Noisy audio on +2, but only when RGB is connected

[jamesh]

There is a similar old thread, which may or may not be related. Unfortunately, it doesn’t help much in my case.

TL;DR: Grey +2, 0500 Issue 3, 8639, audio is practically noise free when using composite signal. However, 15..31Khz noise appears when I connect a proper RGB/SCART cable. Interesting observation: audio signal does not have to be routed to the DIN socket. This is not the tape deck noise, the built-in “Datacorder” is physically disconnected.

Here are all the details. I’ve got a grey +2 few months ago, PCB without any obvious modifications. For a while I was using composite video simply because I had no SCART cable. Computer was in a good working condition when I bought it besides that (normal for unmodified +2, I assume) weak composite signal. I turned TR4 to be in the correct position, replaced voltage regulator, all capacitors, re-wired audio to be properly mixed into RF signal (I guess I'll disconnect the whole RF module anyway), and completed AY volume balancing mod. I am using modern switching power supply unit (9V 2A). I have not serviced the internal tape deck yet, so it remains physically disconnected all the time.

While working on those mods, I was using composite signal from the RGB DIN socket, only two wires connected: pin 1 and 2. No noticeable audio noise, confirmed later with oscilloscope.

Eventually, there was not much left to do and I decided to build a SCART cable since I had all components anyway and wanted a better picture. I decided to go with the circuit at the bottom of Paul’s “Spectrum +2 RGB Differences”, the one that uses LM317 to generate proper blanking voltage and requires jumpers to be set for “Peritel” configuration. However, before wiring the cable I decided to try it on a breadboard.

I soldered eight jump wires to a DIN plug, plugged it in, turned Speccy on and… I was greeted by a relatively loud noise. Using oscilloscope I can see 15..31Khz signal there, frequency changes with the background/border colours. Note, at this stage I had not connected any device/components to the RGB socket besides a few short (~15cm) wires.

First thing I tried – disconnecting audio signal from the DIN socket by removing LK8. Surprisingly, it did not help at all. Re-connecting C31 as it was and removing it completely doesn’t change anything, either.

I went ahead and assembled SCART cable circuit on a breadboard, with LM317, diode and resistors. The picture was nice, but the noise was there. Maybe a properly shielded cable will help, I thought. And I built complete SCART cable. I had a very cheap/flimsy SCART plug cut off from an old DVD SCART cable, but the rest of the components are from a known good supplier. Anyway, the noise was still there. Oh, the picture is crystal clear with the RGB cable, it’s just the noise. To be honest, the noise is less noticeable when Speccy is playing something. But it’s quite annoying when computer sits idle.

Not sure if there is anything else left to try. The other thread linked at the beginning suggests that the noise is expected. I am just surprised that I only get it with RGB cable.

[mark8bit]

the +2 has a few options .. its been a while since i did one :
what are all your link settings ?
if you are using link 8 for audio then link 7 needs to removed or it will be sending the brightness signal to the tv audio input aka din pin 3 and giving a nice buzz
when i make my cables i use the composite for the sync rather than the Csync so it can still be used as composite on non RGB compliant tv
for the blanking i fit link 1 and remove link 3 (i think) to give the 12 volts then fit a resistor in the corresponding position either R9 or R15 at about 680 ohms
i use cheap 8 core alarm cable for my scart leads that seems to work fine
the cables i make don't use the brightness signal / diodes
*** it has been a while since i did a cable so check what i say first

[1024MAK]

Just to be clear, can you please report back answers to the following:

With a plug in the Monitor/RGB DIN socket, how are you connecting the audio?

Are you using the tape/sound socket to connect to the audio? If yes, what happens if you disconnect the Monitor/RGB DIN plug?

What is the status of these links:
LK1 (Vertical Sync) / (TTL)
LK2 (+12V)
LK3 (+12V)
LK4 (Composite Video)
LK7 (Bright Output)
LK8 (Audio output)

Mark

[jamesh]

Thanks a lot for looking at this. The configuration I have is as follows:

With a plug in the Monitor/RGB DIN socket, how are you connecting the audio?

Via SCART cable -- pin 3 of the DIN socket, via LK8.

Are you using the tape/sound socket to connect to the audio?

I was not using sound socket (3.5mm), but since it is directly connected to R37, R45, R80, and LK8 now I am checking the noise level using a short lead attached to it. It makes no difference if 3.5mm socket is connected to anything or disconnected completely, though.

If yes, what happens if you disconnect the Monitor/RGB DIN plug?

When RGB cable is plugged into the DIN socket, the noise is there. When a self-made composite cable (only pins 1 and 2, cut off phono plug and soldered wires to a DIN plug) connected to the DIN socket, there is no noise. When nothing is connected to the DIN socket there is no noise.

*) "no noise" means the signal is roughly the same as signal present on TR6 collector. There is always some minimal noise there, but I assume this is generated by the AY/TR6 circuitry and the volume of that noise is negligible and it only manifests itself as faint hissing sound audible at max volume.

What is the status of these links [...]

- LK8 is set (LK7 is open) for sending audio to the DIN socket, pin 3
- LK3 is set (LK1 is open) for making +12V available on pin 5. There was unpopulated element on the PCB, marked as R19 going from LK3 straight to the +12V rail, I replaced it with a short wire since in my cable that goes straight to FUNCTION SELECT and voltage regulator for BLANKING.
- LK2/LK2 are NOT changed, meaning that LK4 is set (LK2 is open) -- because I am also using composite signal from the DIN socket.

Noise present only on LK8 and connected points -- R37, R45, R80, SK1 (3.5mm audio out), which goes all the way to the DIN socket. So I think it's coming from the cable. Also, I am afraid I made a mistake earlier, when I said that removing LK8 makes no difference. I guess I was checking signal on the wrong side of it. Removing LK8 removes noise from the 3.5mm socket, but it is still there on pin 3, DIN socket.

Oh, another interesting but probably unrelated fact. The tape deck belts were looking good and I found a real tape in a charity shop nearby, so I connected datacorder and (surprise!) it didn't add any more noise or change existing noise. I guess the noise coming from the RGB cable is just much louder Anyway, it is possible to load something from a tape and the tape deck doesn't make it worse. I am going to keep it disconnected anyway.

For now I assume there is something wrong with the SCART cable/plug and the next step would be to build a new one from scratch. I've ordered new high quality SCART plug which should arrive next week. I am just wondering if I should drop the LM317LZ circuit and go with a simpler way of generating BLANKING voltage, but I doubt LM317 is capable of generating anything at exactly 15Khz.

Today I completed "RF mod" -- completely disconnected input wires from the RF modulator and unsoldered resistor from the RCA connector inside the tin, adding a wire from LK4 (75ohm resistor from TR4 emitter) to the RCA connector. I am thinking about removing LK4 and leaving both LK2 and LK4 open. The fewer unused signals go to the RGB cable, the better, isn't it?

Essentially, the situation I have is described by MatGubbins here as normal and I can live with that, since most of the time I use composite output anyway. I am just curious, why I only see it when RGB cable is connected. After fiddling with that two weekends in a row, I guess I made a mistake when wiring the cable and the best part forward is to build a new one from scratch. Any other ideas are appreciated, of course.

[1024MAK]

Based on what you have said, I suspect that is is either a wiring fault with your RGB SCART cable, or there is a solder splash, strand of wire or some other problem that is causing a short circuit connection where there should not be.

Try using your multimeter on the 20k resistance range to see if you can pick up an unwanted connection between the audio (DIN pin 3) circuit and any of the video signals (composite video, R, G, B, or V.Sync.).

Mark

[zx64]

When RGB cable is plugged into the DIN socket, the noise is there. When a self-made composite cable (only pins 1 and 2, cut off phono plug and soldered wires to a DIN plug) connected to the DIN socket, there is no noise. When nothing is connected to the DIN socket there is no noise.

This means something is wrong with the cable. In addition to what 1024MAK said, it can also be cross talk between video and audio wires, but I'm not sure how loud the noise would be. How long is the cable?

[1024MAK]

Crosstalk should not be a significant problem with cables that are less than one metre long. Nearly all commercial all-wired SCART cables use a multicore cable with just an outer braid/screen.

Of course, the better ones have individually screened wires for the video signals to reduce crosstalk.

However, if you are using a twisted pair type cable, with a video signal in one wire of the pair, and the audio signal in the other wire of that pair, that may cause problems.

Mark

[jamesh]

This means something is wrong with the cable. In addition to what 1024MAK said, it can also be cross talk between video and audio wires, but I'm not sure how loud the noise would be. How long is the cable?

The cable is approximately 1m 25cm.

New SCART connector arrived today and I'll be doing one more round of testing on Saturday. But I've triple checked all the connectors and I am certain that there are no problems with soldering. I am using shrink wraps for all the points and have tested continuity with a multimeter. The noise is there even if I am not connecting SCART end anywhere, so yes, I agree, most likely it's the cable.

I am using 9-core RS-232 cable I had, but while it looks decent it may not be the best one. What would be a good source of 8-core individually shielded cable? Preferably not a 25m or 100m reel In the UK.

Thanks a lot!

[Seven.FFF]

What would be a good source of 8-core individually shielded cable? Preferably not a 25m or 100m reel In the UK.

You might just buy a professionally made cable such as this: https://www.ebay.co.uk/itm/260949446667 Retro Computer Shack's cables have great quality and a great reputation.

[1024MAK]

The only relatively cheap source I can think of, is to chop up a VGA cable…

Mark

[Guesser]

The only relatively cheap source I can think of, is to chop up a VGA cable…

I have done this to make SCART cables for my Spectrum +3, and BBC micro.
I wouldn't really recommend it as the cores are fantastically thin and horrendous to strip and solder.

If I was doing it over (which I might if any of the solder joints break again) I think I'd make short (couple of inches) adapter dongles from unshielded but easy to solder wire between VGA and SCART, and the various DIN pinouts to VGA.

[jamesh]

You might just buy a professionally made cable such as this: https://www.ebay.co.uk/itm/260949446667 Retro Computer Shack's cables have great quality and a great reputation.

Yes, sure, that's the plan C. But... It’s not that much fun, isn’t it? Besides, I don’t really like the idea of using CVBS/CSYNC as BLANKING input, that’s why I decided to look for an alternative solution and build Paul’s version with voltage regulator.

[jamesh]

Thanks a lot, everyone, for helping me with troubleshooting. I must admit, I still do not understand why it happens, but I found a workaround.

Well, long story short, yes, it’s the horrible cross talk. The new SCART plug made it possible to test signals everywhere with multimeter and oscilloscope, so I am quite certain that’s the case. And it comes from the RGB lines. Perhaps, (re)using a VGA cable may help. I have only tried a few crazy ideas (the weather did not allow me to spend too much time on this ), but I quickly implemented a simple workaround: soldered a “frankencable” -- mono audio lead (shield + core) connected to DIN pins and SCART plug pins in parallel with the video cable. Presumably because the other, video carrying, cable is shielded and the audio lead is shielded, there is no noise whatsoever.

I am still wondering, why I am the only one experiencing that (or just noticing that?). My current theory is that it may be the wrong choice of wires. Perhaps, when using different wires the cross talk could be reduced (i.e., I made an obvious choice of red as R, green as G and so on, perhaps different wires make different pairs and may behave differently), but I guess I can happily live with my “frankencable”

[Seven.FFF]

I am still wondering, why I am the only one experiencing that (or just noticing that?).

I have no crosstalk from my Retro Computer Shack cable (although that is for +3 not +2 grey, and his cables for those two systems have materially different passive components). All video wires are untwisted and shielded together, audio wires are also untwisted and shielded together in a separate sleeved cable.

[Guesser]

There's no need for a voltage regulator, one 12V pin goes to aspect switching, and the other just needs the right resistance to give the correct blanking voltage, e.g. 560Ω

[jamesh]

Actually yes, a crude simulation I did before soldering anything was showing that the overall load from a simple resistor and LM317 circuit should be roughly the same. However, it is mentioned that "Using a voltage regulator can overcome the loading issue on the power line." And since I had LM317 anyway, I decided to follow that advice.

[jamesh]

I have no crosstalk from my Retro Computer Shack cable [...] audio wires are also untwisted and shielded together in a separate sleeved cable.

That explains everything, this is essentially what I did at the end. Thanks a lot!