DIY Triggered Input Switch

[starfly]

The best crosstalk would come from separate relays for the left and right channels. Beyond that its impossible to say which single relay would have lowest crosstalk. Reed relays won't necessarily be any better. You can lower crosstalk in several ways beyond the relays, such as lowering impedances, but overall its easier to use separate relays for L and R.

Right. Well, I'm even getting a bit of crosstalk on the second input when unswitched. Wondering if there is a better way to isolate that so that absolutely 0 comes through when it's not active. Granted, in practice I'll rarely have both of my sources on at the same time, so it might not really matter in the end.

As for isolating L and R channels, I haven't actually tested how bad that is, I'll take some measurements today or tomorrow. I did end up putting the PCB on nylon stand-offs, and will try using a 10 Ohm resistor for grounding one of these days to see how that measures. Do you think it makes sense to play around with even higher resistance for grounding?

 

[MakeMineVinyl]

Right. Well, I'm even getting a bit of crosstalk on the second input when unswitched. Wondering if there is a better way to isolate that so that absolutely 0 comes through when it's not active. Granted, in practice I'll rarely have both of my sources on at the same time, so it might not really matter in the end.

As for isolating L and R channels, I haven't actually tested how bad that is, I'll take some measurements today or tomorrow. I did end up putting the PCB on nylon stand-offs, and will try using a 10 Ohm resistor for grounding one of these days to see how that measures. Do you think it makes sense to play around with even higher resistance for grounding?

The usual way to stop the un-selected input from bleeding through is to short it to ground when not selected. In early vacuum tube preamps, they literally dead-shorted unused inputs to ground. But you can't do that nowadays because you don't want to risk drawing excessive current from the unused inputs.

The best compromise is to short unused inputs to ground through a suitable resistor - with circuits which have op-amp outputs a safe value would be 2000 ohms, as this is the typical lowest rating for op-amps. I have seen some instances in modern gear where unused inputs are dead-shorted to ground, and in this case they are relying on the fact that most gear has some value of buildout resistor in series with their outputs. But there is no standard value for buildout resistors, and they are typically less than 470 ohms (I routinely use 75 ohms).

Also, grounding is very critical to crosstalk; always use a star ground.

 

[magicwolf]

I'm not an electronics expert, so I can't understand why you don't use transistors to control relays? The usual relay control circuit usually includes a transistor. Although in my case, the transistor was also a source of noise ;(.
If you want, you can put two 100uf and 0.1uf capacitors on the control power input, perhaps this will reduce noise.
To connect the left and right channels, you use wires without a screen, IMHO you need to try well-shielded wires, even at short distances.
P.S. I'm wrong ;(. Capacitors should not be placed between 1 and 8 relay contacts.

 

[MakeMineVinyl]

I'm not an electronics expert, so I can't understand why you don't use transistors to control relays? The usual relay control circuit usually includes a transistor. Although in my case, the transistor was also a source of noise ;(.
If you want, you can put two 100uf and 0.1uf capacitors on the control power input, perhaps this will reduce noise.
To connect the left and right channels, you use wires without a screen, IMHO you need to try well-shielded wires, even at short distances.
P.S. I'm wrong ;(. Capacitors should not be placed between 1 and 8 relay contacts.

Driving a relay with a transistor will not influence crosstalk one way or the other. Transistors are usually used to drive relays when translating between two drive voltages. For instance I do it frequently with microcontrollers which operate either on 5V or 3.3 when they need to drive a relay with a 12 volt coil. Crosstalk in relays is all about capacitive coupling, the spacing of the relay contacts for the separate channels, and grounding.

 

[Iankang]

Hi, I've been following this thread for a while, since I have the similar setup (I bought pre-build relay board though). Seems to me there is a problem worse than ground loop. The balanced output from motu should be hot/cold/shield, where hot/cold are signals with opposite phase. When you tie both pin1 and cold wire to the chassis, you effectively tie cold wire to ground, this is definitely wrong. If the resistance between the two wire is very small (like 1 ohm), you may lose 6dB of signal, plus lot of noise, from your measurement, the signal level doesn't change that much, I guess either the resistance between cold/shield are not that low, or the driving stage of motu is very strong.

Since you amp is balanced input, I think switching both hot/cold wire, and connect all shield (pin1) to chassis is the right way to do. About pop and clicks during signal switching, since the amp is balanced in, I don't think it will be even audible.

 

[magicwolf]

Crosstalk in relays is all about capacitive coupling,

I didn't understand you well, but I read that the coil relay can make noise.
When I soldered a 100uf capacitor between the base of the transistor and the ground (emitter), the noise dropped sharply. I do not know if this is correct or not, but in my case it works.

 

[Hayabusa]

The omron relay I proposed has almost no audible click and I could not notice any audible crosscalk (ear on the tweeter )

 

[MakeMineVinyl]

I didn't understand you well, but I read that the coil relay can make noise.
When I soldered a 100uf capacitor between the base of the transistor and the ground (emitter), the noise dropped sharply. I do not know if this is correct or not, but in my case it works.

Noise on the supply which is powering the coil can couple to the contacts and induce that PS noise. The trick is to have a nice, clean source of power in the first place.

 

[magicwolf]

clean source of power

I absolutely agree with you. I did so. First, I removed the conversion from AC to DC and the 12V stabilizer and connected the signal switch to clean power from the dac board. But alas, it almost didn't help ;(.

 

[MakeMineVinyl]

I absolutely agree with you. I did so. First, I removed the conversion from AC to DC and the 12V stabilizer and connected the signal switch to clean power from the dac board. But alas, it almost didn't help ;(.

I've got a Behringer 1/3rd octave EQ on my bench right now which has a RFI/EMI problem which so far has defied a solution. Sometimes its difficult to trace obtuse problems.

 

[magicwolf]

Sometimes its difficult to trace obtuse problems.

Oh, ho, how right you are. I have a noise problem, but I couldn't solve it:

Ground | DIY Audio Heaven

Is it possible to separate the input ground from the output ground in headphone amplifiers? My pcb is made in such a way that the bottom and top of the board are a solid ground polygon, and at the

diyah.boards.net

Apparently sometimes it's better to do everything from scratch than to try to fix it.

 

[TunaBug]

https://omronfs.omron.com/en_US/ecb/products/pdf/en-g6k.pdf

These are used in the minidsp SHD.. So I guess they must be good...

Reading through the spec sheet that @Hayabusa linked to reminded me of latching relays. Power the coil briefly to switch the relay, but it then stays in place mechanically or magnetically. That would essentially remove any problems with noise related to the relay coil, wouldn't it?

 

[starfly]

Hi, I've been following this thread for a while, since I have the similar setup (I bought pre-build relay board though). Seems to me there is a problem worse than ground loop. The balanced output from motu should be hot/cold/shield, where hot/cold are signals with opposite phase. When you tie both pin1 and cold wire to the chassis, you effectively tie cold wire to ground, this is definitely wrong. If the resistance between the two wire is very small (like 1 ohm), you may lose 6dB of signal, plus lot of noise, from your measurement, the signal level doesn't change that much, I guess either the resistance between cold/shield are not that low, or the driving stage of motu is very strong.

Since you amp is balanced input, I think switching both hot/cold wire, and connect all shield (pin1) to chassis is the right way to do. About pop and clicks during signal switching, since the amp is balanced in, I don't think it will be even audible.

I don't think the cold wire is going to ground/chassis in my design. Only pin 1 is, which is the shield. And pins 2 and 3 just pass through directly without any chassis grounding.

 

[Iankang]

I don't think the cold wire is going to ground/chassis in my design. Only pin 1 is, which is the shield. And pins 2 and 3 just pass through directly without any chassis grounding.

Hi, sorry I may misunderstand. From the schematic, J4 1/2/3/4 is L_hot/L_cold/R_hot/R_cold, and 2/4 of J4 are shorted together then connect to PCB ground (then switched to 2/4 of of one of the input). Doesn't this connect L/R cold together to PCB ground? (and chassis ground when using metal mount)

 

[restorer-john]

I've got a Behringer 1/3rd octave EQ on my bench right now which has a RFI/EMI problem which so far has defied a solution. Sometimes its difficult to trace obtuse problems.

I had an analog EQ which had small value ceramic caps over each diode in the discrete bridge rectifier. They were making a noise, not only audible, but RF as well. They were all in physical alignment with one another and once I bent them at random angles, the noise and the RF went away. Never forgot that one.

 

[magicwolf]

latching relays

It's a good idea. But there are many different types of latching relays:

https://www.mouser.com/datasheet/2/357/105A_755-7090.pdf

For example, TE axicom has bistable relays:

https://www.te.com/commerce/DocumentDelivery/DDEController?Action=showdoc&DocId=Data+Sheet%7F108-98001%7FZ.1%7Fpdf%7FEnglish%7FENG_DS_108-98001_Z.1.pdf

But how to manage them? How to change the polarity?

 

[tomchr]

I had an analog EQ which had small value ceramic caps over each diode in the discrete bridge rectifier. They were making a noise, not only audible, but RF as well. They were all in physical alignment with one another and once I bent them at random angles, the noise and the RF went away. Never forgot that one.

Wow. That's pretty special.

It's remarkable how many just sprinkle capacitors around indiscriminately. It does help to understand the problem first. I wrote quite a bit about rectifiers and snubbers here: https://neurochrome.com/pages/rectification-snubbers

I realize I'm late to the game here, but lemme add my $0.02 CAD: The plot in the first few posts clearly show 60 Hz and its harmonics mixing with the signal. That's why you see all the IMD of 1 kHz ±n*60 Hz. Note that the IMD products show up around harmonics of 1 kHz as well. Fun, fun...
The fact that the hum seems to be predominantly 60 Hz indicates to me that you're dealing with a ground loop and not so much with coupling of anything related to the relay drive signal. If one of the inputs is selected when the power is off, you can always try disconnecting the power supply and see if the hum goes away.

If the hum is related to the power supply, it is possible that the supply has a filter capacitor connected between the mains neutral and the output V-. That can cause trouble. In that case, you can try using a battery to power the box instead. If you still have hum on battery power, the issue is related to the grounding of the circuit within the box.

On the topic of relay selection: You do want to use small-signal relays. Power relays can cause distortion. The relays with fancier contact platings (Ag+AuPd for example) can be worthwhile. I've yet to measure any impact on distortion, hum, or anything else caused by a small-signal relay with gold (or gold alloy) contacts.

Nice looking box, by the way. Who did the engraving (or printing) of the legend?

Tom

 

[starfly]

Wow. That's pretty special.

It's remarkable how many just sprinkle capacitors around indiscriminately. It does help to understand the problem first. I wrote quite a bit about rectifiers and snubbers here: https://neurochrome.com/pages/rectification-snubbers

I realize I'm late to the game here, but lemme add my $0.02 CAD: The plot in the first few posts clearly show 60 Hz and its harmonics mixing with the signal. That's why you see all the IMD of 1 kHz ±n*60 Hz. Note that the IMD products show up around harmonics of 1 kHz as well. Fun, fun...
The fact that the hum seems to be predominantly 60 Hz indicates to me that you're dealing with a ground loop and not so much with coupling of anything related to the relay drive signal. If one of the inputs is selected when the power is off, you can always try disconnecting the power supply and see if the hum goes away.

If the hum is related to the power supply, it is possible that the supply has a filter capacitor connected between the mains neutral and the output V-. That can cause trouble. In that case, you can try using a battery to power the box instead. If you still have hum on battery power, the issue is related to the grounding of the circuit within the box.

On the topic of relay selection: You do want to use small-signal relays. Power relays can cause distortion. The relays with fancier contact platings (Ag+AuPd for example) can be worthwhile. I've yet to measure any impact on distortion, hum, or anything else caused by a small-signal relay with gold (or gold alloy) contacts.

Nice looking box, by the way. Who did the engraving (or printing) of the legend?

Tom

Thanks, nice of you to chime in here The box is from Modushop.biz (Italian manufacturer), their Economica model. It's aluminum, and the outer shell is part silver, part black anodized.

Just now, for the first time ever, I hooked this box up to my amplifier in its unswitched state. And I didn't hear any humming or obvious ground loops. To be fair, right now the PCB is not grounded to the chassis, which seemed to give the best measurements when I was testing it last week.

But, it still bugs me that when properly grounded it causes issues, so I may have to do a redesign of the PCB and use a proper star grounding layout. It's been a really fun project, but I'm an absolute noob when it comes to electronics, and I think so far I've only scratched the surface.

@MakeMineVinyl I tried grounding with a 10ohm resistor in series and that didn't clean up the measurement.

@tomchr For relays, do you have any suggestions for the best relays to use for audio signals, with fancy contacts? During my initial research I couldn't find much at Mouser/Digikey.

 

[MakeMineVinyl]

Thanks, nice of you to chime in here The box is from Modushop.biz (Italian manufacturer), their Economica model. It's aluminum, and the outer shell is part silver, part black anodized.

Just now, for the first time ever, I hooked this box up to my amplifier in its unswitched state. And I didn't hear any humming or obvious ground loops. To be fair, right now the PCB is not grounded to the chassis, which seemed to give the best measurements when I was testing it last week.

But, it still bugs me that when properly grounded it causes issues, so I may have to do a redesign of the PCB and use a proper star grounding layout. It's been a really fun project, but I'm an absolute noob when it comes to electronics, and I think so far I've only scratched the surface.

@MakeMineVinyl I tried grounding with a 10ohm resistor in series and that didn't clean up the measurement.

@tomchr For relays, do you have any suggestions for the best relays to use for audio signals, with fancy contacts? During my initial research I couldn't find much at Mouser/Digikey.

The ultimate solution will probably be grounding the PCB to the chassis like you had it. The grounding on the PCB seemed pretty solid to me from where I sit, but you can always experiment. I routinely ground the signal ground to the chassis when I have problems like this (I never use a safety earth ground to the chassis on my own projects - call me stupid, but that's the way it is. )

 

[tomchr]

A star ground is often the worst for low hum. A properly laid out ground plane provides lower impedance, hence, lower hum.

AES48-2005 says you should have pin 1 of the XLR go directly to the chassis. It also says you should have the board grounded to the chassis. The examples in the standard are pretty easy to follow. AES wants money for the standard (their ~$100/year membership is worthwhile) but if you search for "AES48-2005 pdf" you can sometimes get lucky.

Then again, nothing on the board really needs grounding. You're only switching the two signal pairs. I don't see a need to connect the relay coil voltage to ground unless it's required by whatever controls that voltage.

I've had good results with the Panasonic AGN relays. There are other good ones as well. These days you'll probably have to go with an odd coil voltage to find any in stock.

Tom