3.0 Nano clone no decoupling caps?

I bought a bunch of Nano 3.0 clones off eBay to make some infrared temperature sensors using the MLX9061 and SSD1306 OLED display over i2c. They work fine running off usb, however i noticed their operation is highly unstable when running off a 12v power supply. The clones use the AMS1117 regulator, however i noticed there are no decoupling capacitors at all on the board. I hooked it up to my scope, and discovered a whopping 250mV of ripple on the 5v rail. I then bypassed the stock regulator using a Lm7805 with decoupling caps and everything runs smoothly now.

Was wondering if anyone else experienced the same problem and if there was a easy fix for this? Cause i dont feel like adding Lm7805's to all of them (i bought a dozen of them to make). I figured adding my own 1206 smd caps across the legs of the stock AMS1117 regulator would help?

Thanks!
Mike

The Nano clones i purchased:
http://www.ebay.com/itm/131505652236?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

No caps on the underside? If not then these are worse than junk, they are timebombs that
may work intermittently and fail more or less at random.

Yes adding your own surface mount caps will fix it. Both to the regulator and the processor chip.

1,2,3...7 I count 7 capacitors on bottom side.
I think the AMS1117 has two capacitors.

Those capacitors might be very bad quality. A ripple of 250mV is indeed very large. That is not normal. Does that ripple still exist when nothing else is connected to the Arduino Nano board ? Perhaps the ripple only occurs if the AMS1117 is overheating and shutting down.

What happens if you supply 7.5...9V instead of 12V. With 12V the AMS1117 gets hot very quick.
Do you power the OLED display with the 5V pin ? How much current does it use ?
Could you do some measurements, to determine the current through the AMS1117.

(while I was typing this, MarkT already wrote the same about the capacitors)

While they might be caps are they of the right value?

Peter_n:
1,2,3...7 I count 7 capacitors on bottom side.
I think the AMS1117 has two capacitors.

By the way - when quoting items from eBay, remove the question mark and all characters following. They only refer to your present eBay session.

Paul__B:

By the way - when quoting items from eBay, remove the question mark and all characters following. They only refer to your present eBay session.

Whoops yeah Paul forgot to log out of eBay before i copied the link.

But anyways, looking at the data sheet for the AMS1117 it only has the suggested caps for the adjustable version , these have the fixed 5v version. I assume the cap values would be the same as a lm7805 where its 0.33uf on the input to ground, and 0.1uf on the output to ground. I will test it out later.

And if you look on the top leg of the AMS1117 pictured, it goes right to the VIN pin, no caps at all.

Peter_n:
1,2,3...7 I count 7 capacitors on bottom side.
I think the AMS1117 has two capacitors.

Those capacitors might be very bad quality. A ripple of 250mV is indeed very large. That is not normal. Does that ripple still exist when nothing else is connected to the Arduino Nano board ? Perhaps the ripple only occurs if the AMS1117 is overheating and shutting down.

What happens if you supply 7.5...9V instead of 12V. With 12V the AMS1117 gets hot very quick.
Do you power the OLED display with the 5V pin ? How much current does it use ?
Could you do some measurements, to determine the current through the AMS1117.

(while I was typing this, MarkT already wrote the same about the capacitors)

No the ripple is down to 50mV when nothing is connected. But the OLED on draws 7 milliwatts, and MLX only draws a few milliamps. Originally i planned on running everything on the 3.3v rail, but for some reason when ever i had both connected everything wouldn't turn on at all, even the Nano would fail to boot up. Turns out instead of having an actual 3.3v regulator, they just ran the 3.3v output of the CH3406 USB driver to the 3.3v pin. Couldn't supply enough current to both the OLED and MLX. Ill take some measurements later.

But the OLED on draws 7 milliwatts,

No it draws a current not a power.

Grumpy_Mike:
No it draws a current not a power.

Grumpy_Mike:
No it draws a current not a power.

typo haha. The OLED draws 40mA.

Yeah, that is probably too much. You will need a separate regulator.

OK guys.

Finally got around to looking at both the schematic of the (genuine) Nano and my (presumably clone) board (using an "FTDI" chip rather than a CH340).

What do you know?

There simply is no capacitor on the VIN terminal of the AMS1117-5.0 in any design. The seven capacitors are:

  • 10µF (original design was 4.7µF) plus 0.1µF on VUSB.
  • 10µF (original design was 4.7µF) plus 0.1µF on +5V.
  • 0.1µF on AREF.
  • 0.1µF on 3.3V.
  • 0.1µF in DTR line.

Since VUSB only connects to +5V via the diode, this would appear to represent a major design blunder in this board (and as usual, the blunder was mindlessly followed in the clones!), as the decoupling on VUSB is entirely useless. Either the intent was to power the USB chip from this point - which should have been the case as there is no point in powering it in the absence of a USB connection - or that decoupling should go on VIN.

You are right. Perhaps 100nF at Vusb to filter spikes would make some sense. But the 4.7uF at Vusb is useless and Vin has no capacitor. I have never looked at it this way.

That is really stupid! Gravitech seems to have messed up. :o

There isn't even an AMS1117-5.0 on the original schematic... ???

Fortunately the output capacitor (5volt rail) is more important than the input capacitor.
From my experience, a missing/wrong output cap is what makes them oscillate.
No input cap is bad design practice.
You could add an external input cap close to the board.
Leo..