Electronics > Beginners

DC dummy load circuit calibration

(1/20) > >>

VEGETA:
Dear friends,

I want to create a dummy load which can handle 30v\5A which is going to be based on IRILZ44N mosfet (or several ones in parallel on the same heatsink) with LM358 op-amp and 10-turn pot. I don't have models for them thus I used similar parts. If I let the input voltage for opamps be 5v, it won't let me get the 5A maybe due to the opamps are not rail to rail. I hoped I could achieve this project with only a USB 5v input supply or something.

Circuit simulation in LTspice and it works perfectly, but the only thing left for me is to make the calibration potentiometer. ((attached below))

To make it easier for you, I have 2 potentiometers (modeled as a resistor divider because I don't know how to model a pot in ltspice) in the schematic: one for increasing op-amp gain and the other for decreasing it... because the 1 ohm power resistor might be 1.05 or 0.95 for example.

My goal of this thread is to make it only one potentiometer to do the job of both situations. I have seen many schematics like this one: http://www.openhardwarehub.com/projects/73-Simple-DC-Dummy-Load

but when I try the same circuit (with increasing op-amp gain pot) in ltspice it will solve the problem when the resistor is less than 1 ohm and it cannot do the opposite since it is in non-inverting amplifier configuration. I don't know if it works for them or not, but I need it to work in LTspice to guarantee it works on final circuit which I am gonna make the PCB for it.

calibration method is easy as you expect:
1- turn the 10-turn pot to 0 then turn on the dummy load.
2- adjust the pot to output exactly 1v then turn off the dummy load.
3- connect a power supply with say 5v and put the multimeter in ammeter configuration in series with it... while connected to the dummy load.
4- turn on dummy load... now current could be off a little bit due to resistor tolerence.
5- adjust the calibration potentiometer to make the multimeter display exactly 1A.


You can try adjusting the resistor around the 1R value and then play with positive and negative pots to get the idea.



Looking forward to your replies!

best regards,

Ian.M:
Just tagging in on this topic so I get notified when you update.
I'll need a bit of thinking time before I give a proper reply to your questions above.

Lets get started by finding datasheets for the parts:
IRILZ44N isn't easily found on Google apart from this topic, but there's a datasheet here: https://www.hifituning24.de/downloads/irliz44n.pdf

LM358 is trivial to find - its so popular Sparkfun has the datasheet - https://www.sparkfun.com/datasheets/Components/General/LM358.pdf

It looks like the MOSFET can pass 70A (pulsed) with 5V gate drive.  However with only 3V drive you'll be lucky to get 20A (datasheet Fig. 3), and below that it rapidly drops off to under 2A at 2V, also becoming highly temperature dependant.

Its difficult to find a spec for the LM358's output swing at V+=5V, however the specs for V+=30V indicate it will at best reach 2V under the rail with 3V under being typical, and the internal schematic on page 20 of the datasheet shows a 100uA current source feeding the base of a Darlington pair (Q5, Q6) which is consistent with that, so with a 5V supply you'll be lucky to get over 2.5V out and will never get more than 3V out.   Conclusion: to get more current you need a better OPAMP - maybe a low voltage one with rail-to-rail inputs *AND* output.

The openhardwarehub project you linked above seems a bit dubious - it vastly over-complicates the MOSFET current sink with two OPAMPs in the control loop, then compounds the stupidity by paralleling the MOSFETs (apart from their individual 100R anti-parasitic oscillation gate resistors) which is likely to result in thermal instability and current hogging.

All that is needed is a simple single OPAMP feedback loop - control voltage to +in, and feedback from top of Rs to -in.   The current then becomes Vctrl/Rs.   This can easily be calibrated by using Rs under 1 ohm and feeding in Vctrl via a potential divider so that it can be reduced to get a calibration of 1A per volt.   0.1R for Rs would be a good choice, with a nominal 10:1  potential divider to keep the voltage drop across Rs low.   Its desirable to avoid sudden current increases if the wiper of the calibration preset makes poor contact, so the preset should be wired with its wiper tied to one track end as a variable resistor not as a potentiometer, and it should be in series with the upper limb of the divider.   20K fixed + a 10K preset, with a lower arm of 2K7 gives you about +/-18% adjustment range either side of 10:1 so is a good place to start.

Trying to make the circuit work accurately right down to 0V in, 0A sunk with a single supply requires a *VERY* good rail to rail OPAMP.   If you can offset the ground by even a single diode drop so the OPAMP has a -0.6V negative rail, it will let you use a lower spec OPAMP.   If you use your LM358, a 9V supply split to give -1V and +8V rails will guarantee it will be capable of at least 0V to 5V gate drive swing.

If you are using multiple MOSFETs, each should have its own drive OPAMP and Rs to get stable even current sharing.   The potential divider for Vctrl to in+ can be shared by all the OPAMPs.

VEGETA:
This is the resistor that I have ordered: https://www.aliexpress.com/item/5pcs-RX24-1R-1-Ohm-50W-Aluminum-High-Power-Resistor-Metal-Shell-Case-Heatsink-Resistance-Resistor/32724995615.html?spm=a2g0s.9042311.0.0.MERq4G

it is 1 ohm not 0.1 so I cannot use 0.1, cannot wait another month. So I've got to think about doing it with 1R. I forgot to mention that the panel meter current shunt will be before the Rs so it will add a little bit to it. Panel meter itself will need calibration too.

I tried voltage divider on the positive input of the opamp, but as I said, I need a method to work for 1R resistor.

I can add a diode to make -0.6v, it is easy.

I can make an op-amp for each mosfet but I cannot give them shunt resistor for each one. I guess this is gonna be fine since the gate voltage of each one is gonna be exactly what it needs regardless of others.

Ian.M:
The problem with Rs of 1 ohm is the voltage drop across it  when passing 5A will be 5V.   You'll need at least 0.5V Vds drop across the MOSFET, which will limit the minimum working voltage for the load to 5.5V.   Also, you need enough gate drive to be able to reach 5A.  3V Vgs will probably do it unless you are unlucky and the MOSFET has a higher than expected Vgs threshold voltage.   Therefore the OPAMP output must be able to reach +8V with respect to the load circuit 0V, which means for a LM358, you'll need a 12V control circuit supply, and its impossible to design a 5V supply version, even with a perfect rail-to-rail input & output OPAMP.

Unfortunately you do need separate Rs resistors for each MOSFET to avoid current hogging.  Looking at Fig. 3 of the MOSFET datasheet, below about 20A Id, Id will *INCREASE* with temperature due to gate threshold voltage shift.  This means that if one MOSFET draws slightly more current, it will get hotter and draw even more current until it is drawing the majority of the current the load is set for, or 20A, or it fails due to excessive dissipation, whichever comes first.   You might get away with adding extra 0.1R resistors in series with each individual source pin, but it would be better to have completely separate source resistors.  Therefore I suggest you don't try to parallel MOSFETs until you have ordered a few 0.1R resistors.   Due to the lower resistance, you don't need such expensive power resistors - 3W ceramic will be fine for a max current of 5A per MOSFET.

Accepting these disadvantages, the circuit below should work.   MOSFET and OPAMP substitutions are the same as the sim you posted.

The R2:R3 divider 'taps down' the voltage across Rs so that you can use a similar divider feeding the OPAMP +in to calibrate it.   The Vbe multiplier Q1, R1, R7 acts as a crude shunt regulator to allow you to set the negative rail anywhere between -0.7V and -2.1V, and R8 feeds it >40mA so it doesn't collapse when the OPAMP tries to slew the gate rapidly negative.  You could probably increase R8 if you need to save power - worst case with a 1K gate resistor the OPAMP can draw about 10mA so 680R would leave some safety margin.

I wouldn't go below 1K on the gate resistor - it will limit the fault current into the OPAMP otput if the MOSFET fails.   Similarly the 1K R2 protects the OPAMP -in.   

Refinements for a practical circuit:  Add decoupling from both OPAMP rails to Gnd, a Schottky diode from V- to GND, cathode to Gnd so V- can never be dragged more than 0.3V positive, and a 10K resistor from the OPAMP output to Gnd to hold the MOSFET off if the control circuit isn't powered.  A 10A fast fuse in series with the MOSFET drain would be a good idea - it wont save the MOSFET, but should prevent anything too nasty happening to the rest of the circuit.

N.B. the 12V control circuit PSU *MUST* have a fully floating output.

VEGETA:
My trial is in attachments. TBH, I did it correctly before you post your circuit (which is better overall than mine) but with 12v supply as you mentioned. My original goal was to run this thing from 5v USB (not necessarily from labtop but from wall adapter 5v), now I need to find a way to make 5v to 12v amplification from gelly bean parts. I needed this thing to be as simple as possible, like Dave's one. However, Dave's one is not precise due to no calibration.

I didn't understand what is the floating part of the 12v supply? If I connected 12v wall adapter or say 5v wall adapter + boost converter to it... would it be floating?

I know floating means no ground connection... so what does that leave us with?

Do you know any straight forward way to make this 5v usb to 12v supply?


BTW: I have ordered these boost converters: https://www.aliexpress.com/item/5pcs-lot-MT3608-DC-DC-Adjustable-Boost-Module-2A-Boost-Step-Up-Module-with-MICRO-USB/32834245300.html

I guess they will work if we connect 5v to them then adjusted the pot.


After all this, if we make it 0.1R power resistor... will we get away with just 5v supply from wall adapter?

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod