[time-nuts] 5 Digit Panel meters

[Bob kb8tq]

Hi

Update rate on the specific part mentioned earlier is five a second. Assuming that is correct, it would 
rule out them using the MCP3421 at 18 bits. It will not go that fast at that resolution. It also seems odd that
they would have a +/- capable part with built in ability to auto range, and not utilize those features.

Bob

> On Sep 7, 2020, at 8:24 PM, Robert LaJeunesse <lajeunesse at mail.com> wrote:
> 
> The similar meters I have are based on the Microchip MCP3421, a tiny 6-pin SOT 18 bit ADC with internal 2.048V 15 ppm (typ) reference, differential inputs, and a PGA. Input span is +/-2.048V at minimum gain to +/-0.256V with a PGA gain of 8. So at max gain 1 LSB is about 2uV, although I'd be surprised if it was that quiet (the spec at PGA=1 is 1.5uV typical, no max given). Initial accuracy is not great at 0.35% max, but that can be trimmed out externally or via software. For an 18 bit ADC with a decent reference it's actually a rather good deal at under US $2 (qty 25+). Just don't expect it to be an accurate 18 bits over a wide temp range, references like that don't sell so cheap.
> 
> Bob L.
> 
> ref: https://ww1.microchip.com/downloads/en/DeviceDoc/22003e.pdf
> 
>> Sent: Monday, September 07, 2020 at 1:15 PM
>> From: "Bob kb8tq" <kb8tq at n1k.org>
>> To: "Discussion of precise time and frequency measurement" <time-nuts at lists.febo.com>
>> Subject: Re: [time-nuts] 5 Digit Panel meters
>> 
>> Hi
>> 
>> So, *assuming* there is no other ADC on the board ….
>> 
>> DC performance of the STM8 ADC’s is somewhere in the 8 to 9 bit range as far
>> as ENOB with significant averaging. For the sake of simplicity, let’s say they *do*
>> make it to 10 bits. 
>> 
>> But … do we get the 10 bits?
>> 
>> The range on the meter is 33V. The reference to the DAC is (likely) 5V. You just lost
>> nearly 3 bits from that scale mismatch.  At the 5V level, you would be running just over
>> 7 of your (ideal) 10 bits. 
>> 
>> Net result is that you are trying to check a 16 bit gizmo with a not quite 8 bit device.
>> 
>> Bob
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.