Dummy Load Problem

[awwende]

I'm working on a dummy load vary similar to Dave's from a while back except I'm making it digitally controlled using his cheap DAC idea to give me an analog value from my 10kHz PWM.

The problem that when the fet is disconnect from the circuit I get a low resistance from the drain to source (about 160 ?). I've measured the new fets to be about 8 M? and the one I'm using now is starting to act funny and is about 70 k? from the source to drain.

Does anyone know what I'm doing wrong?

[madworm]

Residual gate charge maybe?

If you short all MOSFET pins together to get rid of anything left, Rds should be several M - if it is still ok. Of course this is only reliable with the device removed from the circuit.

[ikrel]

What kind of mosfet are you using?  Your schematic shows a depletion type mosfet, these require a voltage to turn them off.

I'm new to mosfets though, I could be wrong

[amyk]

What kind of mosfet are you using?  Your schematic shows a depletion type mosfet, these require a voltage to turn them off.

I'm new to mosfets though, I could be wrong

The IRLZ14 is enhancement mode.

[M. András]

you meant over 500Mohm? i tested an irfp4710 with shorted pins the fluke 289 goes to 500Mohm max. however different mosfets can vary on this

[awwende]

The parts in the schematic are the ones that I'm using. I did try shorting the pins, but that didn't help at all. I think my problem could be a transient issue and I'm getting current spikes that are greater than 6 amps (Rds at 5V is ~200 mOhm + 1 ohm at the source so any VCC >7.2V could be causing a break down).

But with the IRLZ14, it is about 8 Mohm give or take a couple hundred kohm.

I ordered a couple more parts, a few fets rated to 10A and a 3.6V zener to attach to the output of the comparator to prevent a short between the drain and source.

[awwende]

Although I should mention it depends on the orientation of the multimeter probes. I don't remember which was which, but in one direction it is 8M the other is >20M (upper limit of my meter).

[blackdog]

Hi awwende,

Your schematic is not verry clear, wera is your load?

First this,
Wat is the Max current u will use?
Because 1A will already will be a problem.

You have to add the voltage over the 1 Ohm resistor + the Gate volgage.
The IRLZ14 is NOT a logic Fet, its need 3.5 a 4.5V + voltagedrop 1 Ohm resistor to switch on.
This is not possible on 5V powersupplu of the OPA2835.

Always use a resistor direct on the gate of the Fet 47 - 220 Ohm.
Always place a resistor between the source of the Fet and the - input of the opamp. (~4K7)
Place a capacitor between the - input and the output of the opamp. ( 1 to 10nF )
Place a good capacitor 10 a 47uF tantal over the powersupply for the opamp.
This is a FAST opamp treat it that way! No long wires, take care of the - input!

My 2 cents :-)
Kind regarts
Blackdog

[HackedFridgeMagnet]

Is it that you are damaging the fet because your control method is not finished yet?

You could add some series R to the fet (drain to Vcc)  to make sure you can't damage the fet due to an unstable feedback loop and or control algorithm.

Start off with something high like 20ohms until you get the control sorted.
Then reduce it down to whatever the fet can thermally handle,  also specify the resistor so that even if the fet shorts the resistor can handle the full voltage.

[awwende]

Blackdog,

My load is VCC with a common ground. Max current I was looking at was around 5 amps. It's controlled from an ATmega328 so that I can do constant current, power and resistance. What is the purpose of the resistor on the gate and the resistor on the feedback loop? I do have a 10uF cap on the power supply of the opamp, forgot to include that on my schematic. It's a ceramic 0805, why should it be a tantalum cap? lower ESR? Wires are pretty short, I had no problems with the breadboard, which is why now that I'm frustrated that I'm having issues on my PCB.

I will try adding a series resistor at the drain. I think I have a massive 50W 10 ohm resistor (over kill but if it helps, i'll find a better resistor when I get another parts order in). I never had this problem in constant current mode, but constant power is when I started noticing the issue.

[HackedFridgeMagnet]

The series resistor is only there to help protect the Fet against any sort of control failure but It does also dissipate power so wouldn't be bad in the final circuit.
If the problem only manifests in constant power mode it does sound like a control/limitation issue.

I think Blackdog thought you were using the Fet in a switching role rather than in the linear region. Because you had no series resistor the only component that could dissipate the power is the Fet so I am sure your design is using the fet in the Linear region. Not fully switched on or off.

If your Fet was to be used in a switching role then you would definitely need the resistor/s to dissipate the power.
If your Fet was to be used in a switching role then all Blackdogs suggestions should be looked at.

Edit: Those statemenet shown to be wrong.

[awwende]

Oh ok. Yes it is in the linear region for the most part. The comparator should really only be high or low, but it switches so fast it looks like it's a fixed dc value. The caps on the drain and source are to help reduce the noise. I have replaced it with a proper low pass filter to my ADC using my 100uf and 0.1uF caps and a 3k3 resistor to give me a -3dB around 0.5Hz because without it there's a large amount of 2Mhz noise.

[blackdog]

Hi awwende, :-)


I made a little schematic for you, maybe it makes it more clear..


The gate resistor is to keep the Fet happy, otherwise it wil oscillate at random!
You Have to use a "Logic Fet" because your opamp VDD is max 5.5V.
Look @ the Source resistor, its now 0.15 Ohm.

The filtecaps C1, C3, is a ~50Hz lowpass filter, adjust this as needed.

C1, C3, R4 and R7, Use a star ground!

For you information, in almost every Fet the is a backward diode connected between Source en drain.
It is not easy to measure a Fet with a multimeter :-)

I hope this wil help you...

Kind regarts,
Blackdog

[frampy]

Hey Blackdog,

What is the purpose of C5 and R6 in the above circuit? Stability reasons?

[Harvs]

What is the purpose of C5 and R6 in the above circuit? Stability reasons?

It adds a dominant pole in the circuits loop transfer function to stabilize it (i.e. rolls off the circuits frequency response early.)

Depending on the drive capability of the which op-amp is used and the fet gate input capacitance R5 may be too low.  If R5 is too low the op-amp output will go into saturation and the C5/R6 feedback will have no effect, leading to the loop still being unstable.  I chose R5 in a similar circuit to be 1k to give some margin (but I was using a FET with high gate capacitance and opamp that couldn't drive more than 1k, I haven't looked at the specs for the actual parts nominated in this circuit.)

Cheers

[blackdog]

Hi,

Harvs explaned the reason for C5 and R6, thanks :-)
The output drive of the OPA2835 is 40mA and 100 Ohm wil be fine.

The change of the input voltage wil be slow because of the PWM filter.
It wil be more interesting if you also want to modulate your current.

Connect a 100K Resistor and a 1uF capacitor to the - input of the opamp.
Connect this point to your Function Generator.
Now is is posible to se how well your powersupply is regulating.

For the best controle, you alway have to tune the values of the compensation components.

Maybe you guys find it interesting to take a look @ the schematic of my Dynamic Load i just finist...


Pictue of the outside


This a is a picture of the load in action.
The top off the puls is 24A @ 15V DC, nice clean puls...


Kind regarts,
Blackdog

[awwende]

I'll try some of those changes and let you know how things turn out. thank you.

[Kevin.D]

Nice case  layout Blackdog .
R5 and R6 ,  you have a constant resistance mode , I was going to do that on one I built then I thought I wouldn't use it in that mode that much so didn't bother ,though nice touch if you ever need it .

[robrenz]

Nice job, very professional look.

[Harvs]

Nice job Blackdog 

How do you go about making your front panel?  Looks really good from the photo.

I notice you've got a reverse polarity protection diode to clamp the input and fuse.  I've been thinking about this myself, it would be really nice to have a method of blocking reverse voltage without adding any voltage drop to the circuit (i.e. a series diode but with out the extra drop.) 

The eLoad I'm currently building and testing I want the capability to be able to test large single NiMH cells at decent loads (like 10A from a 1.2V cell) so a series diode isn't going to cut it.  But I don't really like the idea of just clamping the input and blowing a fuse...  Haven't had any other ideas by chance?

[robrenz]

AFIK the series diode only conducts when a source is hooked up backwards. It has no effect when the source is properly connected. It is the same as a protection diode on the output of a power supply.

[frampy]

Awesome product blackdog, very professional.

Can you show us a picture of the circuit?

[skipjackrc4]

Yet another question for you, Blackdog.  First, that is one of the nicest looking DIY current sinks that I have seen. 

With the external modulation, do you have trouble with overshoot of the control circuitry?  I was planning on adding this feature to mine, but I thought that it might be difficult to distinguish between the sink's affects and the source's.  Is the circuit fast enough to avoid these problem?

Again, that looks like it turned out very well.

[blackdog]

Hi skipjackrc4,

The problem is in the wires between the Current Sink en the Powersupply under test...
This is for AC testing, DC its just some los off power, no problemo.

Faster AC testing of your powersupply is difficult!
I use this Current Sink for FAST! AC testing.
It's a design off Jim Williams and a little off me :-)

Schematic


The box


Inside the box with testpins


Building


3 types of testpins...
The short ones on the right side are still too long at 1MHz square wave


If its possible i use the sortes pins to test the powersupply @ high frequenties.

Regarts,
Blackdog

[Kevin.D]

Blackdog ,I am just building a new dummy load it's nearly finished (it's no where near as posh as yours though)and I may want to add external modulation like yours .
so just wanted to ask you a few questions about your's .Is there any particular reason  you chose to modulate the feedback terminal rather than the + input ?.Also  why did you choose a 200Khz low pass filter ? any particular reason .