Difference between revisions of "Spring 11:QRT-PCR"

Revision as of 19:03, 4 April 2011

Looking for weekly reports?

Or LabView QRT-PCR version info?

To-Do

Last updated 15:50, 1 April 2011 (EDT)

• get used to version 2.712.
• convert RTD code in 2.712 to Thermistor code
• talk to Steven about improving optics.
  • would adding a filter help?
  • what else could be the problem?
• test heat transfer equation with sample vs block thermistor.

Weekly goals

as of 15:50, 1 April 2011 (EDT)

• 3/25: Optics finished. Code started.
• 4/1: Heated lid finished.
• 4/8: No really, heated lid finished. Updated heat transfer function.
• 4/15: Successful PCR.
• 4/22: Optics are ready for QRT-PCR!
• 4/29: Successful QRT-PCR from the side.
• 5/6: QRT-PCR from the top. Now which one was better?

Heated Lid research notes

Desired specs:

• 1/2 inch diameter circle
• can handle up to 3A
• can go from room temp (27C) to 92C in less than 5-ish minutes

OpenPCR first used nichrome wire and silicone tape sandwiched between two plates of aluminum...but it didn't work too well. They said it "smoked and burned out." However, they were designing for 16 samples, not just one (so we won't need to use as much wire). Also, we're not planning on using silicone tape. I still think it's worth trying.

Then they used a peltier device with an aluminum plate + thermal pad.

Another option: using a PCB. Instructions here.

Option 1: Nichrome wire

Researching nichrome wire... looks like NiChrome 80 is the best. Here is the datasheet. Now for calculations:

 I = 3A
 V = 15V
 n = length of wire (feet)
 rpf = resistance per foot (ohms/ft)
 
 I = V/R = V/(n*rpf)

 rpf and n are inversely proportional - and cost goes up with rpf. So we want to minimize that.
 reasonable values seem to be:
 rpf = 6.46
 n = 0.77 feet
 which means a 30 gauge wire, d = 0.0100 inches.

Okay, this turns out to be too long. If we're looking to use only one coil of wire, using nichrome is not feasible.

Option 2: Flexible heaters

Mica, polyimide, silicon These are pretty expensive. Is it possible to just attach leads in random places to these foils, I wonder? That would make it cheaper. But as it is, they're more expensive than a TEC.

It may also turn out to be cheap to order bulk custom-made flexible heaters. I will call and get a quote.

Option 3: PCB

Would it be fast enough? who knows.

Windell Oskay says: "If you have such a specific spec that's not in the off the shelf units, that's a second argument (besides price) for making them yourself.

I don't know about your mechanical design, but you may be more confident in making a heater out of a traditional circuit board-- it can easily handle 110 C, will be rigid, can be custom shaped with mounting holes, and you can dip it in a biocompatible coating if necessary. Final cost will be a few dollars (or lower), per board, exactly meeting your specs in terms of size, shape, voltage input, and power dissipation.

So you need 12 V input, and maybe 4-6 ohm resistance to get 20-30 W. Let's call it 5 ohms, and you'll then need 12/5 = 2.4 A available for the heater. 1 oz copper has resistance of *roughly* 0.05 ohms/inch in a 10 mil trace, so you need about 100 inches of wire. Suppose that you use 10 mil wires with 10 mil spacing, with a simple single-sided back-and-forth pattern. Then one wire plus spacing is 20 mil, and you can fit 50 wires per linear inch. If you have a 1.5" square, then in each linear inch, you get 1.5*50 wires = 75 wires per inch, and in 1.5 inches of that you get 1.5*75 wires = about 112 inches of wire total-- right on target."

 resistance per inch for copper ~ 0.05
 I = 3A
 V = 15V
 n = length of wire (inches)
 
 I = V/R = V/(n*rpf)
 n = V/(I*rpi) = 15/(3*0.05) = 100 inches of wire

That's a lot to fit on a 0.5" diameter circle. Batchpcb.org has a minimum trace width of 8 mils with 8 mils spacing - too big to fit 100 inches. I wonder if other trace materials, with higher resistances per inch, could be used.

Progress Notes

4/4/11

Posted 15:00, 4 April 2011 (EDT)

I now know the reason for the background signal. The fiber optic cable is auto-fluorescing. BOO. It works okay for now (there is a small, but extant difference between H20 and SYBR+dsDNA), but in order to do any serious testing, I'll need to change the design. Or get a new fiber optic

It might, in fact, be time to get rid of the fiber optic all together. But I will also look around for fiber optics that really, really do not fluoresce.

4/1/11

Posted 15:50, 1 April 2011 (EDT)

Officially ordered 29-gauge wire.

The optics *do* detect a difference between water and SYBR dsDNA, but it's a fairly small difference. I think this small difference is due to the HUGE amount of background signal at the carrier frequency. Why is this happening? My current hypothesis is that the dichroic mirror is directly reflecting some light from the blue LED into the photodiode... but with the filters in place, this shouldn't be happening!

Question: how good are the filters we're using? Could they still be letting some other light through? (maybe we need another blue filter in front of the blue LED).

Steven went over with me some of the different existing versions of the LabView code. I will be working with versions 2.700 and later. Not all of them work. I will take notes on this; take a look at QRT-PCR:LabView-versions.

3/28/11

Turns out my calculations were all messed up before! With corrected calculations, and the data from [www.heatersplus.com], I decided on 29-gauge resistance wire for our heated lid.

3/22/11

Getting the temperature calculation right earned me a level-up... meaning I got to graduate to the Current Version of the Code. Got a brief intro to it. It's pretty complicated.

3/20/11

Based on model system (blue LED + ND filter), I don't think gain is high enough. Increased gain of transimpedance amp from 2.5e4 to 2.5e5. Well, it didn't help that much. So it's back the way it was. In any case, we get a significant signal (~0.25 units amplitude) with the model system. Huzzah!

After much wrestling with simple math, the temperature sensing is largely working! yay

3/18/11

Optical setup appears to be working.

3/16/11

Added dichroic mirror.

Question: how can I verify the optical setup? it seems to respond to driving signal, not fluorescence.

TODO: put away 25mm lens