Fixing a vintage Yamaha amplifier

With my wife, we built an old-style stereo hi-fi system, with vintage components, including a turntable, cassette deck, tuner, CD player and of course amplifier. Our amplifier, a Yamaha CA-410 II, gave us a bit of trouble but we managed to fix it. Here I tell the story of how we diagnosed the issue and fixed it.

Note: I didn't use any AI to write this article.

The Hi-Fi System

I have always loved audio cassettes, and I have always been sad that the technology got replaced by CDs and subscription-based services. We wanted to reclaim the past when things were physical, and you owned your tapes and records which had a physical existence. I've always loved touching the tapes and having a physical medium associated with things.

We both were super hyped by this idea and started building our hi-fi system by buying vintage components on eBay.

I especially love my Realistic SCT-90 cassette deck which sits to next to my computer and which I used to record music from my computer to the tapes. We also got an SCT-100 for the hi-fi system but I'll tell the story in another article.

My favorite vintage component we bought: The SCT-90 cassette deck

The Yamaha CA-410 II Amplifier

But let's talk about the amplifier. It's a beautiful Yamaha CA-410 II we bought on eBay.

Our Yamaha CA-410 II, sitting on top of our AIWA AT-9300 tuner, after bulb was changed

It was mostly working correctly except for one meter light which is burnt, and, more annoyingly, sound randomly drops when playing music at a rather low volume.

This gives a feeling of very bad quality when it happens. If we increase the volume, it goes away even after we put the volume back. Then, if the volume is rather low, it comes back after a while, almost like if it happens when "it's cold".

The easy fix: The meter bulb

First, fixing the light bulb was the rather easy part. I wanted to replace it by an incandescent light bulb with similar characteristics. Although it's in theory possible to replace it by an LED by adding an appropriate resistor in series, I wanted to be as true to the original as possible, and keep everything the same.

With the help of ChatGPT, I was able to figure out the correct replacement bulb on DigiKey, which is this one (VCC 7646 / DigiKey PN: CM7646-ND).

I then just needed to open the amplifier, cut the wires leading to the bulb, and solder the new one instead. It all worked perfectly, and looks exactly the same as the meter with the original bulb.

Left is with new bulb, right with original bulb

The hard part: Fixing the sound issue

This issue is nasty, because it has the quality that every engineer hates: it's non-deterministic. It took me a long time to understand if it was just doing it when I was playing tapes (I suspected the SCT-100 cassette deck which had other issues already) or with any sound input. The reproducibility of the issue seems to depend on the type of music sadly, and if the amplifier is "hot" etc.

But I finally found a way to reproduce it from a source that we know was good: a TWRP song played from the CD player. I played track 9 ("International Business Systems") at least a hundred times while doing this diagnosis because I had finally found a way to reproduce the problem!

Then I had the idea to use my oscilloscope to measure the input at the RCA connectors at the back vs the output on the wires that go to the speakers.

Input in green (bottom) and output in yellow (top)

I used ChatGPT to suggest ideas and it suggested a possible relay issue initially, however, based on images like the one above, it led me on a completely different track for about an hour.

As you can see above, the output signal seems to have so noise added to it in the middle, and if we play a constant tune, it's even more obvious (see picture below). The way I did that is I generated a 1 kHz sine in Audacity and recorded it on a tape which I then played it while connected to the amplifier.

ChatGPT suggested a smart thing, which is to use dedicated measure pins that are on the board with the purpose of helping people doing a diagnosis with an oscilloscope like me! Right to repair was on a completely different level at the time. I also have the full electrical schema provided by Yamaha.

I did a lot of measures that ChatGPT suggested, and in the end it suspected an issue in the parts that produce the constant DC voltage from AC, suspecting the voltage was not constant in a clean way. It made sense to suggest that, especially because the interference in question comes 120 times per second, suspiciously 2x the mains frequency here in Canada (60 Hz).

However, when I started to measure different ground points, it was clear that the interference was here even when I measure the ground from the device chassis itself. I therefore realized that this was just a measurement artifact, the oscilloscope was just picking up some of the frequency noise from the mains. This was not showing the real problem.

Back on track

I decided to take back control of the diagnosis strategy from ChatGPT. I needed to find a way to visualize the real issue. By adjusting the settings and doing single captures on the oscilloscope, I was able to see it!

The top signal (yellow) is the input (measures at TP1 to be precise), and you can see the signal is dropped on the left but not on the right. Here I captured this by detecting when the signal comes back in the output (you can actually see the trigger settings on the right), which is why we can see a drop before the t=0 line but the correct signal after. Here we capture the phenomenon right when it happens!

ChatGPT had the merit of suggesting to use those pins to measure if the issue was "already there" at this point. It was not, the signal was still good (it's the yellow above actually), which means the problem was downstream this point, which points to the relay.

Eureka!

I wanted to measure directly the signal pre and post-relay to get the final diagnosis, but I had the idea to just tap on the relay. The issue immediately disappeared temporarily, just like when I increased the volume. That made me laugh "OK that's definitely the relay then!"

At this point the fix was pretty straightforward: order a compatible relay from DigiKey (this one - Omron LY2-0-DC24 / DigiKey P/N: Z977-ND), unsolder the old relay, and solder the new one instead.

The faulty relay we replaced

After soldering in the new relay, the amplifier worked perfectly! And we have been enjoying music of amazing quality with our Yamaha amplifier ever since.

Conclusion

I initially felt helpless when starting to work on this issue, I tried to connect speakers to output B instead of A, deox-it every possible contact on the speakers and amplifier... but it was opening the device, measuring things that makes sense while reproducing the issue that finally helped. I know there are some very talented people like CuriousMarc (YouTube channel) who does amazing repair and would not need AI to get ideas on what to measure.

But such a project was a very first for me. Buying a vintage device, and actually fixing it to make it fully work is something I had never achieved before. Also, having an oscilloscope and knowing how to use it was a decisive factor in reaching a correct diagnosis. I initially bought this oscilloscope when I lived in France (good thing it's compatible for both 50/60 Hz and 115/230V) where I used it for my Arduino projects (debugging serial communication, capacitor charge, buffer level...) and you can see all of that on my Github and YouTube channel, but it's the first time I used it on an analog device.

I learned a lot during this project and it gave me the courage to then tackle the repair of my SCT-100 cassette deck, which involves more mechanics in that case. I'll tell you this story in another article!