Hope you've got deep pockets if you want one of those. Digimesa will sell you them direct...
http://www.xtremesystems.org/forums/...53&postcount=9
Oh and...
For YOUR information, that's not very helpful for anyone already with a D5, Eheim, Iwaki...Originally Posted by Xilicon
Last edited by IanB; 02-06-2008 at 02:00 PM.
Well, they must add some grotesque surcharge for small orders...
The guys from effizienz gurus got them for about 30 euros each on the group purchase.
And I bought mine (similar model, just with G1/2 threads and designed for higher flow rates) from Falmer, which is horrendously expensive, for about 60 euros (£40).
For YOUR information, I mentionned the DDC as one with the ability and not others. In that case, the SF800 is a great and inexpensive way compared to those Digimesa (BTW, the charge they asked is outrageousHope you've got deep pockets if you want one of those. Digimesa will sell you them direct...
http://www.xtremesystems.org/forums/...53&postcount=9
Oh and...
For YOUR information, that's not very helpful for anyone already with a D5, Eheim, Iwaki...).
Ok, I created a new adapter based on a new idea of tapping 5V from a 4 pin or floppy molex. For this, I scavenged the cut off excess wire from the Yate Loon that I did for the radiator fan mod. Here is the cruddy result (I didn't bother sleeving it for a neat look right now for a simple test, I'll do it later) :
Installed in my loop in place of the other adapter :
It works fine and I get exactly the same reading range as the old one so either method works without problems. Martinm210, I think you should try this way in case your 680i is giving funky 12V feeding (maybe due to PWM, which can cause erratical behavior with non-PWM fans from what I heard around) with only the sensor wire going in your motherboard header.
P.S. : Disregard the gross tubing, will be replaced by black Tygon shortly.
Last edited by Xilikon; 12-11-2008 at 08:25 AM.
LOL, the lighting, the tubing or both makes it look like you are pumping sewage water
It look less gross than this picture with your own eyes but stillLOL, the lighting, the tubing or both makes it look like you are pumping sewage waterThe thing that made me say it's the tubing itself and not bad water is the reservoir which remained crystal clean. I guess it's placticizers leeching at his worst and I think I see why Excelon RNT is so cheap (paid 0.75$ per feet).
Cool, I'll try that.Ok, I created a new adapter based on a new idea of tapping 5V from a 4 pin or floppy molex. For this, I scavenged the cut off excess wire from the Yate Loon that I did for the radiator fan mod. Here is the cruddy result (I didn't bother sleeving it for a neat look right now for a simple test, I'll do it later) :
Installed in my loop in place of the other adapter :
It works fine and I get exactly the same reading range as the old one so either method works without problems. Martinm210, I think you should try this way in case your 680i is giving funky 12V feeding (maybe due to PWM, which can cause erratical behavior with non-PWM fans from what I heard around) with only the sensor wire going in your motherboard header.
P.S. : Disregard the gross tubing, will be replaced by black Tygon shortly.
My MB seems to read a fan RPM just fine, so that might just do the trick..
I think I might be doing the bucket test to see if the flowmeter is accurate but since I don't have a second pump, I cannot do this while I connect to the computer. I'm wondering if we can read the sensor output frequency with a multimeter ? I know the SF800 give 100Hz per liter per second, 6000Hz for 1 LPM.
I was trying that a little with a fan and some other test items, but the only thing I could pick up was 60Hz on AC power, not sure it's meant to pick up pulses like this on DC power.
I'll give that a try to though when I rewire it for 5V.
Anyway, I'm thinking about disassembling my loop tomorrow and using the stock Intel heatsink temporarily along with a cheap pci video card. This will allow me to still use my computer (at stock speed with crappy cooling) and clean my components.
With the loop out, I will be able to do some accuracy testing with a bucket/stopwatch while still connected in my computer for rpm reading. I did some calculations with the Flowrate estimator and I should be getting about 4.6-4.7 LPM with my current setup. The flowmeter read half of this and tweaking SpeedFan to multiply the rate by 2 got me in the same range. However, is that correct or not ?? That's why I will do further testing.
Not sure, give the bucket test a shot and check it out.
I just got my crystalfontz working, so I can try that out too.
I finally did the bucket/stopwatch testing with the SF800 plugged to verify the accuracy. I made a very interesting observation of the sensor rpm rate...
I was feeling the flow rate numbers is too slow for my taste, by about 50% compared to Martinm210's flow rate estimator. Just for fun, did the sensor/3000 instead of the sensor/6000 as expected by the SF800 specifications and it approached within 10% of the estimation.
With the bucket testing, before I started it, I see the SF800 is reading about 3.5 LPM so I expected it to fill 4 liters in a little more than 1:10 mins. Instead of this, it filled 4 liters in 35 seconds. 2nd try gave me 34 seconds and 3rd try gave me 34.5 seconds. By using simple maths (60x4 / stopwatch value), I got exactly the double of the sensor rate so this confirmed my suspiction that the flowrate reading is slow by 50%.
That's when I remembered reading somewhere (forgot where I saw) that to get the fan rpm accurately, don't poll for faster than once each 2 seconds. This mean that the fan rpm sensor will read each 2 seconds so each other second, the value is 0 causing it to read 50% slow.
Basically, if you plug it on the motherboard, be sure to configure it so you divide the rpm output by 3000, not by 6000. Remain to see if this also affect the BigNG/Sensor hub reading as well. I would like someone who own both of them (ladderman, I think) to check the reading.
EDIT : While poking Google for fun about that, I found a thread about the SF800 with a CrystalFontz display. The link is http://www.crystalfontz.com/forum/showthread.php?t=4106 and it mentionned that some motherboard will have a divider of 2 so you need to get around that divider to get a accurate number. This show how important it is to calibrate the sensor first against another mean of calculating the flowrate like the bucket/stopwatch method or using the flow rate estimator.
Last edited by Xilikon; 02-09-2008 at 03:57 PM.
Another reference about the motherboard dividing by 2 : http://forums.procooling.com/vbb/sho...&postcount=144 (BTW, the thread there is talking about the various flowmeters including the SF800 so it's a wealth of knowledge already).
Bah, I think my SF800 is now dead... While doing more accuracy testing last evening, it stopped reading all of a sudden then after a few mins of messing, it read in sporadically for 1-2 seconds then nothing at all.
This afternoon, I redid the connection thinking it might not be soldered right. Even with a perfect job, it doesn't work at allMaybe Martinm210's sample is also dead.
I just wrote a email to Dirk of Swissflow to ask if failures like that is common and to check if they are willing to send a replacement sample (it has a 10 years warranty).
Last edited by Xilikon; 02-10-2008 at 03:01 PM.
Debris or air bubbles? Both can kill a flow meter faster than you can say "Scooby-doo".Bah, I think my SF800 is now dead... While doing more accuracy testing last evening, it stopped reading all of a sudden then after a few mins of messing, it read in spoadically for 1-2 seconds then nothing at all.
This afternoon, I redid the connection thinking it might not be soldered right. Even with a perfect job, it doesn't work at all
I just wrote a email to Dirk of Swissflow to ask if failures like that is common and to check if they are willing to send a replacement sample (it has a 10 years warranty).
Are you using it with a filter?
No filter but with clean distilled water only. I didn't notice any debris in it and I'll investigate the guts but it's not easy to determine.
Martinm210 is DOA anyway.
Bummer....
I wonder if it's the 10V or so coming out of the fan header sensor connection? That's the only thing I can think of. If testing connected to a BigNG works, maybe it's the fan header??
Regardless, give the other one on the way a try and take a look at my wiring again to make sure I didn't mess it up. It could still work, I had trouble getting a DDC2 pump to register as well.
Yeah, if yours work with my setup, I'll be very careful with it and continue the testing while I try to deal with the dead SF800 with Dirk. If yours is dead as well, I'll let Dirk know...
I will also ask him if voltage difference can kill a SF800 or it's just current. If the voltage difference can cause issues, then my 2nd connection method with tapping on 5V off a molex would be safer.
Xilikon, did you ever try to screw the 800 directly into your PA 120.3? I guess you don't have the room, but would that work otherwise? In my book, less adapters/fittings = more reliable and leak-free.
I didn't but technically, it's possible. It's even a great idea as long as you can fit the tubing after the SF800. In my case, it's not possible because there is no room.
So 2 units dead because (probably) of an incorrect electrical mod.
I hate to say it, but this really just proves my point that it would be better to use your existing links to Dirk or whoever in the company to persuade them that there is a need for a version of this already wired correctly for a computer mainboard fan header and plumbed to match our PC tubing standards. There is a market here and they should be made aware of it. There really shouldn't be a need to mod resistors and diodes onto these things to make them work with standard, highly ubiquitous hardware, especially when the mod appears to be sheer guesswork anyway...
My SF800 is not dead after all
I decided to give a close inspection with a lightbulb behind the SF800 and I noticed there is some gunk in it. It turned up it was some hairI managed with a jeweler screwdriver and tweezers to remove 98% of it and I replugged it to test (with very very clean distilled water this time, along with a cleaned pump). It works again
It is very important to make sure your loop is very very clean or you will have a dead SF800. A good idea would be to do a leak test with clean distilled water for 24 hours, flush it then refill with clean distilled water again before prolonged use. Alternatively, use a good filter but I'm not sure anyone would want a filter in the loop for prolonged use.
Before we dig the grave for the 2nd SF800 I loaned to Martinm210, I would test it myself since it might be possible there is something fishy with their own fan headers or bad connection.
hehe. I actually have a appointment with a optometrist this thursday (booked since 1 month because my eyeglass was broken due to my son. I'm currently using a older eyeglass in the meanwhile).
When I'm working to remove the gunk, I had the opportunity to look at the mechanism very carefully and I see the impeller is struck between a plastic wall and a cross frame to center the impeller. There is not a lot of room between the impeller and the side support so even a bit of gunk can lock it but a good tapping on the side (to move the gunk from the hub) and tweezers will fix it.
when I think about how it happened, I realized that this happened while I'm doing the bucket testing with 2 buckets. I may not have cleaned them perfectly so it's possible the hair fell in the water then gunked it in the time to say "scooby doo'
That's another point in favour of the Digimesa flow meter. Notice the clearance between the impeller and the walls? You'd need a rock to jam it
And the 10 mm ID is a bonus too.
Downsides:
- It's huge.
- It's ugly.
- With 65 ppl the accuracy will suck with any monitoring option that doesn't let you set a sample window of 8-10 secs. (1/65)*60*60/2=27.69 lph deviation with the usual 2s sample window.
- Double the cost and may be difficult to source.
- You need adaptors, it's G1/2.
Yes, the Digimesa might be better suited for our needs, especially for loops with dyes and additives. It looks free flowing since water will go in a straight path. The SF800 have a wall so it will have collision and water will be forced to get around the wall (this is how it make a vortex which turn the impeller and generate the signal). the accuracy of the Digimesa might be a turn off too because I don't know of any software who can take a 8-10s sample and generate a value.
Well, it's up to everyone which one is best. For myself, I find the SF800 great as long as you have a clean loop, which shouldn't be hard if you take precautions. My mistake was doing the bucket/stopwatch testing with dirty buckets (not that dirty but not cleaned thorougly).
Posting Permissions
Reply With Quote
Bookmarks