Author Topic: Replacing some very hot resistors  (Read 1239 times)

0 Members and 1 Guest are viewing this topic.

Online AndyC_772

  • Super Contributor
  • ***
  • Posts: 3425
  • Country: gb
  • Will design for cookies
Replacing some very hot resistors
« on: May 24, 2018, 06:39:04 pm »
Hi all,

I'm repairing a PA speaker, which has died a couple of times for various (unrelated) reasons, and before I give it back this time I'd like to pre-empt another failure which I can see is bound to happen before long.

On the power supply board there's a resistive dropper which provides a start-up supply to an op-amp. This chain of resistors connects directly to the positive side of the rectified mains at one end (about +340V), and at the other end is the supply to an LM393, which must of course be low voltage.

The combined resistance of the chain is ~30k, and since most of the mains voltage is dropped across it, the total dissipation is around 3 to 4W which is excessive. The PCB is blackened and there's no way those little SMD resistors are going to last, even though there's a fan nearby.

(It's possible that there's another fault in the PSU, but since I know where both ends of the chain are, and roughly what the voltages must be, I can't imagine what any fault might be to cause a significant increase in dissipation. Suggestions invited!)

I'd like to replace them with a suitable ceramic resistor, but I'm concerned about mechanical strength. Those little surface mount pads won't reliably support the weight of a heavier component, and if the low voltage end becomes detached, there's a wire at +340V flapping about which I'm not keen on.

Another option I briefly considered was to cover the resistors with a thermal pad and a small heat sink, but again, it needs attaching in a reliable way.

Can anyone please suggest a safe, reliable way I might retrofit (say) a 10W resistor - given that it's at high voltage and inside a unit that'll likely get dropped from time to time?
 

Online wraper

  • Supporter
  • ****
  • Posts: 9988
  • Country: lv
Re: Replacing some very hot resistors
« Reply #1 on: May 24, 2018, 08:02:08 pm »
My advise is to not touch them. Those resistors are rated for at least 1W and likely 2W heat dissipation.
« Last Edit: May 24, 2018, 08:03:59 pm by wraper »
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2681
  • Country: gb
Re: Replacing some very hot resistors
« Reply #2 on: May 24, 2018, 08:25:08 pm »
My advise is to not touch them. Those resistors are rated for at least 1W and likely 2W heat dissipation.
The resistors might be OK but the board looks as though it is getting toasty so might well be a long term reliability issue.

Given the voltages involved I would be wary about using a thermal pad/heatsink.

Any single replacement resistor needs to be rated for the voltage as well as the power.

What is on the rear of the board? Is there scope to drill completely through the board to take/support a through hole component (ideally two holes - passing the lead through, then back up to solder to the SMD pads).
« Last Edit: May 24, 2018, 08:28:39 pm by grumpydoc »
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2681
  • Country: gb
Re: Replacing some very hot resistors
« Reply #3 on: May 24, 2018, 08:26:58 pm »
PCB darkened but not that much to cause an issue IME.
Depends how long it's taken to darken that much

If years I'd agree

If months or weeks I'd say there was a problem.
 

Online wraper

  • Supporter
  • ****
  • Posts: 9988
  • Country: lv
Re: Replacing some very hot resistors
« Reply #4 on: May 24, 2018, 08:29:28 pm »
PCB darkened but not that much to cause an issue IME. I would call it somewhat concerning but not enough to make half-assed modifications. Certainly not:
Quote
no way those little SMD resistors are going to last
 

Offline duak

  • Frequent Contributor
  • **
  • Posts: 589
  • Country: ca
Re: Replacing some very hot resistors
« Reply #5 on: May 26, 2018, 01:15:34 am »
Three ideas come to mind:

- could an aluminum bracket with a power resistor be attached to the PCB using the mounting hole in the upper right corner?  One resistor in the chain is removed and the new resistor is connected to the ends of the chain using wires.  If the end resistors in the chain are left in place, the wires will be better anchored and less likely to lift the traces.

- those resistors would likely only meet their power capability with larger copper areas to dissipate the heat.  My rule of thumb is to use 1 square inch of copper per watt dissipated in still air.  Could some vertical flat copper heat sinks be soldered  to the ends of the resistors?  Each heat sink would connect to two resistors and stand over the common trace between them.

- replace each resistor with a vertically mounted axial leaded resistor that is stood off the PCB.  If some of the scorched traces look as if they will delaminate, then use a terminal strip with the mounting lug attached to the PCB with the mounting hole.

I would also use a lead free solder as its melting temperature is higher than that of a tin-lead solder thus reducing the chance of a part unsoldering itself.

Best o' luck,
 

Online AndyC_772

  • Super Contributor
  • ***
  • Posts: 3425
  • Country: gb
  • Will design for cookies
Re: Replacing some very hot resistors
« Reply #6 on: May 26, 2018, 01:40:42 am »
Some good suggestions here, thanks everyone.

As much as anything, I just find that circuits which get unnecessarily hot offend me. This circuit could have been designed more efficiently, or it could at least have been constructed using a suitably rated ceramic wire-wound resistor in the first place, but it wasn't. That's just penny pinching, IMHO, and is one of the reasons I'm not too keen on designing consumer products.

That said, clearly it has worked for some time, even though it doesn't look like it. It's survived a seized fan (the speaker's first failure) and a fried amplifier IC on another board (which is why I've had it in bits this time).

I think I like the "do nothing" suggestion best, at least for now.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 8657
  • Country: us
Re: Replacing some very hot resistors
« Reply #7 on: May 26, 2018, 02:01:49 am »
You could attach a heatsink to the top of the resistors with a suitable insulator to prevent a short, or glue a heatsink to the bottom of the PCB depending on what's there. Personally I would probably just leave it alone though. When it comes to crappy designs you'll fix one thing and then likely find something else that isn't done well.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 5970
  • Country: de
Re: Replacing some very hot resistors
« Reply #8 on: May 26, 2018, 02:15:26 am »
Higher power resistors could be tricky, as it would increase the mechanical load to the board.

There might be the option to change the comparator to a lower power one and than increase the resistor values - if the resistors are not also needed as bleeding resistors to discharge a large filter cap.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 8657
  • Country: us
Re: Replacing some very hot resistors
« Reply #9 on: May 26, 2018, 02:25:22 am »
The motor drive in my washing machine has a similar setup, a big string of power resistors dropping the mains voltage to a shunt regulator that produces the 5V logic supply. It was broken when I got it, fixed the solder joints and noticed the silly resistor arrangement that was turning the PCB brown. Well that was 10 years ago and it still works, so while the design is lousy it does the job.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 15059
  • Country: za
Re: Replacing some very hot resistors
« Reply #10 on: May 26, 2018, 03:03:07 am »
Had a timer for an industrial fan that used a similar arrangement, and the resistors cooked the power supply capacitors. Solution was to simply replace the original 3W resistor with 2 10W ones in series, and replace the capacitors with new ones. The resistors dissipate 4W between them, and the larger surface area of the new ones means they run a lot cooler. As there was room in there to fit them it worked out fine.
 

Offline Doctorandus_P

  • Frequent Contributor
  • **
  • Posts: 573
  • Country: nl
Re: Replacing some very hot resistors
« Reply #11 on: May 26, 2018, 03:41:03 am »
How about this idea:

1). Remove SMD resistors.
2). Solder a regular Through hole resitor to each SMD pad. ( Only one side, so 10 resistors total.)
3). Solder the other sides of the Through hole resistors together.
4). You now have a string of 10 resistors in series.

 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 9747
  • Country: us
  • DavidH
Re: Replacing some very hot resistors
« Reply #12 on: May 26, 2018, 04:34:12 am »
My first choice would be to use a chassis mount power resistor and attach it to the board through leads.  If necessary, an in-line connector could be included also for easier maintenance.

My second choice would be to drill some holes through the board for a through-hole power resistor.  The solid wire leads can be run through more than one hole to act as a strain relief before attaching the leads to the pads.
 

Offline bdivi

  • Regular Contributor
  • *
  • Posts: 99
  • Country: bg
Re: Replacing some very hot resistors
« Reply #13 on: May 30, 2018, 12:23:34 am »
How about using a capacitor dropper instead of resistors. The required 30k can be achieved by 100nF at 50Hz with virtually no heating.
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2681
  • Country: gb
Re: Replacing some very hot resistors
« Reply #14 on: May 30, 2018, 04:47:52 pm »
How about using a capacitor dropper instead of resistors. The required 30k can be achieved by 100nF at 50Hz with virtually no heating.
According to the description the top of the resistor chain was fed from rectified DC so a capacitor dropper isn't possible.
 

Online AndyC_772

  • Super Contributor
  • ***
  • Posts: 3425
  • Country: gb
  • Will design for cookies
Re: Replacing some very hot resistors
« Reply #15 on: May 30, 2018, 07:09:30 pm »
Correct, the resistive dropper is providing a DC supply from a rectified and smoothed high voltage rail.

In the end I replaced the faulty amplifier IC, and a couple of other parts that it had taken with it, and returned the speaker unmodified on the basis that:

- if I modify it, and a high voltage wire ends up coming loose, it's my fault if something else gets fried or someone gets a shock

- if I don't, the worst that'll happen is the speaker stops working again, which is much less bad

If I do end up having to modify the unit, I think I'll drill holes in the PCB to mount a resistor using nuts and bolts.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf