Re: LM5066 low- and high-current limit scaling

[Guenter Roeck <linux@roeck-us.net>]

On Fri, Jan 25, 2019 at 03:13:39PM -0500, Matt Wilbur wrote:
> Hi
> 
> First off, this is the first time I've ever tried something like this,
> so I fully anticipate I will make some etiquette / process mistakes.
> 
> In a project I am working on, we make use of the LM5066 hot-swap
> controller.  I believe the kernel module for this chip
> (drivers/hwmon/pmbus/lm25066.c) has an error where the scaling
> parameters for the low- and high-current limits are swapped.
> 
> What's the best way for me to make my case?  Just create a patch and
> try to submit it?  A free-form discussion on the mailing list?
> 

Normally you would send a patch and make your case in the description.

However, be careful.

PSC_CURRENT_IN_L is used if DEVICE_SETUP[4]=1, which (hopefully) translates
to CL=VDD. Per LM25066 datasheet, table 41:
	CL=GND:		M=13661, B=–5200, R=-2
	CL=VDD:		M=6854, B=-3100, R=-2

Current code:
	DEVICE_SETUP[4]=0: M=13661, B=0, R=-2
	DEVICE_SETUP[4]=1: M=6852, B=0, R=-2

This looks mostly correct to me, except for the offset and the
slight difference if CL=VDD. Looks like that was changed at some
point in later datasheet versions.

Overall, the problem (or confusion) is that the logic for
DEVICE_SETUP[4] in the datasheets for the various chips changes.

5066i:
	0 = high setting (50 mV)
	1 = low setting (26 mV)

25066, 25066i:
	0 = low setting (25 mV)
	1 = high setting (46 mV)

... so of course it may well be that the data for some of the chips
is wrong, since we use PSC_CURRENT_IN_L for DEVICE_SETUP[4]=1
throughout (in other words, it is likely that either the data for
LM5066 is wrong, or that the data for LM25066 is wrong - the only
question is which one). It may also be that the datasheets are wrong
or misleading, and that CL=VDD is not (or not always) equivalent
to DEVICE_SETUP[4]=1.

Guenter