Author Topic: DIY Current load, keeping it as simple as possible, thoughts?  (Read 1393 times)

0 Members and 1 Guest are viewing this topic.

Offline SpottedDickTopic starter

  • Regular Contributor
  • *
  • Posts: 91
  • Country: ie
OK, first off, I understand the issues with MOSFETs in parallel, so this discussion isn't about that.

I want to try and build a 24V 20A capable current load that anyone can replicate using as few parts as possible. I'm planning on using it with the test controller software.

My main goal was keeping component count and cost to a minimum, at the moment the only things I think are questionable are:

1. Using a 2 output DAC instead of a single output DAC, but the single output 4821 is suffering shortages at the minute.
2. The LCD display could be deemed unnecessary, as you'll be able to see everything on test controller, but I decided some sort of local feedback was needed, so kept it in the base design.

Things like buttons, keypads, buzzers etc... can be added later. The point here is to get a stable base design and then the user can add "mods" like (for an example) a 3X4 keypad later.


As a base design, does anyone see any major issues here?

EDIT: Forgot to add, the Nano is getting its' supply via USB.

(Note, I'm aware that there is a GND2 labeled on this image, it should just read GND)

J2 is the Voltage sense for the load, J3 is the current input. 4 wire system for accurate voltage detection.

The LM324 is used as a control loop for the base input to the MOSFETs, and also as an amplifier for the current shunt which outputs 75mV at 20A.


« Last Edit: June 25, 2022, 02:14:17 pm by SpottedDick »
 

Offline ledtester

  • Super Contributor
  • ***
  • Posts: 3032
  • Country: us
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #1 on: June 25, 2022, 02:28:35 pm »
There have been a lot of electronic load projects discussed here. At the moment I'm only able to find this one:

https://www.eevblog.com/forum/projects/smart-electronic-load/msg3602103/#msg3602103

Note the following:

- each MOSFET is driven by its own op-amp
- each MOSFET has its own shunt resistor
- a compensation network for each op-amp (e.g. R19, C23) to prevent oscillations
« Last Edit: June 25, 2022, 02:33:38 pm by ledtester »
 

Offline SpottedDickTopic starter

  • Regular Contributor
  • *
  • Posts: 91
  • Country: ie
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #2 on: June 25, 2022, 02:53:05 pm »
- a compensation network for each op-amp (e.g. R19, C23) to prevent oscillations


Thanks. I have to do a bit of reading on this and it looks like I will need to add this across U3A at least.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #3 on: June 25, 2022, 09:10:15 pm »
If you want to keep things simple: consider using switched power resistors. Resistors can dissipate much more power versus cost and have little frequency depedent behaviour.

Also: don't use any analog parts from Microchip. If you look at the spec you'll see it basically is an 8 bit DAC. And the LM324 is a horrible opamp to use for almost any circuit.
« Last Edit: June 25, 2022, 09:12:18 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: edavid

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3194
  • Country: us
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #4 on: June 25, 2022, 09:32:54 pm »
With a 3.75 mohm shunt, the offset voltage of the LM324 could easily create over an amp error!! With that low a shunt, you'll need a very low offset Op-Amp instead of the LM324.

Edit: Also as mentioned, with the massive delay caused by the large number and size NMOS devices...and with the Miller effect, your circuit is almost guaranteed to be a oscillator. A well crafted frequency compensation network is a must.

Best,
« Last Edit: June 25, 2022, 09:37:58 pm by mawyatt »
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline Hexley

  • Regular Contributor
  • *
  • Posts: 197
  • Country: us
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #5 on: June 25, 2022, 10:40:56 pm »
Consider the MOSFET drive requirements:
  • You want 20 A split across 6 devices. So 3.3A each.
  • You can see from Figure 1 of the IRFP260N data sheet that it takes more than 4.5V Vgs to get 3.3A of drain current (at 25 deg. C)
  • The LM324 can only pull up to within 1.35 volts (typ.) below the positive rail, even when it is supplying negligible current.
  • Your circuit runs the LM324 from +5 volts. So it is good for 3.65 volts out. Not the required 4.5V.
  • That means the load won't reach 20 A at full tilt.
You will probably want to run the opamps from +12V, so that you have plenty of oomph to drive the FETs at full load. And don't forget that you need one opamp section for each FET, as mentioned by ledtester.

An active load is a fun design project, but there are lots of hidden issues. Good luck.

 

Offline MarkF

  • Super Contributor
  • ***
  • Posts: 2523
  • Country: us
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #6 on: June 26, 2022, 12:44:01 am »
OK, first off, I understand the issues with MOSFETs in parallel, so this discussion isn't about that.


Well, it should be!
Clearly you don't understand because you are planning to do it anyway.
The rest of the discussion is irrelevant because your MOSFETs are just going to burn up.
You will never achieve a stable design with MOSFETs paralleled like that.

Other than this fundamental flaw, take a look at Jay_Diddy_B's Electronic Load design:
  https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/

Just add a few more MOSFET groups (each with its own op-amp and sense resistor) for the current capability
and then switch out his front-end with your Arduino  control.
« Last Edit: June 26, 2022, 12:52:05 am by MarkF »
 

Offline pqass

  • Frequent Contributor
  • **
  • Posts: 706
  • Country: ca
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #7 on: June 26, 2022, 01:36:49 am »
Reconsider 4 or 5 of these modules:  https://www.eevblog.com/forum/testgear/example-of-why-people-say-you-sholdnt-use-mosfets-in-parrallel-as-dummy-load/msg4251337/#msg4251337  and see reply #17 too.

All module opamp positive inputs are fed from the same DAC output. For example: 4 modules, each with a 0R1 shunt, therefore, a DAC output voltage of 500mA (after your divider) passes 20A at the output terminals.

Each module's readback output is fed into an ADC input of its own or a non-inverting summing amplifier then a divide by four (# of modules) divider into one ADC input.  If you use a LM358 then Vee (the negative rail) can be tied to GND; no negative supply required.  However, there is still going to be <10mA flowing through the load. So not quite 0.000A unless you get a better (rail-to-rail) opamp.  Having two opamps in the loop will double the offset.

In my implementation, I measure the gate voltage at 3.4V, 4.7V, 4.9V, 5.1V, 6.5V when set to pass 0A, 3A, 4A, 5A, 6A, respectively (with a 0R1 shunt).  So a 5V opamp supply won't cut it.

You're going to need the resistor and cap in the feedback loop otherwise it will oscillate.

A fuse on the positive terminal may be a good idea.

How does the Nano get its power?  Vin?

My approach was to first work on the pure analog parts of the load; opamp(s), MOSFET(s), shunt(s). Use a simple pot to provide the set current to test it.  Then work on the MCU, ADC, DAC, etc. parts.

« Last Edit: June 26, 2022, 02:00:04 am by pqass »
 

Offline SpottedDickTopic starter

  • Regular Contributor
  • *
  • Posts: 91
  • Country: ie
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #8 on: June 26, 2022, 02:00:36 am »
Reconsider 4 or 5 of these modules:  https://www.eevblog.com/forum/testgear/example-of-why-people-say-you-sholdnt-use-mosfets-in-parrallel-as-dummy-load/msg4251337/#msg4251337  and see reply #17 too.

You're going to need the resistor and cap in the feedback loop otherwise it will oscillate.

A fuse on the positive terminal may be a good idea.


I actually have the fuse holder sitting there, I just didn't remember to put it in the schematic. This is just a first pass!

Yes, feedback loop needs dampening, will be implemented.

As someone just before you noticed, yes, the OP-AMPs will need more voltage.

The Nano is getting its' 5V from USB.

I like your modular idea!

Consider the MOSFET drive requirements:
  • The LM324 can only pull up to within 1.35 volts (typ.) below the positive rail, even when it is supplying negligible current.
  • Your circuit runs the LM324 from +5 volts. So it is good for 3.65 volts out. Not the required 4.5V.


Huh, I'm simulated this in Falstad circuit.js using the LM324 model and never noticed this somehow. Thanks for spotting this. I'll move them to the 12V rail.

With a 3.75 mohm shunt, the offset voltage of the LM324 could easily create over an amp error!! With that low a shunt, you'll need a very low offset Op-Amp instead of the LM324.

Edit: Also as mentioned, with the massive delay caused by the large number and size NMOS devices...and with the Miller effect, your circuit is almost guaranteed to be a oscillator. A well crafted frequency compensation network is a must.


Could you go into the voltage offset on the LM324 or point me in the right direction?
EDIT: See the offset issue. It's closer to a 2A error than a 1A...


Yes, oscillation dampening will be added.

« Last Edit: June 26, 2022, 02:06:04 am by SpottedDick »
 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3194
  • Country: us
Re: DIY Current load, keeping it as simple as possible, thoughts?
« Reply #9 on: June 26, 2022, 02:49:17 pm »

Could you go into the voltage offset on the LM324 or point me in the right direction?
EDIT: See the offset issue. It's closer to a 2A error than a 1A...


Yes, oscillation dampening will be added.

You can think of input referred offset voltage as a voltage source in series with the op amp input. In a closed loop condition like you have (which BTW is not a good design) the op amp will produce an output to drive the negative feedback loop to null out the offset, thus making the total offset approach zero. So in your case the op amp will drive the NMOS devices to produce a voltage so the net result will null the input referred offset of the op amp(s).

Yes the current error could be well beyond 1A as the spec for the LM324 show 7mv max offset, and you have two of such op amps in the feedback loop. Analyze the loop and you'll soon see how this large error occurs, and with a little research you'll find numerous op amps with very low offsets.

Your best path forward IMO is to spend some serious time studying analog circuit design, semiconductor devices and control theory. Once you are comfortable with these concepts then all the good advise you've received from others will begin to make sense and show you the proper path to producing a working circuit. What you have presented and seem to "defend" won't produce a workable and reliable circuit.

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: Performa01


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf