R116 with no speaker load, resistor not even getting warm after 10 minutes, will hook it up with some tunes and speakers and run it for a while.

any thoughts on the updated data?
thanks,
Mick

[Mick Duprez]

Ok, so something interesting perhaps...

I was holding my hand over the pcb to see if I could feel any abnormal heat, I noticed a heat sink on an output circuit and I put the back of my finger to it to see if it was hot and got a good zap, like static but a bit harder. I noticed printed on the board "Heatsink (KG: +27.5V) and thought ok and measured with the DMM to chassis and it was 27.5V but thought it strange that it was 'live'.
I then hovered over some cap's in the output section (63V for the 55V rails at grids E:6-7) which seemed like they were warm, again I touched the 'top' of the cap's (where the 'x' relief is, not the terminals ) with the back of my finger and got a good zap again.
I took a reading on the 'tops' to chassis and got maybe a volt or +0.9 on one and -55V on the other.

Is this normal or do I have a bad earth/short somewhere I need to find?
If it is I'll be taking even more care than usual when repositioning live test circuits

Thanks,
Mick

[Tomorokoshi]

Ok, so something interesting perhaps...

I was holding my hand over the pcb to see if I could feel any abnormal heat, I noticed a heat sink on an output circuit and I put the back of my finger to it to see if it was hot and got a good zap, like static but a bit harder. I noticed printed on the board "Heatsink (KG: +27.5V) and thought ok and measured with the DMM to chassis and it was 27.5V but thought it strange that it was 'live'.
I then hovered over some cap's in the output section (63V for the 55V rails at grids E:6-7) which seemed like they were warm, again I touched the 'top' of the cap's (where the 'x' relief is, not the terminals ) with the back of my finger and got a good zap again.
I took a reading on the 'tops' to chassis and got maybe a volt or +0.9 on one and -55V on the other.

Is this normal or do I have a bad earth/short somewhere I need to find?
If it is I'll be taking even more care than usual when repositioning live test circuits

Thanks,
Mick

Be very careful with power supplies like that. Particularly because it's on 240V, which I didn't realize until the other post. I will try to reply to that message separately.

Anyway, C137 and C138 are found in the schematic (which is a bit hard to read) in E6 and E7. Note that they share a common ground, with C137 going to +55V and C138 going to -55V. Those are polarized aluminum can capacitors. The negative terminal is connected directly to the metal case. That is why you felt the -55V, which is plenty to make you notice, particularly if it's a bit humid. Measure everything and don't touch anything. That kind of supply can have heatsinks at high voltages, etc.

[Mick Duprez]

...

 Measure everything and don't touch anything. That kind of supply can have heatsinks at high voltages, etc.

Roger that 

I just uploaded a better copy of the pdf, I tried to upload the whole doc but it's too large.
btw, are the 'TL###' points noted on schematic 'test locations'? There is no reference or mention of them anywhere in the docs but there are little through hole like spots on the pcb coming off the tracks at random places.

cheers.

[Tomorokoshi]

...

 Measure everything and don't touch anything. That kind of supply can have heatsinks at high voltages, etc.

Roger that 

I just uploaded a better copy of the pdf, I tried to upload the whole doc but it's too large.
btw, are the 'TL###' points noted on schematic 'test locations'? There is no reference or mention of them anywhere in the docs but there are little through hole like spots on the pcb coming off the tracks at random places.

cheers.

Much better!

Those numbers are about right.

Now we can take an overall view. Instead of trying to diagnose a failure, we can review the circuit.

It seems the power rating of the resistor may have been based on operation at 100V or maybe 120V. This kind of oversight is very easy to get into a design.

Because of your comment about it being unloaded, then in loaded operation if the current gets to be more than about 12mA the rating will definitely be exceeded. It's unclear what the loading is that would result in that, because it seems to be running the "PWM CONTROLLER" which drives the "Pulse Transformer". While much of this power supply is a switching power supply, the bridge rectifier to the 15kOhm resistor is a linear supply, which is difficult to efficiently tune over a wide voltage input range.

It is certainly possible that thorough testing for 240V was not done.

There are a few options:
1. Replace the 15K 2W part with a 30K, either 5W or 10W part. This would tune the circuit for 240V operation. Power dissipation in R116 would be about twice that in the 120V case, but about half of what you are getting now, for about 4W. Not sure how to fit it in.
2. Run off a 240V to 120V transformer.

If possible get a picture with a clear view of R116.

[Mick Duprez]

It seems the power rating of the resistor may have been based on operation at 100V or maybe 120V. This kind of oversight is very easy to get into a design.

Yeah, I was surprised to see the info on the back read 100-240V, most of these active speakers I've seen have a 110/240 switch to deal with this rather than a 'one size fits all'.

Because of your comment about it being unloaded, then in loaded operation if the current gets to be more than about 12mA the rating will definitely be exceeded. It's unclear what the loading is that would result in that, because it seems to be running the "PWM CONTROLLER" which drives the "Pulse Transformer". While much of this power supply is a switching power supply, the bridge rectifier to the 15kOhm resistor is a linear supply, which is difficult to efficiently tune over a wide voltage input range.

It is certainly possible that thorough testing for 240V was not done.

There are a few options:
1. Replace the 15K 2W part with a 30K, either 5W or 10W part. This would tune the circuit for 240V operation. Power dissipation in R116 would be about twice that in the 120V case, but about half of what you are getting now, for about 4W. Not sure how to fit it in.
2. Run off a 240V to 120V transformer.

I've ordered a 15k 5W backup but will order a 30K. He bought these units as a pair for a PA, the other one works fine without trouble (so far).

If possible get a picture with a clear view of R116.

See attached.
I left it high off the pcb and bent it away from the fuse to give it more room, same for the fuse (I can't get it in shot sorry). I ran some music through it for a while, I didn't run it hard just idling really but no noticeable changes in voltages or heating of the R116 resistor.

Thanks again!

[Mick Duprez]

Just for completeness if anyone happens upon this error:

Ok, I got the another speaker (of the pair) in recently so I inspected the same components and F103 (thermal fuse) was blown so I replaced both this fuse and the 2W resistor but no lights/fan etc and it was getting hot quick. It turns out that the cap from VDD to ground was low capacity and esr so I replaced that and still no go (but I bet it is why the first unit failed and I need to re-repair it now and make good on the old circuit). I figured it was holding onto the startup circuit by affecting timing and sensor voltages, I think this was the case but...
It turns out that Q105 was being clamped on as there was a leak or something around the 2 digital transistors that feed its base. I reflowed the Q103 and Q104 around these components and it now fires up fine. I'm guessing the extra current/heat and general corrosion (we live near the ocean) was the culprit but at least it's running fine now and the startup network is being released.

I still think the bad cap was the cause of all this and the need for reflow was collateral damage, we'll see after some soak tests I guess.

cheers.