thunderaxe
Active member
over spec by how much? and how stable would that make them?
Getting Started with PCBA at JLCPCB
- Thread starter z2amiller
- Start date
- Featured
z2amiller
Active member
I think it matters less for AC (signal), but for DC there's a pretty strong de-rating curve - the capacitance gets less as the voltage gets higher.
The package size is actually relevant here as well, with physically larger capacitors losing less capacitance with a DC offset.
I use X7R 1206 50V 1uF's (basic: C1848) when I do use 1uF ceramics in the audio path.
Murata has an online tool called 'SimSurfing' where you can look up the exact capacitance derating curve for any of their caps.
For example, that 1uF 50V 1206 capacitor is only 500nF when it's at ~40V DC bias. I imagine that the derating curve is probably
at least similar for other manufacturers with the same dielectric/size/voltage rating.
Since the caps in the audio path would usually be biased at 1/2VCC (4.5V or 9V if you're running 18V) I don't think it matters
much for this cap though -- it is stable at ~1uf until ~11V.
But if you were to try to use an 0603 1uF 25V capacitor there, it would be ~750nF at 4.5V and ~500nF at 9V!
(Sorry I'm on my work computer at the moment so I can't upload the screenshots but #trust)
X7R's are also pretty stable temperature-wise, ±15% over their whole temperature range, which is like -55C -> 85 or 125C.
I imagine at the human-tolerable temperatures in that range the difference will be pretty negligible though maybe
if your pedal consumes a ton of power and heats up to 45C or something it could drift by a measurable percent.
(but audio-perceptible? eh)
The package size is actually relevant here as well, with physically larger capacitors losing less capacitance with a DC offset.
I use X7R 1206 50V 1uF's (basic: C1848) when I do use 1uF ceramics in the audio path.
Murata has an online tool called 'SimSurfing' where you can look up the exact capacitance derating curve for any of their caps.
For example, that 1uF 50V 1206 capacitor is only 500nF when it's at ~40V DC bias. I imagine that the derating curve is probably
at least similar for other manufacturers with the same dielectric/size/voltage rating.
Since the caps in the audio path would usually be biased at 1/2VCC (4.5V or 9V if you're running 18V) I don't think it matters
much for this cap though -- it is stable at ~1uf until ~11V.
But if you were to try to use an 0603 1uF 25V capacitor there, it would be ~750nF at 4.5V and ~500nF at 9V!
(Sorry I'm on my work computer at the moment so I can't upload the screenshots but #trust)
X7R's are also pretty stable temperature-wise, ±15% over their whole temperature range, which is like -55C -> 85 or 125C.
I imagine at the human-tolerable temperatures in that range the difference will be pretty negligible though maybe
if your pedal consumes a ton of power and heats up to 45C or something it could drift by a measurable percent.
(but audio-perceptible? eh)
Rpschultz13
Well-known member
I asked this over on AllAboutCircuits. Great thread, especially the Cyril Bateman articles. Also the TI article in post #2 is fantastic.
thunderaxe
Active member
i've got another question: how can i be certain i have the right footprint for the PCBA part? i wish that was a feature of the bouni JLCPCB plugin.
for instance, for a reverse-polarity protection diode, there are no 1n5817s available in basic, but there is this 1n5819 which as far as i can tell is essentially the same thing:
jlcpcb.com
okay, so it's a SOD-323 package, and i take a glance at the easyEDA footprint. should be enough to go on, right? except that there are three different SOD-323 footprints in KiCad, and none of them look like the easyEDA one. the hand-soldered one looks close, but it's not quite right either. worse still, the recommended footprint on the datasheet is different from both the easyEDA footprint and any of the ones in KiCad. i ended up creating my own, but i don't want to have to do that too much. i just want to be certain that i have the right one to make sure the assembly is done correctly.
for instance, for a reverse-polarity protection diode, there are no 1n5817s available in basic, but there is this 1n5819 which as far as i can tell is essentially the same thing:
1N5819WS | Guangdong Hottech | Schottky Diodes | JLCPCB
1N5819WS from Guangdong Hottech - Schottky Diodes is available for JLCPCB assembly, check the stock, pricing and datasheet, and let JLCPCB helps you assemble the part 1N5819WS for free.
okay, so it's a SOD-323 package, and i take a glance at the easyEDA footprint. should be enough to go on, right? except that there are three different SOD-323 footprints in KiCad, and none of them look like the easyEDA one. the hand-soldered one looks close, but it's not quite right either. worse still, the recommended footprint on the datasheet is different from both the easyEDA footprint and any of the ones in KiCad. i ended up creating my own, but i don't want to have to do that too much. i just want to be certain that i have the right one to make sure the assembly is done correctly.
vigilante398
Authorized Vendor
Your SMD footprint doesn't have to 100% line up with the one on EasyEDA, the part just has to fit, oversizing so you have a hand-soldering option isn't a problem. I usually make my own footprints for things, but Mouser and DigiKey have an ever-growing number of footprints for every software out there, and they even offer 3D models. I've never used KiCAD so I don't know what the process for loading a footprint in from a file looks like, but their footprints are pretty reliable.i've got another question: how can i be certain i have the right footprint for the PCBA part? i wish that was a feature of the bouni JLCPCB plugin.
for instance, for a reverse-polarity protection diode, there are no 1n5817s available in basic, but there is this 1n5819 which as far as i can tell is essentially the same thing:
1N5819WS | Guangdong Hottech | Schottky Diodes | JLCPCB
1N5819WS from Guangdong Hottech - Schottky Diodes is available for JLCPCB assembly, check the stock, pricing and datasheet, and let JLCPCB helps you assemble the part 1N5819WS for free.
okay, so it's a SOD-323 package, and i take a glance at the easyEDA footprint. should be enough to go on, right? except that there are three different SOD-323 footprints in KiCad, and none of them look like the easyEDA one. the hand-soldered one looks close, but it's not quite right either. worse still, the recommended footprint on the datasheet is different from both the easyEDA footprint and any of the ones in KiCad. i ended up creating my own, but i don't want to have to do that too much. i just want to be certain that i have the right one to make sure the assembly is done correctly.
If you're making your own footprint (which was the only option in the early days) you typically rely on the datasheet from the part which will give you a recommended footprint. I would usually try to find a 3D model from GrabCAD to drop onto the footprint to make sure the part would land the way I expected.
As a final check, during checkout at JLC they have a visualizer where they'll drop their 3D models onto a rendering of your board, and that makes it pretty easy to see what will fit and what won't. If you order something that won't fit, they'll email you and tell you, and you can either cancel the order and try again or select a different component that may fit better.
EDIT: also if you're looking for an SMD version of 1N5817, try B5817.