In Summed Pricing, when the base price makes up a relatively small % of the total price for products (after summing prices across the other attributes), then the total summed price tends to be more correlated with the other attribute levels. Multicollinearity is a bad thing for precise utility estimates. So, to counteract that, larger random shocks to price should be chosen.
We conducted some simulations to test how much random shock was needed under different conditions: specifically, when the base price makes up a large proportion of the final summed price vs. when it makes up a relatively small proportion. The results are reported in this white paper: http://www.sawtoothsoftware.com/download/techpap/price3ways.pdf
The paper gives some general recommendations for the amount of price shock, based on the relative size of the base price:
If base price is 3/4 of total average price: +/-10%
If base price is 1/2 of total average price: +/-20%
If base price is 1/3 of total average price: +/-30%
You say for your situation that base price is 8 Euros and the summed price could go as high as 28 Euros. But, I suspect the average base price is closer to 20 (for argument's sake). If that's the case, then your base price is 8/20 = 40% of total average price. So, we'd recommend random price shock around +/-25%. As you add more random shock (say, +/-40%) the statistical precision of your utility estimates increases a bit, but you also lose task realism and the product combinations don't seem as realistic to respondents as before.