search

keirf / greaseweazle

Tools for accessing a floppy drive at the raw flux level
The Unlicense
949 stars 93 forks source link

AMS1117 on V4 PCB is not ceramic stable #129

Open tdaede opened 2 years ago

tdaede commented 2 years ago

There's no github for the V4 PCB so I'm filing the issue here. The regulator on the V4, the AMS1117, is attached to a ceramic bulk cap directly, but it's not rated for ceramic stability. I haven't noticed any problems yet but it could easily cause transient problems. It's fixable by either putting a resistor in series with the output cap or switching to a ceramic stable regulator.

keirf commented 2 years ago

Good point. We borrowed this part from the Gotek schematic but I think that does use an electrolytic at the output. If I make another batch I will have to fix this. However I currently have about 900 so it won't be soon. Luckily I haven't heard anything attributable to instability. Presumably this would be triggered by load transients at the reg output? And that is fairly unlikely in the Greaseweazle design (after startup)

tdaede commented 2 years ago

It's more likely to happen at light load. I haven't observed it personally so you're probably OK, but it can vary depending on batch and temperature so it's good to fix for the future. I just observed it when I was looking at adding trimode support.

keirf commented 2 years ago

Interested question: What about all the smaller ceramics on the 3v3 output rail? Do they not matter so much? Due to size? Distance?

tdaede commented 2 years ago

Distance, mostly, but they can in theory cause problems too. See section 7 of https://www.ti.com/lit/pdf/snva167

I've never seen bypass capacitors cause a problem myself, but also most designs with a high density of ceramic caps I've worked on also go with switching regulators or ceramic stable LDOs.

keirf commented 1 year ago

Thinking about a rev of the GW V4 board. What would be the easiest fix here? A series resistor I'm guessing. Do you have a value in mind that would work well? Is this solution as good as replacing with an electrolytic? And any recommendations on exact placement schematically or in physical placement? Thanks!

tdaede commented 1 year ago

Yeah, using a series resistor works. The exact value depends on the 1117 variant, but eg. the lm1117 wants between 0.3 and 22 ohms. A 0.47 or 1 ohm resistor is probably good. The output cap should be placed as close to the regulator as possible for low inductance.

Using a resistor is arguably better than an electrolytic because it won't leak or wear out. Only other thing to watch out for is that ceramic capacitors have lower capacitance with higher DC bias voltages. Something to watch for if you're near the max voltage rating of your cap.

On Thu, Oct 27, 2022 at 4:50 AM Keir Fraser @.***> wrote:

Thinking about a rev of the GW V4 board. What would be the easiest fix here? A series resistor I'm guessing. Do you have a value in mind that would work well? Is this solution as good as replacing with an electrolytic? And any recommendations on exact placement schematically or in physical placement? Thanks!

— Reply to this email directly, view it on GitHub https://github.com/keirf/greaseweazle/issues/129#issuecomment-1293409620, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJ7NGTFML33KX5MOXJPVA3WFJT67ANCNFSM5H4BPB5A . You are receiving this because you authored the thread.Message ID: @.***>

keirf commented 1 year ago

Thanks. I assume there's no reason to prefer a particular capacitor pin for the resistor (output side vs ground side)?

tdaede commented 1 year ago

Nope, it doesn't matter.

(off topic, did you see my email about pc88 support? if not I can make it in the form of an issue instead)

On Thu, Oct 27, 2022 at 5:21 AM Keir Fraser @.***> wrote:

Thanks. I assume there's no reason to prefer a particular capacitor pin for the resistor (output side vs ground side)?

— Reply to this email directly, view it on GitHub https://github.com/keirf/greaseweazle/issues/129#issuecomment-1293444013, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJ7NGQA32U2JJPXZ2VLNPTWFJXTJANCNFSM5H4BPB5A . You are receiving this because you authored the thread.Message ID: @.***>

keirf commented 1 year ago

I don't recall your email. Always better as an issue so it can be tracked.

kaunomedis commented 1 year ago

I bough mine PCB on ebay. Noticed some unstability, added electrolytic caps on both sides of the 1117. I am using 1117 for my projects and I noticed, that there are very cheap clones with stability issues, some even have even worse problems (like shorting when reversed). The "original" are much more stable. If someone is designing new PCB, it cost nothing to add place for small electrolytic cap. I even suggest polymer ones. There is still some problems with my PCB, but I can blame very wearied out USB connectors.