TEC pulsed load causing voltage sags

[OBCblackhawk]

hello, I am no electrical engineer by any means but I've taken up a interest in electronics and whatnot. Now that we have my novice status out of the way here is my problem.

I have a GE Medical TEC water chiller, uses peltier junctions to cool or heat the water. It works via a PWM circuit of some type, and it works very well. The problem is it causes some severe flickering of the lights on the same circuit. I am just wondering if there is a way to suppress these sags as my parents are about to kill me because the flickering is quite annoying. I have tried a 40uF motor run cap wired parallel to the load but that did nothing, according to my father it seemed worse. I've tried hooking it to a isolation transformer albeit a small one and hooked the run cap on either side; still didn't help suppress the sags. My fluke meter reads inrush on the pluses to 4-4.5amp for about a second and my voltages only sag to 119v from 121.5v according to the meter. I would like to see what a scope says but I don't have one so the DMM is the best I can do.

Anyways do you guys have any ideas on how to suppress these voltage sags, hopefully on the cheap side.

BTW the chiller is being used for computer water cooling and does an excellent job of it, just wish it wouldn't annoy my parents with the flickering lights. Plus I am somewhat concerned about the PF coming into my computers PSU.

 

[n.karthick]

You can use Un-interruptible power supply (UPS) to protect your sensitive loads from voltage sags and interruptions, if it is affordable.

 

[OBCblackhawk]

I can give it a try if I can acquire one to test tomorrow, I have a old one with a shot battery and a really low current rating.. stupid thing has a 2amp breaker, my computer alone trips it.

Im open to more suggestions tho if you guys have them, I would really like to suppress these sags. But I hope a simple UPS will do the job, anyways thanks to anyone who looks at this and gives it some thought.

 

[MATLABdude]

Welcome to PhysicsForums!

Can you copy the information off of the electrical plate on the chiller? It should tell you how much power it uses, current draw, voltage, etc. As you surmise, what you see is probably just what the multimeter can catch.

Easiest way: dedicate a circuit to the cooler. (Possibly) cheapest method: buy a proper water cooler!
http://www.greenwayhp.com/water_dispensers/

Okay, my curiosity is piqued: why do you have a chiller unit at home?

 

[OBCblackhawk]

As I had stated earlier I use it to cool my computer, and it does a excellent job of it. Instead of using normal old watercooling I can get sub ambient temps with this and it appears to handle fairly high heatloads fine. Cools two different highly overclocked CPU's and I know of a guy cooling laser diodes with this. Plus why not?

Anyways the inrush on my fluke meter is pretty damn accurate I am referring to the voltage measurements, doesn't refresh nearly fast enough. The unit has a rating of 100-240VAC 8A Max. other than that I have no information on the unit as the manufacture will not release it to me.

Edit: I also don't want a water chiller, like you posted. They are pretty weak and low capacity compressor type units, I had one already from a water fountain and it doesn't handle the loads well. This TEC chiller is pretty big and I estimate it has anywhere from 600-800W of cooling power at its lowest temp setting which is 4C.

Edit 2: Also I already have the unit plumbed and set up as well as paid for, last thing I want to do is buy something else. Also this does not answer my original question.

 

[Ken Freeman]

The answer to your original question is not simple. People are asking you for more information in order to better understand your problem. Engineers usually do not "just know" the answer, we rather have skills at diagnosing situations and developing an explanation.

Usually the first step is to "get the obvious stuff out of the way, but correctly." That's why the request for nameplate information. Can you find any information on-line at GE's site? It would be good to have their actual model number, which should be marked somewhere on your unit. If not, find the correct area in GE's sites, and browse until you find something that looks like yours. See if any manuals or info sheets can be downloaded. They need to provide a technical data sheet for potential customers.

Next is usually to better understand the symptom.

Have you confirmed that the flickering does not occur due to the computers alone? If you've modified their power supplies for some reason, they could be putting out more line noise than previously.

Do you know the frequency of the flickering, i.e. once per second, too fast to count?

What kind of lights are flickering?

Can you measure the voltage change where the flickering is occurring?

Is your chiller connected on a bedroom power circuit? Is the circuit shared with the computers? What happens if you plug in the TEC on a circuit that has no lights on it? Are the same lights still flickering?

 

[Ken Freeman]

I have a GE Medical TEC water chiller, uses peltier junctions to cool or heat the water. It works via a PWM circuit of some type, and it works very well. The problem is it causes some severe flickering of the lights on the same circuit. I am just wondering if there is a way to suppress these sags as my parents are about to kill me because the flickering is quite annoying. I have tried a 40uF motor run cap wired parallel to the load but that did nothing, according to my father it seemed worse. I've tried hooking it to a isolation transformer albeit a small one and hooked the run cap on either side; still didn't help suppress the sags. My fluke meter reads inrush on the pluses to 4-4.5amp for about a second and my voltages only sag to 119v from 121.5v according to the meter. I would like to see what a scope says but I don't have one so the DMM is the best I can do.

Plus I am somewhat concerned about the PF coming into my computers PSU.

Your power factor question is a very good one. Will GE tell you what the nominal power factor is?

Does GE recommend or provide any ancillary equipment to filter the input, suppress line noise, or ensure that the cooler power factor is benign?

Overall I suspect the problem is due to unusual currents in your house wiring, rather than any broadcast electromagnetic noise. These could be caused by misbehavior of your chiller, either of your computers, or an interaction between that set of equipment. There could also be interaction between your equipment and the other stuff in the house, including the lights themselves if they use compact fluorescent bulbs.

What happens if you unplug one of your overclocked computers? What happens if you substitute unmodified computers?

 

[OBCblackhawk]

It does not matter where you plug this chiller in, the problem is it that it requires fairly large pulses of current. Similar to when a large motor or compressor kicks on like your AC or fridge. This thing only draws about 300-500mA at 120VAC when not pulsing the TEC (peltiers) and draws 4-4.5A, the thing is it pulses constantly in order to maintain a temp. Similar to PWM motor drive, except this is PWM temperature control and no GE will not provide anything. Technically I shouldn't have these; but they are custom made units by SMC for GE for cooling medical equipment and or lab equipment to my knowledge anyways. The only thing I could personally think of that would stop this is a large capacitor bank motor run caps, but this isn't very practical.

So basically it is a very uneven load that pulses 4-4.5A every 1-3 seconds and drops to .3-5A afterward, plays hell on residential wiring. All our rooms in this house are wired with 12ga wire instead of the recommended 14ga so the wire isn't really undersized.

 

[OBCblackhawk]

So the UPS I have is completely shot, and I am not about to go buy one to find out it will not work. Nobody has any ideas whatsoever on how to prevent this, I can't believe there isn't some solution. Bleh, this is why I hate AC if this was a DC load a simple old cap would completely fix this problem. maybe I can get a large 24v cap or something to add into the PSU on the chiller if possible...

 

[Ken Freeman]

actually there are a lot of ideas, but none are simple. One possibility is that your TEC is shot. I don't think your current pulses are what's causing the flicker on incandescent bulbs, because a 3 or 4 volt drop is really not very much.

You don't know that noise problems are inherently complex, just wanting simple answers.

You think it's running as a pulse width modulated power supply. I think there may be an impedance interaction between the three PWM converters you have connected in your system: the TEC and the two computers. But there's no way you are going to diagnose that without good knowledge of the closed-loop dynamics of such converters. You might understand that after 4 years of engineering school, but most BSEEs don't learn about it until after years of practice in the field.

My recommendation: de-clock your computers, and buy some that are actually fast enough to run your games or whatever, and are properly engineered for this usage. Bandaids are not going to fix this problem.

People who hot-rod cars sometimes find their engines blowing up. Same with electronics. You've found the limit.

One of the other good ideas is to get a good UPS. But you have already cast that out twice. Nobody has any ideas you'll accept.

Think about trying an LC filter. Make sure it's not underdamped, in fact it should be overdamped. If you're interested ask the right questions and be receptive and we can help you see if such a design is feasible.

 

[OBCblackhawk]

LOL, this is a inherently uneducated response to overclocking... Look I have 3 of these chillers they ALL do the same thing even with every computer off in the room. A 3v sag in line voltage is plenty to cause a noticeable dimming of the lights. I fully understand the problem is complex, AC and inductive loads are immensely complicated with power factor with all the out of phase voltages and currents. But when you pulse a 4amp load for only a second then turn it off it will dim ANY residential lighting circuit. I don't see where you get off telling me these are broken, I am silly for overclocking or whatever.

Look at it this way forget the computer for get it is a chiller. SIMPLE terms I have a 4 amp load that pulses every 3 seconds with ~1sec duration. Offer a solution, infact I've pretty much come up with my own solution to the problem. I am not a complete novice to electronics and I figured I would get some "engineers" insight into this. But the best thing I can come up with is a rectifier that charges a cap at 12-24V range then back to an inverter hooked to the chiller in order to suppress burst or inrush loads on the main circuit. VERY similar to industrial motors with start caps... man.. I had higher hopes for solutions here, it isn't that complex of an issue I would think.

Take a moment and think about that tho, take a 1-5HP motor and hook it to your home wiring flick the switch on and off... its going to dim the damn lights. does that mean the motor is broken, no it has a high burst load or inrush current or whatever you guys want to call it. Causes a voltage sag, might only be 2v but you will notice it. Fridge, AC, microwave any high power device is your house will cause the same dimming of the lights, especially if it is on the same circuit. It just isn't much of a problem as it isn't a constantly PULSED load. when you have the lights dimming just a little ever 2-3 seconds even just a tiny bit it is quite noticeable and annoying. Bleh I am kinda annoyed by these comments, don't worry about the why I need it or my overclocking it has nothing to do with the problem. ONLY reason I mentioned it is constant voltage sags couldn't be healthy for the switching PSUs in my computers.

Anyways thanks to anyone who offers actual solutions rather than buy better computers, don't overclock, or buy a different cooler.. Also one other thing there might be some commercially available continuous UPS units that could fix this or even some line conditioners that could work but that are designed to suppress sags coming not from the applied load; but I see no reason it couldn't work, hell could maybe even add a large 1F cap in parallel with the charging circuit on the UPS.

 

[OBCblackhawk]

Just wanted to add I haven't cast out a UPS completely I just wanted to be assured it could suppress these sags. I am not into spending money on could helps... hell nobody mentioned there are different types of UPS's which would make a HUGE difference in this application... Really though, why so concerned about the computer, don't knock something you don't understand. Overclocking is a hobby, I've been overclocking for years now infact every computer in this house has been overclocked. I can assure you they have NOTHING to do with the problem at hand.

Basically what I have determined, you must convert into DC, store that energy then invert it back to DC and it needs to be always on, unlike switched UPS's.. So a continuous UPS which is VERY expensive, several hundred dollars for some, or some kinda line filter, which also seems very expensive. although I have found some cheaper ones around 100 bucks but I have no idea of how they work. They could simply call a isolation transformer or some kinda LCR filter a line filter. What I need is energy storage, with some kind of AC reproduction and I highly doubt these 100 dollar units will do the trick.

Sorry for blowing up but man it annoys me when people give me lame answers like stop overclocking or whatever and dodge the problem at hand.

 

[Ken Freeman]

Just wanted to add I haven't cast out a UPS completely I just wanted to be assured it could suppress these sags. I am not into spending money on could helps... hell nobody mentioned there are different types of UPS's which would make a HUGE difference in this application... Really though, why so concerned about the computer, don't knock something you don't understand. Overclocking is a hobby, I've been overclocking for years now infact every computer in this house has been overclocked. I can assure you they have NOTHING to do with the problem at hand.

Basically what I have determined, you must convert into DC, store that energy then invert it back to DC and it needs to be always on, unlike switched UPS's. So a continuous UPS which is VERY expensive, several hundred dollars for some, or some kinda line filter, which also seems very expensive. although I have found some cheaper ones around 100 bucks but I have no idea of how they work. They could simply call a isolation transformer or some kinda LCR filter a line filter. What I need is energy storage, with some kind of AC reproduction and I highly doubt these 100 dollar units will do the trick.

Sorry for blowing up but man it annoys me when people give me lame answers like stop overclocking or whatever and dodge the problem at hand.

I think you said at the outset that the 4 amp pulses resulted in a voltage drop of just a few volts. I doubt that would cause an easily noticeable flicker in an incandescent light. (By the way, is the light actually pulsing on and off, or is it a very subtle flicker?). You can test this by plugging 5 100 watt incandescent lights into a power strip, turning them all on, then controlling them all simultaneously using the on/off switch on the power strip. That will inject pulses of about 4 amps into the branch circuit. If it causes flickering in sync with your switching, it confirms your hypothesis about pulsed voltage drop, regardless of what I might think. But my thought at this point at least is that a few volts drop shouldn't be causing a lot of flickering. I don't know off hand if a little flickering is irritating and causing family complaints.

I was then thinking there could be coupling between the power supplies of the TEC and the two computers, leading to a low-frequency oscillation, since you have multiple second order active systems connected together at their input ports. Due to answering you very late last night, I wasn't at all clear, and mis-spoke about the overclocking. I was thinking you might have modified the power supplies. The potential for a coupling problem is straightforward to explain, but you have to understand Bode plots, stability theory and port impedances. But its premature to assume this is the problem.

It bugs me when someone who really doesn't know much electronics, as you seem not to, makes an incomplete problem statement asking for solutions, gets irritated when people who do know how to deduce possible solutions ask simple questions, and then says he knows the answer anyway.

It's pretty simple to test your current pulse hypothesis as I described, or some other method you might come up with. It should be the next step.

 

[OBCblackhawk]

the lights are not turning off, they are just dimming a small ammount, but the pulses are so short and rapid it appears to make the lights flicker. It is very irritating and very noticable, my father has complained about it 3x now so effectively I have had to shut down several hundreds of dollars of equipment b/c of this flickering. About the only cheap solution I am coming up with is actually running a dedicated line. I called tripp-lite and APC about their power backup and correction equipment. I was told the only thing that will correct this type of load might be a ON-Line UPS or a always switched UPS, where it is always drawing from a battery and is always being charged. The battery and optionally a 12 1F cap that I could add in parallel could potentially smooth out the sags enough to where they would not be noticeable.

I don't understand why this is being made more complex of a problem than it is, engineers always seems to do this. They always seem to never think in real world situations, just an example you make a spa, but put the drains 2" above the lowest point in the spa.. There are thousands of things like this I and my father always seem to run into. Anyways, I said before, I don't have a scope to hook up to the line.. I am sure it is sagging a good 5v or more for a couple ms, but my fluke meter refreshes maybe every second, it is impossible to measure it. The inrush function works great and without it on I never see the current go above 3.5a typically it read 3 or so for a fraction of a second and refreshes, just to prove a point...

I is a simple problem, no crazy voodoo out of sync current and voltage transients or anything like that. I am not sure the exact terminology for what causes these sags, but common sense tells me unless you WAY WAY oversized the wire sags will occur when high loads come on. I already explained a dozen times that it is very similar to a large motor, microwave, space heater etc.. it requires a high burst of current then immediately subsides, if this wasn't pulsed it wouldn't be an issue as the lights would dim then once the starting current passes it would return to normal. In fact when the chiller runs a 100% duty cycle this is exactly what happens, the lights dim the instant it demands current then everything appears fine.

So the simple solution here, is to lessen the instantaneous current demand on the AC line, and the only way to do this that I can think of is storing DC in caps or a battery then inverting it back to AC. As long as the charge circuit is regulated and doesn't follow the load, think of it like a flash circuit in a camera.. minus the high voltages, the battery could never supply the burst current needed therefore the flash would be dim or not even flash. The battery can supply the energy just not quick enough. Simple, I don't get why this is being made so complex. Anyways if this AC to DC storage then back to AC thing sounds like it would work let me know. Also if you guys have any other solutions let me know, I don't really want to spend 1k bucks on a On-Line UPS system... Seeming the "easy" solution would be to make a dedicated circuit for this which would require a ton of work but would be much cheaper in the end.

 

[Ken Freeman]

When I hear of a noise problem, I think of all the potential noise causes and coupling methods, and try to gather the information I would need to narrow down that list.

If that looks too complex to you, that's your lack of perspective. It's also false to look at a problem as complex, then when it is appropriately simplified, find fault for having gone through a diagnostic thought process.

I think we're in agreement about the solution. Being someone who can design the kind of UPS you're talking about, I think you'd be much better off buying one than designing one. See if your dad wants to have a homemade device on the circuitry of his house, designed to uncertain safety requirements and not tested in a certified safety lab, or one that is already designed and tested to be safe under a wide range of normal conditions. For my house I'd consider the $1000 a sound investment.

I think your AC to DC to AC thing is the same solution I outlined. I think my opinion about it has already been made clear. You are correct in assessing that an energy storage bank will need capacitors and batteries, due to the potential speed of the pulses and the limited bandwidth of the battery. I don't think its being made too complex, I think it's being made complex enough.

If you want to talk about whether you should try to design such a UPS, or if you find some plans you'd want reviewed, I think that's a reasonable question. Asking why it has to be so complex is not a reasonable question.

 

[OBCblackhawk]

Installed a new wiring circuit.. easy stuff, anyways I am plenty competent enough to build a simple little UPS like that. I mean there isn't anything that complex about building a rectifier and voltage regulator circuit and or step down transformers and I can easily buy a 1KW inverter. I don't get how something so simple could cost so damn much, but then again this is an enterprise class UPS and it has a 2yr warranty which these companies charge insane amounts for.

Anyways.. problem solved thanks for your time.