#2000 Post : Teardown and study of Fluke 5700A Calibrator [56K warning!]

[Kleinstein]

I recently read through the following patent US11316528 PWM DAC with improved linearity and insensitivity to switch resistance that could be interesting for some of you?

Edit: Attached also the previous patent US5402082 for completeness.

-branadic-

Thanks, but that patent looks like it has been written by a robot, not a lawyer.  Skimming over it, it looks like Fluke prevents anyone else to use PWM with two switching networks without actually explaining how this addresses errors of (varying) switch resistance, imperfect, varying switch timing, stray capacities in the switches and charge injection.  But then, perhaps it is obvious to those skilled in the art.  Did anyone grok it and can translate it here to English, pretty please?

I totally agree that that patent is awfully hard to understand - like written to be not understandable. From what I understood it does not really sensible. It kind of has 2 parts: one is claiming good linearity by having some switches insider the OP's loop - however this substitutes the switch resistance by the OPs settling as a weak point. E.g. it is not uncommon to have settling depending on the current and this could cause INL. So I don't think this works.  The second point is somehow getting a larger range: not fully sure how it works but somehow the gain would likely depend on the resistors and maybe RC time constant - hard to believe that this would be stable and linear. In that part the switch resistance is not compensated for.
I would not consider the patent relevant, more like smoke screen type to repeat old, expired claims.

[deepfryed]

Hello branadic,

as you already should know I used the EDN-Design

https://www.edn.com/dc-accurate-32-bit-dac-achieves-32-bit-resolution/

some results are here:

https://www.febo.com/pipermail/volt-nuts/2010-October/000537.html

The linearity shurely could be optimized if not using the whole range of the PWM.
E.g. cutting off the low and high end by adding a negative offset and amplification of the middle part.
(see last diagram here: https://www.eevblog.com/forum/metrology/stacking-ad587s-for-higher-output-voltage-(20v-30v-etc-)/msg1320513/#msg1320513 )

Alternatively one could do a mathematical correction. (of both linearity + temperature)
As controller any controller can be used which has at least 3 synchronizable 16 Bit PWM outputs which should be a multiple of the AC mains frequency to use the NMR of a multimeter. (e.g. 244 Hz or 300 Hz PWM frequency)

As controller I tested ATMega1284, PIC24FV32KA302 and one DSPIC.

with best regards

Andreas

Thanks Andreas, that thread was a very interesting read. I've been thinking about building a discrete PWM DAC as a learning exercise, looks like there's lot of rabbit holes going down that path. Did you end up with anything relatively stable / linear ?

[Andreas]

Thanks Andreas, that thread was a very interesting read. I've been thinking about building a discrete PWM DAC as a learning exercise, looks like there's lot of rabbit holes going down that path. Did you end up with anything relatively stable / linear ?

Hello,

no further development.
but perhaps also the LM399 10V PWM DAC is interesting.
around here:

https://www.eevblog.com/forum/metrology/lm399-based-10-v-reference/msg2271156/#msg2271156

with best regards

Andreas