Help with PSU (Temp control fan, load bank, & PWM circuit)

[()blivion]

Yeah, Agreed. Changing BOM.

DigiKey BOM (Bill Of Materials):

10 x STP80NF12....................= $11.20
10 x AC07000001307JAC00...= $7.84
5 x LM358P......................... = $2.25
2 x P160KN-0QD15B1K..........= $1.68
1 x RNF14FTD23K2................= $0.15

Total minimum cost: $23.12 + S&H

*optional*
5 x small bypass capacitors for the Op-Amp's ≈ $2.50
10 x ALSR10R1300FE12....= $17.60 (instead of 10 x AC07000001307JAC00)
1 x Better Pot that comes with a knob.
More than the minimum amount of each part, In case you break a few.

Extended Cost: Should not need to exceed $50

I'm using 1458's and it works fine though. And they are not rail to rail. But that means nothing.

 

[ronv]

Extras

If you change the sense resistors to .1 ohms you could use a poor mans current limit to limit the current to less than 6 amps each.

Coukd amplify the voltage across the sense resistor (1) and read amps with a volt meter. Maybe one of those liitle ones you see on Ebay. It needs to be the kind that can read voltage relitive to it's own ground. Some can't.

I think I would go with 2 supplies. Who knows what evil lurks in a supply that doesn't pass.

 

[ronv]

Maybe the 2N2222's should have 1K or so in the base

 

[jocanon]

I will hold off on ordering parts until we get it all ironed out.

I did get the copper piping and fittings from home depot. I decided just to go with the garden hose method, but not really a garden hose...I got clear vinyl tubing that I am going to connect to the faucet and run the water through and just drain it into the laundry tub in our basement. I was just thinking as I was buying the copper pipe at home depot, what if I get condensation on the pipe? It doesn't seem like that would work too well with the electronics. It probably won't happen though since I don't think the pipe will get cold enough when the load bank has current running through it. It might though if I turned on the water and let it run for a while before I turned on the power on a humid day...so maybe I just need to be careful not to do that. Or maybe I should put a fan on the copper piping to keep it dry, hopefully that's not needed...any thoughts? Is this a valid concern?

 

[()blivion]

Yeah, I've dumped my whole unit in the water tank a few times, shake it off, runs fine. A little FRESH water won't hurt it. Just gota keep the sensitive op-amp inputs from getting wet or you may have problems. Fresh water doesn't really conduct electricity. A lot of metal soldering fluxes are acid core, which would make the water hundreds of times more conductive. A 99 cent can of generic acrylic clear coat works wonders for conformal coating.


Current limit

I would really like a current limit. I would like a latching shut down though, Like whats in a PC PSU. And even though it's not the greatest idea, we *COULD* have such things trip off of just one of sense resistor... can't we? It won't protect all the FET's, but if things are working, it should keep the OP from turning the current up too much by accident, that's all we really need. Cheep slot style car fusses can work for global over current IMHO. And maybe get and use a spare op-amp for the shutdown?

Edit: Misinformation, older circuits are irreverent.

 

[jocanon]

Your Questions
#4 Get two RNF14FTD23K2 and any random switch I guess. That will give you a 24v mode and a 12v mode. Don't run it off 24v in 12v mode.

You lost me there...can you explain what you mean, why I would need two of those resistors and a switch?

 

[ronv]

Current limit

I would really like a current limit. I would like a latching shut down though, Like whats in a PC PSU. And even though it's not the greatest idea, we *COULD* have such things trip off of just one of sense resistor... can't we? It won't protect all the FET's, but if things are working, it should keep the OP from turning the current up too much by accident, that's all we really need.

Yea, I think so to. I was thinking about that during my nap. If a FET shorts nothing will save it anyway.

Cheep slot style car fusses can work for global over current IMHO. And maybe get and use a spare op-amp for the shutdown? So we don't have to use 100 milliohms resistors (my math says LOTS of power dissipation with 100 milliohms)

Lets see what we can come up with. But we could just limit the maximum current with the adjustment.

Did I miss a post? I was puzzeled by your comment about 1 current sense resistor? Are there still 10 130 milliohm resistors?

 

[()blivion]

OH wait.... Never mind, I'M the one who's confused ronv. You were right.

My thinking on the 100 milliohm resistor idea was from THE OLD circuit with a single current sense point... Which would explode with that. But that's apples and oranges from the new circuit. 130 milliohms to 100 is fine. I'll change the BOM.

You lost me there...can you explain what you mean, why I would need two of those resistors and a switch?

Sure thing, here is how we will change the schematic. Unless any objections come forth.

View attachment 66002


Sorry I'm being confusing as hell today people. I have more than a few projects going right now.

 

[jocanon]

()blivion, I think it's me, not you, but I still don't quite get it...so with the new schematic are you saying that we need to have two of the resistors RNF14FTD23K2 and the switch will make it so when I am connecting to 24v I only use one of the resistors, then when I am connecting to 12v I flip the switch and I am now connecting both resistors and that is the only difference?

 

[()blivion]

()blivion, I think it's me, not you, but I still don't quite get it...so with the new schematic are you saying that we need to have two of the resistors RNF14FTD23K2 and the switch will make it so when I am connecting to 24v I only use one of the resistors, then when I am connecting to 12v I flip the switch and I am now connecting both resistors and that is the only difference?

Correct. It's all about altering the "programing" voltage getting sent to the op-amp control loops. What this modification does exactly is halves the resistance to the pot, which doubles the programing voltage for a given load input voltage. This doubles the current the device draws, because the control line is telling it to do so. Since we are going from 24v to 12v, or exactly half, it makes sense to be able to double the control voltage to balance things out... right?

 

[jocanon]

Ah ha, light bulb just went on. Got it, thanks!

 

[ronv]

If you go with a separate supply for the op amps the program voltage would not change so you would have the same current with either supply, 12 or 24.
Nothing wrong with the 130 milliohm resistors if we can come up with a good current limit/latch. I've been looking at one with an scr, but having trouble with it.
How about just making it so the pot can't be adjusted for more than 5.5 amps per FET?

 

[ChrisP58]

All commercial electronic loads use an independent supply to power the op-amps and control circuit. That way they can be used down to almost zero volts.

But there is another problem with deriving the control pot voltage from the 'Load' voltage. That is a positive feedback path that can cause the system to oscillate. If the load voltage sags a little, either from reaching a limit in the power supply, or just from wire resistance, it will cause the control voltage to sag a little, which reduces the current. That, in turn, lets the load voltage rise back up, which increases the control voltage. Repeat.

At the very least, you should stabilize the voltage feeding the control pot with a 5v zener. This will also make it easy to go between 12V and 24v testing.

 

[jocanon]

I think it might be better to just go with the separate supply for the op amps, it would prevent accidentally connecting a 24v supply and forgetting to change the switch from 12v and blowing everything up, right?

 

[()blivion]

Sorry I wasn't here last night, busy with DTV stuff.

@jocanon: A separate supply for the op-amps, and the addition of the 12v-24v switch are slightly different and unrelated things. Even with a separate supply on JUST the op-amps, one can still over current a 24v DUT with the switch in the 12v mode. *HOWEVER*, using a separate supply for both the op-amps AND the control line (instead of the switch) should/could make it a constant current system, (draws the same amps regardless of the voltage), thus eliminating any need to modify the existing load control schema.....I think we are maybe on to something with this...

@ChrisP58: I *HATE* correcting people that are trying to help, makes me seam like a gigantic ass. And this is just a small thing anyway, so take no offense. But what your describing is actually negative feedback that does not have enough hysteresis. Positive feedback is like a snow ball rolling down hill, the bigger it gets, the faster it gets bigger. Negative feedback instability is more like the "fish tailing out of control" or "over correcting" process that you're describing. In any case, what you're describing is certainly possible, however mine is wired that way and does not appear to do this. Though I have not actually scoped it to know 100% for sure. We should still anticipate things like this though, you're correct. Another option is a RC circuit across the control voltage. This would add the before mentioned hysteresis.

I will probably do some more experiments with my setup and see what we can do with a separate supply hooked up to different things.

Interesting thoughts, unrelated to OP's needs
I had decided early on in this project that part of my objectives here (other than helping the OP) are to make a DIY load that is useful for many things. That being said, it occurred to me that if we changed the MOSFET's to P-Channel instead, we could still be able to solder them all on to the same single cooling pipe, but be able to make separate loads for independent power rails. We can't do this with the existing design because the pipe has to be connected to the positive of the DUT. It would be more preferred to someone wanting to make a dummy load for computer PSU's or similar to use use the common ground setup a P-MOSFET would provide. The current sense system would have to be changed of course, since we would no longer be able to insert the sense resistor between the FET and ground, but this is a small thing to figure out I would think. We would just use the FET as the current sense probably. Possibly to the point of even being able to remove the current sense resistor all together! I have only though about this a little bit though, so don't quote me. Only theoretical at this point.

Finally, the OP has different needs, and thus the cost savings of getting N-Channel MOSFET's are preferred. So this is not the place to discuss it. Maybe a new thread some time.

 

[jocanon]

@jocanon: A separate supply for the op-amps, and the addition of the 12v-24v switch are slightly different and unrelated things. Even with a separate supply on JUST the op-amps, one can still over current a 24v DUT with the switch in the 12v mode. *HOWEVER*, using a separate supply for both the op-amps AND the control line (instead of the switch) should/could make it a constant current system, (draws the same amps regardless of the voltage), thus eliminating any need to modify the existing load control schema.....I think we are maybe on to something with this...

Yeah, I thought about it after posting that question and was thinking they were not quite the same. I do like the idea of not having to switch between 12v and 24v by having a constant current system, the separate power supply for the op-amps and control input doesn't bother me so much, in fact it seems like the better way to go, from the limited understanding I have and based on the other posts...but just to make sure I am understanding, you are saying we would need 2 power supplies in total (not 3 right?), one to power the DUT and another to power the control input to the op-amps and the op-amps themselves? I would think that the power supply to the op-amps could just be a simple ac-dc plug.

 

[()blivion]

just to make sure I am understanding, you are saying we would need 2 power supplies in total (not 3 right?), one to power the DUT and another to power the control input to the op-amps and the op-amps themselves? I would think that the power supply to the op-amps could just be a simple ac-dc plug.

Correct. One power supply WOULD BE the DUT (Device Under Test) and the other would be a wall adapter (ac-dc plug) of some kind. The wall adapter would probably need to be regulated (stabilized) when it enters our circuit, to prevent any power line noise from causing the load bank to go nuts. A few mV change on the control is several amps change in the load. Such things *COULD* be disastrous if not properly designed out of the system. Good filtering and some regulation on the op-amp power should prevent this.

 

[jocanon]

So, here's an idea...maybe if I can heat things up enough instead of just draining the water into the sink...it can double as a tankless water heater, haha just kidding, I don't think it will get that hot...hopefully it won't get that hot

 

[ronv]

I'm still back looking at the features list.

Here is what I am thinking about current limit:

Possible failure modes:
1-FET shorts.
2- Op amp goes bad.
3- Logic supply goes bad.
4- Sense resistor shorts.
5- adjusted to high.
I don't think we can actually protect against 1 - 4 except to crowbar the test supply. We could I guess do that but I think the damage will have already been done.
5 can be avoided by using the 2 supplies.
My vote would be no overcurrent detect. What do you guys think?

Over temperature: Attached is a schematic that uses an LM35 temperature sensor to shut off the load at 100C and turn it back on at about 75C.
Also shown is an op amp to scale the current up to 5 volts = 50 amps. If we add a little voltmeter we could put in a switch to look at current, voltage and temperature. The temperature sensor is .01 volts per degree C. I still need to go back and check that the FETs are okay for 5 amps at that temperature.
I'm thinking a kitchen timer or something like that for the timer, but I'm not sure what exactly you want it to do.

I haven't thought about closed loop temperature control. I'm guessing this is to keep the basement from filling up with water when nothing is running?? Is that it?

 

[()blivion]

My vote would be no overcurrent detect. What do you guys think?

Yeah, let's just use fuses I guess. That should save us from just about any problem. Though.... of the three PSU's I broke, not a single one blew it's fuse... lol.

Attached is a schematic that uses an LM35 temperature sensor to shut off the load at 100C and turn it back on at about 75C.

I have no hard objections to your thermal shut down circuit. Just minor "what if" things.

First, I like the idea of things turning off and screaming/blinking at you to telling you SOMETHING'S WRONG!!! Rather than getting stuck in a loop where you may not notice anything. I had a project that would work for a while, screw up randomly, then stop completely. It was low power so for several days I didn't even think to check the temp of the LM7805 that had no heat sink. It now has a heat sink and works 100%. Would have been a lot easier to figure out the problem if there was some obvious indicator of what was happening. Even if such before mentioned loops work to keep things from getting destroyed. Indicators are nice.

Secondly, the OP has purchased a bunch of NTC thermistors, and knows how they work for the most part. It would be less accurate, but we could use something like that for over temperature. And I agree 100% that we need some over temperature protection. Even on mine at 50 Watts it got hot as hell after only a few seconds when the water was off. And even with clear hose and a some what noisy pump, there is almost no indicator that water is flowing. Things you gota check for that are easy to forget are EXACTLY what we want safety designs installed for. Takes the human element out of it.

Third, I know nothing about the LM35 temperature sensor. It would be your thing to help the OP with that, not that I'm incapable of learning.

Also shown is an op amp to scale the current up to 5 volts = 50 amps. If we add a little voltmeter we could put in a switch to look at current, voltage and temperature.

This is good, was thinking of this too. This allows us to do a wider range of things, based off of current level. One of those little analog needle point meters, like the ones you find in old voltmeters can run off less than 5 milliamps. I have one here, it's 85 Ohms, and maxes out at 0.002A. Something like this would be quite functional as an indicator. The OP needs to make a back plate with the right markings on it. This can be done with a LASER printer and sticky back labels. With analog we don't have to make the scales exact. We can just get it close and some what even, then adjust with trimmer pots. With digital, we need a stable reference and exact maths. More complex, but cleaner and more sharp looking results.

If I take this project farther in the future, I plan on doing something like this using a PIC and a text LCD module. Very flashy results can be had going this route. (warning!! Annoying music)