StillDragon® Community Forum

Welcome!

Be part of our community & join our international next generation forum now!

In this Discussion

My PID Experiments

2

Comments

  • edited January 2016

    Every once in a while you come across a CNI16D53 ethernet on eBay for under $100.

    It's not a cheap unit, by far, but they resale for big discounts to the retail price. At the discount price, I think the value is good, as the ethernet gives you a really nice way to data log, and it's got some good features.

    The 3 I posted, they all have very good documentation.

  • these are close to @smaug over in hialeah... more than 10 available...

    Temperature process PID - OMEGA CNi16D53-Ei 1/16 DUAL DISPLAY CONTROLLER @ eBay

  • edited January 2016

    Newport i800-ei is a nice unit if you just need a temperature probe with ethernet data logging. It has no controls.

    Would be nice for basic vapor temp. I grabbed one for $40.

    Here is one for 50 - I believe it supports RTD even though they say strain.

    NEWPORT STRAIN/PROCESS CONTROLLER W/ POWER SUPPLY 90-240V~I800-EI @ eBay

    The one I use for vapor temp is the Newport INF-B (Beyond Infinity) ultra precision meter.

    This one has a mac address sticker on it, so it might have ethernet (but the model number indicates not, it's a field installed option). If it has ethernet - killer deal - this is a $700 meter if it's got web.

    Newport Electronics INF-B-C24 Digital Gauge @ eBay

  • edited January 2016

    Alright, @grim 's valve arrived but since I was in the middle of a run and clean up I wasn't able to install and test right away.

    Valve bodies are hole compatible but somehow my valve has threads while grim's has through holes. Nothing that can't be fixed though. Reassembled the plumbing and leak tested. Quick check of behavior in manual mode shows a much more gentle ramp up of the water flow over the range compared to 1.9 Cv.

    Next weekend I'll do another tuning session and post the results.

    Pics are the two valves in comparison

    image

    image

    1.jpg
    800 x 595 - 83K
    2.jpg
    800 x 600 - 76K
  • edited January 2016

    Damn, I sent you a slip valve? I didn't even know I had that, I know I have the same valve in thread somewhere around here. You can sweat on threaded adapters. I didn't even think to look, because I had no idea I had that.

    Ooh, nevermind, now I see it, the valve flange. I think this valve came attached to a thermal barrier unit, and had screws up through the bottom into a big plastic spacer. Feel somewhat better, thought I was losing my mind, I'd have never bought a slip valve.

    The side by side is a nice visualization of the characterization disc and really shows why a plain ball valve really doesn't work in this application.

    The smaller Cv looks like it constrains the flow to the narrow gap to at least 50% of the valve stroke.

  • edited January 2016

    I was able to make it work with the machine screws I had laying around, so no big issue on the installation.

    Yes, the discs give a good representation of how the constraint is happening.
    One more thing to note:
    If you install the valve body, make sure that the actuator movement is opening in the right direction. When I re-assembled first, the valve opened from left to right (looking at the pics) which generated a massive flow with very little movement and to me questioning my sanity. Once properly aligned, it starts with a trickle and slowly increases.

  • So how long to fill a 10lt bucket when 100% open?

  • edited January 2016

    Here were my numbers with that 1/2" 1.2Cv valve from a post back in 2014:

    Valve position vs flow (At 60psi municipal water pressure, full open) - 4 meters total hose (2 in, 2 out), 3/8th ID with push-fit fittings, zero head (with municipal mains it's probably irrelevant anyway).

    25% -> 120 L/h

    50% -> 450 L/h

    75% -> 600 L/h

    100% -> 800 L/h

    At this higher PSI, the flow curve appears to be very nicely sloped at the lower end 0-50%, very good control of the flow rates, but flattens out above 50%. There is some variability there, since this was done by filling a big 2 liter graduated cylinder with one hand, and a stopwatch in another.

  • Also make sure your flow is from A to B, and not the other way, it works but the valve is noisy as all heck.

  • So at those flow rates I'm thinking perfect for a 20-25kW boiler, depending on Dephl efficiency???

  • edited January 2016

    @CothermanDistilling can chime in, he is running about that ballpark power input (not sure which specific VG model he is running), I know it's a 1/2" though.

    This is the smallest Johnson valve though, there is no smaller that I am aware of. I think 60psi mains is about the top end of normal, so also keep that in mind, we're lucky enough to have really strong water pressure. I think you would be very hard pressed to replicate those flow rates with a pump, unless you are fine burning up KW to push a pressure pump for hours on end.

    When I used this little guy with a small pond pump (Rio 2100), the max flow rate was about a quarter of the mains (around 4l per minute), with good controllability from 0-100%. I swore the pump wouldn't last to see the light of the next day, but the damn thing never gave up. I could dial it back to a dribble, but these valves get a little bit tricky around the 0-5% open positions. Specifically, the input current/voltage that cracks the valve, you would think if you were using 0-20ma, 1ma would open the valve a tiny crack, that's not the case. The minimum open will vary a bit, from valve to valve, and run to run. If you have a good controller, you can play with this a bit by setting the minimum mA range at 1-20ma if you are using 0-20ma on the valve, or 5ma if you are using 4-20ma. Worst case, you get a very small amount of bleed by. You need to be very careful if you think the critical range for the valve is going to be in the 0-5% open range.

    In almost all my cases, the valve pretty much never goes over 50%, and seems my sweet spot is always about 25-50%. Though we haven't yet run the big rig in the hot hot summer, with warmer cooling water, which I'm sure will swing that range much higher.

  • I will have to measure my PSI(assume 5-10psi, but I am using a small centrifugal pump (slightly larger than a 809, not as big as an 825 march pump.... it and an 809 max at about 4gpm, ~960l/hr , I use about .5gpm for each 5500w element that is on for a 60-70F rise in temperature... you know, a year or so ago, I put out here a .xls , I need to look at my data ans see if the math holds up to what I saw...

  • @CothermanDistilling said:...I use about .5gpm for each 5500w element

    Sounds about right. I use about 1 lt/min for around 2kW. You can calculate it pretty easily.

  • Here is the thread and spreadsheet.

  • I think this is about right.

    FlowRate in Lt/min =
    60 * kW / (SHw * DeltaT)

    SHw = 4.185
    DeltaT = Temp out - Temp in
    kW is power in the vapour which is probably 80% to 90% of what your controller says depending how well you insulate and how cold your workshop is.

    You shouldn't need to worry about the vapour side properties too much for this other than the temp.

    So 2kW boiler with 25° feed water and 55° water out (an ave efficiency HEX) gives abound 1litre/min which is about right from my experiences.
    A bigger (more efficient) HEX could get that 55° closer to the vapour temp which would reduce that flow rate down further.

    One very inefficient way to get a bigger valve to work is by putting in a smaller dephl which I did with one of my designs.

  • edited January 2016

    Alright, I installed the 1.2Cv valve and decided to give it a whirl today.

    My thought was to try to tune at 100% power with a low set-point (= need for higher flow). With the current weather, my supply water comes in at a little under 8C/46F.

    First off, I wanted to see where the temp levels off at various levels of output. For that, I put in an offset to the output that water would flow @ 5% output level. Looking at the huge jump from 10% to 5% I'm already not feeling to confident that this exercise is going to be more stable.

    I then went on to run an auto-tune. After two cycles, it switched back to PID mode and over the next couple of minutes, it stabilized on the set-point. I then wanted to check how the controller would react to a change in power and switched to 60% power. Result was that I now had an oscillation :(

    Parameters at this point:

    • P 135
    • I 36
    • D 9

    I then continued to try and tune at different power levels and set-points and was unable to get to a stable temperature level. My goal would have been to maybe get a parameter set for reflux and another 1-2 for different levels of temperature while taking product off (my controller has 4 settings that can be stored with easy switching). I no other setting was I able to get a stable temperature for the rest of the day... crap.

    My result so far:

    • At low temperatures (=high flow) and high power input (=high flow) tuning seems to be achievable.
    • Once power is reduced or set-point increased, even the 1.2Cv valve has no enough control on the lower flow.

    @CothermanDistilling you mentioned that you think it's tuneable. My observation at my take-off setting (65% Power / 55C RC) was that the valve's minimal opening is already leading to oscillations. I don't know enough about practical tuning of PID's but it seems to me that to maintain higher temperatures, the valve would have to be moving pretty much permanently. How would you tune in a situation like this?

    image

    image

    20160117 PID Tuning - P100 - Out 50 to 0.jpg
    800 x 581 - 33K
    20160117 PID Tuning - P100 - AT1.jpg
    800 x 581 - 42K
  • edited January 2016

    So, after my set back on PID tuning, I figured I install a meter in my power box to be a little more exact. Tested at 110V - worked. Installed with 230V worked for 5 seconds the smoke coming out of the box. Crap!

    Looks like a resistor didn't quite make it to the rated 300V...

    image

    image.jpg
    600 x 557 - 58K
  • Cap looks slightly cooked too. It's not one of the ones that need to go on the supply side is it?

  • Not based on description on website

    Unlike the other voltage gauge on the market, Auber DVA-120 does not require any external DC power supply; it will be self-powered when connecting to a circuit with testing range between 80 ~ 500 VAC.

  • Try again with significantly more integral. Pick a few arbitrary points and do some trial, error. For example, try 125, 250, 500, and 1000. Since you have data logging, this should give you a pretty easy way to compare.

    On municipal water, I did run this off a spigot, and I did dial back the flow rate on the valve input. On pump, I didn't need to, but the flow rate was very very low in comparison.

    What was the big head scratcher for me was that with the little rig, my integral numbers were nowhere near as high as I am running on the big rig, and it wasn't until I pegged the integral higher that I was able to reduce the oscillation.

  • @grim I'll give it a whirl. Might be a couple of days though as one of my fermenter cracked and I did an emergency racking into the pot. Don't want to run white experimenting

  • I hate autotune... OK, now that that is said, do what @grim says... I say start with 1000 for the integral, it will not bounce with that, it may take a minute or two to get to temp, though..

    your 8 degree C water may be an issue, that is CHILLY... I did a run yesterday here in Florida and water was 20C and I had half the flow I normally do when it is 28-30C, my flow meter was shutting off because it was going under .1gpm...

  • Municipal water temps are great here for cooling applications. In summer, I'm not getting higher than 16C.

    On the downside, we have about 6 month of winter and I'm usually done after January...

  • @Unsensibel said: So, after my set back on PID tuning, I figured I install a meter in my power box to be a little more exact. Tested at 110V - worked. Installed with 230V worked for 5 seconds the smoke coming out of the box. Crap!

    Looks like a resistor didn't quite make it to the rated 300V...

    You're supposed to keep the magic smoke inside of the box

  • your 8 degree C water may be an issue, that is CHILLY...

    We are seeing issues with very cold water this time of year now too, the system is entirely more stable when the recirculation tanks warm up a bit.

    Issue we're seeing is low flowrate into the dephlegmator causing significantly more reflux than the dephleg temperature would lead you to believe, primarily because of the very cold injection water, and the stratification at the bottom of the dephleg. I think a number of guys have indicated the same over the years.

  • Maybe reconfigure or create a by pass that allows the dephleg to take water from the PC discharge during this time of year?

    StillDragon North America - Your StillDragon® Distributor for North America

  • This is why I keep pushing the temperation loop angle. You could be running >0° glycol and it would make no difference.

  • We had a kind of attemperation loop on the delpheg originally but reconfigured because the dead time was impossible for the PID to deal with. The oscillation was impossible to control, even with very high pBand/low gain.

    The big difference though, was our plumbing lengths were long and system volume was very large.

  • Dead time? Sounds like the layout was the issue not the concept.
    What flow velocity? What was distance between RC and where the coolant feeds into the loop?
    It should be pretty easy to minimise. One of the benefits of a loop is it should drastically reduce the lag time IMO.
    I really need to get my shit together and build mine. I'm all talk at the moment.

  • edited January 2016

    Thinking out loud.....With a PC large enough to hold a measure of "reserve" water, you would be able to use preheated RC water supplied by the PC with out affecting the knock down capability of the PC.

    StillDragon North America - Your StillDragon® Distributor for North America

Sign In or Register to comment.