Control precision/significant decimals in ADEXL results

Dear Shankar,

First, as you probably aware, ADE-XL is no longer supported by Cadence, and hence I might suggest you consider migrating to Assembler and Explorer. The transition is very easy as Assembler has a very similar interface to ADE-XL and provides functions to import your ADE-XL based state files into its common maestro view. 

Secondly, with respect to your question concerning fixing the number of decimal characters in the Output tab, have you considered changing the format of the output setting? As shown in Figure 1, if I change the format to "scientific" and select 8 digits, the format of the output values shows 

Does this address your question or did I not understand correctly?

Shawn

Figure 1

My apologies Shankar as the firm posted my reply before I finished it!

In any case, the output format with scientific selected as the format will provide N-1 fractional digits if you specify N as the number of significant digits. With N = 8 selected in Figure 1, there are 7 digits representing the fractional part in Figure 2.

Shawn

Figure 1

Figure 2

Hi Shawn

You understood my question correctly.

Your solution works, however in scientific format it is very difficult to read the data. for e.g 14.3 becomes 1.43e+02

An engineering format works best, but in that mode it seems no other option to control the display of digits in fraction.

We are migrating to Assembler, is there an easy way to achieve this results there?

Regards

Shankar

Hi Shankar,

No. Most things in engineering are more concerned with precision which would be affected by the number of significant digits rather than the number of decimal places, so there is no control to display the data with a fix number of decimal places.

By the way, you can also set the number of digits and notation per output by using the "Digits" or "Notation" columns in the Outputs Setup table, but that also doesn't allow you to set a fixed number of decimal places.

I couldn't find any request for this - to be honest I'm rather surprised that you'd be happy with 1.23 and 12832.43 as the first only has 3 significant digits and so has very poor accuracy and the second has 7 significant digits so much more precision. What kind of quantity would benefit from this fixed number of decimal places? (I can think of currency, but that tends not to come up very often in simulation).

Regards,

Andrew

Hi Andrew

The reason I wanted to have a control on the fraction part is when I am dealing with relative quantities in a log scale(eg Gain in dB, noise figure in dB). In those cases 1.23 and 12832.43 have the same level of accuracy.

In a display column lets say I have the following dB quantities, it is easier for me to read the following

    0.91

    1.91

  10.91

100.91

instead of(deluge of numbers)

0.91234

1.91234

10.9123

100.91

This request is something tantamount to what is used normally in printf statements("%5.2f") and in spread sheets where in you can control the decimal places.

Thanks & Regards

Shankar

I'm not sure how often it would be necessary to compare the two decimal places of something in dB which are 100dB apart, but I can certainly see the need to compare more easily 0.01dB and 1.02dB without the first ending up as 10mdB. You should contact customer support with this reasoning so that we can consider it as an enhancement.

As a workaround, you could have an output expression such as this:

lsprintf("%5.2fdB" 1000*pi)

(yes, I know outputting 1000*pi in dB doesn't make sense; the expression itself isn't the point here - I just wanted something with lots of digits). The idea is that if you convert the output to a string and it has some non-numerical characters in (hence the dB in the format string), then ADE will output it literally. If it looks like a number still, it will be reformatted to the desired precision.

The downside of this workaround is that you can't then put a spec limit on that output because it's a string not a number. So it may not fully do what you want, but it is a workaround that's possible right now.

Regards,

Andrew

Hi Andrew

Thanks for the tip on converting into string, I have already been doing that as a temporary workaround.

As you said, I lost the spec limit feature and color coding of the cells.

Will take it to the customer support as an enhancement request.

Regards

Shankar

Hi Andrew

Thanks for the tip on converting into string, I have already been doing that as a temporary workaround.

As you said, I lost the spec limit feature and color coding of the cells.

Will take it to the customer support as an enhancement request.

Regards

Shankar