Dextran 40 USP/EP

[shaun78]

Anyone out there in Chromforum land have experience running the Dextran 40 USP/EP monograph?

I am having extreme difficulties meeting the 5% accuracy check for Mw of standards run.

IMHO ... the USP standards are not of narrow distribution and the USP does not provide enough information to produce a broad standard calibration plot (USP gives Mw and MP moments only). I believe that when using a narrow distribution as a calibrant the CDS (Empower in my case) assumed MP=Mw, which is a massive source of error.

Can anyone help clear the muddy waters with respect to how this analysis is supposed to be successfully run?

TIA...

Shaun

[Gerhard Kratz]

I would recommend to try also other Dextran standards than USP. Have a look at the company Ludger LTD in UK they are producing also high MW Dextran standards. For sure you will find other sources for Dextran standards, maybe Restek, or PSS. Good luck.

[shaun78]

Well ...

I have been informed by the USP that no CDS software performs the calculations in accordance with the Dextran 40 monograph.

Therefore, I must calculate manually. Anyone know how to perform Hartley modified Gauss-Newton calculations? Anyone even know what I need to plot in order to start the approximation?

Calculations outside of the above are fairly straight forward...

[Alfred88]

Hi.
I am in CA, USA. I have got same problem as you for Dextran 70 USP. I have solved the problem successfully (look up my old posts on this forum).

I offer my service as a computing consultant for your problem. I will let you know my fee schedule (if you are interested!). I cannot do for free, because I am busy.

Let me warn you that non linear optimization (NLO) is an advanced topic in computer science. You need to take courses in programming, linear algebra, etc to be "ready" for this analysis. Note: I have done these!

Common programs such as MS Excel or Lotus 123 etc cannot handle the NLO problem.

Thanks and best wish.

Alfred

[shaun78]

Luckily for me I was a math major as well as a chemistry major, so the math involved in this really does not scare me.

I have also re-read your older posts re:dextran 70. Looks like I am at the same point you were: the series is not converging. Though I have found one equation to model all 5 curves, my sum of residuals is anything but minimized.

Regardless, we need to get this done. I will ask my boss if he is willing to pay for you to consult us on this.

[shaun78]

Alfred88 please feel free to email me at [email removed].

I will remove my email address from this post once I hear from you.

FYI ... we are under a tight deadline and need to report this data to the client by end of business on Tuesday at the test.

[shaun78]

Sorry to bump an old post... but what is good for the goose is also good for the gander.

I figured this out for our client a few months ago...

A few weeks ago we received additional sample.

Then I did a lit search and came across this. Wish I would have found it before I reinvented the wheel.

https://www.dropbox.com/s/5dpvjj2he1xyk ... s.pdf?dl=0

If anyone is in a bind with respect to this analysis, get in touch and I'll send you the excel sheets I use for both USP and EP analysis.