What to use for a 45 amp load

[trilerian]

I'm posting this in the beginner forum since my project thread isn't really helping.

The basics here, I am designing a 1s/2s lipo discharging system for RC.  I have my schematic in my project thread.  But I am struggling on picking a load for the system.  This unit is taking a lot of inspiration from a unit that was made by someone else, however that someone else does not make them anymore, and I am making some subtle changes to it.  The unit that was previously made used a cpu heatsink as the load, then had a 60mm fan and a couple of 40mm fans to help cool it off.  Also the drains of the mosfets were directly screwed to the heat sink, so the mosfets were also using it for cooling.  My design will be a bit different, the mosfets will require their own heat sink, but the drain tabs will be on the pcb for better current flow.  The load will be externally connected via 10 or 12awg wire. 

So the easy way of doing this is to wire 6 0.5Ω 100watt resistors in parallel as the load.  But that is boring...  And sort of copy catish of resistor banks that are used currently.  I'd like to come up with a better idea for a load, but price is also a factor.  I'd like to ideally be able to make the load for less than $30.

There are a bunch of cpu heat sinks with LED fans, I am leaning is this direction, but how do I determine how big of a heat sink I need if it is used as the load? 

[Grandchuck]

https://coolingsourcethermal.com/heatink-size-calculator/

There are several on-line calculators out there.

[trilerian]

Those calculators are used for a heat load.  IE, dissipate the heat. I mean using the heat sink as a resistive load to sink the 45amps.  Really this will be programmable from 5-45 amps. 

[IanB]

So the easy way of doing this is to wire 6 0.5Ω 100watt resistors in parallel as the load.  But that is boring...

Why is "boring" a problem? "Boring" is usually a good property for well-engineered solutions. The one thing you don't typically want is "exciting".

I'd like to come up with a better idea for a load

How do you define better? Smaller? Cheaper? Lighter? More reliable? To define "better" it helps to define what is unsatisfactory about existing solutions.

There are a bunch of cpu heat sinks with LED fans, I am leaning is this direction, but how do I determine how big of a heat sink I need if it is used as the load?

This does not make sense. CPU heat sinks are not designed with electrical resistance, which is the primary property that a load should have.

[trilerian]

Why is "boring" a problem? "Boring" is usually a good property for well-engineered solutions. The one thing you don't typically want is "exciting".

You're right, I don't want "exiting" in the way you imply.  But exiting in the way of "oh, that looks cool". Flashing lights, not sparks mind you. 

How do you define better? Smaller? Cheaper? Lighter? More reliable? To define "better" it helps to define what is unsatisfactory about existing solutions.

Easier to package and make it look good.  Also casing needs are a concern, I don't have access to any cnc equipment, just a few 3d printers. So any case for this would probably have to be outsourced for me.

This does not make sense. CPU heat sinks are not designed with electrical resistance, which is the primary property that a load should have.

Correct.  But the mosfet duty cycle will be controlled by a current feedback from the current sensor.  I will program the mcu to use the current draw and adjust the duty cycle as needed to maintain a constant current.  My assumption is this will work, because that is how it looks like the unit somebody gave me works. 

[IanB]

Correct.  But the mosfet duty cycle will be controlled by a current feedback from the current sensor.  I will program the mcu to use the current draw and adjust the duty cycle as needed to maintain a constant current.  My assumption is this will work, because that is how it looks like the unit somebody gave me works.

No, this won't work. The goal of a dummy load is to dissipate power, which means the power has to go somewhere safe. The ideal states of a MOSFET used for switching is to be "off" and not conducting, or to be "on" with as close to zero resistance as possible. This keeps the MOSFET cool and prevents it destroying itself, but either way, the power is not going to the MOSFET. If the heat sink does not have significant resistance, then the power is not going into the heat sink either. Which means you don't have a load, you have a short circuit, which is just the "exciting" situation you don't want.

What will happen without proper resistance is that as soon as the MOSFET turns on, it will conduct maybe hundreds of amps through the short circuit (especially if there is a beefy LiPo pack on the other end), and something will quickly blow up. Probably the MOSFET, which will let out the magic smoke and perhaps explode.

[IanB]

Correct.  But the mosfet duty cycle will be controlled by a current feedback from the current sensor.  I will program the mcu to use the current draw and adjust the duty cycle as needed to maintain a constant current.  My assumption is this will work, because that is how it looks like the unit somebody gave me works.

No, this won't work. The goal of a dummy load is to dissipate power, which means the power has to go somewhere safe. The ideal states of a MOSFET used for switching are to be "off" and not conducting, or to be "on" with as close to zero resistance as possible. This keeps the MOSFET cool and prevents it destroying itself, but either way, the power is not going to the MOSFET. If the heat sink does not have significant resistance, then the power is not going into the heat sink either. Which means you don't have a load, you have a short circuit, which is just the "exciting" situation you don't want.

What will happen without proper resistance is that as soon as the MOSFET turns on, it will conduct maybe hundreds of amps through the short circuit (especially if there is a beefy LiPo pack on the other end), and something will quickly blow up. Probably the MOSFET, which will let out the magic smoke and perhaps explode.

[trilerian]

Correct.  But the mosfet duty cycle will be controlled by a current feedback from the current sensor.  I will program the mcu to use the current draw and adjust the duty cycle as needed to maintain a constant current.  My assumption is this will work, because that is how it looks like the unit somebody gave me works.

No, this won't work. The goal of a dummy load is to dissipate power, which means the power has to go somewhere safe. The ideal states of a MOSFET used for switching are to be "off" and not conducting, or to be "on" with as close to zero resistance as possible. This keeps the MOSFET cool and prevents it destroying itself, but either way, the power is not going to the MOSFET. If the heat sink does not have significant resistance, then the power is not going into the heat sink either. Which means you don't have a load, you have a short circuit, which is just the "exciting" situation you don't want.

What will happen without proper resistance is that as soon as the MOSFET turns on, it will conduct maybe hundreds of amps through the short circuit (especially if there is a beefy LiPo pack on the other end), and something will quickly blow up. Probably the MOSFET, which will let out the magic smoke and perhaps explode.

Then I am completely at a loss on how the current unit works.  Because it uses a 60mm cpu heat sink with the mosfets drain on the heatsink the source pin goes to a 12awg wire which then goes to a low side current sensor, and the positive of the battery is bolted to the heat sink.  The mosfets are controlled by a microcontroller.


Edit:  Essentially what I am taking from your posts, is that it is a bad design, don't replicate it, use the power resistors I think are boring. 

[IanB]

Then I am completely at a loss on how the current unit works.  Because it uses a 60mm cpu heat sink with the mosfets drain on the heatsink the source pin goes to a 12awg wire which then goes to a low side current sensor, and the positive of the battery is bolted to the heat sink.  The mosfets are controlled by a microcontroller.


Edit:  Essentially what I am taking from your posts, is that it is a bad design, don't replicate it, use the power resistors I think are boring.

There do exist special power MOSFETs that can operate in the linear region, which means the transistors themselves can act as variable resistors and can dissipate power into the heat sink. Such MOSFETs do not simply get turned on and off with PWM, but instead get controlled somewhere in between. It is possible to build a dummy load that works this way.

However, such MOSFETs are not that common, and the circuit has to be carefully designed to keep them inside the safe operating region.

It is possible your example unit is designed to work that way.

[trilerian]

The fets used in the unit given to me are IRL2910PBF
The fets I have in my design are IRLB8721PBF

The unit has the fets hooked to a pin of a 328p with a series resistor in between to the gate.  It really is a simple circuit. 

Edit:

My assumption was that with a 5v mcu controlling the gate, it would be getting around 4.5-5v.
4.5v gate with a 20-25 amp for current keeps it in the linear section of the graph in the data sheet.
4.5v has an rds on of 13mΩ.

Edit 2:

Of course, the unit that was given to me has a predisposition of burning up fets... But again, I assume other units he build do work.

With all that said, I can still use the same idea with power resistors as the load though, right?

[Terry Bites]

Make a lava lamp.
Bolt your resistors into a to an IP68 alloy box.
Feed multicore cable through IP68 gland
Coat box with a few layers od epoxy.
Sink to the bottom of a clear tank.
Add water, a bit of salt and washing up liquid and parffin wax. Seal it up tight.
Wow.

[RoGeorge]

A bucket of water with wires or resistors in it:

Dummy load in a bucket
mikeselectricstuff



 

[Eraldo]

youre not gonna have much luck in designing a reliable 45 A variable electronic load with your budget.

As for the transistors i would go with bjts in parallel. they are more reliable but at these currents the will go 5+ $ a pop and you'll need at least 2 of them in parallel.
The bigger problem for you would be the heatsink which would need to be quite big and would need active cooling (a fan or 2), and the current shunt whic would need to be between 1 and 5 miliohm. in these resistances you would need quite a specialised shunt with 4 terminals, 2 for current and 2 for sensing the voltage ( the resistor itself would set you back half the budget at the very least). you would also need some precision opamps with low offset, noise etc to measure these small voltages accurately.

it,s not as impossible as you might think, but neither as easy as you might think to make this project. Innovation requires efforts.

in the end maybe just a wire in a bucket of water might be the easiest and best solution 

[trilerian]

youre not gonna have much luck in designing a reliable 45 A variable electronic load with your budget.

As for the transistors i would go with bjts in parallel. they are more reliable but at these currents the will go 5+ $ a pop and you'll need at least 2 of them in parallel.
The bigger problem for you would be the heatsink which would need to be quite big and would need active cooling (a fan or 2), and the current shunt whic would need to be between 1 and 5 miliohm. in these resistances you would need quite a specialised shunt with 4 terminals, 2 for current and 2 for sensing the voltage ( the resistor itself would set you back half the budget at the very least). you would also need some precision opamps with low offset, noise etc to measure these small voltages accurately.

it,s not as impossible as you might think, but neither as easy as you might think to make this project. Innovation requires efforts.

in the end maybe just a wire in a bucket of water might be the easiest and best solution 

The budget I listed is just for the load.  The project budget is to stay under $100 per unit, in parts.  I already have the circuit designed, it is in my project thread.  I'm using an ACS772 in the 50amp variety.  I will sample that 2-300 times in my mcu, which I already do in another project, and use that to set my duty cycle on the the mosfets to maintain the variable current draw.  The voltage will be measured using simple voltage dividers that I will sample 100 times, plus over sample by at least 2 bits, to get 12 bit resolution for my adc.  I think that will suffice.  I'll take the measurements off all the resistors I use for the dividers to get a pretty accurate voltage. This part of it, I'm not too worried about.

My inquiry was for what to use as a load instead of power resistors.  I know power resistors will work, we discharge at 40 amps to power resistors all the time.  But I wanted to see if something different would work. I'm not going to create a water bath, I'm not going to create a massive box.  I'll attach a pic of the dummy load we use, what I want can't be any more bulky.  These are simple and effective, I was just hoping for something a little more stylish.  RC Racers are, eccentric.

[alligatorblues]

I have a 12A electronic load for testing batteries. I bought for about $30. It works exactly how the OP wants his to work, except it has digital readout for V/A/W, and a push/rotary control to set those where you want. It has a heat sink and a fan that is on a thermostat. And it displays what it is dissipating in V/A/W.

[ArdWar]

My inquiry was for what to use as a load instead of power resistors.

I think you're running into fundamental physics at this point. If you're going to dissipate heat into the same medium with the same constraint it doesn't really matter how or what mechanism you pick, the end result will be about the same.

The only meaningful way to reduce size is if you can tolerate higher delta T (using higher temperature element), using more mass flow (higher airflow), using higher heat capacity fluids, or using phase change heat exchanger.

[Aurgelme]

I have used water heating elements mounted in a small tank and some switches, and with the possibility of water circulation. Has worked flawlessly for many years.

[p.larner]

why not use a bunch of different wattage car bulbs 55w-21w-5w etc and switch them as to the needed load,i use an old heater element and alter the current draw using a number of taps and crock clips,works fine,use it in series with an avo8 ok for 10a or more with a shunt.,use that setup for ebike battery testing.

[Jwillis]

I think what's happening here is a bit of confusion on how the OP's load works. The evidence presented seems to fit.
Indeed the drain of the MOSFET is connected directly to the heat sink through the body of the MOSFET , But the heat sink is acting as a Current Path not as a resistor. Aluminum is a conductor and has very little resistance.The Mosfet is acting as a current Sink producing Heat that is dissipated by the the heat sink and fans. The resistor is a current shunt,  likely giving feed back to the Arduino 328p. The Arduino is delivering a PWM pulse to the gate of the MOSFET. Probably 480hz or around 2 millisecond pulse. This depends on the model. The duty cycle will also plays an important role. 
It can be done this way,  but there is a draw back. Because it is pulsing the current through the MOSFET current sink, it will give an inaccurate measurement of the battery capacity over time  because it will be nonlinear. This is why a Linear MOSFET is used with a a DC safe operating range so the battery capacity over time can be measured accurately. Even the cheap loads found on Ebay or Aliexpress work in linear mode and not PWM.

The fets used in the unit given to me are IRL2910PBF
The fets I have in my design are IRLB8721PBF

The unit has the fets hooked to a pin of a 328p with a series resistor in between to the gate.  It really is a simple circuit. 

Edit:

My assumption was that with a 5v mcu controlling the gate, it would be getting around 4.5-5v.
4.5v gate with a 20-25 amp for current keeps it in the linear section of the graph in the data sheet.
4.5v has an rds on of 13mΩ.

Edit 2:

Of course, the unit that was given to me has a predisposition of burning up fets... But again, I assume other units he build do work.

With all that said, I can still use the same idea with power resistors as the load though, right?

The Fets IRL2910PBF that are used in the load giving to you have different safe operating characteristics than the IRLB8721PBF at the 2millisecond pulse of the PWM . The RL2910PBF would indeed sink 20amps safely up to 20 or even 30V drain to source at 2 millisecond pulse if you look at Fig. 8 of the data sheet.  https://www.infineon.com/dgdl/irl2910pbf.pdf?fileId=5546d462533600a40153565b9013250b.
The IRLB8721PBF  will probably not handle more than 15 Amps at 10V drain to source . And that is even difficult to predict at 2millisecond pulse when you consider what it will do at 10 milliseconds. Look at fig. 8 of data sheet.
https://www.infineon.com/dgdl/irlb8721pbf.pdf?fileId=5546d462533600a40153566056732591 

If either the IRL2910PBF or the  IRLB8721PBF are operated in linear mode they would most likely fail because the Safe Operating Area in DC cannot be predicted. For either of them, it's is highly unlikely that they can handle more than 5Amps in linear mode depending on drain to source voltage . And most certainly not the expected 10 to 20amps.
Understanding  that the unit given to you have probably 5 MOSFETs , the amount of current it can handle is very dependent of the drain to source voltage. The duty cycle of the PWM is also a factor. The higher the Duty cycle the lower the Safe Operating area becomes. This is because with a high duty cycle the MOSFET is kept in the linear region longer.
ALL pulsed Mosfets operate in the linear region but only for short periods of time. https://www.infineon.com/dgdl/Infineon-ApplicationNote_Linear_Mode_Operation_Safe_Operation_Diagram_MOSFETs-ApplicationNotes-v01_00-EN.pdf?fileId=db3a30433e30e4bf013e3646e9381200 https://eepower.com/technical-articles/understanding-linear-mosfets-and-their-applications/#

If you look at a Linear MOSFET designed for that purpose like the IXTH30N60L2,for example, you will see that it has a DC Safe Operating Area. Look at Fig. 14 of data sheet https://www.littelfuse.com/media?resourcetype=datasheets&itemid=5795824c-b355-4cbe-a6c0-e0782d30498c&filename=littelfuse-discrete-mosfets-n-channel-linear-ixt-30n60-datasheet
Many MOSFET will have a DC linear Safe operating area. these are the type that are usually chosen for electronic loads because their linear characteristics can be predicted.

Resister banks can be used as a Load. But unless the load current is monitored, the the measurement of the capacity of a battery will be inaccurate because the resistance changes with temperature. This is why Electronic loads can produce accurate measurements because the current is kept stable in relation to the voltage over time. 

[trilerian]

I think what's happening here is a bit of confusion on how the OP's load works. The evidence presented seems to fit.
Indeed the drain of the MOSFET is connected directly to the heat sink through the body of the MOSFET , But the heat sink is acting as a Current Path not as a resistor. Aluminum is a conductor and has very little resistance.The Mosfet is acting as a current Sink producing Heat that is dissipated by the the heat sink and fans. The resistor is a current shunt,  likely giving feed back to the Arduino 328p. The Arduino is delivering a PWM pulse to the gate of the MOSFET. Probably 480hz or around 2 millisecond pulse. This depends on the model. The duty cycle will also plays an important role. 
It can be done this way,  but there is a draw back. Because it is pulsing the current through the MOSFET current sink, it will give an inaccurate measurement of the battery capacity over time  because it will be nonlinear. This is why a Linear MOSFET is used with a a DC safe operating range so the battery capacity over time can be measured accurately. Even the cheap loads found on Ebay or Aliexpress work in linear mode and not PWM.

The fets used in the unit given to me are IRL2910PBF
The fets I have in my design are IRLB8721PBF

The unit has the fets hooked to a pin of a 328p with a series resistor in between to the gate.  It really is a simple circuit. 

Edit:

My assumption was that with a 5v mcu controlling the gate, it would be getting around 4.5-5v.
4.5v gate with a 20-25 amp for current keeps it in the linear section of the graph in the data sheet.
4.5v has an rds on of 13mΩ.

Edit 2:

Of course, the unit that was given to me has a predisposition of burning up fets... But again, I assume other units he build do work.

With all that said, I can still use the same idea with power resistors as the load though, right?

The Fets IRL2910PBF that are used in the load giving to you have different safe operating characteristics than the IRLB8721PBF at the 2millisecond pulse of the PWM . The RL2910PBF would indeed sink 20amps safely up to 20 or even 30V drain to source at 2 millisecond pulse if you look at Fig. 8 of the data sheet.  https://www.infineon.com/dgdl/irl2910pbf.pdf?fileId=5546d462533600a40153565b9013250b.
The IRLB8721PBF  will probably not handle more than 15 Amps at 10V drain to source . And that is even difficult to predict at 2millisecond pulse when you consider what it will do at 10 milliseconds. Look at fig. 8 of data sheet.
https://www.infineon.com/dgdl/irlb8721pbf.pdf?fileId=5546d462533600a40153566056732591 

If either the IRL2910PBF or the  IRLB8721PBF are operated in linear mode they would most likely fail because the Safe Operating Area in DC cannot be predicted. For either of them, it's is highly unlikely that they can handle more than 5Amps in linear mode depending on drain to source voltage . And most certainly not the expected 10 to 20amps.
Understanding  that the unit given to you have probably 5 MOSFETs , the amount of current it can handle is very dependent of the drain to source voltage. The duty cycle of the PWM is also a factor. The higher the Duty cycle the lower the Safe Operating area becomes. This is because with a high duty cycle the MOSFET is kept in the linear region longer.
ALL pulsed Mosfets operate in the linear region but only for short periods of time. https://www.infineon.com/dgdl/Infineon-ApplicationNote_Linear_Mode_Operation_Safe_Operation_Diagram_MOSFETs-ApplicationNotes-v01_00-EN.pdf?fileId=db3a30433e30e4bf013e3646e9381200 https://eepower.com/technical-articles/understanding-linear-mosfets-and-their-applications/#

If you look at a Linear MOSFET designed for that purpose like the IXTH30N60L2,for example, you will see that it has a DC Safe Operating Area. Look at Fig. 14 of data sheet https://www.littelfuse.com/media?resourcetype=datasheets&itemid=5795824c-b355-4cbe-a6c0-e0782d30498c&filename=littelfuse-discrete-mosfets-n-channel-linear-ixt-30n60-datasheet
Many MOSFET will have a DC linear Safe operating area. these are the type that are usually chosen for electronic loads because their linear characteristics can be predicted.

Resister banks can be used as a Load. But unless the load current is monitored, the the measurement of the capacity of a battery will be inaccurate because the resistance changes with temperature. This is why Electronic loads can produce accurate measurements because the current is kept stable in relation to the voltage over time.

This is a lot of information, and thank you!   

The unit that was given to me actually only has 2 of the IRL2910PBF in it, and there is an ACS758 current sensor used for feedback to drive the duty cycle.  I plan on using a new ACS772, but the result should be the same.  Sample the current and vary the duty cycle to maintain the selected current.  Capacity, that will be measured, but capacity isn't a big factor.  What we will be looking for is the voltage curve vs time at a given discharge rate.

If I am operating in the SOA, does it really matter if I am using a purpose built linear mosfet for the job?  IE, can I just go with PWM and use these cheaper mosfets, (not the IRLB8721PBF, as you pointed out it won't work), but another that I find?  To be honest, I can wrap my head around using pwm and dumping to a resistor bank.  I'm not there quite there for the linear mosfets design. 

Unfortunately, I have pressure on me to get this design done.  Which is insane, because I never said I could do it in the first place, but I would look and see what I could do.  But now I have multiple people wanting progress reports a few times a week.  I figured replicating what was already made, should work, but having to redesign it, I'm in over my head...

Don't get me wrong, I'm not giving up, I am going to figure this out.  I have other design ideas that I want to implement that will be using mosfets. 

[trilerian]

So, let me go back and do some math here.  If I use 6 0.5Ω resistors in parallel.  That will give me a rough resistance of 0.083Ω.
Let's add the worst case scenario 10% error to the resistors. So that would be 0.0917Ω.  I want up to 45a discharge and this setup should max out around 90 amps.  So that is half duty cycle @45amps.  If I use this fet: IPP014N06NF2S https://www.mouser.com/datasheet/2/196/Infineon_IPP014N06NF2S_DataSheet_v02_02_EN-3084802.pdf It has a DC line in the SOA, @ 9v it looks to be about 8 amps, but I will be operating somewhere in-between DC and 2ms.  So if I used say 5x of this part, I should be fine? Running 1s the fets will be running at close to 100% duty cycle, but at 4v in the graph it looks like it can handle 11 amps.  So within the SOA if using 5 in parallel. 

The datasheet also shows in the graph (diagram 6), with 5v vgs, there will be roughly 1.8mΩ resistance.  So I would guess not a lot concern thermally, especially using 5.

Now, can I run 5 off the same PWM pin from my 328p?  The input capacitance is 13800pf, what series resistors am I looking at here on the gate?

[trilerian]

There is also maybe this possibility, PSMN1R0-30YLDX. https://www.mouser.com/datasheet/2/916/PSMN1R0_30YLD-2938848.pdf

These are used in a low, race level ESC.  Pretty much the most entry level race esc that is used.  That ESC uses 12 of these, 4 parallel on each phase of the motor.  Supposedly the ESC is rated for 50amps continuous @8.4v. Looking at the data sheet it is a logic level, which is nice.  Also smb, which I would like, then I will move to the ACS724 model for the current sensor and get rid of some of my through hole components.  Using the DC line in the SOA looks like 7 amps @8v, which as soon as you apply a load you will be down to 8v.  So maybe 8 of these in parallel?

These are the mosfets I was considering for another project, simply because I know they work in that application.  Being able to use them over multiple projects would be nice. 

[macboy]

There is also maybe this possibility, PSMN1R0-30YLDX. https://www.mouser.com/datasheet/2/916/PSMN1R0_30YLD-2938848.pdf

These are used in a low, race level ESC.  Pretty much the most entry level race esc that is used.  That ESC uses 12 of these, 4 parallel on each phase of the motor.  Supposedly the ESC is rated for 50amps continuous @8.4v. Looking at the data sheet it is a logic level, which is nice.  Also smb, which I would like, then I will move to the ACS724 model for the current sensor and get rid of some of my through hole components.  Using the DC line in the SOA looks like 7 amps @8v, which as soon as you apply a load you will be down to 8v.  So maybe 8 of these in parallel?

These are the mosfets I was considering for another project, simply because I know they work in that application.  Being able to use them over multiple projects would be nice.

Are you thinking of getting rid of the load resistors then? If you are thinking of operating these at 7A@8V(V_DS) then that is the implication. But I don't think so right?

If you are still thinking of doing PWM for the load resistors, then don't look at the DC line and the 8 V point on the I_DS vs V_DS curve. When the MOSFET is on, the V_DS will be very low (determined by I_DS * R_DSON), so you can push a very large current through this device. The main trick is turning it on and off very quickly. Any time spent in between full on and full off will result in more power dissipation. In fact, it's likely that most of the heat in this FET will result from switching on/off, not from passing current when on. It is a logic level device which can be directly driven by a (5 volt) microcontroller output, but it isn't that simple. The micro's GPIO pins will be able to supply some peak amount of current, which in combination with the gate capacitance, will determine how long the device spends in that transition between on and off. If using more than one FET, try to use a separate pin for each FET, or use a buffer (per FET) to boost current. A logic bus driver usually has higher current output than a generic logic gate. Look for high current drive and fast transition time (e.g. 74AC series, maybe a 74AC245 or 244). If you have a 3.3V microcontroller then  you must boost the output to 5 V for the gate; use TTL-input CMOS-output logic like 74ACT245 ("T" for TTL) powered from 5 V; the logic levels for 5V TTL inputs are very similar to and very compatible with 3.3 V CMOS outputs. If you can manage to switch quickly, and use a not-to-high PWM frequency (fewer switching cycles per second), and keep it cool, then you could probably use just one of these. But they are not expensive, so more would give some breathing room. For switching applications like PWM, paralleling FETs is simple; connect Drains together, connect Sources together, then drive all the gates appropriately (maybe connected in parallel, maybe not).

Make sure to use a snubber to absorb the inductive kick from turning off the current flow. This is important! Otherwise you are effectively making a high frequency ignition coil, and it is the FETs that will ignite/burn.

[trilerian]

There is also maybe this possibility, PSMN1R0-30YLDX. https://www.mouser.com/datasheet/2/916/PSMN1R0_30YLD-2938848.pdf

These are used in a low, race level ESC.  Pretty much the most entry level race esc that is used.  That ESC uses 12 of these, 4 parallel on each phase of the motor.  Supposedly the ESC is rated for 50amps continuous @8.4v. Looking at the data sheet it is a logic level, which is nice.  Also smb, which I would like, then I will move to the ACS724 model for the current sensor and get rid of some of my through hole components.  Using the DC line in the SOA looks like 7 amps @8v, which as soon as you apply a load you will be down to 8v.  So maybe 8 of these in parallel?

These are the mosfets I was considering for another project, simply because I know they work in that application.  Being able to use them over multiple projects would be nice.

Are you thinking of getting rid of the load resistors then? If you are thinking of operating these at 7A@8V(V_DS) then that is the implication. But I don't think so right?

If you are still thinking of doing PWM for the load resistors, then don't look at the DC line and the 8 V point on the I_DS vs V_DS curve. When the MOSFET is on, the V_DS will be very low (determined by I_DS * R_DSON), so you can push a very large current through this device. The main trick is turning it on and off very quickly. Any time spent in between full on and full off will result in more power dissipation. In fact, it's likely that most of the heat in this FET will result from switching on/off, not from passing current when on. It is a logic level device which can be directly driven by a (5 volt) microcontroller output, but it isn't that simple. The micro's GPIO pins will be able to supply some peak amount of current, which in combination with the gate capacitance, will determine how long the device spends in that transition between on and off. If using more than one FET, try to use a separate pin for each FET, or use a buffer (per FET) to boost current. A logic bus driver usually has higher current output than a generic logic gate. Look for high current drive and fast transition time (e.g. 74AC series, maybe a 74AC245 or 244). If you have a 3.3V microcontroller then  you must boost the output to 5 V for the gate; use TTL-input CMOS-output logic like 74ACT245 ("T" for TTL) powered from 5 V; the logic levels for 5V TTL inputs are very similar to and very compatible with 3.3 V CMOS outputs. If you can manage to switch quickly, and use a not-to-high PWM frequency (fewer switching cycles per second), and keep it cool, then you could probably use just one of these. But they are not expensive, so more would give some breathing room. For switching applications like PWM, paralleling FETs is simple; connect Drains together, connect Sources together, then drive all the gates appropriately (maybe connected in parallel, maybe not).

Make sure to use a snubber to absorb the inductive kick from turning off the current flow. This is important! Otherwise you are effectively making a high frequency ignition coil, and it is the FETs that will ignite/burn.

Yes, to make this easier I will just use the resistor bank as a load and use pwm to control it.  MCU is a 328p, the pwm pins are about 500hz, so nice and slow, and not alot of switching.  Originally I was planning to run 2 IRLB8721PBF off one pwm pin with a 100Ω series resistor between the pwm pin and the gate. 

I guess being new still, mosfets are kind of confusing.  I assumed, bad of me, that Vds was the voltage being applied to the drain, but your reply makes it look like it is a voltage drop across the drain to source.

Anyway, I'll attach my original schematic.  Substitute cpu heatsink for resistor bank for the load. I think there is another pwm pin available if I need to use a separate pin.

There are also some other changes I would like to make.  For instance, I would like to use a shunt resistor instead of the ACS772 for my feedback.  But I have a design using the ACS772 in another project already and maybe I am taking the easy way out.  In other words, I have not made a design using a shunt resistor yet.  So choosing components and layout is a thing. Also, I may remove the temp sensors, which are just thermistors.  I originally was going to use them to monitor the heatsink temperature and shut the fets off if the heatsink got too hot, but if I am using an external resistor bank it makes adding the thermistor more challenging.  The other sensor was to keep track of battery temp and again, shut off incase the pack got over a threshold.  Also I am bucking a bit at adding the blutooth printer.  But I am keeping the TFT touchscreen display.  That has been fun learning how to program and use.

[Eraldo]

This is not a good way of approaching variable electronic load.

Your aproach of using power resistors as current load and a mosfet to switch the rms current on the load,  doesn't take into consideration:

1) the resistance of power resistors dictates the maximum current sinked in them and also the current spike of the electronic load

2) because of these current spikes, the electronic load can't be used in power supplies/batteries that are not able to source at leas 45 A per your specification (also depended on the supplies voltage) , and the power supply/battery would basically either explode/burn or activate whatever security system in overcurrent scenarios. This means your supply would only be useful for that pack of batteries you have there and would rarely see any use in other projects.

3) your current load would be very bad in constant current at the higher current ratings because the batteries change in voltage based on their capacity. So you can't really maintain something like 40A constant over time. (This will be less of a problem in lower current ratings because you can compensate with your duty cycle)

4) you need to find a method to measure the rms current of your load, either by using a specialised ic or by programming the microcontroller to calculate the rms voltage.


These are just some of the disadvantages that came to my mind and some things you might and might not have considered before starting the project

The superior solution would be to use a completely linear solution (which will be a bit more expensive) but at least will be more reliable/useful of a project.

[Mira Rautio]

t's great to see you working on your RC project, and I understand your desire to come up with a unique and cost-effective load solution. Using a CPU heatsink with LED fans is an interesting idea, and it can add a cool visual element to your project.

To determine the size of the heatsink you'll need, you should consider a few factors:

    Heat Dissipation: Calculate the heat dissipation requirement of your load. This will depend on the power you intend to dissipate and the duration of your discharging process. You mentioned 1s/2s LiPo, so you can calculate the power based on the nominal voltage and discharge current.

    Thermal Resistance: Check the thermal resistance of the CPU heatsink. This value indicates how well the heatsink can transfer heat away from the load. The lower the thermal resistance, the better.

    Airflow: Ensure that the fans you're using can provide adequate airflow to cool the heatsink effectively. You may need to consider the CFM (cubic feet per minute) rating of the fans and the specific heat sink design.

    Material and Design: Consider the material of the heatsink. Aluminum is commonly used for its excellent thermal conductivity. The design of the heatsink, including fins and surface area, plays a crucial role in heat dissipation.

    Mounting and Attachment: Make sure that the heatsink can be securely attached to your setup and that there is good thermal contact between the load and the heatsink.

    Budget: As you mentioned, cost is a factor. Keep an eye on your budget and try to find a heatsink and fan combination that fits within your $30 target.

Remember that heat sinks are designed for specific thermal loads, so you'll need to match the heat dissipation requirements of your load to the heatsink's capabilities. If you can find a heatsink with LED fans that meets your needs, it can be a unique and visually appealing solution.

If you provide more specific details about your expected power dissipation and load characteristics, the community can help you with more precise recommendations. Good luck with your project!

[trilerian]

This is not a good way of approaching variable electronic load.

Your aproach of using power resistors as current load and a mosfet to switch the rms current on the load,  doesn't take into consideration:

1) the resistance of power resistors dictates the maximum current sinked in them and also the current spike of the electronic load

2) because of these current spikes, the electronic load can't be used in power supplies/batteries that are not able to source at leas 45 A per your specification (also depended on the supplies voltage) , and the power supply/battery would basically either explode/burn or activate whatever security system in overcurrent scenarios. This means your supply would only be useful for that pack of batteries you have there and would rarely see any use in other projects.

3) your current load would be very bad in constant current at the higher current ratings because the batteries change in voltage based on their capacity. So you can't really maintain something like 40A constant over time. (This will be less of a problem in lower current ratings because you can compensate with your duty cycle)

4) you need to find a method to measure the rms current of your load, either by using a specialised ic or by programming the microcontroller to calculate the rms voltage.


These are just some of the disadvantages that came to my mind and some things you might and might not have considered before starting the project

The superior solution would be to use a completely linear solution (which will be a bit more expensive) but at least will be more reliable/useful of a project.

1.  I did take into consideration the resistance of the power resistors to determine maximum load that could be put on the batteries.
1s battery, 4.2v charged, with 6 0.5Ω resistors in parallel.  So that is 0.083Ω and 4.2v gives a max load of ~50amps, and at 3.8v it is ~47 amps.  Add in tolerance for the resistors, and it should still be fine. 2s lipo could be discharged at 100 amps with this setup, so I see no issue there.  Also, my intent is to start the program at something like 5% duty cycle and increase from there, so any spikes will be just that, momentary spikes. 

2. All batteries that will be used can withstand at least 100 amp spikes, and according to the manufactures, much more than that.  This project is specific to use for RC racing batteries.  Not flight, not bashing, but for 1s and 2s racing applications.

3 & 4. Current will be measured by the ACS772x current sensor.  I have already designed and use a telemetry module that can measure the current and programming required for my microcontroller.  Current will be monitored by the mcu and duty cycle of the pwm will be adjusted to maintain the cc discharge.  Not really sure why this is a bad idea...

A linear solution very well may be the better way to go.  But it is a ground up redesign as I have only a faint idea of where to start based on some of the responses to my posts.

Is there a reason why my project will not work? 

Edit:  Why RMS to measure current?  This is DC current with PWM, wouldn't I want to average?  And if I know my duty cycle and my pwm frequency, I should be able to get pretty accurate results without any filtering.

[IanB]

4) you need to find a method to measure the rms current of your load, either by using a specialised ic or by programming the microcontroller to calculate the rms voltage.

Why RMS? You would use an RMS measurement if you are measuring or regulating power. If you want to measure or regulate current, then an average measurement is appropriate.

Edit:  Why RMS to measure current?  This is DC current with PWM, wouldn't I want to average?  And if I know my duty cycle and my pwm frequency, I should be able to get pretty accurate results without any filtering.

Good question. I meant to respond to this at the time when I saw the earlier post.

[Eraldo]

4) you need to find a method to measure the rms current of your load, either by using a specialised ic or by programming the microcontroller to calculate the rms voltage.

Why RMS? You would use an RMS measurement if you are measuring or regulating power. If you want to measure or regulate current, then an average measurement is appropriate.

well since we are working with a square wave the rms current would be 'easy' to calculate (you would still need to keep track of the battery voltage so that you would get a more accurate reading)

as for keeping track of the rms current, we would need it if we were to try and keep a constant current on the load. since both the batteries and the power resistors are not really stable with charge/power, you could compensate by varying the duty cycle until the reference RMS current is reached. so if you were to go this route you would want a relatively accurate rms current and that's not easily achieved by just taking into consideration the duty cycle and calculating the rms current with it. but i doubt the precision is a matter of grave importance in this design. a -+ 1 or 2 A wouldn't' be that terrible and just connect a clamp meter to know the exact current

Did OP also say about measuring the capacity ( I'm not sure so I'm just leaving it here) ? in that case at least a 5 % accuracy would be necessary for meaningful measurement.

1.  I did take into consideration the resistance of the power resistors to determine maximum load that could be put on the batteries.
1s battery, 4.2v charged, with 6 0.5Ω resistors in parallel.  So that is 0.083Ω and 4.2v gives a max load of ~50amps, and at 3.8v it is ~47 amps.  Add in tolerance for the resistors, and it should still be fine. 2s lipo could be discharged at 100 amps with this setup, so I see no issue there.  Also, my intent is to start the program at something like 5% duty cycle and increase from there, so any spikes will be just that, momentary spikes. 

What i meant was that this type of electronic load wouldn't be appropriate for general use and as you mentioned it will be only used with 1s and 2s lipo batteries so in that case its fine.

3 & 4. Current will be measured by the ACS772x current sensor.  I have already designed and use a telemetry module that can measure the current and programming required for my microcontroller.  Current will be monitored by the mcu and duty cycle of the pwm will be adjusted to maintain the cc discharge.  Not really sure why this is a bad idea...

as i said in my previous comment it will work fine in 0 to 95% of the current rating of the electronic load. after that there would be nothing to compensate for the resistor drift and based on their temp coefficient they will either increase or decrease in resistance and you would no longer have a constant current load but a load that varies with temperature and the voltage of the battery. Also as i said above by using the duty cycle of the waveform to calculate the rms current only,  the accuracy wouldn't be that great but if it doesn't bother you then its fine.


honestly finishing the project first would be the major goal. and if there are things in which its lacking you could add them with time or even build a V2.0.
I just wanted to make sure you knew the small shortcomings that might have bothered you in the future and it seems you already thought about them.

[IanB]

well since we are working with a square wave the rms current would be 'easy' to calculate (you would still need to keep track of the battery voltage so that you would get a more accurate reading)

as for keeping track of the rms current, we would need it if we were to try and keep a constant current on the load. since both the batteries and the power resistors are not really stable with charge/power, you could compensate by varying the duty cycle until the reference RMS current is reached. so if you were to go this route you would want a relatively accurate rms current and that's not easily achieved by just taking into consideration the duty cycle and calculating the rms current with it. but i doubt the precision is a matter of grave importance in this design. a -+ 1 or 2 A wouldn't' be that terrible and just connect a clamp meter to know the exact current

It is not so much that the RMS current would be 'easy' to calculate or not, but rather that the RMS current would be an incorrect measurement. For instance, if you had 10 A pulses with 50% duty cycle, then the average current would be 5 A, which would be the correct value, whereas the RMS current would be 7.07 A, which is clearly not correct.

[Eraldo]

well since we are working with a square wave the rms current would be 'easy' to calculate (you would still need to keep track of the battery voltage so that you would get a more accurate reading)

as for keeping track of the rms current, we would need it if we were to try and keep a constant current on the load. since both the batteries and the power resistors are not really stable with charge/power, you could compensate by varying the duty cycle until the reference RMS current is reached. so if you were to go this route you would want a relatively accurate rms current and that's not easily achieved by just taking into consideration the duty cycle and calculating the rms current with it. but i doubt the precision is a matter of grave importance in this design. a -+ 1 or 2 A wouldn't' be that terrible and just connect a clamp meter to know the exact current

It is not so much that the RMS current would be 'easy' to calculate or not, but rather that the RMS current would be an incorrect measurement. For instance, if you had 10 A pulses with 50% duty cycle, then the average current would be 5 A, which would be the correct value, whereas the RMS current would be 7.07 A, which is clearly not correct.

Yeah a blunder on my part. what I meant all along was the True RMS of the signal.

[trilerian]

New question!
I replaced the mosfets (IRL2910PBF x 2) on the original unit that was given to me for inspiration.  When replacing them I saw that thermal compound was being used between the drain and the heat sink.  Makes perfect sense, need to transfer the heat.  The only problem here is that the tab to the heat sink is the current path as well and all the thermal compound I can find is non electrically conductive.  Since the heat sink is aluminum, I can't use liquid silver either.  Pin 2 on the mosfets are not actually used, only the tab for the drain.  Should I put thermal compound on half of the tab?  I think this maybe the reason the unit was blowing mosfets, because thermal compound was coating the tabs and creating a high IR from the mosfets drain to the battery + side.

I replaced the mosfets without putting any thermal compound on the tabs at all.  I was also curious and ran the unit with my scope probing the gate pin.  Definitely a linear DC signal and not chopped with PWM.  At a 10 amp discharge, the voltage on the gate is ~2.6V. This kind of confused me since the 328p outputs a pwm signal.  I followed the traces from the gate to the 328p and found an RC filter on the 328p to the gate, so this must be filtering the pwm to a DC signal.  Which means the mosfets are operating as variable resistors and sinking the current themselves, right?  Basically how y'all are saying to design it, lol.  But, this is using cheap mosfets and not the more expensive linear fets.

Oh well.  I already ordered some prototype pcbs and components to try my pwm version with an external load.  I'll see how that turns out. 

What a learning experience this is becoming... 

[shapirus]

Which means the mosfets are operating as variable resistors and sinking the current themselves, right?  Basically how y'all are saying to design it, lol.  But, this is using cheap mosfets and not the more expensive linear fets.

FWIW, those cheap-ish Chinese constant current loads use generic off-the-shelf MOSFETs as power sinking elements (no way they obtain those special MOSFETs designed for linear operation), and they work, as long as heat sinking stays adequate. So at least this is something to keep in mind.

The 45A, however, poses some challenges. First off, heat dissipation. For a 2s lithium battery, worst case 4.2V*2 = 8.4V, this means you'll have to dissipate 8.4V * 45A = ~380W of power. This means a *big* heat sink with a fan (or several fans). To get an idea, check some of the more expensive CPU heatsinks, see their sizes and what levels of TDP they are designed to deal with.

Next, finding a suitable transistor won't be easy, if at all possible, at least to fit your budget with a good margin. In fact, I highly doubt that anything short of a resistor bank in a bucket of water will fit the budget, when you add up all the parts. Budget considerations aside, availability of such a transistor that can sink a continuous 45A and dissipate just below 400W is a big question as well.

Now, what do the CPU makers do when they hit the 5GHz barrier and fail to improve single-core performance any further? Right, they begin to scale performance horizontally by adding more (and more) cores and forcing poor programmers into writing programs that utilize parallelizing algorithms. See the point? You could do the same. Parallelize it. Use *several* cheaper/more easily available transistors. Each of them will have its own current sensing shunt. Each of them can have its own smaller heat sink, or they can share a single big one (good luck finding something suitable though). Since you are going to use a microcontroller, controlling them becomes trivial: read the voltage drop values from each of the shunts (amplifying them before feeding to the ADC, if necessary) and control each of the transistors' gates (or bases, if you use BJTs) individually to: a) make currents flowing through all of them equal; b) make sum of these currents equal to 45A.

If you use, say, 5 transistors, which means only 9A/75W per transistor, you enter the area where those cheap Chinese electronic loads do the job easily, meaning you can solve it using cheap components too.

The only (major) caveat here is that your microcontroller must have a sufficient number of outputs and ADC inputs to read current values and control all the transistors. You might need to use some additional controller(s) to read raw values and send them over a digital line to the main one.

Another caveat: even when current flowing through each transistor is equal, it doesn't necessarily mean that power dissipated on each of them is equal, because their characteristics may vary to some extent. However, it should be close enough, if the transistors you use are the same part number coming from the same batch. You might also read voltage drops across the transistors and control them to make power, instead of current, equal across all of them, but that'll be more difficult and smell like overengineering.

Of course, you'll also need to implement not only total overcurrent protection, but also overcurrent protection for each of the transistors to prevent a cascading magic smoke escape if one of them fails.

[EPAIII]

The biggest load I have ever seen was at a used generator company. They had to test run generators up to some that occupied a 40 foot semi trailer. The load was a pool of salt water about the size of a van with electrodes in it. When it was in use, the energy went into boiling the water. I am not sure how they controlled the current, perhaps by the depth the electrodes were submerged.

When it got too hot, they had to shut it down and let it cool.

So yes, I would suggest a plastic bucket, some water, some salt, some wire coat hangers for electrodes and an Amp meter and experiment a bit. That should be in your budget.

A bucket of water with wires or resistors in it:

Dummy load in a bucket
mikeselectricstuff



 

[trilerian]

I cobbled together a start.

6 0.5 100w resistors in parallel. Thermal glued to the cold side of 3 TEC1-12706 plates which are thermal glued to a 120mmx70mmx27mm heat sink with 2 60mm fans on top.
With a 240w load the resistors hit 50C. The heat sink also got to about 45C. I need a bigger heat sink or fans. Not really sure if the peltier did much.

[shapirus]

With a 240w load the resistors hit 50C.

That's barely warm for them.

[trilerian]

So I finally go my prototype pcb in, soldered all my components on the board.  Figured out some code to run, and made some test passes.  First thing... Remember to anlogWrite the pin and not digitalWrite the pin, lol.  Since my resistor bank is low enough, it can pull over 90amps from a fully charged lipo.  I kept running the program and it would instantly trip my safeguards, lol.  Oops...  So analogWrite.  Being that the control is via pwm, the output of the ACS sensor is also pwm, and since the 328pb, I changed my rc filter on the output to a 1µF ceramic capacitor and a 5k1Ω resistor.  This helped the output of the ACS and stabilized the feedback loop a bit.  Working well at 10 amps.  However, my pcb design couldn't handle the stress test at 35 amps, let alone 45 amps.  The load will sink it, lol, but the board couldn't handle it.

However, I am super excited that I got it to work at all, and my next attempt will be better.  Speaking of.  I think I want to go the linear route, yes I know, that's what y'all said.  But you know, stubbornness and all...  I need help choosing the correct mosfet(s) for this design going the linear route. 

Back to the drawing, err, KiCad...

[Aurgelme]

I have experimented a lot when it comes to electronic load and linear solutions are what I use.

I learned a lot from this design.
http://www.kerrywong.com/2017/01/15/a-400w-1kw-peak-100a-electronic-load-using-linear-mosfets/

[trilerian]

Another update on this project...

So the first stress test I did after verifying the unit would work, was at 30 amps and a 2s lipo.  No heatsinks on the mosfets since they are just switching.  But, they got so warm that they started to unsolder themselves since I put the drain tab down on the board.  Next test was to flip them over with the drain tab up and rig 2 new mosfets so the legs were soldered where I wanted them.  I was able to get to 10 amps and keep the temps below 100°C.  So the next test was to add 2 more mosfets in parallel to the mosfets already there. With 4 mosfets, all running on the same pwm pin, I did a stress test at 40 amp discharge for 2 minutes.  The mosfets got to 125°C without a heatsink.  So I am getting close here.  Maybe add a couple more and add a heat sink to them...

Beyond that, I think I need to figure out how to do a PID control for the pwm.  Just averaging the current I am fluctuating a few tenths of an amp below to above my target discharge rate.

As to my load, I don't think the peltiers help much.  While the resistors start out around 10°C, the heatsink is already up to 50°C.

[trilerian]

I have experimented a lot when it comes to electronic load and linear solutions are what I use.

I learned a lot from this design.
http://www.kerrywong.com/2017/01/15/a-400w-1kw-peak-100a-electronic-load-using-linear-mosfets/

I have looked at the writeup multiple times and watched the video. The mosfets being used are beefy and pretty pricey.  Just curious on how much they can really handle independently.  The writeup says due to thermal conditions, they can only handle about half of what they are rated for.  I need a max of about 350W.  But I did order 2 of them to try and test with, but would be nice if I only needed 1 per unit.

Here is a link to the data sheet on the mosfet:  https://www.mouser.com/datasheet/2/240/media-3321496.pdf

Also, do you think this can be controlled via an ESP32 DAC output?

[Jwillis]

Start simple first. By taking on to much at one time, things can get messed up pretty fast. Leave the Micro controlling for the last step. Once you have the basic design working, then adding a micro-controller won't be that hard. If the basic electronic load is designed right, then it will manage it's self without the need of a computer. The Micro-computer should only take your input to change the operation of the of the load, monitor to give you a visual readout and log data.If you use it to micromanage something that is capable of managing it's self, you will run into problems.
Do not use PWM to operate the mosfets on an electronic load. PWM will make noise which will introduce errors and takes a lot of extra compilations to remove that noise and the errors it creates. The more complicated you make things means more room for error.
When using MOSFETs in linear designs, pay close attention to how they de-rate with temperature. So you will probably have to parallel more Mosfets than you initially expect to get the desired power.
You don't have to get the most expensive "Linear" mosfets on the market. As long as it has a DC rating for the SOA in the datasheet you can use it for a electronic load. In fact any mosfet could be used but it's better if there is some information about it's use in Linear mode. A  FDL100N50F  for instance is nearly half the price of a IXTK90N25L2.
I can forward my design if you like. There are others here as well that are just as good if not better.

[pdenisowski]

I used a standard automotive battery tester to show that one of our supplies really could output 40 amps.  I could only run it for a few seconds at a time, but it did work

(And don't try this at home)