Anyone have a schematic for a Constant Current DC load tester?

[spec]

Wow that's an impressive response! Went to bed with 3 posts, wake up and we're up to 19! Thank you to all who are interested in helping me

No probs EF.

I appreciate the efforts to keep costs to a minimum, however the very reason I'd like to build something myself is not to save money (well maybe a little) But more to the point it's about building something rock-solid reliable. And also I enjoy the learning process along the way rather than buying manufactured end products. So don't worry so much on the most cost effective solution, rather I'd like the best solution to build a constant current adjustable DC load tester.

That sounds great

#8, The only practical heat management for my application will be the use of very large heatsinks and fans

That will be fine

#16, That does depend on the size of the bank I'll be testing, It should be designed in such away that it can handle hours of abuse.

That sounds great too.

#18, That looks like a real nice idea you have there! While I'm not knowledgeable enough to create a program for an adruino for such an application, I certainly have experience editing and know all the basics when it comes to working with arduino's.

Sounds like you know your way around the Arduino, so writing the PWM sketch should not be a problem, besides which there are quite a few Arduino coding experts on ETO who will no doubt help to sort any problems.

Spec, I could be wrong but I think you're liking this man-sized project aren't you? After reading my latest post I imagine you have a fair idea of what I'm after design wise. Do you think the idea you posted in #18 is the best solution? It sure looks great to me!

Too true: I would love to be there helping to build your megga machine.

I do think that the post #18 approach would be the one to go for if you want to build from the ground up. It will also give good performance and maximum flexibility. If you like, I will have a look at a more detailed circuit for you to consider.

If I can give a piece of cracker barrel advice: the first stage of your project would best be to sort the mechanical side: case, heat sinks, fans, high-current cable, terminals/connectors etc, so that you have a solid platform for developing the electronics.

Also, may I suggest a solid walnut cabinet with titanium fittings to give the unit a touch of gravitas and an optimum balance between traditional and modern.

spec

 

[ElectronsFlow]

The walnut and titanium sound like a smashing combo! Subtitle enough yet not understated I've got an old 2U rackmount case sitting around. It should provide enough room and enough ventilation? A detailed circuit would be absolutely fantastic Spec! Really do appreciate your help on this project (and everyone else's input too for that matter) When it's all done and working I'll be sure to show you how your creation performs!

 

[spec]

The walnut and titanium sound like a smashing combo! Subtitle enough yet not understated

I've got an old 2U rackmount case sitting around. It should provide enough room and enough ventilation?

OK, I will work on that basis.
Are you planning on mounting the case in a rack?

A detailed circuit would be absolutely fantastic Spec!

Will do.

Really do appreciate your help on this project (and everyone else's input too for that matter)

No sweat EF.

When it's all done and working I'll be sure to show you how your creation performs!

Pictures and a full report on how it works would be very interesting, but it will be your creation.

spec

 

[spec]

CONSTANT CURRENT LOAD VERSION 3

Issue 07 of 2016_10_16

(1) ERRATA
(1.1) Connect LM5510 Vee to lower 0V heavy line
(1.2) Connect LM5510 0V ref to lower 0V heavy line
(1.3) Change R18 from 100K to 47R
(1.4) Connect a 1N4007 diode in series with the voltage regulator input (thanks Ron Simpson)
(1.5) Connect a 100A, minimum, Schottky diode in series with the 48V line at the input (thanks Ron Simpson)
(1.6) The MOSFET schematic symbols show PMOSFETs. These should be NMOSFET symbols (thanks ChrisP58)
(1.7) Connect four 20A minimum high speed diodes: cathode to 48V supply line and anode to each NMOSFET drain. (thanks CrisP58)
(1.8) Connect a 4m7F (4700uF) minimum, 60V minimum, aluminum electrolytic capacitor from the 48V line, physically near to the junctions of the four 3R9 resistors, and the thick 0V line (thanks ChrisP58)
(1.9) Separate the two drive outputs from the driver chip and connect so that each output connects to just two NMOSFET gates (thanks ChrisP58)
(1.10) Place a high pass filter in the opamp feedback loop (thanks Ron Simpson)

(2) NOTES
(2.1) All capacitors, unless otherwise stated (UOS), are ceramic X7R dielectric (not surface mount)
(2.2) All resistors, UOS, are 250mW, or greater, +-5% or better, metal film (not surface mount)
(2.3) The physical layout is critical and should be as indicated on the schematic.
(2.4) The gate resistors should be connected physically on the NMOSFET gate terminal. The gate resitors have the following functions:
(2.4.1) Prevent the NMOSFET from oscillating at around 6MHz due to parasitic components (gate stoppers)
(2.4.2) Shape the gate drive voltage
(2.4.3) Protect the gate driver chip
(2.5) As there will be high currents involved, substantial wires and connectors will be required.
(2.6) Wiring should be as short as possible while maintaining a good layout.
(2.7) Wire should be high temperature type
(2.8) The capacitors on the schematic that may not appear to have a function are decouples and are essential for the correct and reliable operation of the circuit. Don't be tempted to omit these capacitors, even on a prototype circuit.
(2.9) For good heat conduction the NMOSFETs should be mounted on heat sinks using ceramic insulating pads (not mica, film, or rubber). Anodized aluminum washers would also be suitable, but are very expensive.
(2.10) The NMOSFETs should not share a heat sink with the power resistors.
(2.11) As there will be a huge amount of heat generated, powerful cooling fans will be necessary. The cooling arrangements, and especially the air flow, will need careful consideration.
(2.12) Each MOSFET will only dissipate about 10W maximum (estimate), depending on current setting.
(2.13) Each 3R9 resistor will be dissipating 600W maximum, depending on current setting.
(2.14) IC7b (spare op amp) should have its output connected to its inverting input and its non inverting input connected to the thin wire 0V (IC7b is not shown on the schematic).
(2.15) R29 has a four-terminal connection, as shown on the schematic (R29 can be a normal two-terminal resistor, but wired as a four terminal resistor) .

(3) DATA SHEETS/SOURCES
(3.1) Microcontroller
Arduino UNO clone: **broken link removed**
(3.2) Voltage regulator
TL783: https://www.ti.com/lit/ds/symlink/tl783.pdf
(3.3) Gate driver
Texas Instruments LM5110: https://www.ti.com/lit/ds/symlink/lm5110.pdf
https://www.digikey.co.uk/product-s...rcuits-ics/pmic-gate-drivers/2556427?k=LM5110
(3.4) Resistor
(3.5) NMOSFET
Vishay SUP50020EL: https://www.vishay.com/docs/68273/sup50020el.pdf
https://www.digikey.co.uk/product-d...onix/SUP50020EL-GE3/SUP50020EL-GE3-ND/5729951
(3.6) Opamp
Texas Instruments OPA2192: https://www.ti.com/lit/ds/symlink/opa192.pdf
https://www.digikey.co.uk/product-detail/en/texas-instruments/OPA2192IDR/296-42106-1-ND/5252859
(3.7) TO220 insulating pad
Avid 4170G: https://www.digikey.co.uk/product-detail/en/aavid-thermalloy/4170G/4170G-ND/1625435
(3.8) Fan
(3.9) Heatsink

 

[spec]

Hi EF,

You are considering using a 2U, 19 inch case. That gives 2 * 1.75 = 3.5 inch height.

I suggest that you intake cool air from the front of the case and expel it from the rear with fans mounted vertically along the front. But a 2U case height only allows 3.5 inch maximum fans.

With fans, the bigger the better: quieter, more efficient, and more air flow.

How wedded are you to just 2U height?

spec

PS: I have a plan in mind, which I have used successfully before, for the case and cooling. If I get the time I will do a sketch for you to have a look at.

 

[ElectronsFlow]

Hey Spec, that looks awesome! Thanks so much for your efforts and time. I'm not fixed on the idea of a 2u case, Really I could work with anything. The thing is going to be a beauty! Could I trouble you to add pin# to the LM5110 on the schematic? I'd hate to overlook something on the wiring side of things!

(2.12) IC7b (spare op amp) should have its output connected to its inverting input and its non inverting input connected to the thin wire 0V

I couldn't find IC7B? Either I'm blind (my mother says I can't find food in the pantry) or did you mean IC7A?

 

[spec]

Hey Spec, that looks awesome! Thanks so much for your efforts and time. I'm not fixed on the idea of a 2u case, Really I could work with anything. The thing is going to be a beauty! Could I trouble you to add pin# to the LM5110 on the schematic? I'd hate to overlook something on the wiring side of things!

I couldn't find IC7B? Either I'm blind (my mother says I can't find food in the pantry) or did you mean IC7A?

Hi EF,

I won't tell you the sort of day that I have had today- or perhaps I will. It started when I got up and found that the flue jab we had has given me and missus flue-like symptoms.

Then windows decided it was going to do a major update and messed up access to the NAS.

Then Office decided that my software needed configuring.

Then when I was doing your schematic, EAGLE decided it needed to update itself and as a result I lost half of the component libraries. And also the library editor would not work; that is why there is only a green box for the driver chip. And IC7B is not shown (IC7 is a dual opamp and only one of the opamps in the pak is used).

Then my missus reported that she had lost one of her favorite earnings- she located it at the out of town Next store.

After sorting that lot out I just sat down at the laptop for some serious computing when the phone rang and a shifty sounding chap said that they suspected that my Barclay card had been compromised and could I give him my card number. Of course I didn't fall for that! But to cut a long story short, I managed to confirm that the call was from from Barclaycard and was genuine so both my missus'card and my card will be cancelled.

Then, would you believe it, I just dropped a rich tea biscuit crumb in the mouse left button, so that has stopped working... ArrrraGH

Anyway back to earth: the circuit at the moment is a first stab. I haven't analysed it in detail or optimized the values, but I thought you would like to see the extent of it and the general form. I am thinking of changing the voltage regulator and possibly the driver chip for instance.

I will progressively refine the schematic and let you know when it is in a final state.

spec

 

[throbscottle]

I've only skimmed the thread 'cause my eyes are going funnier than my head, but didn't flyback post a design for something like this a while ago? Maybe it wasn't enough W's for this.

 

[ElectronsFlow]

Oh boy Spec, Sounds like you had more than a full day! You're a trooper to muck around with my schematic while juggling all those issues!

Then, would you believe it, I just dropped a rich tea biscuit crumb in the mouse left button, so that has stopped working... ArrrraGH

At this point I lost it and just had to laugh

No rush for a finished schematic my end, just do it whenever

 

[spec]

Oh boy Spec, Sounds like you had more than a full day! You're a trooper to muck around with my schematic while juggling all those issues!


At this point I lost it and just had to laugh

No rush for a finished schematic my end, just do it whenever

As John Denver says: 'Some days are diamonds and some days are dogs'

spec

 

[spec]

I'm not fixed on the idea of a 2u case, Really I could work with anything.

That's good. I'm envisaging a unit about 8 inches high by 19 inches wide with two 8 inch high flow-fans on the front mounted vertically.

The case would be around 12 inches deep and the rear would essentially be open to exhaust the hot air into the local environment.

The bottom of the case would comprise an aluminium plate to which the four large resistor heat sinks would mount.

A vertical partition, front to back would isolate the four NMOSFETs, probably just bolted to the bottom plate, and the electronics.

Of course, you would need a huge titanium knob mounted on a sub panel at the front to control the current. And a few flashing LEDs would be quite nice. Later it would be quite simple, and cheap, to add a small graphics display to indicate actual amps.

Then you could get really fancy and measure the actual battery voltage and then do all sorts of calculations in the Arduino.

Just some initial thoughts really.

spec

 

[ronsimpson]

Spec,
Like your circuit.
Comment(1): It is a little hard to measure current. If you divide "48V input" down to <5 volts then the Arunino can measure the voltage. If you know voltage and resistance then current is also known. The voltage divider needs a capacitor to remove switching noise.
Comment(2): I am thinking about driving each MOSFET separately. Example: Turn 1&2 on all the time, 4 off all the time and 3 on 50%.
(3.9//3.9 is about 2 ohms)//(3.9 at 50%= about 8 ohms) This will reduce noise and give a more constant current flow. The switching frequency could be very slow.
Comment(3): Diode (1N4002) in series with R23. You know some fool will put the battery in backwards! Don't ask how I know.

 

[atferrari]

Then, would you believe it, I just dropped a rich tea biscuit crumb in the mouse left button, so that has stopped working... ArrrraGH

It is of a good baker to top a cake with a nice berry! I wonder if your life on board, in spite of slipping cocktails, wasn't less eventful...?

 

[spec]

It is of a good baker to top a cake with a nice berry! I wonder if your life on board, in spite of slipping cocktails, wasn't less eventful...?

Yes, life can be a challenge some times!

spec

 

[ChrisP58]

spec,

One more issue that your eagle glitch caused was to screw up the mosfet symbols. They're p-channel parts in the schematic, but n-channel in the bom (as they need to be)

Also, you've chosen a dual gate driver and paralleled the inputs and outputs to drive the four mosfets. Personally, I would parallel the inputs, but separate the outputs so that they each drive two mosfets.

But, I do have a concern for doing this load by PWM. The battery will not see a constant 50 amp current, but on-off pulses greater than 50 amps. Whether this represents a valid load to test the battery is hard to say. But, with such high current pulses, it wont take much connecting wire inductance to create some pretty big voltage spikes. C10 an it's fellows will absorb some of that, but will it be enough?

 

[spec]

Spec, Like your circuit.

Thanks Ron; I hope it works OK

Comment(1): It is a little hard to measure current. If you divide "48V input" down to <5 volts then the Arunino can measure the voltage. If you know voltage and resistance then current is also known.

Yes, measuring 50A is a bit difficult. The physical layout of the NMOSFETs, the drain resistors, and the four-terminal current-monitoring resistor is key to the correct operation of the circuit. That is a very clever approach, measuring the voltage and letting the Arduino do the current calculation. It took me some time to work out what you meant. My main worry there though would be loss of accuracy.

The voltage divider needs a capacitor to remove switching noise.

Quite right

Comment(2): I am thinking about driving each MOSFET separately. Example: Turn 1&2 on all the time, 4 off all the time and 3 on 50%. (3.9//3.9 is about 2 ohms)//(3.9 at 50%= about 8 ohms) This will reduce noise and give a more constant current flow. The switching frequency could be very slow.

Another good idea. Do you think the Arduino Uno will be fast enough to do that in real time?

Comment(3): Diode (1N4002) in series with R23. You know some fool will put the battery in backwards! Don't ask how I know.

Brilliant, and so simple and cheap to do! In theory, the NMOSFETs would not worry about a reverse battery connection, but all the same it may be wise to put a 75A diode in series with the NMOSFETs too.

spec

 

[spec]

spec, One more issue that your Eagle glitch caused was to screw up the mosfet symbols. They're p-channel parts in the schematic, but n-channel in the BOM (as they need to be)

Thanks Chris; more of a spec glitch.

Also, you've chosen a dual gate driver and paralleled the inputs and outputs to drive the four MOSFETs. Personally, I would parallel the inputs, but separate the outputs so that they each drive two MOSFETs.

Good idea. I started out with the plan to use two driver chips with separate drives for each NMOSFET, but in the end opted for the easy way out. I haven't done an analysis of the gate charge etc, so the 22R base resistors are just nominal at the moment. Quite often, the gate resistors need adjusting (and diodes need to be fitted in parallel with the resistors) on the day when you can compensate for the layout parasitics etc. Rather than a 5A driver, I was considering a 7A version.

But, I do have a concern for doing this load by PWM. The battery will not see a constant 50 amp current, but on-off pulses greater than 50 amps. Whether this represents a valid load to test the battery is hard to say.

These were my thoughts too when considering this approach, but a lead acid battery itself is a pretty big integrator. A relatively high switching frequency would help to smooth the current pulses. There is just one small point: the current pulse can never exceed 50A by virtue of the four current defining resistors in the NMOSFET drains.

But, with such high current pulses, it wont take much connecting wire inductance to create some pretty big voltage spikes. C10 and it's fellows will absorb some of that, but will it be enough?

Yes, this is a good observation; the wire-wound resistors will also add inductance, but low inductance types are available. The NMOSFETs are 60VDS; maybe that needs to be increased, and maybe some 20A schottky diodes to catch positive voltage swings are required (any negative voltage swings will be caught by the NMOSFET drain source diodes). The values of C10 & Co will probably need to be increased, as you imply, and a hefty electrolytic is needed I would think. C10 & Co will probably need to be polypropylene metal film types too.

I am thinking of an alternative approach: turn the NMOSFETs on and off relatively slowly and suffer higher NMOSFET dissipation to reduce the di/dt.

Keep the comments coming.

spec

 

[spec]

POST ISSUE 04 of 2016_10_11

I'm now fancying a linear approach.

What do you think of this approach EF?

What's the consensus?

spec

ERRATA
(1) The 12R resistors in the NMOSFET drains need to be 200W rating rather than 100W rating.

NOTES
(1) This is an outline circuit only, to illustrate a concept
(2) Instead of using Pulse Width Modulation (PWM), this circuit is linear. The circuit might be a bit agricultural, but it would be simpler to implement (in practical terms) and lower risk. It would also be great fun.
(3) The performance would be better and it would eliminate the current pulses mentioned by ChrisP58 in post #35
(4) ElectroMagnetic Interference (EMI) would be virtually eliminated
(5) The Arduino could be eliminated and a simple analog control loop formed with the opamp, but that would not allow the current monitoring technique mentioned below. Also, eliminating the Arduino would reduce flexibility.
(6) The circuit could be greatly simplified (in practical terms) by using Ron Simpson's current calculating technique (modified) of post #32. This technique would be more accurate with the proposed linear approach
(7) As no switching is involved standard, rather than non-inductive, resistors would do. This would half the cost of the resistors.
(8) Each NMOSFET dissipates around 48W maximum
(9) Each drain resistor dissipates around 200W maximum

 

[dougy83]

I'm now fancying a linear approach. What's the consensus?

Yes.

For a previous dummy load project, schematic and results, see https://www.electro-tech-online.com...emp-control-fan-load-bank-pwm-circuit.128753/

 

[ronsimpson]

I'm now fancying a linear approach. What's the consensus?

Linear is simple. I just don't like to dump 1000s of watts into silicon. Probably because I spend hours every day trying to not put heat into silicon.

30 years ago I made truck/bus alternator testers. (12/24 volt) Used car head lights as a load. Could not stand the heat. Pointed the light up and heated the sealing. Kept me cool. At that time head lights were cheaper than resistors.