I decided to do some research with Google:
Water Hardness is one reasonable sounding article. It says 30.02 mg/liter is the amount of NaHCO3 needed to get 1 dKH, because dKH refers to carbonate from CaCO3, not from just any old source. (This is in the right direction to explain my problem.)

Carbonate hardness - Wikipedia, the free encyclopedia is Wikipedia's explanation. It says 50 mg of NaHCO3 in a liter of water gives about 18 mg/l of carbonate hardness, which is about 1 dKH. (That too is in the right direction to explain my problem.)

KH reference solution/ baking soda is where one of our members is asking similar questions. No answers, of value, yet.

I suppose I need to take the 30.02 number in the first reference, make up my 4 dKH solution based on that and test to see if it works. But, that doesn't really make any case for doing more than just adding sodium bicarbonate until the same test kit says it is at 4 dKH.

Consider the equation that is used to determine ppm of CO2 by measuring KH and pH: ppm = A*KH*10^(B-pH) (A and B can have various values, but the ones used the most are A=3 and B=7) In any case, the ppm calculated is proportional to KH. So, a 10% error in KH gives a 10% error in ppm. That is much, much less significant than a .3 error in pH, which gives an error in ppm of 2X or 1/2 X (+100% or - 50%). I am having a hard time seeing why I need great accuracy in setting my reference KH. Even a 20% error would seem to not be that big a deal.

In a chemistry lab, where a measurement of CO2 ppm really, really has to be dead accurate, a very carefully mixed reference KH solution would be essential. But, we (at least I, if not we) are trying to establish that the ppm in our aquarium is in the ballpark of 20-40 ppm, and not 5 ppm or 60 ppm. A micrometer isn't always the best way to measure a length!