Chromatography Forum

Does anyone have any suggestions as to how to validate a Waters Satin box (A/D converter). It's linked to Millennium 32 and just takes the detector signal to the PC.

I've got one. But what do you mean by validate.
what detector is linked to it ?

It's linked to LDC/Milton roy SM3000 series detectors and we need to validate the Satin box. Waters will do it for you at great cost, but we need to validate it to prove it's fit for it's purpose and that whatever data it is converting, is being done correctly. I personally don't see what we can do other than to ensure that a signal is being received, that it changes with increase/decrease in absorbance and that the absorbance scale shown on Millennium is similar to that displayed on the detector.

If anyone else has any ideas to add then I'd appreciate it

Sure ! It is as if you validate the detector. the procedure keep the same ( specificity, linearity...).
for my part, I think you can't validate the SATIN alone.An operational qualification is like measuring intering and outsiding signal.

eSATIN validation involves the use of internally(? may be available) or externally generated test signals. These signals are of a well characterized waveform nature and the corresponding reports of their collection via the eSATIN serves to demonstrate its correct operation (if they match the theoretical waveforms).

Josimar,

I think what you need is a device called a peak simulator or a chromatogram generator. This device generates various chromatogram signals that can be used to test the accuracy, precision and linearity of Analog-to-Digital devices such as the SATIN.

We purchased ours from SSI which is now Agilent (no, I don't work for them).

Check out this link:

http://www.chem.agilent.com/scripts/gen ... &prodcol=Y

We use this to calibrate our Analog-to-Digital devices that are used for our CDS. We don't have the SATIN device but the principle is the same.

Regards,
Dan

You need 2 things:

1. A signal generator that can produce Gaussian/2 patterns like the SS120C mentioned above.
2. A validation procedure (OQ) to perform the validation.

ad. 1. Can be bought.
ad. 2. You need to figure out or get somewhere. What I used to use is a check for linerarity and reproducibility using an Excel spreadsheet. We ran 5 or 6 Gaussian/2 runs and let Excel do the math. The Excel Report would serve as an appendix to the OQ report.

Linearity: how good the Gaussian/2 pattern was calculated as Area. The 2nd peak needs to be 50% of the Area of the first and so on.
Reproducibility: how reproducible is the Area of each Peak with each run.

JW
Applied Instruments
http://www.applied-instruments.com

Is there really a need to simulate gaussian curves to validate the analogue to digital converter ?

I'm assuming that the A/D converter just reads a voltage and sends that number to the chromatograhpy software, and the software then logs all the data and works out what is a curve and what isn't.

We use Chromeleon, which has a similar A/D unit called a UCI-100. We validate (and regularly calibrate) the A/D unit using a verified voltage generator (A Martel model). We have a set protocol that generates linearity steps of voltages, reproducibility of different voltages, noise and bias and crosstalk between different channels. Chromeleon is even clever enough to be able to control the voltage generator so the entire test can be automated, then reviewed and signed electronically to geenrate a PDF report.

Paul.