-
BLACK FRIDAY SALE!
20% off all PCBs - No coupon code required - Valid through 12/01/2024
Components and sale items excluded
HELP! PCB Layout Guidance
- Thread starter finebyfine
- Start date
finebyfine
Well-known member
JamieJ
Well-known member
I’ll try that thank Alex. I even updated to the newer version of diptrace but that didn’t work but I managed to loose all my libraries and saved preferred settings
finebyfine
Well-known member
I’ll try that thank Alex. I even updated to the newer version of diptrace but that didn’t work but I managed to loose all my libraries and saved preferred settings
I gotta imagine it’s a cross compatibility error between OSs. The number of mac bug reports on their forum with replies suggesting OP buy a windows pc does not inspire me with hope on finding a reasonable solution for ya.
Why do you need to use the not connected pin attribute anyway? AFAIK it doesn’t make any difference when converting to pcb
Robert
Reverse Engineer
The Mac version of DipTrace leaves a lot to be desired... It's a really poor adaptation in my opinion.
With that said, I've never marked an unconnected pin as "not connected", it's really not necessary, but I'd want access to context menus regardless.
Is there not a key on the keyboard that brings up context menus on Mac? (Fn, Control, Option, Command)?
With that said, I've never marked an unconnected pin as "not connected", it's really not necessary, but I'd want access to context menus regardless.
Is there not a key on the keyboard that brings up context menus on Mac? (Fn, Control, Option, Command)?
finebyfine
Well-known member
I’m not on my computer so I can’t confirm you can do this without right clicking on a pin but if you show the op-amp component’s pin names you could at least get “NC” to show up on your page. This option is disabled by default for me on most components I think. But for example, diode patterns I use have anode and cathode labeled by default as opposed to showing “1” and “2” like how pins are displayed on a resistor or cap.
Edit, because I am vicariously frustrated through/for you
If you wanted to get as anal about it as I would, and I suppose I now have, you could try using BetterTouchTool and remapping your right click. What I think is happening here is diptrace not inheriting some of mac’s global keyboard shortcuts. The same thing happens for me on windows, and windows key global shortcuts don’t activate with diptrace active.
Holding down CTRL before clicking on a mac, (and I would hazard to guess what mac calls ‘secondary click’ on trackpads is the same) technically switches left and right clicks - but still has to check for a “left click” before remapping. The short of it is that it doesn’t really fire what a regular mouse would send for “right click”, which seems absolutely negligible in general except for in diptrace.
I think this screenshot of bettertouchtool’s documentation basically proves this by having ctrl click and right click be different.
Bettertouchtool, last time I used it (which was everyday when my work computer was a mac), is pretty good at making macros global/supercede others and might not run into the same problem as with ctrl click
Holding down CTRL before clicking on a mac, (and I would hazard to guess what mac calls ‘secondary click’ on trackpads is the same) technically switches left and right clicks - but still has to check for a “left click” before remapping. The short of it is that it doesn’t really fire what a regular mouse would send for “right click”, which seems absolutely negligible in general except for in diptrace.
I think this screenshot of bettertouchtool’s documentation basically proves this by having ctrl click and right click be different.
Bettertouchtool, last time I used it (which was everyday when my work computer was a mac), is pretty good at making macros global/supercede others and might not run into the same problem as with ctrl click
Last edited:
xefned
Well-known member
If you're on a MacBook, two-finger click on the trackpad should do it, in most apps anyway. I don't know for sure about Diptrace.
szukalski
Well-known member
Just get what's cheapest if it's not for commercial.Hey, I'm about to order some little utility PCB's and I'm reading about this HASL surface finish for the first time.
My stupid question is; should I order the leaded HASL option since I use leaded solder?
I had no idea some PCB's were coated in lead
dawson
Well-known member
Thanks @szukalski, will do.
Hey, I should also mention I pilfered your diptrace library a few hours ago!
That was a huge time-saver, I really appreciate you posting that resource.
Hey, I should also mention I pilfered your diptrace library a few hours ago!
That was a huge time-saver, I really appreciate you posting that resource.
szukalski
Well-known member
That’s why it’s there! To save others my pain..Thanks @szukalski, will do.
Hey, I should also mention I pilfered your diptrace library a few hours ago!
That was a huge time-saver, I really appreciate you posting that resource.
Reviving an old thread to discuss a couple (three) PCB layout rules of thumb that I'm finding to be somewhat contradictory:
So my first question: the handful of through-hole analog Boss pedals I've looked at appear to blatantly violate all those rules (thinking specifically of the OD-3). Perhaps I'm missing something obvious, but it looks to me like Roland used single-layer PCBs. All the traces are fairly long, generally pretty fat, and running mostly in parallel to each other. Unless there are inner layers I can't see, they have no real ground plane, just ground traces. And they're not afraid to use jumpers on the component side (presumably as a way to achieve the equivalent of a second layer of traces). And on top of that, they have a not-too-short ribbon cable between the main PCB and the pots (i.e. essentially more long(ish) parallel traces).
Maybe emulating Boss isn't a good idea? I just assumed you could do a lot worse, given their sheer volume and reputation for longevity and reliability.
I've found the easy way to satisfy all the above requirements (and more) is to use a 4-layer board. What I've done is put signal routing on the top and bottom outer layers, and one inner layer for power lines, and another inner layer for an wholly unbroken ground plane. This is of course more costly, and I've been told overkill for analog guitar effect circuits.
But if we move back to two layer boards, it gets virtually impossible to satisfy all three of the above best practices. So I'm wondering which rules take precedence? In other words, we know we can't follow all rules to the letter, so which rule(s) should be the firs to be broken?
For example, let's say I have a trace to run. My options are:
Stated another way, if I care about neither long traces nor traces running in parallel, then I can run all traces on one side of the board (like Boss), leaving the other side for a single, continuous ground plane. Or I can keep traces short and minimize/eliminate parallel runs, but my ground planes will literally be a patchwork of vias, and traces will likely run through IC pins.
Thoughts?
- Try to have a large unbroken ground plane (which I interpret to mean the ideal is a single continuous pour that fills the entire area of the board)
- Avoid having traces run in parallel to avoid crosstalk
- Keep traces as short as possible
So my first question: the handful of through-hole analog Boss pedals I've looked at appear to blatantly violate all those rules (thinking specifically of the OD-3). Perhaps I'm missing something obvious, but it looks to me like Roland used single-layer PCBs. All the traces are fairly long, generally pretty fat, and running mostly in parallel to each other. Unless there are inner layers I can't see, they have no real ground plane, just ground traces. And they're not afraid to use jumpers on the component side (presumably as a way to achieve the equivalent of a second layer of traces). And on top of that, they have a not-too-short ribbon cable between the main PCB and the pots (i.e. essentially more long(ish) parallel traces).
Maybe emulating Boss isn't a good idea? I just assumed you could do a lot worse, given their sheer volume and reputation for longevity and reliability.
I've found the easy way to satisfy all the above requirements (and more) is to use a 4-layer board. What I've done is put signal routing on the top and bottom outer layers, and one inner layer for power lines, and another inner layer for an wholly unbroken ground plane. This is of course more costly, and I've been told overkill for analog guitar effect circuits.
But if we move back to two layer boards, it gets virtually impossible to satisfy all three of the above best practices. So I'm wondering which rules take precedence? In other words, we know we can't follow all rules to the letter, so which rule(s) should be the firs to be broken?
For example, let's say I have a trace to run. My options are:
- Keep it short as possible, but it will definitely break up my ground planes, and require ground vias to prevent copper islands from being formed. It might also run through IC pins, or require vias itself (i.e. the trace is partially on the top layer, and partially on the bottom layer).
- Make it substantially longer, possibly even routing around the board perimeter. This keeps the ground plane more contiguous, avoids vias in the trace, and avoids routing through IC pins. But doing this even a few times results in parallel traces (Boss-style), and many traces being longer than otherwise necessary.
Stated another way, if I care about neither long traces nor traces running in parallel, then I can run all traces on one side of the board (like Boss), leaving the other side for a single, continuous ground plane. Or I can keep traces short and minimize/eliminate parallel runs, but my ground planes will literally be a patchwork of vias, and traces will likely run through IC pins.
Thoughts?
Robert
Reverse Engineer
#1 is too generalized, there are situations where you'll break this rule on purpose. Ideally you wouldn't use a ground plane at all and would hand-route the ground traces. You don't want any "islands" where there is a copper pour with no connection to the rest of the circuit.
#2 is good advice but realistically there will be times where you have no choice. What you have to do is become selective about which traces run in close proximity to other traces.
I wouldn't try to emulate Boss at this point. What might look like a random mess is actually a very well planned layout done by a qualified engineer.
#2 is good advice but realistically there will be times where you have no choice. What you have to do is become selective about which traces run in close proximity to other traces.
I wouldn't try to emulate Boss at this point. What might look like a random mess is actually a very well planned layout done by a qualified engineer.
Chuck D. Bones
Circuit Wizard
These "requirements" are goals. Engineering is about balancing conflicting requirements. True, you can't keep every trace away from every other trace, As Robert said, you need to be selective and the way you accomplish that is by knowing which traces are sensitive to noise pickup. A good starting point is to keep input and output away from each other. Of course, we routinely violate that rule right at the stomp switch.
Look at the original BMP boards, they are laid-out like the schematic. Input and output are on opposite sides of the board. The ground rail runs along one side of the board and the power rail along the other.
Single-layer boards are done that way for one reason: cost. Used to be multi-layer and thru-plated holes were tres expensive. Not any more. Low-gain boards like phase-shifters don't need a ground plane. When you make a single-layer board, there's a pretty good chance you'll have to put in jumpers. An extreme case of a single-layer board is a Vero. Traces are in close proximity. Grounds are distributed. When I lay out a Vero, I make an effort to keep input and output away from each other and I am generous with the ground traces. Jumpers are inevitable.
If the circuit demands it, then you make a 4 layer board with one or more continuous ground planes. The only pedal circuits which require that are the purely digital ones.
"Hand-routing" the ground traces makes sense with high power circuits like tube amplifiers because you need to steer the power return currents away from the signal return currents. In a small, low-power circuit like a pedal, a single-point ground where the ground plane is the single point constitutes the best design. Besides providing a common ground path, the ground plane(s) provide isolation between sensitive traces and shielding from pickup of unwanted signals.
When it comes to board layout and routing, you have to be willing to rip stuff up and iterate to get a good solution. The opamps in a dual or quad package are interchangeable. Swap them around as req'd. I see some board layouts where the input opamp and output opamp are in the same package and there are other gain stages in between. I do not recommend that, even if the original production boards are routed that way. True, Boss and Ibanez are huge companies that can afford to hire a team of qualified engineers. They have a ton of money on the line because they build huge quantities. With the boutique pedal builders, it's another story. Some of them know what they're doing. Others do not. There may be one guy designing boards with no one looking over his/her shoulder. We don't know how many board that roll off of their production line end up in the trash.
We do what we can to minimize the risk that the board routing causes problems and has to be done over. I've seen boards where the designer simply piled the parts on with no regard to signal flow and then let the auto-router connect it all together. The auto-router can only follow the rules that the designer imposes and it does not know anything about how the circuit works.
Look at the original BMP boards, they are laid-out like the schematic. Input and output are on opposite sides of the board. The ground rail runs along one side of the board and the power rail along the other.
Single-layer boards are done that way for one reason: cost. Used to be multi-layer and thru-plated holes were tres expensive. Not any more. Low-gain boards like phase-shifters don't need a ground plane. When you make a single-layer board, there's a pretty good chance you'll have to put in jumpers. An extreme case of a single-layer board is a Vero. Traces are in close proximity. Grounds are distributed. When I lay out a Vero, I make an effort to keep input and output away from each other and I am generous with the ground traces. Jumpers are inevitable.
If the circuit demands it, then you make a 4 layer board with one or more continuous ground planes. The only pedal circuits which require that are the purely digital ones.
"Hand-routing" the ground traces makes sense with high power circuits like tube amplifiers because you need to steer the power return currents away from the signal return currents. In a small, low-power circuit like a pedal, a single-point ground where the ground plane is the single point constitutes the best design. Besides providing a common ground path, the ground plane(s) provide isolation between sensitive traces and shielding from pickup of unwanted signals.
When it comes to board layout and routing, you have to be willing to rip stuff up and iterate to get a good solution. The opamps in a dual or quad package are interchangeable. Swap them around as req'd. I see some board layouts where the input opamp and output opamp are in the same package and there are other gain stages in between. I do not recommend that, even if the original production boards are routed that way. True, Boss and Ibanez are huge companies that can afford to hire a team of qualified engineers. They have a ton of money on the line because they build huge quantities. With the boutique pedal builders, it's another story. Some of them know what they're doing. Others do not. There may be one guy designing boards with no one looking over his/her shoulder. We don't know how many board that roll off of their production line end up in the trash.
We do what we can to minimize the risk that the board routing causes problems and has to be done over. I've seen boards where the designer simply piled the parts on with no regard to signal flow and then let the auto-router connect it all together. The auto-router can only follow the rules that the designer imposes and it does not know anything about how the circuit works.
Last edited:
Similar threads
