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

#### jocanon

##### Member
Just in the way of back ground, I am working on a project that I hope to be able to develop and then replicate so I can sell power supply units online to RC hobbyists. I decided, after reading the research that has already been done on the topic on RCGroups, that the DPS-600PB by HP pulled from file servers would be my best bet as a PSU to convert to bench power to be a supply for high end lipo battery chargers such as the PowerLab 8, iCharger, ect...
Now to the reasons I am posting in the Electro Tech forum:

Temp Controlled Fan
1. I want to make a temperature controlled circuit that powers the fan speed as the fan is very loud when running on full speed (which I am 99% of the way done with this part, many thanks to ()blivion...). There is a way to make the fan go down to an idle speed by shorting pins 4 & 8 and some have reported that the fan idles up when the load is increased by doing this, but there are three reasons I do not want to slow the fan this way. First, according to what others have reported, not all DPS-600PBs will idle up when current increases (there must have been multiple versions made, well, at least two). Secondly, I want it to be controlled by temperature, not simply current since there are other things that could cause the PSU to run hotter, for instance, if you are powering it outside on a generator in Phoenix, AZ in 120 degree weather (OK, that's extreme, but you get the point). Thirdly, by making it temperature controlled, instead of just shorting pins 4 & 8, I can slow the fan to a complete stop when it is not needed and have greater control over the range of speed when it is needed. So, as I mentioned above, I am 99% of the way done with this part, with much thanks to help from ()blivion from his write up called "Stupid Simple Temperature Controlled PC box fan driver." located at this link:

Sorry for the lengthy back ground, but now here is the point to question 1, I need to determine where is the best place to put the thermistor that will control the fan speed based on temperature. I took some measurements and it looks like the back heat sink, the one closest to the mains, the largest one, is what heats up the most between the two heat sinks, but the yellow cylinder (what is it called, I don't know???) right above the heat sink also seems to heat up pretty quick under load so I am thinking those are my two best bets. I will circle these two spots so you know what I am referring too. Can anyone tell from the pictures where they think the best place to put the thermistor would be? Can I use epoxy to glue the thermistor down? Can I glue it directly to that yellow cylinder or the heat sink, which would be best, or is a third option better?

PWM Circuit (possibly)
3. I want to also be able to use this dummy load to tune the fan settings, which means I will need to have different loads other than just max power in order to test the fan at a low current draw, a mid range current draw, and max. Ideally, I would like to make a PWM circuit that could handle this type of load (1200 watts) so that I could just turn the dial and turn the load up or down. I need help doing this though as I have never built such a circuit. So, does anybody know where I could find detailed instructions on how to build a PWM circuit that could control this high load? If the PWM circuit is too difficult or too pricy, I have another idea. Could I simply put bullet connectors on the resistors so that I could connect more or less of them in parallel and thus have 3 or 4 different points of resistance to get different loads? Like some sort of jumper cable where I could connect only one resistor, two resistors, three, or all four for max current...I think that might be the easiest way to go.

Those are all the questions I have for now. Thanks for reading and thanks in advance for any advice/answers. Please let me know if you think there is a better way to accomplish anything I have mentioned above (keeping in mind I am on a small budget), or if you see anything I am doing incorrectly, please let me know. Thanks again!

Last edited:

#### ()blivion

##### Active Member
The yellow tube thing in picture #4 is a Transformer and should only need minimal cooling. They get hottest when under low load if I recall correctly. The first heatsink (on the left) is for the Schottky diodes and can get moderately hot. The second heatsink (on the right) is for the power switch, and in a perfect world, would not get hot at all. This is because it should be in an always ON or always OFF state. In reality, they switch these at a really fast pace, and it takes time to switch from ON and OFF and back, which lets it linger in the middle and heat up some.

The problem is, all parts get hottest at different times, and for different reasons. So picking a heat sink to put your thermal probe on and sticking to it *MAY* not work out well. Though generally, if you must attach it to something, the heatsink that gets hottest under heavy load conditions would be optimal. Which should be the farthest back one, though I can't be certain about this.

Also, you could try and put the heat sensor in the air directly behind and above your hot parts (with respect to air flow and gravity). This should let the combined heat from all sources average together to determine the fan's speed. The only conceivable problems with this is that there may be a larger delay from when the heatsink heats up, to when the fan starts to speed up. But this delay is more or less going to exist anyway and should not effect the performance of your cooling system too much. Heating and cooling is usually a slow process anyway. Just make sure it's close to at least one heatsink, almost touching it. Another minor thing is, if the air gets turbulent enough, it may start to move cool air past your thermal sensor. Insulating it from the things that are actually hot. But this is much less likely as there is also a radiant (light) component to heat, and turbulence can't move this away from your sensor.

You could also attach to the farthest back heatsink and hope they share enough heat from air movement to be more or less the same temperature.

Or, you could try and bridge the two heatsinks with a NON CONDUCTOR, and attach your thermal sensor to that. Though this will be difficult to mass produce.

In the end, sensor placement is not too critical so long as you get it somewhat near things that gets hot, then tune the sensor's sensitivity appropriately. If things heat up, and there is no air moving, the whole PSU will heat up, eventually reaching the sensor and starting the cooling fan. Whether it will break before this time or not is debatable. The worst thing you could do is thermally attach your sensor to something large that doesn't really get all that hot, or cools unnaturally well. Such as the outer case. Then your almost certain to over heat. Don't do this.

-()blivion

#### jocanon

##### Member
There was a guy in the RCGroups forum that works in IT where they used a bunch of these PSUs when they first came out. He said in the first release of them, the fan was coompletely off when the PSU was in standby mode (no DC voltage on the rails, but plugged into AC mains). He said they left them like that for a while, weeks maybe days, I can't remember...then when they went to power them on the PSUs were all fried from being overheated. They ordered replacements of the same PSU and the new versions had the fan stay on at a low speed even when the PSU was in standby so they wouldn't have this problem. So, what part of the PSU would heat up in idle? Do you think the temperature controlled fan would have prevented them from frying? I would think so since it is temperature controlled, so once they start to heat up the fan should come on (assuming it had power). Another thing to note, I will be hard wiring the pins for the DC 12 volt rail to always be on, so it will never go into standby. I haven't started tuning the fan speed yet, but I envision it not coming on once I plug in the PSU, but then slowly starting to rev up as it sits for a while, even when no load is present, or should I tune it to always be on at a low speed even at room temperatur and then increase speed as it heats up? I would prefer to have it be completely silent when it's first plugged in. Maybe I am over thinking this. I probably just need to start experimenting with it, test it in real life and see if I fry anthing.

#### ()blivion

##### Active Member
Maybe I am over thinking this. I probably just need to start experimenting with it, test it in real life and see if I fry anything.
That's what I would do, with a good idea in your head of what probably is going to work and what probably won't though.

Do you think the temperature controlled fan would have prevented them from frying?
Yes. Being an independent control loop for your "fan + sensor" means that no matter what is happening with the PSU the fans will most likely spin when things get hot. So the "standby + no fan" scenario you mentioned should not be a problem with this modification. However, this is only true provided the fans still have power of course. If when in standby there is no power, and things do make a good deal of heat, then your going to have problems. On that note, it is highly unusual for there to be significant heat made when in standby mode with an somewhat ATX compliant PSU. The +5 V standby wire (purple) should be the only wire with power when the PSU is in standby mode. And this wire should only be able to draw an amp at most, which is only 12 Watts. This should not cause significant heating in the power supply even without cooling. I can't think of a reason why one would overheat and die in this scenario at all.

In any case, keep the fans running when in idle if you can. Safest bet.

#### jocanon

##### Member
Is this a link I could use to build a PWM ciruit to control a really high current?

http://www.freewebs.com/acselectronics/dc_controller.html

I found a bunch of MOSFETs on mouser that are capable of handling 300 watts or more and they are surprisingly inexepensive. So please let me know if I am thinking of this correctly... as long as I can make a circuit that can drive the MOSFET on and off really fast (the PWM circuit) then the only thing I really need to worry about handling the large current is the MOSFET itself as this is the only part that the huge current will be passing through, the other parts of the circuit are essentially just there to control the MOSFET, shutting it off and on, so again, as long as I can find the right components and wire them correctly to shut the MOSFET off and on at the right times, and the MOSFET can handle at least 300 watts, if I put one such PWM controller on each 300 watt resistor I should be good to go, right? I am thinking I will need 4 PWM controllers, one for each 300 watt resistor, because I didn't see any MOSFETS that could handle much more than 300 watts, let alone 1200 watts .

#### ()blivion

##### Active Member
Yes, you have every thing about perfect. That circuit is quite nice, though you can achieve the same effect with a slightly smaller circuit and a 555 timer IC. There is nothing wrong with that circuit though, and it is well explained.

Would you kindly put up a link to the MOSFET's you thinking of using? There is slightly more to a good MOSFET than just the Watts. And in some cases, you don't need to consider the Wattage at all. For this project and energy levels, it's best to get something with a high wattage. But the more important thing is that it can handle > 24 volts and > 12.5 Amps, which is easy for most MOSFET's. Also, A unit that is avalanche rated is preferred.

Normally, we would be talking about controlling a motor. Which is an inductive load prone to generating back EMF. However... your load bank/dummy load is going to be more or less a purely resistive load. This means we can eliminate the normally needed Schottky protection diodes from the circuit. I envision the system something like this...

View attachment 65854

Note: 1500 Watts is as much as an electric space heater. Your resistors WILL GET HOT. And will need a fan on them.

Last edited:

#### jocanon

##### Member
Thanks ()blivion, I see now, it looks like from your diagram you are saying all I need is one PWM control board to control all the MOSFETs. That is good. I was thinking I would need a separate PWM control for each resistor, but that makes sense, I can just have one PWM control and all 4 MOSFETs hooked up to it.

I am all about simplicity! If there is an easier way to accomplish the same thing then I want to know about it. I will google a 555 timer and PWM and see what I can find.

As far as what MOSFETs, I don't really have one particular one in mind yet. I will look for one that has the characteristics you mentioned and post it here.

BTW, that's a cool diagram you made. What program did you use to make it?

Last edited:

#### jocanon

##### Member
I think this is an example of a simple 555 PWM circuit:

http://www.dprg.org/tutorials/2005-11a/index.html

So ()blivion, are you saying I can eliminate D1 and D2 in the above link?

Also, per this link (below) I think I may need to step down the voltage to the 555 timer:

http://www.555-timer-circuits.com/555-on-24v.html

My question is, will this decrease the voltage running through the resistors too or will it not effect them in anyway so the resistors will still be getting 24 volts. I am almost certain the resistors will still have 24 volts because they are connected to 24 volts, the PMW control just turns the MOSFETs off and on. But I just want to make sure. I am also wondering if I will be able to rig this load bank to test 12 volt PSUs too so I will have to figure out a way to bypass the voltage step down when running on 12 volts to begin with. I wonder if I could just put a simple switch in there to bypass the voltage step down?

Last edited:

#### ()blivion

##### Active Member
I think this is an example of a simple 555 PWM circuit:

http://www.dprg.org/tutorials/2005-11a/index.html
Yes, a perfect example of what you need to do. Do note that you also need to have a push-pull buffer on pin 7 because the 555 IC can only pump out ~200ma max. This is not enough to reliably switch several MOSFET's ON and OFF. They will end up being switched to slow and heat up. The push pull from the first circuit you linked to in post #5 should work. The parts BC337 and BC327 make the Push-pull driver.

So ()blivion, are you saying I can eliminate D1 and D2 in the above link?
Not in the 555 circuit, the high current PWM motor driver circuit you linked to in post #5.

What I'm talking about in general is the "clamping" diodes that are often connected directly across motors and directly across MOSFET's to protect the MOSFET's from the generator effect of the motor. Here is a link Of what I'm talking about. This effect can easily destroy a power MOSFET and needs to be shorted out so voltages don't rise above what the MOSFET can handle. However, this effect is only seen with motors and other inductive loads. Your load on the other hand is a purely resistive load, which negates the need for such safety. I only mention it because 98% of the PWM control circuits on the internet are for motor control, and thus incorporate this safety. Point of fact, such diodes can be seen at the top of the schematic you provided in post #5, they are on the far right and far left. You would not need them for the reason stated.

D1 and D2 in the 555 PWM schematic you linked to *DO NOT* provide this function, they serve a different purpose. You will still need them if you use that circuit.

Also, per this link (below) I think I may need to step down the voltage to the 555 timer:

http://www.555-timer-circuits.com/555-on-24v.html
Yes, most 555 timer IC's have a max working voltage of only 18v. Your best bet is a three terminal voltage regulator, as they are simple to work with and require little external parts. However, any option on that page should do just fine. Do note that you may want to drive the push-pull stage off of the 24v. The more voltage you can feed the gates of your power MOSFET's, the faster and farther they will switch the the ON and OFF states.

My question is, will this decrease the voltage running through the resistors too or will it not effect them in anyway so the resistors will still be getting 24 volts. I am almost certain the resistors will still have 24 volts because they are connected to 24 volts, the PMW control just turns the MOSFET's off and on. But I just want to make sure.
Yes, the load will still be running on the 24v like you want it to be.

I am also wondering if I will be able to rig this load bank to test 12 volt PSU's too so I will have to figure out a way to bypass the voltage step down when running on 12 volts to begin with. I wonder if I could just put a simple switch in there to bypass the voltage step down?
If you use your 24v load with a 12v supply, you will only be running at half the Watts. It will function as far as the circuitry and the way it works, but it will not draw the same power. If you want to be able to draw ~1500 watts from 12v, you will need half the resistance.

#### jocanon

##### Member
If you use your 24v load with a 12v supply, you will only be running at half the Watts. It will function as far as the circuitry and the way it works, but it will not draw the same power. If you want to be able to draw ~1500 watts from 12v, you will need half the resistance.

Thanks for your detailed response, you have truley been very helpful! In the above, are you saying that if I run it at 12v then I will have half the current (that would make sense since I have half the voltage). If so, this is actually what I would want because the max current the PSU is rated for on 12v is 575 watts. But what about the stepping down of the voltage I need to do on the 24v supply to power the 555 timer, if I am running it from a 12v supply do I need to somehow bypass the stepping down of the voltage to the 555 timer so it stays at 12v?

Last edited:

#### ()blivion

##### Active Member
Not a problem. It happens to be immensely helpful when the person your assisting is as naturally competent as you are though. It makes my part much easier. And it surprises me that no one else has chimed in on this project yet. I know that it's simple from an electrical engineering point of view, but many other people have similar projects that get significant help. I still think someone sooner or later will jump in to say a few words.

In the above, are you saying that if I run it at 12v then I will have half the current (that would make sense since I have half the voltage). If so, this is actually what I would want because the max current the PSU is rated for on 12v is 575 watts.
Yes, you have it right. Actually now that I think about it... that will be 1/4 the Watts after all!!!

OK, so, it's the same resistance, but half the voltage, That sounds like half the Watts at first. BUT it halves the current as well as the voltage. So... originally... to get 1128 Watts with 24 volts, we need ~0.51 Ohms, this will draw about 47 Amps in accordance with Ohms law. Then we drop the voltage to 12v, but this only draws 23.53 Amps at 0.51 Ohms. So Yes, half the Volts AND half the current!!! This totals out to only 282 Watts, NOT the full 575 like you want. So I was wrong with my first estimate. You will have to get twice as many resistors, or get resistors with half the resistance if you want a full power 12v dummy load.

Keep in mind, it is still possible to make a ~1500 Watt Load at 24 volts even with such changes. We just need to make sure that when we make the PWM circuit, we keep the duty cycle always lower than 50%. We can modify a circuit that will have a hard lockout at 50% fairly easily I believe.

what about the stepping down of the voltage I need to do on the 24v supply to power the 555 timer, if I am running it from a 12v supply do I need to somehow bypass the stepping down of the voltage to the 555 timer so it stays at 12v?
A 555 timer will run all the way down to ~5v reliably. And in most cases, what ever you use to drop down 24 volts to 12 should not be to much of a hindrance if you run it off 12 volts either. In other words, it should work fine on both voltages, though it may dip below 12v. As long as the push-pull stage is powered from the 24v rail, there should be no change in functionality. But I will need to confirm this as I'm not 100%.

Edit:
Here are some good MOSFET's. I'm designing a circuit based off these as your power switches.
http://www.digikey.com/product-detail/en/PSMN4R4-80PS,127/954-PSMN4R4-80PS127-CHP/2596675

Last edited:

#### jocanon

##### Member
It happens to be immensely helpful when the person your assisting is as naturally competent as you are though.
Thanks for the compliment. This has been quite a learning experience for me. I feel like I have gotten the bug or something...I mean learning what I have about electrical circuits has made me want to build more things, I am beginning to realize that it is not as mysterious as I once thought it was. I was looking at the inside of my computer the other day and could actually recognize a lot of the components whereas before it was just a bunch of electrical stuff.

Keep in mind, it is still possible to make a ~1500 Watt Load at 24 volts even with such changes. We just need to make sure that when we make the PWM circuit, we keep the duty cycle always lower than 50%. We can modify a circuit that will have a hard lockout at 50% fairly easily I believe.
So if I am understanding correclty, I will still need it to be 100% when on 12v but then a max of 50% when on 24v, right, meaning I would not be able to hardlock it out permenantly at 50%. Now, I have a watt meter I was planning on keeping plugged inbetween the PSU and the dummy load at all times, so I could just manually make sure I don't adjust the pot above the rated wattage.

#### ()blivion

##### Active Member
So if I am understanding correclty, I will still need it to be 100% when on 12v but then a max of 50% when on 24v, right, meaning I would not be able to hardlock it out permenantly at 50%.
Ah! That is true. I think we can make it sense that it was attached to greater than 12 Volts. Then I think it should function properly still.

Now, I have a watt meter I was planning on keeping plugged inbetween the PSU and the dummy load at all times, so I could just manually make sure I don't adjust the pot above the rated wattage.
That can work, chances are that if you do happen to go over the rated Wattage, it will just trip the over current protection so we are "safe" to go over the limit.

I have possibly found another problem with this idea that I had overlooked.
All computer PSU's uses a "switch mode" approach to create regulated voltages, not entirely unlike our PWM load. This approach works well for maintaining and approximating an ideal battery up to the rated load. The problem is that using PWM as our method for controlling power works by switching between max load and minimum load for fractions of a second. This *MAY* cause a problem for us if the PSU ends up seeing this max load, and not our averaged out "virtual load" like we want it to. I have never PWM using a PSU before, so I don't know what exactly is going to happen. But in theory, I think as long as we make the frequency high enough, the output filter capacitors will average it out as far as the power supply sees it. Currently I am playing with the idea of a LC or RC filter on our load, to isolate the pulses from the supply entirely. I will be posting this circuit in a little while.

Last edited:

#### jocanon

##### Member
That can work, chances are that if you do happen to go over the rated Wattage, it will just trip the over current protection so we are "safe" to go over the limit.
I believe you are right, the PSU does have over current protection.

#### ()blivion

##### Active Member
OK, did some playing around and THIS is what I have so far (opens a simulator Java applet, Java must be enabled in your browser). Know that although this circuit is 100% functional as far as I can tell, it is also not very economical to build because the LC filter + 5 x 300 Watt resistors make the price steep. I estimate the total BOM (Bill Of Materials) to be around $175~$200 + S&H + Tax if applicable. $75 of that price is the resistors and$50 is for the filter network. You could significantly reduce the price tag by 1: Winding the filter inductor yourself (-$20). And 2: Using Ni-Cr wire for your resistors (-$70). You will need a fire proof non conductive form for your Ni-Cr resistors (ceramic rod would be ideal), and a large powdered ferrite core for your inductor. You also need heavy magnet wire for your inductor, which you will probably need to then double up on to get it to handle the 50+ Amps of current. And you will need Ni-Cr wire, and need to know how to make 5 lengths of approximately 2 Ohms each. *DO NOT* try to measure it with an Ohm meter, it's much to small for anything but a NASA priced meter to measure accurately. Instead figure out the resistance per foot and work out a length for a given resistance from that.

The circuit.
The circuit is exactly what we had discussed. Use the link above to see a schematic in the simulator in real time.

Your two power supply's are represented by the two battery symbols on the right hand side of the schematic. You can change it to 12 volts at any time by left clicking the switch on top of them. Your voltage is shown above the top wire, your current is shown above the bottom wire. Take note that although the power inside the circuit is pulsing quite wildly, the power drained from the supply is steady and smooth.

Directly left of this is the filter network, the diode is a low forward voltage drop type Schottky diode. This diode is there to prevent the LC circuit from oscillating. The capacitor is actually three(3) 1,500uF high ripple-current capacitors tied in parallel. The inductor value is not critical, so long as it's equal to or bigger that 750uH. All parts need to be able to handle at least 50 Amps, which is quite difficult to pull off economically.

Left of this is your five(5) Load resistors, five(5) gate buffer resistors, and five(5) N-Channel MOSFET's. This simulator is not the greatest, it does not simulate MOSFET's well when making high current circuits. So instead of MOSFET's I used the "analog switch" model to represent The equivalent action. So remember that the five switches directly below the five load resistors are actually your MOSFET's, not physical switches. They are basically the only wires of the simulation that move on their own. Your MOSFET's and your load resistors *WILL* need significant cooling. I suggest a medium to large squirrel cage fan and that you stand off and separate the resistors a good amount.

Left once more is a complementary MOSFET push-pull driver. This provides high current drive to the main MOSFET's so as to switch them quickly. It is necessary as the main MOSFET's have high gate capacitance that needs to be filled, and this 555 timer is only able to supply ~20mA. The 1k resistor is there to bleed off the small charge on the line when it is in the OFF position. This *MAY* not be needed in a real circuit. (Again, this simulator does not do MOSFET's very well) 1K resistors are cheep and will not effect the functionality of the circuit otherwise.

Left of this is the 555 timer circuit. It consists of the 555 IC it's self, (shown in schematic form). The two(2) resistors, for controlling the charging and discharging currents of the 15nF capacitor. The 15nF capacitor, for the timing base. and the 1K-10K potentiometer, which is meant for controlling the duty-cycle. You can change the duty-cycle, and thus the virtual load with the horizontal slider on the left side of the simulator window. More left is more current/load, more right is less. The frequency of operation is about 60Khz, which is far above human hearing. The right graph is the frequency and the duty cycle of the timer.

The farthest left section of the schematic is the voltage regulator for the 555 timer circuit. It consists of a generic NPN transistor, capable of up to 500mA current. Another 1k resistor, for biasing the transistor. A Zener diode, which is chosen so it breaks down at ~12 volts, creating our voltage reference. And a 1uF capacitor, to clean up any self oscillation. The transistor may need cooling. A small heatsink should be enough, forced air would help also.

That's it. I am still working out bugs and trying some things. So it's probable that this circuit will change somewhat over time. but as is, if you can afford it this should make a functional dummy load for 24 volts at ~1300 Watts and a ~320 Watt load for 12v. And it is trivial to add more resistors and MOSFET's to make it go larger... up to a point.

Last edited:

#### dougy83

##### Well-Known Member
Is there any reason you don't just use a 'linearly' controlled (rather than PWM) load? This has a few advantages including
* not applying the 50A peak load across the source for even low duty
* not requiring filtering
* not requiring a PWM controller

It does mean that the transistors will have to bear a greater proportion of the load, but they can be heatsinked and fanned. The resistors are not strictly necessary, although they will take some of the load and also provide some current-sharing for the transistors. The resistors can be replaced by some lengths of wire cable for cheapness also. The controller can just be an opamp regulating the current passed through the load.

#### jocanon

##### Member
()blivion...good work. I assume I could increase the resistors so that I can go up to 600 watts on 12 volts?

dougy83 I am intrigued by cheaper and simpler if it's possible...what do you mean by "linearly" controlled, would that be using a pot somehow to control voltage going to the resistors and thereby change the current?

Last edited:

#### dougy83

##### Well-Known Member
By linear I meant the MOSFETs are turned on partially. For PWM they are either fully on or fully off.

Being on partially means they will be dissipating a lot of heat and require adequate [fan-forced] heatsinking.

Being fully on/off means they won't be dissipating that much heat (the resistors will be instead); fully on/off means the only way to adjust the load is to bang it on and off e.g. using PWM, or to switch in stepped resistor legs (e.g. 15 lines of 100W resistors).

If the stepped load is ok ,you could just use e.g. 15 lengths of wire (as the resistors) and 15 switches (or screw terminals) - just note that the resistance will increase as temperature does depending on the wire material.

#### jocanon

##### Member
()blivion I wonder if the schematic is showing up differently for me than it is for you. I don't see the battery symols on the right, the horizontal slider seems to be on the right instead of the left side of the simulator window, and I don't see graph on the right that is the frequency and the duty cycle of the timer. Maybe you made edits to it already. Attached is the picture of what I see, let me know if it is the same.

#### jocanon

##### Member
By linear I meant the MOSFETs are turned on partially. For PWM they are either fully on or fully off.

Being on partially means they will be dissipating a lot of heat and require adequate [fan-forced] heatsinking.

Being fully on/off means they won't be dissipating that much heat (the resistors will be instead); fully on/off means the only way to adjust the load is to bang it on and off e.g. using PWM, or to switch in stepped resistor legs (e.g. 15 lines of 100W resistors).

If the stepped load is ok ,you could just use e.g. 15 lengths of wire (as the resistors) and 15 switches (or screw terminals) - just note that the resistance will increase as temperature does depending on the wire material.
So, please forgive me, I am a complete newbie here, but how would I adjust the level of "on"ness of the MOSFETs, with a pot? It seems to simple, but could I just attach a pot to adjust the amount of current running through the MOSFETs? Lastly, by stepped load I take it you mean just having e.g. 15 lines of 100W resistors and then manually plugging in more of them to get less resistance and more current or unplugging some to get less current and more resitance?