GND Scheme questions

[Davebassman]

The silk screen overlay shows that the signal grounds are already connected via the PCB. Wiring the XLR/unbalanced connections exactly as per the diagram is how I'd wire it. I took a look at the original rack mount Dimension D, and pin 1 of the XLRs are all connected to signal ground, whilst the XLR shells are connected directly to chassis ground. Roland perhaps expected the use of cables where pin 1 is not connected to the shell. The pictures of the prototype unit you're building show the boards isolated on nylon standoffs, though that unit is wall-wart powered. It may be worth considering connecting the PCB signal ground to chassis ground via a single connection using 'loop breaker' as per Figure 5 in the link. I've used this method with good success (though using antiparallel diodes rather than a bridge).

https://sound-au.com/earthing.htm#s6

I understand that in the original the signal gnd is not connected to chassis. What happen if I don't connect the signal gnd to chassis? I use internal transformer, not a wall mart.

And if the pin 1 is not connected to the shell is unsense the use of shield cables. No?

Many many thanks Mick!

[Mick Bailey]

The connection of signal ground to chassis is to provide additional protection in the event of short that connects mains live to signal ground, either within the unit, or by means of a faulty connected device.

Pin 1 connects internally to signal ground, and pin 1 is where the cable screen is wired in order to screen the balanced twisted pair. The shell connects to chassis to provide a) screening of the XLR pin assembly and b) to ensure that an exposed metal part is connected to mains earth.

[Davebassman]

The connection of signal ground to chassis is to provide additional protection in the event of short that connects mains live to signal ground, either within the unit, or by means of a faulty connected device.

Pin 1 connects internally to signal ground, and pin 1 is where the cable screen is wired in order to screen the balanced twisted pair. The shell connects to chassis to provide a) screening of the XLR pin assembly and b) to ensure that an exposed metal part is connected to mains earth.

Finally I will do the things simple, and I will don't connect the signal ground to earh. Like the original.

Many thanks for all!

[g1]

Roland perhaps expected the use of cables where pin 1 is not connected to the shell.

That's how I was always taught to wire up XLR cables.

[Davebassman]

That's how I was always taught to wire up XLR cables.

Seems that is not the correct way...

[g1]

Seems that is not the correct way...

That is the jacks on the unit, not the cables.

[dai h.]

That's how I was always taught to wire up XLR cables.

I think that usually the shell to pin 1 connection is made inside the chassis (a wire from XLR pin 1 to chassis), but I've noticed that on some plastic XLR connectors there is no contact for the shell and I think this might be an issue sometimes. There is a certain mic guru who seems adamant that the correct way to wire (balanced XLR cables) is to wire the cable shield to the shell (despite other (seemingly informed) opinions) since a) Neumann did it that way (he gives example of a tube mic), and b) customers have reported that that method solved RF problems. (I'm guessing) what may have happened on the mic end is that the particular mic (or mics exhibiting RF problems) had a wire connection from pin 1 of the incoming XLR to mic body(chassis) that was too long (giving opportunity for RF to get inside), and wiring pin 1 to XLR shell inside the connector end minimized the connection from shield through to pin 1 to chassis (mic body). (And after thinking about it a little bit maybe the same thing on the mic pre input end). I've also noticed (maybe more on less expensive gear?) that some plastic XLR connectors don't have a connection for the shell, and that might possibly be an issue in some situations. (To be clear my understanding is that the correct way is to *not* connect pin 1 to shell and let the incoming device decide but that it may make sense to wire pin 1 to shell inside the connector depending on the specific situation.)

Neutrik makes an EMC XLR connector which is set up (if you look at it inside) to shunt (AC ground through a super low impedance path for high frequencies made up of paralleled surface mount multi-layer ceramic caps (the paralleling according to my understanding increases the self-resonant frequency for the composite cap and makes it more effective at very high frequencies) on a ground plane inside a little PCB inside the connector (with a ferrite bead over pin 1--so it's sort of impeding (with the FB) and shunting (with the MLCs) any RF on the shield to the gear of the chassis that it's being plugged into, but if the input connector has no shell contact, it wouldn't appear that the connnector is going to work as it should. Also (according to someone who actually used one and seems to be technically literate (I bought one so I could look at the innards but haven't tried using it)), the connector might be lacking in effectiveness apparently from the shield not making a good enough (low impedance enough) contact (the cable shield connection is a sort of "press fit" connection to this metal piece (actually made of two pieces) using the (screw-in) plastic "gland" on the end, and that physical configuration is maybe not providing enough pressure). (I'm guessing (since there was an earlier EMC type connector but (IIRC) the shield connection was crimped (and crimping tools typically seemed to be expensive plus I think the operator needs to be competent) this was an attempt to make something easier to assemble but it didn't quite work out?) Even if you don't use these connectors, they do seem to help point how things should more (ideally) go. (Besides stuff already mentioned), (for example) it has circumferential extra spring contacts. So maybe make sure whatever you are using has clean non-oxidized contacts (so you aren't unwittingly creating choke points for RF), use a good quality (Neutrik, Switchcraft, Cannon, etc.) all metal connector that has a good low impedance contact to the chassis. Very short pin 1 connection inside to chassis. I don't quite remember the specifics, but there might be special cases where there isn't a direct DC connection and it's lifted through a relatively low value resistor (but shorted to chassis AC-wise with a cap). The whole thing could be a big headache (or interesting puzzle I guess), thinking about grounds, loops, RF, possible interconnectivity issues.

[Mick Bailey]

The original Dimension D is wired to the 'wrong' connection shown in post #20.

[Davebassman]

It seems this is the most obvious way to wiring. Changing TS jacks with TRS. But doing this, they short the ring to gnd and I think this is not a good idea with a power transistors output stage. This way short one phase of the output to gnd.

[dai h.]

On an original (bit fuzzy to make out where the grounds are going completely clearly, but) it kinda looks like XLR and 1/4" input and output grounds are connected to chassis (there's a daisy chain across the connectors and what appears to be two lug connections to chassis at the input end (and skinny black wires from that side to PCB)), plus a thick wire to the PCB over at what looks like the power input end (kinda looks like a round-shaped bridge diode, relatively big (primary caps?) and regulators(?).

https://www.bn1studio.co.uk/on-the-bench-dimension-d/

From looking at the service manual:

https://usermanual.wiki/Collections/...anual.pdf.html

it looks like the fat wire (lug #29 on the layout) is tied in to the power end ground at a quieter part (tied in with its own trace where the second filter cap ground is where there should be less ripple--generally this is the point where you'd tie the rest of the circuit ground to and keep the earlier part off of the chassis). It might be going to a chassis connection on the bottom left (?). Not sure what to make of what I'm seeing so far.

[Davebassman]

On an original (bit fuzzy to make out where the grounds are going completely clearly, but) it kinda looks like XLR and 1/4" input and output grounds are connected to chassis (there's a daisy chain across the connectors and what appears to be two lug connections to chassis at the input end (and skinny black wires from that side to PCB)), plus a thick wire to the PCB over at what looks like the power input end (kinda looks like a round-shaped bridge diode, relatively big (primary caps?) and regulators(?).

https://www.bn1studio.co.uk/on-the-bench-dimension-d/

From looking at the service manual:

https://usermanual.wiki/Collections/...anual.pdf.html

it looks like the fat wire is tied in to the power end ground at a quieter part (tied in where the second filter cap ground is where there should be less ripple--generally this is the point where you'd tie the rest of the circuit ground to and keep the earlier part off of the chassis). Not sure what to make of what I'm seeing so far.

Thanks. But the problem with this unit is the "pin 1 problem". It's wired incorrectly or correct for the year 1979. I can wire like the original unit, but it's a nonsense in 2022 year and I want to wire it with AES48 standard.
With the XLR connectors I see clearly. With TS jacks, the solution would be change to TRS, but the problem is that this unit have a power transistor output stage and If I change to TRS jack's I will short pin 2 to GND, and I don't know if I should do.

I read in some notes, that a solution would be a RF network. 100ohms resistor with 100nF cap. But I don't know.

[g1]

(To be clear my understanding is that the correct way is to *not* connect pin 1 to shell and let the incoming device decide but that it may make sense to wire pin 1 to shell inside the connector depending on the specific situation.)

This is also how I was trained to wire xlr cables.
For the fellow you mentioned that insists the shell should connect to pin1 on the cable itself; connecting such a cable to a unit that has a ground lift switch or strap can be problematic if you want to use the ground lift.

[dai h.]

Thanks. But the problem with this unit is the "pin 1 problem". It's wired incorrectly or correct for the year 1979. I can wire like the original unit, but it's a nonsense in 2022 year and I want to wire it with AES48 standard.
With the XLR connectors I see clearly. With TS jacks, the solution would be change to TRS, but the problem is that this unit have a power transistor output stage and If I change to TRS jack's I will short pin 2 to GND, and I don't know if I should do.

I read in some notes, that a solution would be a RF network. 100ohms resistor with 100nF cap. But I don't know.

(I don't understand the specific grounding scheme of the Dimension D, but very generally speaking) the incoming shield needs either a DC or AC connection to chassis (or perhaps very low impedance ground plane area that is set up in a way to not cause problems by coupling RF into sensitive grounds). So if isn't some sort of issue it should be okay to conform to the modern standard and tie pin 1 to chassis immediately at the connector (but as I said I'm not really sure how to make sense of the grounding scheme).

re: the 100 ohms and 100nF: I think the 100 ohms (I think this may be a ballpark value like 10 ohms or 51 ohms or whatever depending on the situation) is that it presents enough impedance to some problem current, while the cap is to short out the shield high frequency AC-wise (to shunt away any RF on the cable shield). The emphasis (from my understanding) is this measure should be *immediately* at the input or output. (So don't run some skinny wire for a length inside the chassis and then connect to chassis.) I think the ideal is making the connection on the outside of the chassis, but immediate inside seems to be okay.

re: the TS and TRS jacks and the output (I don't know if it matters but) the instructions say only use one or the other (only use balanced or unbalanced).

re: the pic the ground wire running sideways: some might not be connected to circuit ground straight from the input and might be part of some switching circuit using an FET which appears to affect the gain (looks like a gain R in the feedback loop of an op amp is shorted).

I found a passage in an Elliott Sound Products article that talks about connecting shield pins to power supply common ("Design Checklist" towards the bottom,

https://sound-au.com/articles/balanced-2.htm

but it wouldn't seem to apply here, so still confused. (Is the fat wire that is connected to power supply common also connected to chassis with a screw on the lower left side of the PCB (in the service manual layout)?)

Also if you read pin 1 related articles, it's not recommended to connect the shield to circuit(signal) ground, but it doesn't seem unusual to do so. For example, in the mixer section of the Tascam 488mkII cassette multi-track I mentioned, in the input area, pin 1, phantom ground, 1/4" inputs and outputs, and RCA connector grounds are all connected by a trace running sideways, and all of these (insulated connectors) are attached to a steel panel, and the connection for the PCB ground to this panel is about at the center of the ground trace. (I wondered if it would just be better to use non-insulated against RF but I never got as far as taking it apart and following exactly how the grounds are connected.)

I also found a couple of more guts pictures on Reverb:

This is also how I was trained to wire xlr cables.
For the fellow you mentioned that insists the shell should connect to pin1 on the cable itself; connecting such a cable to a unit that has a ground lift switch or strap can be problematic if you want to use the ground lift.

Somebody did mention a possible ground loop problem in the thread I am thinking of (with the guru guy). He also seemed to think that RF measures resulted in a catastrophic effect on the sound (despite other seemingly informed opinions). (My novice understanding) it is maybe possible to have some sort of interaction resulting in a problem (weird peaking or whatever), but it is possible to have measures that don't effect the sound, and if well implemented should ensure the audio is not affected.

[Davebassman]

(I don't understand the specific grounding scheme of the Dimension D, but very generally speaking) the incoming shield needs either a DC or AC connection to chassis (or perhaps very low impedance ground plane area that is set up in a way to not cause problems by coupling RF into sensitive grounds). So if isn't some sort of issue it should be okay to conform to the modern standard and tie pin 1 to chassis immediately at the connector (but as I said I'm not really sure how to make sense of the grounding scheme).

re: the 100 ohms and 100nF: I think the 100 ohms (I think this may be a ballpark value like 10 ohms or 51 ohms or whatever depending on the situation) is that it presents enough impedance to some problem current, while the cap is to short out the shield high frequency AC-wise (to shunt away any RF on the cable shield). The emphasis (from my understanding) is this measure should be *immediately* at the input or output. (So don't run some skinny wire for a length inside the chassis and then connect to chassis.) I think the ideal is making the connection on the outside of the chassis, but immediate inside seems to be okay.

re: the TS and TRS jacks and the output (I don't know if it matters but) the instructions say only use one or the other (only use balanced or unbalanced).

re: the pic the ground wire running sideways: some might not be connected to circuit ground straight from the input and might be part of some switching circuit using an FET which appears to affect the gain (looks like a gain R in the feedback loop of an op amp is shorted).

I found a passage in an Elliott Sound Products article that talks about connecting shield pins to power supply common ("Design Checklist" towards the bottom,

https://sound-au.com/articles/balanced-2.htm

but it wouldn't seem to apply here, so still confused. (Is the fat wire that is connected to power supply common also connected to chassis with a screw on the lower left side of the PCB (in the service manual layout)?)

Also if you read pin 1 related articles, it's not recommended to connect the shield to circuit(signal) ground, but it doesn't seem unusual to do so. For example, in the mixer section of the Tascam 488mkII cassette multi-track I mentioned, in the input area, pin 1, phantom ground, 1/4" inputs and outputs, and RCA connector grounds are all connected by a trace running sideways, and all of these (insulated connectors) are attached to a steel panel, and the connection for the PCB ground to this panel is about at the center of the ground trace. (I wondered if it would just be better to use non-insulated against RF but I never got as far as taking it apart and following exactly how the grounds are connected.)

I also found a couple of more guts pictures on Reverb:








Somebody did mention a possible ground loop problem in the thread I am thinking of (with the guru guy). He also seemed to think that RF measures resulted in a catastrophic effect on the sound (despite other seemingly informed opinions). (My novice understanding) it is maybe possible to have some sort of interaction resulting in a problem (weird peaking or whatever), but it is possible to have measures that don't effect the sound, and if well implemented should ensure the audio is not affected.

The output of the dim d is a true balanced output, if you use trs jack's and use unbalanced cables, you will short pin 2 to gnd, and this is not a good idea. This is the reason Roland use separte connectors for balanced and unbalanced.

[Davebassman]

From Haible,

https://electro-music.com/forum/topic-26527-275.html

"Modern" balanced outputs are either electronically balanced in a way that they behave like a transformer (tie one output to GND, and the other output will double its voltage!), or, more often, they are very simple and cheap, balancing only the output impedance: that is, just connecting the ring to GND with the same resistor value you find between the tip and the (unbalanced) output amp. The former is great (at least as long as it doesn't clip), the latter is not really a balanced output if you're asking me; it's just an unbalanced output providing balnced impedance for a balanced input making the best of the connection to an unbalanced output.
What Roland did in the Dimension D is a real balanced output, driving the two wires with opposite polarity - but not like a transformer, i. e. not allowing a short of one end of the output. This is why Roland had separate jacks for balanced and unbalanced outputs. And if you're looking at my schemos, I suggest doing the same, for the "vintage" version.
My "modern" version is less critical by far. Same topology as Roland, but no power output stage to drive 600 Ohm loads. Series resistors large enough to allow a short without damage, and still strong enough to drive modern 10kOhm or higher inputs. It's still not recommended to short these outputs, but it won't cause much harm.