0
\$\begingroup\$

Some years ago I connected a relay inline with the DC input of some old computer speakers so they could be powered on and off by an Arduino. This was working perfectly for a few years, but then one day the audio started getting really distorted and garbled. Things would often be fine when first turning them on, but after a few seconds to a few minutes, the audio would degrade severely. It also seemed to be worse at higher volumes. I initially thought the speakers had blown, but after some tinkering I found the relay to be the culprit. After swapping it out with a new relay things worked great again for a few years, but now the issue is back.

The relay is one of those common blue Songle relays (SRD-05VDC-SL-C) that, if I'm reading the specs correctly, should handle a 30V DC load. The speakers are powered by an 18V wall wart, so I think everything should be within tolerances. The audio source is an entirely separate device with its own power supply, so I don't think that is a factor.

Any thoughts as to what might be going wrong with the relays and how I might prevent this from happening again?

\$\endgroup\$
2
  • 3
    \$\begingroup\$ contact corrosion ... polish contacts with a piece of white laser printer paper ... white paper contains clay and is abrasive (that's why it dulls scissors) \$\endgroup\$
    – jsotola
    Commented Nov 12, 2022 at 23:52
  • 3
    \$\begingroup\$ You can verify jstola’s theory (before polishing!) by measuring the voltage across the relay contacts when the audio is distorted. \$\endgroup\$
    – Bryan
    Commented Nov 13, 2022 at 0:36

3 Answers 3

Reset to default
2
\$\begingroup\$

Some contacts are designed for a minimum current (sometimes termed "wiping current"), as well as maximum. The tiny bit of heating and arcing at the microscopic point of contact, though just a few mA, is enough to clean off oxidation or dirt from contacts.

If the current draw is very low, use a self-wiping relay, or a sealed device such as reed relay, which is one of the most durable mechanical devices. You can also clean the existing relay, and mount it in a sealed enclosure, perhaps adding a silica gel capsule from a bottle of medicine (or even an oxygen-absorbing packet from food) to the enclosure.

\$\endgroup\$
2
\$\begingroup\$

Sounds like the contacts are getting corroded. The speaker power supply is probably a nasty capacitive load that contributes by mucking up the contacts when they close, so the obvious solution of replacing the relay with one with gold-plated contacts or a reed relay (which may blow off the gold or weld on the first operation) may not work.

One possibility might be to use a higher power relay which really slams the contacts together such as a T90 type. Naturally they draw more current, so the drive circuit or power supply may or may not be up to the task.

enter image description here

Alternatively, and this would almost surely work, replace the relay with a MOSFET DC SSR.

\$\endgroup\$
6
  • \$\begingroup\$ There is usually a fairly loud pop from the speakers when the power cycles. Would this be due to the capacitive load you speak of? I had previously thought about placing a capacitor between the output of the relay and ground thinking that it might ease the voltage spike. Any thought if that would do anything useful? I do like the idea of an SSR just to get rid of the click, but all the ones I've seen seem way pricier than I would have expected for what I need. Maybe I'll just have to bite the bullet and get one. \$\endgroup\$
    – Pak
    Commented Nov 13, 2022 at 5:56
  • 1
    \$\begingroup\$ The amplifier probably isn't designed to be switched that way (or if it is, it isn't well designed to suppress the pop). I don't think a capacitor will help. Amplifiers sometimes wait until the output stabilizes before closing a relay to connect the speakers. I don't know your current/voltage requirements, but a part like this one is less than $1.50 USD one-off and can handle 60VDC and some hundreds of mA.. Don't try to use an AC-output SSR, it will close but never open. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Nov 13, 2022 at 6:05
  • \$\begingroup\$ Or maybe the CPC1706Y 4A at up to 60VDC and less than $4. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Nov 13, 2022 at 6:07
  • \$\begingroup\$ Perhaps I need to rethink my overall strategy. There is a push button switch on the front of the speakers for controlling the power that doesn't exhibit the same popping. I always assumed the switch was directly connected to the DC input, but maybe there is another layer between. I'm almost certain it's not as sophisticated as having another relay to wait for the output to stabilize (there is no delayed relay click like you would find in a home theater receiver). \$\endgroup\$
    – Pak
    Commented Nov 13, 2022 at 16:43
  • 1
    \$\begingroup\$ @Pak Sure, you might be able to use a MOSFET. The details would depend on the exact connections and whether isolation is required. It might be easy or not so easy. But if you want to go this way, best ask another question with lots more details. Typically a high-side switch (P-channel) is more appropriate. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Nov 13, 2022 at 16:56
0
\$\begingroup\$

Most likely the power supply is of poor quality and does not have current limiters at the time of connection to the network (inrush current limiter). The fact is that there is a capacitor at the input of the pulse power supply. When connected, this capacitor acts as a short circuit. It turns out a very large current and at this moment the relay contacts burn out. Sometimes this current exceeds the operating current by 30 times. The simplest thing is to put in series a resistor or an NTC (for example, an ametherm or any other manufacture).

Your voltage is 18V. Find the maximum relay current I in the documentation. Install a resistor or NTC with a resistance of at least 18/I. I advise you to read https://en.wikipedia.org/wiki/Inrush_current

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.