Dummy Load II

[jocanon]

I am calling this thread Dummy Load II because I already built dummy load 1, now this is version 2. Dummy Load 1 thread was here:

https://www.electro-tech-online.com...emp-control-fan-load-bank-pwm-circuit.128753/

So the dummy load we built the first time worked great! I say "we" built because, as will see if you read the first thread, I had a lot of help from ()blivion and ronv. Thanks again guys! So, it worked great until I had to go and mess it up . I was testing the voltage on one of the op-amps and accidentally shorted the Vcc to the output while the load was connected...I know, stupid...! Anyway, that pretty much fried a few things in there. In actuality, I can still use that first dummy load, I just disconnected the damaged components. Initially there were 10 MOSFETs that burned off the current, now there are just 8...it's actually just barely enough for my purposes with 8.

So, partially breaking the first one got me to thinking that I should build a second as a back up anyway, just in case I fully break the first one someday. I run a business where I have to test power supplies on a regular basis and cannot afford to be without the dummy load and do not really want to spend thousands of dollars on one at retail.

Attached to this post you will find the schematic of the original design. Essentially it is a variable dummy load capable of loading up a PSU up to about 1500 watts or so, maybe higher. I am testing two 12v PSUs connected in series for 24v that go up to about 50 amps. We bleed off the current directly through the MOSFETS instead of using PWM and burning the current in resistors. In order to cool the MOSFETS I soldered them to a copper pipe that I then hooked up to my basement laundry tub sink and I run cold water from the tap through it. I have a temperature reading hooked up to the copper line and the load will shut down if it gets above about 88C and then turns back on again once it gets back down to about 75C. An LED and alarm also will sound when the load shuts off due to getting too hot. It will only ever get that hot if I forget to turn on the water. With the water running it stays a cool 20C in the winter when loading it up to about 1100 watts and a little hotter in the summer (water is pretty cold out of the tap in the winter here in Buffalo, NY).

The other schematic I am attaching is a temperature controlled circuit that controls the fans inside the dummy load. The fans are there to provide cooling for the current sense resistors.

There are only a couple things I want to tweak in the original design:

1. I would like to make it so the current sense resistors do not get so hot. I emailed ronv and he suggested using 10 WATT resistors instead of 5 and tweaking the OHMS a little. We will also need to swap out some of the resistors for different values to get the current readings right once we change the current sense resistors.

2. Also, the voltage readings AND current readings are very twitchy, they jump around. I would like to add something to the schematic to even them out and reduce the noise so that they are a little more stable readings.

. Lastly, if possible, I would like to make the dummy load capable of 1700 watts. I think we could just add a couple more MOSFETS to increase the capacity.

I contacted ronv via email and he said he would be willing to help me make these modifications. The attached schematic was his creation, so he still has it and is working on it right now.

So this thread will be covering the build of this new slightly modified dummy load. The attached schematic is the un-modified original version. If anyone out there has any input or ideas to improve the design all are welcome.

Thanks!
Jeremy

 

[()blivion]

For mod 1, simply reducing the resistance to lower than 0.1 Ohms would be easy enough. The Op-Amps have plenty of gain to work with a lower feedback signal. I'm thinking 0.01 Ohms, which you can purchase as 10 milliohm resistors. This is not that hard to do and will waste 1/10th the power in the resistors. It also has the potential to solve problem #2 automagically.

For mod 2, you can hook a simple RC lowpass filter up in series with each of your meters. This will take any erratic waveforms and calm them down a bit. If you were to look at the signal, it would look something like a heart monitor, big spikes every now and then. The RC filter will chop off the tops of the spikes and use that to fill in the valleys, making the signal more flat.

Another option for mod 2 is to examine the nature of the noise, determine where it is coming from, and try and correct it. The most likely place would be the Op-Amp feedback loop. It's nature is highly susceptible to "fish-tailing" out of control (oscillation). This will not effect the performance of the dummy load very much, but it will screw with any sensors that are in the loop, as you are no doubt seeing. We could redesign it such that the Op-Amps have slightly less gain if need be. As is they are setup to have infinite gain. The difference would be a few more resistors per Op-Amp.

Mod 3. The current revision of the device can probably reliably go up to ~2000 Watts, if you're still using 10 STP80NF12. Technically, each FET can be made to safely dissipate up to 300 Watts, which with 10 FET's would suggest up to 3000 Watts. But I wouldn't count on this, you always want a safe margin of error for breathing room and parts tolerance. If you are running at 3000 Watts and one FET decides to blow out... All the rest will have to carry the weight and won't be able to. But 2000 Watts should be fairly safe provided that the cooling system can remove that much heat that quickly. Which I can not even speculate on. 2000 Watts is quite a lot of power.

Anyway, that's my two cents for now. I'll of course be monitoring this thread and helping out as time goes on. Good luck.

-()blivion

 

[fernando_g]

Do you have access to a scope?
If so, probe the opamp output that drives the Mosfet gate. One needs to understand the nature of the noise or oscillation to fix it.

I also think that is feedback causing the oscillation, but that is only a guess...... scope images are certainly required.
If that is the case, simple compensation networks will cure the problem.

 

[()blivion]

Do you have access to a scope?

I can virtually guarantee that he does not own or have access to an O-scope.

 

[jocanon]

Hey guys, thanks for the responses...()blivion you are right, no scope. I have considered it, but still have not purchased one. Is there a way to trial and error it to fix the problem without a scope...I think that would be the less expensive way to go.

 

[Mosaic]

He should slow down the NFEt response a bit with a 470 to 1000 ohm gate resistor.

 

[fernando_g]

Trial and error?
Let me give you a personal experience.

Many, many years ago, while on high school I built a vacuum tube amplifier......yes, I've been around electronics that long.
One day, while listening to a record, my sister came to my room and announced: "I can listen to your record on the AM radio." I went to her room and indeed, the radio was playing my music!!

I didn't know what was happening. But I knew an acquaintance who had a TV repair shop, and asking around it was suggested that the amp was oscillating and that I should tweak the feedback circuit. Asking how to do that, I was told: by trial and error.
So, using an AM radio as my test equipment, I fiddled around until I could no longer listen to it anywhere in the AM band.

Several years later, while studying my EE, I decided that I should measure my little amp's characteristics. In those days, scopes would cost as much as a small car, so I had to request permission from the lab supervisor.
Upon hooking up the amp to the scope, I was surprised to see an output sinewave without anything connected to the input and the volume control fully down! With the scope, I measured the frequency, and it was about 80 Khz. What I had done back then, is that by trial and error I had managed to shift the oscillation frequency to a value which couldn't be detected by the radio. And I had assumed that the problem had been fixed.

So, with the help of the scope and a better understanding of feedback theory and stability criteria, I was able to completely suppress the oscillations.
But the fascinating thing was that the amplifier DOUBLED its output power, and that the output transformer's heating completely went away.

My recommendation? Get yourself a scope. If one is seriously pursuing electronics as a career, and it appears that you do, an oscilloscope will be your best friend, after your multimeter. And, one can find excellent deals online.

 

[jocanon]

hmmm...I am looking at this one:

**broken link removed**

I could justify $66 if it would do what I need. It kind of looks like a little iPhone lol, I wonder if it would work? I won't be using it for much other than this dummy load. This is more of a second career/hobby. I work full time at a CPA firm as a tax accountant which keeps me busy most of the time .

 

[fernando_g]

That is a possibility. Also search USB scopes.

 

[jocanon]

maybe this one would be better, it's only about 10 miles from me too...maybe I could do local pickup, unless it's just a drop shipper:

**broken link removed**

 

[()blivion]

He should slow down the NFEt response a bit with a 470 to 1000 ohm gate resistor.

This is probably a good idea. You normally do put some resistance between the gate and whatever is driving it. That will probably fix any oscillation problem. I approve.

maybe this one would be better, it's only about 10 miles from me too...maybe I could do local pickup, unless it's just a drop shipper:

**broken link removed**

That isn't too bad, provided you get all the accessories with it. I myself don't own a scope yet. I want one, I just don't want to get one unless it has a really really high sample rate. However... such scopes are supper pricey, which is why I don't have one yet.

 

[jocanon]

...That isn't too bad, provided you get all the accessories with it.

It says it comes with "Oscilloscope Probe kit including one pair of switchable( x1, x10) clip Probes and a deluxe accessory pack."...would I need anything else (other than a laptop of course)?

edit:
here's a list of the accessories:
S/W and Driver CD, 2 Clip Probes, User Manual On the CD, USB cord

 

[dougy83]

That first scope is very limited.

As far as filtering the monitor signal goes, you can connect a capacitor across R14 and R11 to form a first-order filter. The value will be quite large, so you can scale the resistor values (R10/11/14/15) to reduce the required capacitance.

Make sure that a proper star-connection has been used for the 0V reference rail so that the load current doesn't affect the reference voltage and therefore the operating point of the opamps.

Re reducing the sense resistors, ensure that the voltage across them is much greater than the input offset voltage of the opamps; otherwise the MOSFETs will not be sharing current evenly, and the current monitor may be very inaccurate.

 

[jocanon]

Hey dougy83, welcome back...you had helped with the first dummy load, good to have you here too. So how about that second scope I linked, do you think it would be worthless too, or is it better?

 

[dougy83]

Hi. The second scope has a more usable sample rate, for sure. I have the Hantek DSO2250 (250MSps, 60MHz), which is good for playing around with, but has many shortcomings including excessive background noise (seems to be coupled into the scope via the USB power from my laptop) and severe aliasing at large timescales. The software is uninspiring and unprofessional but usable for basic tasks.

I'm not sure how the model you referred to compares, perhaps someone else has some knowledge of its performance (you could check for reviews on the net)

 

[ronv]

Lots of ideas. Here is a new schematic that includes most of them and a few notes on the simulation.

I added some filters for the +24 and the current meter.
I added pull downs on the 358 to draw some current. They have a tendency to have little oscillations when driving purely capacitive loads. Pretty small but easy enough to make go away.
Change the .13 ohm resistors to .1 ohm 10 watt. They will still be hot but not nearly as bad as before. I didn't want to go to much lower for fear of loosing low current settings in the mud.
I ran bode plots and there is good margin (about 10 db) however if I added gate resistance it was reduced due to the gate capacitance. I added a little inductance for the sense resistors - looking for trouble- and it was actually better, so I think we are ok.
Added a bigger filter cap to the +12. Along with your bypass caps. The circuit is sensitive to noise on the 12 volts as it sets the reference.
I'm not sure if something funny is happening or if it is just the cheap little meters.
Don't know what to say about the scope. If you plan on more "stuff" maybe nice.
Thanks everyone for the input.

 

[jocanon]

Hey ronv...welcome! for some reason I can't seem to open the attachment. It just has a paper clip and says Attached Thumbnails, but nothing happens when I click it???

 

[ronv]

Schematic

Try again as gif.

 

[jocanon]

Hm, still not seeing anything on my end?

 

[ronv]

Schematic

If at first..... and all that.