Poor performance of old diff pump.

Every fusor and fusion system seems to need a vacuum. This area is for detailed discussion of vacuum systems, materials, gauging, etc. related to fusor or fusion research.
Post Reply
Marcin Dziedzic
Posts: 6
Joined: Thu Dec 20, 2018 12:57 pm
Real name: Marcin Dziedzic

Poor performance of old diff pump.

Post by Marcin Dziedzic »

Hi,

Sorry for my bad english, I'm from Poland. I'm using a research vacuum furnace which is an old used unit transferred to laboratory. Sadly my company decided to refurbish an old furnace instead of making a new one for R&D purposes. It its equipped with old CVC-20 diffusion pump manufactured by Consolidated Vacuum Corporation of Rochester, NY. The company does not exist anymore, as far as I know. The pump is 29 years old. It's a 18 000l/sec diffusion pump connected to furnace via 20" ASA flange. Backing pump (foreline) is Edwards Stokes 412J mechanical pump coupled with Stokes 615 roots booster, capable of going down to 8e-3 mbar. Pump has its original heaters broken long ago, it's now retrofitted with six heating elements from Varian HS-20 (total 12kW vs 11kW original). It is filled with DC-704 oil euqivalent, I've got only photograph of old manual page which says the pump has to be filled with 1 US gal of oil, preferrably DC704. Sadly the pump doesn't have oil themperature T/C. I managed to install thermocouple on pump base near heating elements (outside of pump chamber), and it shows absurdly high values like 510 deg C. I had never operated such small and such old pump. The pump starts to work when heater temperature is about 400 deg C, so it probably corresponds with oil temperature above 200..210 degrees.

I managed to get the pump working, but despite the throughput between 5e-2 mbar and 8e-4 being satisfactory, pump starts to behave poor below this pressure. Pumping speed falls quickly and about 30 minutes are needed to reach 2e-4 in quite small (2,4 cubic meter) chamber. 1e-4 seems to be ultimate vacuum - the pumping speed falls to zero. I managed to get 9,8e-5 after six hours of pumping. The chamber tightness may not be ideal, but I managed to go down to 1e-5 or 1e-6 with bakeout on bigger chambers, having worse leak rates. The pump itself was tested and it's helium-tight. I tried changing the quantity of oil (original level gauge was broken) from 3.8 (1 US gal) to 9 liters, it gives no effect whatsoever. I'm maintaining very big cooling water flow, much bigger than recommended, with 20degC inlet and 33deg outlet, despite this pump runs very hot, you can't touch side walls in half of the pump height. Maybe there is scale in cooling coil preventing heat transfer ?

The pump was visually inspected and flushed with organic solvent and new oil, but the jest assembly has been not disassembled. Can this poor speed be due to some mechanical defect in the jet ?
User avatar
Richard Hull
Moderator
Posts: 14991
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Poor performance of old diff pump.

Post by Richard Hull »

That is a monster diff pump! I would work on the jet stack. If you take it apart, make absolutely sure you lay all parts out on a table in the exact order of disassembly clean one item at a time.

You may just have some clogged jet holes.

I assumed you have clean pump oil and are not using the old oil.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
Bruce Meagher
Posts: 148
Joined: Thu May 12, 2011 11:25 pm
Real name: Bruce Meagher
Location: San Diego

Re: Poor performance of old diff pump.

Post by Bruce Meagher »

Marcin,

Your english is excellent.

I personally have no experience with huge diffusion pumps like yours, but a couple things to consider. Could the DC-704 oil just need to be degassed a bit longer? On smaller diffusions pumps I’ve experienced long initial pump down times with a new charge of oil (getting all the water out). I’d image with a gallon of DC-704 it could take even longer, but this is just a guess. Does the pumpdown time decrease with each successive run, or does it always take 5+ hours to get to 1e-4 mbar? Also, it probably takes at least 30 mins to get the diffusion fully running. What is the diffusion pump’s startup time listed in the manual?

It would be instructive if you tracked the foreline pressure along with the chamber pressure. With the throughput of your diffusion pump, your foreline pressure must be pretty high even when the chamber is at 10-4 mbar. Does your foreline have any gauging, and if not, is this something you can add?

Another thing to verify is to ensure the ejector on the diffstack is pointing at the foreline and not rotated to some other place. I would do some more runs before pulling it apart though.

I’d love to see some pictures of your setup for reference. Also take some temp readings along the outside of the diffusion pump when it’s fully running so we can see what the temp profile along the side looks like.

Finally, what pressure does the chamber reach with your two backing pumps before you turn on the diffusion pump?

edit:

The obvious problems could be a large leak, a throttle valve not properly opened, a bypass valve not properly closed, or contamination in your chamber.

Bruce
Last edited by Bruce Meagher on Fri Dec 21, 2018 3:51 am, edited 1 time in total.
John Futter
Posts: 1848
Joined: Wed Apr 21, 2004 10:29 pm
Real name: John Futter
Contact:

Re: Poor performance of old diff pump.

Post by John Futter »

Big diff pumps suffer from bad thermal connection between heater and bottom of pump
this area tends to get a large rust build up that stops the heat getting to where it is needed.
An IR thermometer is useful to prove that the heater is making good contact to the base of the pump.
most big diff pumps i have seen have used a temperature controller to set the boiler temp ie they are not left open loop for the vapour to set the temp gradient.
Marcin Dziedzic
Posts: 6
Joined: Thu Dec 20, 2018 12:57 pm
Real name: Marcin Dziedzic

Re: Poor performance of old diff pump.

Post by Marcin Dziedzic »

Hi Guys,

Thanks for the replies!

Again sorry for my bad english, I'm quite embarassed about it. (also I'm a bit drunk after friend xmas party). Also please pardon me that I'm using metric units on american forum, I'm slavic to the bone, and I'really don't have an idea how much "cfm" or "lb*ft" is.

I'd love to post some pictures of my setup, but my company has a quite stringent "secrecy" policy and considerations, as this is a R&D unit, blah blah, so I must ask my bosses if I can post photographs of the furnace, I kinda don't want to get fired. (And the bosses are on holidays of course, of the whole company only I'm not on holiday because I've used all of my off days climbing in Himalayas).

The basic setup is as follows: The furnace has two chambers separated by an isolation valve (helium-tight 10e-7 mbar*dm3/sec measured by PhoeniXL-300 leak detector). To the lower chamber are connected four ordinary vacuum pumps with roots booster, and to the upper chamber - diffusion pump (or fine vacuum pumps). Diffusion pump is joined with the system by means of a pneumatically operated poppet valve. And:

- chamber low vacuum is measured by means of a single pirani gauge (leybold TTR90)
- chamber high vacuum is measures by means of two bayard-alpert gauges (Leybold Ionivac ITR90)
- diff pump foreline pressure is measured by two pirani gauges, one near diff pump, and one near roots booster.

Finally, what pressure does the chamber reach with your two backing pumps before you turn on the diffusion pump?
I usually turn on the diffusion pump when chamber pressure is about 3e-2, in worst case 5e-2, as the diff pump tends to "choke" on higher pressures as We all know. It takes about 15 minutes to reach it from atmospheric pressure. If I let backing pumps running for whole night they go down to 7e-3 mbar.
The obvious problems could be a large leak, a throttle valve not properly opened, a bypass valve not properly closed, or contamination in your chamber.
Poppet valve is fully opened as I can hear it banging on upper limit ring when it's opening. Leaks - well, the problem is, that the furnace chamber has about 20 vacuum feedthroughs including rotary ones, moveable vision ports, thermocouples, pyrometers, water hoses, RF power cables, and a huge (10") hydraulic actuator. As far as I know the furnace chamber is not ideal, but is quite good. Leak rate on air is 1,9e-2 mbar*dm3/sec. This is of course not very good but with worse leak rates and similiar diff pumps (Varian HS-20, Leybold DIP18 000) I'm easily achieving 2e-5 in about 30 minutes.

Now I discovered a new leak: From the air cylinder actuating the poppet valve. Normally it's powered by a 8 bar (about 100psi) of compressed air. I managed to power it with helium (!!) from gas cylinder with leak detector connected to the foreline and I see there is a leak straight to the diff pump. We'll work on that (probably the cylinder needs to be replaced which is pain in the a**). It certainly does deteriorate vacuum system performance.
I personally have no experience with huge diffusion pumps like yours
And I have no experience with such small diffusion pump :( Been only operating 35 000l/sec and 50 000 l/sec pumps ;)
Could the DC-704 oil just need to be degassed a bit longer? On smaller diffusions pumps I’ve experienced long initial pump down times with a new charge of oil (getting all the water out). I’d image with a gallon of DC-704 it could take even longer, but this is just a guess.
I'm using a equivalent of DC704 called something-704 (I really don't remember company name, not Kurt Lesker). As far as I know original 704 is now withhdrawn from the market due to some absurd worries about diffusion uranium enrichment and US export control (DC704 is resistant to fluorides, and thus HF and UF6 essential to uranium enrichment). (Of course every buyer of DC704 wants to enrich uranium, the bad guys surely will purchase it in legitimate sources ;) ) The diff pump has been turned on for about 40 times now, I thinks it is degassed. The pump itself is VERY tight, if it's pumped down to 7e-3 (foreline), and left over for 24 hours, the foreline pressure is still below 9e-3 (!) which I'm quite proud of, as I had tailor-made new sight glass and oil inlet stoppers. (The pump was really a scrap when I got it). I've made all the gaskets from SIGRAFLEX (tm) board.

The pumpdown time to 5e-4 is about 10 minutes.
The pumpdown time 1e-4 is about 1 hour.
Pumpdown times DECREASE with each subsequent pumping, but they increase back to beginning when the furnace is idle over the weekend (despite it is left under vacuum).
It would be instructive if you tracked the foreline pressure along with the chamber pressure. With the throughput of your diffusion pump, your foreline pressure must be pretty high even when the chamber is at 10-4 mbar. Does your foreline have any gauging, and if not, is this something you can add?
So I did. Sadly, there is evident rise in foreline pressure while pumping down, but when the chamber pressure is down to about 2e-4 the foreline pressure is very low. I think this is evidence of poor pump throughput - pump is behaving poorly not because it is "choking" on gases in the chamber, but simply the throuhput is mediocre. When the chamber pressure is 1e-4 the foreline pressure is about 1..2e-2. When the pump is isolated from the chamber the pressure drops to 7e-3. You have to remember, that backing pumps have a huge througput, they are capable of pumping the entire chamber to 7e-3 (!)

As You have suggested I will dismantle the pump and check the jet assembly.

Cheers,
Martin.

(The green/blue line is foreline pressure. The orange line is chamber pressure.)
48987529_1232168530267173_1734176724607827968_n.jpg
John Futter
Posts: 1848
Joined: Wed Apr 21, 2004 10:29 pm
Real name: John Futter
Contact:

Re: Poor performance of old diff pump.

Post by John Futter »

You still have not given a boiler temp via an IR thermometer or even better K type thermocouple
and have you degassed the Bayard alpert gaugehead
Marcin Dziedzic
Posts: 6
Joined: Thu Dec 20, 2018 12:57 pm
Real name: Marcin Dziedzic

Re: Poor performance of old diff pump.

Post by Marcin Dziedzic »

I don't have a thermocouple in the boiler, pump was not equipped with such a feature. I've got J type thermocouple on heating elements. As I have read in the manual pump was meant to be operated in an open loop. However i was running heaters thru PID controller due to strange high temperature reading, raising temperature 10 degrees at a time. As you said the pump suffers from poor contact between heaters and boiler. Now the heaters T/C shows 510 (!) degrees C and the boiler temperature via IR thermometer is 221 C which seems to be correct for DC704 oil. The pump is now being run in open loop, as it was intended. The gauges have been baked out and double checked, also penning gauge has been used.
Bruce Meagher
Posts: 148
Joined: Thu May 12, 2011 11:25 pm
Real name: Bruce Meagher
Location: San Diego

Re: Poor performance of old diff pump.

Post by Bruce Meagher »

Thoughtput on a diffusion pump is closely related to the input power transferred to the boiler and the cooling of the pump. An analogy of a flowing river has been used to described how this works. A slow boil (either too little power transferred to the boiler or too much cooling) will produce a small amount of oil vapor through the jets heading towards the wall (much like a small river). The ability to entrain molecules in this small river of oil vapor is limited. Now a vigorous boil will produce a large volume of oil vapor (much like a large flowing river) and will be able to entrain many more molecules (i.e. increasing the throughput). Of course it’s a balancing act with a diffusion pump because you have to remove the amount of input power with the cooling or you’ll have a thermal runaway.

John mentioned poor contact of the heater element to the pump as one potential issue. Since you recently replaced the heater elements maybe this should be rechecked. Are you using some type of thermal paste? Also have you verified the new heaters are actually consuming 12kW?

Next would be your cooling loop. You mentioned you’re running at a higher flow rate than recommended. Can you slow the flow rate to the recommended value and see if you notice a difference?

On a closed system with minimal leaks, like our fusors, I believe the throughput of the diffusion pump isn’t very important. The region when this becomes an issue is of such a short duration for us it really doesn’t impact our pump down times. For larger leaks or processes with continuous flow, throughput is very important but also requires properly sized backing pumps to handle the load.

Bruce
User avatar
Dennis P Brown
Posts: 3159
Joined: Sun May 20, 2012 10:46 am
Real name: Dennis Brown

Re: Poor performance of old diff pump.

Post by Dennis P Brown »

If I read your post correctly, you say the chamber drops lower when the running DP has the fore line closed? If so, that indicates that the mechanical pump's oil vapors are back streaming for some reason into the chamber; this event could be caused by very contaminated mechanical oil (maybe burned oil) or very poor performance of the DP (i.e. its back stream pressure by the devices oil jets is too weak (possibly because of too much oil or too little or possibly due to the jet stack in the DP assembly being lose and/or the top cap not on properly.) Yes, poor heater contact could be an issue but if the heater is in good shape, that shouldn't be an issue - checking power consumed should help clarify that issue.
Bruce Meagher
Posts: 148
Joined: Thu May 12, 2011 11:25 pm
Real name: Bruce Meagher
Location: San Diego

Re: Poor performance of old diff pump.

Post by Bruce Meagher »

I looked up your roots booster backing pump and it does have a tremendous throughput. From the paper specs when the diffusion pump is at 1e-4 Torr and the foreline is at 5e-3 Torr the throughputs look similar at 2 Torr*liter/sec (assuming your DP is similar to the Varian HS-20). Your realized throughput will be different than the specs based on all the plumbing.

Have you thought about modeling your vacuum furnace to see what you should expect (see viewtopic.php?f=10&t=12114)?

Also if you have access to an appropriately sized MFC you could measure the throughput of your system by plotting the chamber and foreline pressure vs flow rate of the MFC. If your diffusion pump’s throughput is really the issue then adding a 1 mbar*liter/sec flow will be clear. Your leak detector only measured a 1,9e-2 mbar*liter/sec leak for the entire chamber.
Attachments
HS-20.jpg
Marcin Dziedzic
Posts: 6
Joined: Thu Dec 20, 2018 12:57 pm
Real name: Marcin Dziedzic

Re: Poor performance of old diff pump.

Post by Marcin Dziedzic »

The pump has been dismantled, and turned out to be heavily damaged. The jest assembly is probably incomplete. The slits in jest assembly are bent in many places, and it's evident that the pump had been cleaned many times, marks from scratching with knife (!) are evident. :( The pump is not equipped with water-cooled coldcap and huge oil backstream destroyed two of my vacuum gauges (ITR90 and PTR91). I'll post a short story with photographs, for now I'm trying to replace it with new varian HS-20.
Jerry Biehler
Posts: 975
Joined: Tue Nov 24, 2009 3:08 am
Real name:
Location: Beaverton, OR

Re: Poor performance of old diff pump.

Post by Jerry Biehler »

The gauge heads ought to be able to be cleaned to get them working again. I would probably start by washing them out with mineral spirits to dissolve the oil and then acetone. I would be surprised if the filament on the ion gauge ever even turned on at the pressure you were getting.
Marcin Dziedzic
Posts: 6
Joined: Thu Dec 20, 2018 12:57 pm
Real name: Marcin Dziedzic

Re: Poor performance of old diff pump.

Post by Marcin Dziedzic »

Ion gauge in Leybold ITR90 turns on at 2.4e-2 and turns off at 3.2e-2. In range between 2.4e-2 and 1e-3 output signal is combined from both Ion and Pirani gauges, hence old sensor name "combivac". Cleaning attempts turned out to be unsuccesful, the sensor was turning on ion gauge even at 5e-2 mbar, so the Pirani must have also been defective. The gauge was 6 years old, maybe it was just worn out, i replaced it with a new one.

My new pump has just arrived :) I'll see how high vacuum can I get.
Attachments
59964569_459963034811539_2532766884743348224_n.jpg
Bruce Meagher
Posts: 148
Joined: Thu May 12, 2011 11:25 pm
Real name: Bruce Meagher
Location: San Diego

Re: Poor performance of old diff pump.

Post by Bruce Meagher »

Wow that’s an impressive pump! Thanks for sharing and updating your post. I look forward to hearing how your system performs with your new pump. Many of the combo gauges have replaceable Pirani sensor. I’m not sure about your particular gauge (or if it’s really worth your time), but something to consider before you trash it.

Bruce
Post Reply

Return to “Vacuum Technology (& FAQs)”