Need help with a TLM 102 sized PCB for DIY LDC mic

[MidnightArrakis]

Okay... SO, I'm really not sure what else to do for this board, I feel like there's still problems but I obviously don't have enough knowledge to address them on my own, so I'm sort of at a crossroads.

If anyone would like to, let me know if anything is wrong - and specifically how I can address it if so, I'm not getting any DRC errors and can't figure out what else is going to be problematic, let me know in any case

if there are any other better views or screenshots I can give of the board, lmk!

[If anyone would like to, let me know if anything is wrong - and specifically how I can address it] -- I am getting the impression by looking at the routing of your layout that this PCB is being "auto-routed". Are you using an "Auto-Router"??? I say this because a "normal PCB Designer" just wouldn't route a PCB in the fashion as I see how this board has been routed. Weird!!!

Since you don't have, or at least aren't showing, any component REF DES silkscreen (i.e., C2, R3, etc.), it is difficult to specifically address where your errors are in this layout, let alone how to address on fixing them. But, I'll try.....

>> NOTE: Place your brain into a "mentally-envisioning" mode before proceeding here!!!

In the image below, imagine moving the "large" component that is located down in the lower-left had corner over to the right and "over" the letters -- A R -- in "PARK". By doing so, its top pad will be able to be routed directly up to "OPA-102", its bottom pad will be shorter to the "horizontal" component and the "small" component directly above the -- P -- in "PARK" will be able to be routed directly to XLR-2 -- WITHOUT -- having to be routed in-between the pads of the lower-right hand component.



>> NOTE: If you would use and place component -- REF DES -- silkscreen on your PCB-layout, I *COULD* have written what I wrote above like THIS instead..... >> In the image below, imagine moving R4 over to the right and "over" the letters -- A R -- in "PARK". By doing so, its top pad will be able to be routed directly up to U1, its bottom pad will be shorter going to C7 and R9 would be able to be routed directly to XLR-2 -- WITHOUT -- having to be routed in-between the pads of R4. -- Isn't that easier?

Frankly.....I would suggest that you redesign your schematic and PCB-layout using KiCAD and then send me your KiCAD files. That way, I can help you out much more easily.....as your layout does have multitudes of different issues. Like what??? Like.....1) your tracks being too close to component pads, 2) your component outline silkscreens having different "line-weights, 3) The track entrance/exit to component pads looks to me as being rather "awkward". And, 4) I see that you have 2 different PCB-layouts, which one are you actually using?

>> Is this circuit being powered by "Phantom-Power"? It looks to me like it is. Is it???

/

 

[edenooo]

I see you're very stubbornly refusing to read, and/or comprehend and/or accept my tips. Sorry to be so blunt, but that board is a bit of a dog's breakfast, as they say... And you can't say I didn't try to help

And that's even before the high-impedance concerns that @rogs is referring to...

It's not that I'm not listening, I just don't know what I'm being asked to do, sorry if I'm coming off as stubborn, I'm really trying to listen and learn.

 

[edenooo]

To be brutally honest, trying to do a decent PCB layout for an LDC condenser mic is not really a very good choice for a first project!

The input impedance required for interfacing a true condenser capsule is incredibly high..... In the order of 1000 million Ohms!
That kind of impedance is rarely encountered elsewhere in electronics, and to get it to work well requires some special attention.
It's really not a good 'starter' project.

If your layout now matches your corrected circuit, it should work .... but for how long, it would be difficult to say. The track spacings around R1 and R9 are likely to give problems, with noise from dirt and flux contamination, within a fairly short period.

The highlighted part of the attached image shows how Rode address the high impedance part of their circuitry. The highlighted flat white resistor, with what looks like a stripe, is the IG resistor. Note that it is surrounded by a pcb track which passes through it's terminals.
That's known as a 'guard ring', and is included to surround the input signal with a much lower impedance signal path, to minimise 'leakage' problems.
But it's a specialised subject (just google 'pcb guard ring' to get some ideas on the subject )

The alternative to that is to stand the appropriate terminals of the 1G resistors off the board (as I described in my earlier posts). That technique is adopted by many commercial mic manufacturers.

Don't want to seem negative, but continuing with the layout as you have it may not give you very good results.....

View attachment 147410

Thanks for being honest about it, I know (and acknowledged in my initial post) that this was probably going to be a pretty difficult first project, so I don't take absolutely any offense to your honesty.

I wanted to take your advice about the 1GOhm resistors, there just genuinely aren't any other 1GOhm resistors in stock, I'll wait until there are any, but I guess these aren't exactly the most common.

and I'm not married to the layout I have at all, I just want whatever I end up with to work and fit within the dimensions I need it to.

 

[edenooo]

[If anyone would like to, let me know if anything is wrong - and specifically how I can address it] -- I am getting the impression by looking at the routing of your layout that this PCB is being "auto-routed". Are you using an "Auto-Router"??? I say this because a "normal PCB Designer" just wouldn't route a PCB in the fashion as I see how this board has been routed. Weird!!!

Since you don't have, or at least aren't showing, any component REF DES silkscreen (i.e., C2, R3, etc.), it is difficult to specifically address where your errors are in this layout, let alone how to address on fixing them. But, I'll try.....

>> NOTE: Place your brain into a "mentally-envisioning" mode before proceeding here!!!

In the image below, imagine moving the "large" component that is located down in the lower-left had corner over to the right and "over" the letters -- A R -- in "PARK". By doing so, its top pad will be able to be routed directly up to "OPA-102", its bottom pad will be shorter to the "horizontal" component and the "small" component directly above the -- P -- in "PARK" will be able to be routed directly to XLR-2 -- WITHOUT -- having to be routed in-between the pads of the lower-right hand component.

View attachment 147409

>> NOTE: If you would use and place component -- REF DES -- silkscreen on your PCB-layout, I *COULD* have written what I wrote above like THIS instead..... >> In the image below, imagine moving R4 over to the right and "over" the letters -- A R -- in "PARK". By doing so, its top pad will be able to be routed directly up to U1, its bottom pad will be shorter going to C7 and R9 would be able to be routed directly to XLR-2 -- WITHOUT -- having to be routed in-between the pads of R4. -- Isn't that easier?

Frankly.....I would suggest that you redesign your schematic and PCB-layout using KiCAD and then send me your KiCAD files. That way, I can help you out much more easily.....as your layout does have multitudes of different issues. Like what??? Like.....1) your tracks being too close to component pads, 2) your component outline silkscreens having different "line-weights, 3) The track entrance/exit to component pads looks to me as being rather "awkward". And, 4) I see that you have 2 different PCB-layouts, which one are you actually using?

>> Is this circuit being powered by "Phantom-Power"? It looks to me like it is. Is it???

/

I got KiCAD and am actively trying to use it instead of EasyEDA, I'll get back to you once I've learned where everything actually is in the software.

What you said regarding positioning the components, I realize now how poor the locations of everything in both my layouts are, I understand why they were bad, I'll use what I learned.

And regarding which layout I'm using - neither of the layouts I made are ideal... so I hopefully won't be using either of them, I'm still trying to learn how all of this is supposed to work

And yes, it's powered by Phantom Power

 

[edenooo]

To be brutally honest, trying to do a decent PCB layout for an LDC condenser mic is not really a very good choice for a first project!

The input impedance required for interfacing a true condenser capsule is incredibly high..... In the order of 1000 million Ohms!
That kind of impedance is rarely encountered elsewhere in electronics, and to get it to work well requires some special attention.
It's really not a good 'starter' project.

If your layout now matches your corrected circuit, it should work .... but for how long, it would be difficult to say. The track spacings around R1 and R9 are likely to give problems, with noise from dirt and flux contamination, within a fairly short period.

The highlighted part of the attached image shows how Rode address the high impedance part of their circuitry. The highlighted flat white resistor, with what looks like a stripe, is the IG resistor. Note that it is surrounded by a pcb track which passes through it's terminals.
That's known as a 'guard ring', and is included to surround the input signal with a much lower impedance signal path, to minimise 'leakage' problems.
But it's a specialised subject (just google 'pcb guard ring' to get some ideas on the subject )

The alternative to that is to stand the appropriate terminals of the 1G resistors off the board (as I described in my earlier posts). That technique is adopted by many commercial mic manufacturers.

Don't want to seem negative, but continuing with the layout as you have it may not give you very good results.....

View attachment 147410

I've replaced the SMT 1GΩ Resistors for R1 and R9 with THT ones, those should hopefully not be as problematic in terms of impedance? I've also redone all the tracks, and insured there are no 90 degree turns for any routing, and have made the overall position of things much better I THINK.

it's still not there yet, but I think it's an improvement

 

[MidnightArrakis]

I got KiCAD and am actively trying to use it instead of EasyEDA, I'll get back to you once I've learned where everything actually is in the software.

[I got KiCAD and am actively trying to use it instead of EasyEDA] -- GREAT!!! That's good!!! -- NOW!!!.....don't ever tell anyone that I've never given you anything, OK??? The link that you see immediately down below goes to a folder that I have on an online web-storage site. This folder contains just over 8GB of -- KiCAD 6.0 Training Videos -- that I bought when I first got into KiCAD. I would suggest that you make a folder under your main KiCAD program installation folder and simply title it to be -- KiCAD Training Videos -- and then download this entire folder into it. From there, you can go through these training videos at your own pace and properly learn how to use both the KiCAD schematic and PCB-design programs. It'll do you good!!!

>> NOTE: Even though KiCAD is at Release 9.0 now and these training videos are for KiCAD Release 6.0, the essence of how the program works and operates still remains the same. By going through these training videos, you will learn what's inside each of the pull-down menus, how to use the schematic program, how to transfer the schematic data over to the PCB, how to use the PCB program and how to generate the GERBER and N/C Drill data output files for fabrication. You will be better-off by taking your time and learning how to properly use KiCAD, than you will be trying to stumble through it on your own!!! Meanwhile and in addition, you will also be learning the PCB-design lingo and terminology, as well as what to do and what NOT to do when designing a PCB.

Here's my personal "KiCAD Training Videos" folder link:

https://mega.nz/folder/MmJFlA6Q#tN2tZ9fY5M8d5CJyg2WauQ

An-n-n-n-n-n-nd.....here's a link to download and install a rather neat high-resolution -- FREE -- video-player program to watch the "KiCAD Training Videos" on:

https://www.5kplayer.com/#5KPlayer

Let me know how well everything downloaded and worked out, OK???

>> This type of routing is a -- DEFINITE NO-NO!!! -- and should be
avoided at all costs!!! This type of routing creates "acid-traps" which
can capture etching solution to be trapped underneath the copper
tracks, potentially causing the tracks to peel-away from the laminate:


>> Same type of problem, only different:


/

 

[edenooo]

>> This type of routing is a -- DEFINITE NO-NO!!! -- and should be
avoided at all costs!!! This type of routing creates "acid-traps" which
can capture etching solution to be trapped underneath the copper
tracks, potentially causing the tracks to peel-away from the laminate:
View attachment 147477

>> Same type of problem, only different:
View attachment 147478

I'm using JLCPCB, which uses a more modern standard for PCB manufacturing that is photosensitive, PCBWay and the other services like these also use this process I believe, but this means it doesn't have the same issues with acid-traps as more "vintage" styles of PCB manufacturing had. I'm still going to address these corners regardless, but as far as I know acid-traps are mostly a problem of the past, isn't technology amazing? (I'm still going to address these corners regardless, but I thought I'd mention it)

Other than that, do you notice anything worrisome..?

 

[MidnightArrakis]

I'm using JLCPCB, which uses a more modern standard for PCB manufacturing that is photosensitive, PCBWay and the other services like these also use this process I believe, but this means it doesn't have the same issues with acid-traps as more "vintage" styles of PCB manufacturing had. I'm still going to address these corners regardless, but as far as I know acid-traps are mostly a problem of the past, isn't technology amazing? (I'm still going to address these corners regardless, but I thought I'd mention it)

Other than that, do you notice anything worrisome..?

[it doesn't have the same issues with acid-traps as more "vintage" styles of PCB manufacturing had] -- OK, but.....

Why Acid Traps Are a Problem

https://gesrepair.com/pcb-acid-trap-causes/

https://www.pcbcart.com/article/content/commonly-seen-PCB-design-issues.html

-- Your earlier layouts had "Issue #3" and "Issue #5" listed here:

https://www.mclpcb.com/pcb-guide/

>> And, just in case you would ever want to try "some other" PCB design program:

https://resources.pcb.cadence.com/blog/2025-how-to-download-orcad-for-free

JBW

/

 

[Itsme]

Yet another suggestion: when using an opamp it won't hurt to add a feedback divider to the opamp to have the option for a higher gain than just one. Having a fairly low polarization voltage and single ended Output, more gain would be beneficial: Single ended, not coaxial transmission would pick up significantly more noise, which together with a low signal level is a recipe for disappointment... The actual low signal level would also emphasize the input noise of the receiving input amp.

Cheers
Andreas

 

[MidnightArrakis]

I've replaced the SMT 1GΩ Resistors for R1 and R9 with THT ones, those should hopefully not be as problematic in terms of impedance? I've also redone all the tracks, and insured there are no 90 degree turns for any routing, and have made the overall position of things much better I THINK.

it's still not there yet, but I think it's an improvement

[I think it's an improvement] -- How's this??? Did I manage to get everything placed and connected up correctly, even though it is almost completely different than yours? This schematic was created using KiCAD 9.0.


(HINT: -- REMEMBER -- ....."capacitors FIRST"!!!).....

/

 

[rogs]

[I think it's an improvement] -- How's this??? Did I manage to get everything placed and connected up correctly, even though it is almost completely different than yours? This schematic was created using KiCAD 9.0.

View attachment 147491
(HINT: -- REMEMBER -- ....."capacitors FIRST"!!!).....

/

Schematic looks OK - apart from one major error ..... remove R6. (Join C7 negative directly to R5 )

 

[Itsme]

More suggestions: Put 2.2uF between R7 and R8 and you can omit R9 and C8. R7 and R8 could also be larger for a lower cut-off frequency, if necessary..
Hooking the Opamp supply to the phantom power the way you did, will blow the poor guy into oblivion... Reason: recommended is +/- 18V or 36 single supply with an abs. max. rating of 40V. with 5kOhms to the phantom power you will exceed the 40V!
Below an alternative for R3, R4 and C3, C4. Saves another electrolytic to gain space and reliability:




Cheers
Andreas

 

[rogs]

More suggestions: Put 2.2uF between R7 and R8 and you can omit R9 and C8. R7 and R8 could also be larger for a lower cut-off frequency, if necessary..
Hooking the Opamp supply to the phantom power the way you did, will blow the poor guy into oblivion... Reason: recommended is +/- 18V or 36 single supply with an abs. max. rating of 40V. with 5kOhms to the phantom power you will exceed the 40V!
Below an alternative for R3, R4 and C3, C4. Saves another electrolytic to gain space and reliability:

View attachment 147498


Cheers
Andreas

I think you may have missed the point on a couple of things?..... C8 and R9 are fitted to provide a passive impedance balance for the usused 'cold' side...
Yes you can remove them, but you will unbalance the line. Not a good idea.

The feed resistors for the op-amp supply are selected to ensure the max supply to the op-amp cannot be exceeded.
Remember that the 48v phantom power feeds are via the series 6.81K phantom power resistors fitted within the connected pre-amp.
Each line provides about 1mA to the system, via a 16.81k resistance (10k +6.81k).
Therefore there is a minimum voltage drop of 16.8v to the op-amp supply pin: 48v - 16.8v = 31.2v - which is within the opamp spec.
In reality it's usually a bit more of a voltage drop - phantom power supplies do not have an infinitely low impedance.

(Some more details about the original circuit on page 3 of the project notes HERE )

 

[Itsme]

Each line provides about 1mA to the system, via a 16.81k resistance (10k +6.81k).

Oh thanks, I missed that one... I probably still wouldn't rely on the resistance alone and add a 20V Zener in parallel to the opamp just for peace of mind .
In this example the output is imbalanced anyway... I wouldn't expect any harm in this case - what am I missing?

 

[Khron]

In this example the output is imbalanced anyway... I wouldn't expect any harm in this case - what am I missing?

Rookie mistake...

https://sound-au.com/balance.htm#intro2

Firstly, it is commonly (but very much mistakenly) assumed that signal balance is important, but this is not correct. In reality, it doesn't matter at all if a balanced line has all its signal on one lead, and none on the other. What is important is impedance, and the signal leads of balanced lines must have the exact same value of impedance to earth/ground/common at all frequencies of interest. Ideally, this should be as high as possible for both the send and receive circuits.

 

[Itsme]

ok - face palm... it's about the receiving side.

 

[rogs]

Oh thanks, I missed that one... I probably still wouldn't rely on the resistance alone and add a 20V Zener in parallel to the opamp just for peace of mind .
In this example the output is imbalanced anyway... I wouldn't expect any harm in this case - what am I missing?

You're making the common mistake that a 'balanced line' must have audio on both lines. A balanced line should have the same impedance applied to both lines, but it doesn't care whether there is signal on one line, both lines - or no signal at all ! - the line stays balanced.

The fitting of an uncecessary Zener diode in this application will have 2 effects: It will limit headroom, and it might become a potential source of noise.

The idea of using single sided audio on balanced lines is quite common. Rode use it, as do Neumann - and I suspect many others.
It reduces the component count and keeps the noise to a minimum.
Adding a differential audio line driver to the second output will add 6dB to the output signal.. but will also add more noise as well... and draw more current form the phantom supply, further reducing the headroom.

 

[Itsme]

The fitting of an uncecessary Zener diode in this application will have 2 effects: It will limit headroom, and it might become a potential source of noise.

In this case I would expect the Zener would help (given, a bypass cap of about 1uF is added in parallel): Opamps have limited PSRR (Power Supply Rejection Ratio) which is important in this case as the opamp has to drive a (fairly-) low impedance which would modulate the supply rail. As the PSRR becomes lower with higher frequencies the higher frequencies would potentially become more distorted in an unpleasant way. It also may cause oscillations under some circumstances. The loading of the signal lines is symmetrical (2x10k) and well decoupled in this circuit thus very low impact on the noise level.
The output swing of the OPA1641 is only 1V less that the supply voltage which gives an output swing of 19V if a 20V Zener is used...

 

[Khron]

In this case I would expect the Zener would help (given, a bypass cap of about 1uF is added in parallel)

Just too bad zeners' impedance is way lower than capacitors', so you'd have to go up to ridiculous (or at least impractical) capacitor values before they put a dent in the zener noise.

 

[rogs]

The output swing of the OPA1641 is only 1V less that the supply voltage which gives an output swing of 19V if a 20V Zener is used...

No, the OPA1641 has a 'rail ro rail' output swing -- see here: https://www.ti.com/lit/ds/symlink/opa1642.pdf

At this kind of output swing most mic pre-amps will have overloaded way earlier!