-
- Posts: 2
- Joined: Sun Aug 27, 2023 7:35 pm
I recently assembled a basic HPLC-UV system for hobby use (mostly agricultural, like capsaicin for superhot breeding), at surprisingly low cost--the first generation of practical instruments is now in the hands of general surplus equipment traders, one step from the junkyard. I wanted to share my experience here for the benefit of anyone attempting a similar goal. I'd also appreciate any advice, especially if anything I'm doing here looks totally wrong. I'm pretty obviously not a professional chromatographer (or chemist of any kind), but I'm comfortable enough with most of the engineering work.
I bought a Waters 510 pump, on eBay as is. It was missing most of the plumbing, which I replaced with 0.010" PEEK and the finger-tight ferrules. This was easy to work with. I didn't use any special tool to cut the tubing, just a new sharp razor. It also had no draw-off valve, just a hole for a 17 ga needle to prime the pump near the inlet. That hole needs to be covered in normal operation, and I'm not sure how that was supposed to work. For now I'm just leaving the needle permanently inserted, with a Luer stopcock to close it off when the priming syringe isn't attached. I was able to prime the pump with a standard 5 mL syringe and drugstore 91% IPA; I understand its relatively high viscosity should make that easier. Priming with methanol also seems to work, perhaps with a bit more effort. After priming, I found that while each pump head delivered the expected (half) flow rate alone, flow decreased and became inconsistent when the two pump heads were connected together through a tee. I traced this problem to the output check valves, which looked very worn--so the two heads were trading the same solvent back and worth. I was able to find a rebuild kit and rebuilt both valves, apparently solving the problem. The pump delivered ~2500 psi into the column (per below) around 1 mL/min of pure methanol, which seemed about right. The pressure fluctuated though, and the left head slowly leaked. I replaced its seal, and both problems resolved. The pressure sensor needed to be zeroed, using the trimpot on the adjacent printed circuit board. I paid $150 for the pump, plus $50 for the rebuild kit and $30 for the seal. I'd chosen the 510 partially for its abundance of spare parts, and that turned out to be important.
I bought a Waters U6K injector, again as is. It arrived with no sample loop, but I bought a new 20 uL loop and installed it. The back-pressure when injecting was very high, and the needle sealed poorly, with the result that most of the sample leaked out. I disassembled the injector completely (which requires snap ring pliers for the cam that actuates the microswitch that marks the injection time, but no other special tools) and tried cleaning all the ports with a thin wire. This may have helped, but I still got the high back-pressure. After some study, I slightly loosened the male-threaded round part that retains the valve on the vent port; the back-pressure problem was solved, and the valve still didn't leak under pressure. I tightened the nut around the injection port, and that solved the leak around the needle, I guess by forming the PTFE seal more tightly around it. I paid $60 for the injector, plus $20 for the loop. It's possible that I should have just bought a Rheodyne 7125 or similar; they're pretty cheap too, and seem more convenient.
I bought an HP 79853C, the variable wavelength detector from their Series 1050, advertised as working. Upon receipt, the optics package was bashing around inside, I believe because the screws that are supposed to retain it during shipment weren't installed. That hadn't visibly broken anything, though. The deuterium lamp was dead (possibly from the bashing?), but I bought a new one and it ignited. The detector then passed all self-tests and seems to give plausible output. I paid $160 for the detector, plus $150 for a new lamp.
I bought a 4.6x250mm C18 column for $80, and a guard column for $40. I hadn't realized how high the pressure would get with the long column; even 1 mL/min of 70/30 water/methanol is beyond my max working pressure, which seems to be limited practically by the PEEK tubing (which burst around 5000 psi, though maybe at a weak spot from a pre-existing kink). I'll probably get a shorter one, especially since I'm considering a switch to ethanol for safety reasons. In any case, that makes $740 total for the major components. I also bought various tubing and fittings, a vacuum filtration setup and some 0.45 um membranes, microliter syringes, my solvents and some Type I water, etc., for a total just under $1K.
Assembling everything, an injection of 2 uL of 100 ppm caffeine showed the expected peak, plotted on an oscilloscope connected to the detector's analog output. My oscilloscope has cursors and can integrate peaks, though it's obviously less convenient than a real chromatography data system. (Is there any cheap option for that? I see stuff like Clarity Lite, but that's >2x my total spend so far.) I haven't actually fit a calibration curve yet, but so far the system looks more than repeatable enough for my purposes. I'm not degassing beyond the vacuum filtration, and I've seen no trouble with bubbles. I connected a 40 cm 0.005" PEEK capillary to the detector output per HP's recommendation, but it doesn't seem to make a difference.
In any case, I've at least learned more here than I did from handing the TA an autosampler vial when we did our "HPLC lab". Any further advice appreciated, or perhaps this will help someone else enjoy these historically significant machines for another few years.
I bought a Waters 510 pump, on eBay as is. It was missing most of the plumbing, which I replaced with 0.010" PEEK and the finger-tight ferrules. This was easy to work with. I didn't use any special tool to cut the tubing, just a new sharp razor. It also had no draw-off valve, just a hole for a 17 ga needle to prime the pump near the inlet. That hole needs to be covered in normal operation, and I'm not sure how that was supposed to work. For now I'm just leaving the needle permanently inserted, with a Luer stopcock to close it off when the priming syringe isn't attached. I was able to prime the pump with a standard 5 mL syringe and drugstore 91% IPA; I understand its relatively high viscosity should make that easier. Priming with methanol also seems to work, perhaps with a bit more effort. After priming, I found that while each pump head delivered the expected (half) flow rate alone, flow decreased and became inconsistent when the two pump heads were connected together through a tee. I traced this problem to the output check valves, which looked very worn--so the two heads were trading the same solvent back and worth. I was able to find a rebuild kit and rebuilt both valves, apparently solving the problem. The pump delivered ~2500 psi into the column (per below) around 1 mL/min of pure methanol, which seemed about right. The pressure fluctuated though, and the left head slowly leaked. I replaced its seal, and both problems resolved. The pressure sensor needed to be zeroed, using the trimpot on the adjacent printed circuit board. I paid $150 for the pump, plus $50 for the rebuild kit and $30 for the seal. I'd chosen the 510 partially for its abundance of spare parts, and that turned out to be important.
I bought a Waters U6K injector, again as is. It arrived with no sample loop, but I bought a new 20 uL loop and installed it. The back-pressure when injecting was very high, and the needle sealed poorly, with the result that most of the sample leaked out. I disassembled the injector completely (which requires snap ring pliers for the cam that actuates the microswitch that marks the injection time, but no other special tools) and tried cleaning all the ports with a thin wire. This may have helped, but I still got the high back-pressure. After some study, I slightly loosened the male-threaded round part that retains the valve on the vent port; the back-pressure problem was solved, and the valve still didn't leak under pressure. I tightened the nut around the injection port, and that solved the leak around the needle, I guess by forming the PTFE seal more tightly around it. I paid $60 for the injector, plus $20 for the loop. It's possible that I should have just bought a Rheodyne 7125 or similar; they're pretty cheap too, and seem more convenient.
I bought an HP 79853C, the variable wavelength detector from their Series 1050, advertised as working. Upon receipt, the optics package was bashing around inside, I believe because the screws that are supposed to retain it during shipment weren't installed. That hadn't visibly broken anything, though. The deuterium lamp was dead (possibly from the bashing?), but I bought a new one and it ignited. The detector then passed all self-tests and seems to give plausible output. I paid $160 for the detector, plus $150 for a new lamp.
I bought a 4.6x250mm C18 column for $80, and a guard column for $40. I hadn't realized how high the pressure would get with the long column; even 1 mL/min of 70/30 water/methanol is beyond my max working pressure, which seems to be limited practically by the PEEK tubing (which burst around 5000 psi, though maybe at a weak spot from a pre-existing kink). I'll probably get a shorter one, especially since I'm considering a switch to ethanol for safety reasons. In any case, that makes $740 total for the major components. I also bought various tubing and fittings, a vacuum filtration setup and some 0.45 um membranes, microliter syringes, my solvents and some Type I water, etc., for a total just under $1K.
Assembling everything, an injection of 2 uL of 100 ppm caffeine showed the expected peak, plotted on an oscilloscope connected to the detector's analog output. My oscilloscope has cursors and can integrate peaks, though it's obviously less convenient than a real chromatography data system. (Is there any cheap option for that? I see stuff like Clarity Lite, but that's >2x my total spend so far.) I haven't actually fit a calibration curve yet, but so far the system looks more than repeatable enough for my purposes. I'm not degassing beyond the vacuum filtration, and I've seen no trouble with bubbles. I connected a 40 cm 0.005" PEEK capillary to the detector output per HP's recommendation, but it doesn't seem to make a difference.
In any case, I've at least learned more here than I did from handing the TA an autosampler vial when we did our "HPLC lab". Any further advice appreciated, or perhaps this will help someone else enjoy these historically significant machines for another few years.
