Be careful where you buy your MCP1700 regulators

The MCP1700 is one of my favourite low drop out regulators for a very low sleep current node. These regulators should only consume around 1uA or so when the processor on a node is put into deep sleep. Total deep sleep current of around 2.0uA is the norm.

I was working on a small node with a DIP version of the ATmega328 and a MCP1700 regulator. The deep sleep current worried me, it was maybe 50uA. I could not work it out. I pulled the node to bits and eventually worked out the problems was the so called MCP1700 regulator.

I had recently bought some ‘MCP1700s’ from that web site that does not offer ‘express’ delivery and perhaps does not sell ‘A’ class products.

I tested the bag of MCP1700s I had recently aquired and these were the queiscent currents, the node was a bare bones ATMega328P, LED, MCP1700 regulator and capacitors;

36uA,15uA,24uA,9uA,9uA,26uA,18uA,11uA,37uA, 17uA,25uA,7uA,13uA.

Ouch not quite to MCP1700 spec which is 1.6uA typical.

So I tested some MCP1700s I had bought from Farnell\Element14, the results of testing these was;

2.2uA,2.1uA, 2.2uA,2.0uA,2.0uA.

Interesting …


Whoa! Thank you very much for your contribution. This saves many others the frustration of searching for the mistake and shows once again that you should be careful where you buy your parts.

Interesting! If one is really targeting this could be worth either putting a regulator on a test coupon with no load (or a known high value resistor) or having a 0-ohm to disconnect the rest of the circuit, so that the performance of the regulator itself can be evaluated apart from anything else that might have a leakage path.

(It’s also a lot simpler to measure if you can just plug it in through a sensitive ammeter, and not need something that can pass enough current for an MCU to start up and then go into sleep mode)

I had problems with MCP1700 also this summer. My board in middle of transmission was restarting all the time. My conclusion was that it drops voltage under BOD level so I solved it with some bigger capacitor. My regulators had 662K code on itself and I saw somewhere that some other regulator use that code and not MCP. So I bought from other seller, they came with other code on itself (CS97 I think), and they worked great even without big capacitor. So, also watch out where are you buying because of this reason.

I measured quiescent current of regulator with Current Ranger (@cslorabox it works as sensitive ammeter). I put only regulator and 1uF capacitor on board without any load and I got around 5.7uA. I was using 3.6V battery as power supply.
According to MCP1700 datasheet in chapter 6.2.1 Power disipation -
"The power dissipation resulting from the quiescent current draw is so low it is insignificant (1.6 µA x VIN). The following equation can be used to calculate the internal power dissipation of the LDO."
I’m not sure how do you get 2uA. Maybe I have some bad regulators, but I tested from 3 different suppliers and got almost the same results. @LoRaTracker how do you measure quiescent current? You are using resistor shunt? What power supply do you use? :smiley:

The data sheet spec says Input Quiescent Current 1.6uA ‘typical’.

So allowing for some small leakage in the capacitors used, and 0.1uA taken by the processor 2uA seems about spot on I would suggest.

With one of two multimeters, either my OWAN B35T or my Fluke 87V.

I would say that they are better quality and more precise :smiley: