Hacking A Dynamic Electronic Load Circuit To Use Function Generator? Solved

[t1d]

EDIT: This thread results in two Auxiliary Function Generator input circuits... One, as an additional option, for Jay's Dynamic Electronic Load Design. And, two, another that does not use Jay's Function Generator option. Many, many thanks, to those that participated, in the work. And, of course to Jay, for the original E-Load circuit, itself. The hack would not have come together, without them!

I am working toward building Jay Diddy B’s Dual Supply Dynamic Electronic Load. You can find it, here: https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/. The schematic, for my adaptation, is attached, below.

Note About My Schematic
Jay has told me that the values of my resistors for the Meter Output are incorrect… As is, they, in total, produce 25mV. I am working on getting them changed. (If you know the formula, feel free to calculate the value.)

And, I mislabeled the op amps... I used the quad symbol, of the LT1014, but labeled it as the dual LT1013. Creating a final schematic, for Jay's project, and for my personal preferences, is in the proofing stage. This is how this was just discovered.

Disclaimer: I have posted these questions, within Jay’s thread. They have not been answered. I think, because they add project creep. To avoid that, I think it better, to post them as a separate thread, because this is for my own singular purpose.

A prominent feature, of the design, is an on-board square wave oscillator that can be used to test the slew rate of power supplies. I want this feature, but I have several Function Generators, that have auxiliary outputs. I’m thinking that I would have many more options (a sign wave, in particular, variable frequency, etc.,) if I removed the on-board oscillator and added an input, for a Function Generator. I am an intermediate noob and these design questions are beyond my knowledge base. So, I would appreciate your help.

Adapted from previous post:
If using an independent function generator would add functionality greater than the on-board oscillator, where, and how, should it be injected, in the circuit? Just use the type jack that breaks the circuit? Break Before Make? What issues might there be?

If using an independent function generator is not wise, then I would keep the existing oscillator. If I wanted to add a switch, to bring in, or cut out, the oscillation section, without spinning its adjustment knob, to zero, where, and how, would it go?

Would there be any great advantage to using ten-turn pots? I would think that, for home-lab use, things just aren’t going to need to be so precise. Thoughts?

It is most likely that I would be using the function generator that is in my Tenma Universal Test Center 72-1005. The auxiliary output is selectable; 50 Ohm, or 600 Ohm.

Its operation manual and schematic are too big to post. I have placed them in a DropBox, so that you can know exactly what we are working with. https://www.dropbox.com/h   If you have trouble getting them from DropBox, just send me a PM with your email address and I will send the files to you, direct.

I am concerned that some of its functions might damage the Electronic Load circuit… Like the pulse feature… Or, too much input gain. The operation manual explains all of its features.

Thank you, in advance, for your help. In appreciation of your support, I would be glad to post the completed KiCad schematic file, for my version. (I am already set up to release Jay Diddy B's design, on his thread.)

[t1d]

I discovered the answer to my question, in Jay_Diddy_B's original thread:

"If I replace the square wave oscillator say with wave gen that capable of producing various pulse shapes like sawtooth, triangle, shark fin, sine and etc, again, as above, will it affect the loop stability ?"


"The U3 -input (on the LTspice model) is a summing node. You can add the output from your signal generator, through a 100K resistor, to this input. You can then use all the waveforms in signal generator."

[David Hess]

And, I mislabeled the op amps... I used the quad symbol, of the LT1014, but labeled it as the dual LT1013. Creating a final schematic, for Jay's project, and for my personal preferences, is in the proofing stage. This is how this was just discovered.

They are the same operational amplifier though so it does not matter.  I actually prefer using duals for easier layouts.  The LT1006 is the single version of the LT1013/LT1014.  Note that these are improved direct replacements for the 358/324.

A prominent feature, of the design, is an on-board square wave oscillator that can be used to test the slew rate of power supplies. I want this feature, but I have several Function Generators, that have auxiliary outputs. I’m thinking that I would have many more options (a sign wave, in particular, variable frequency, etc.,) if I removed the on-board oscillator and added an input, for a Function Generator. I am an intermediate noob and these design questions are beyond my knowledge base. So, I would appreciate your help.

That makes sense but the design at your link is not intended for fast dynamic operation.  Power supply slew rate is tested with a much faster open loop circuit.

If using an independent function generator would add functionality greater than the on-board oscillator, where, and how, should it be injected, in the circuit? Just use the type jack that breaks the circuit? Break Before Make? What issues might there be?

I would probably inject it as another signal into inverting summing amplifier U3.  This would allow combining it with the existing DC load control signal although most function generators can generate a DC offset.

If using an independent function generator is not wise, then I would keep the existing oscillator. If I wanted to add a switch, to bring in, or cut out, the oscillation section, without spinning its adjustment knob, to zero, where, and how, would it go?

There is nothing wrong with the function generator idea.  Some 2 quadrant power supplies support this and I assume some 2 quadrant dynamic loads do as well.

While not ideal, it is probably best to open R18 to disable the square wave function.

Would there be any great advantage to using ten-turn pots? I would think that, for home-lab use, things just aren’t going to need to be so precise. Thoughts?[/i]

I prefer coarse and fine adjustments however greater precision requires the coarse adjustment or both to be replaced with a multi-turn potentiometer.  Dynamic load testing does not require great precision.

I am concerned that some of its functions might damage the Electronic Load circuit… Like the pulse feature… Or, too much input gain. The operation manual explains all of its features.

The auxiliary input could be clamped to limit its range to safe values.

[t1d]

Good stuff, David. Thank you, so much.

That makes sense but the design at your link is not intended for fast dynamic operation.  Power supply slew rate is tested with a much faster open loop circuit.

Faster speed and different wave shapes shapes is why I wanted to explore this.

I am concerned that some of its functions might damage the Electronic Load circuit… Like the pulse feature… Or, too much input gain.

The operation manual explains all of its features.

The auxiliary input could be clamped to limit its range to safe values.

For a clamp, just a typical zener diode application?

[David Hess]

I am concerned that some of its functions might damage the Electronic Load circuit… Like the pulse feature… Or, too much input gain.

The operation manual explains all of its features.

The auxiliary input could be clamped to limit its range to safe values.

For a clamp, just a typical zener diode application?

There are lots of ways to clamp a signal and a zener diode (or two) is certainly a candidate.  It does not have to be complicated or precise.

[t1d]

Thank you!

[t1d]

David,

RE: Zener Clamp

David,
I have been reading about zener clamps. Best I can tell...

- I would inject the Function Generator signal, before Pot #2. This would allow the FG signal to be attenuated, by the pot.

- I would base the calculations of the clamp on the fixed resistor value of 100 ohms, as this is the minimum resistance of the pot and resistor, together. But, wouldn’t this resulting variable resistance mess up the resistor to cap relationship?

- I need to calculate the cap value to accommodate the upper speed of any frequency that I would inject. My FG only goes to 20MHz, but I want to make the E-Load as versatile as possible. What would be a good number, for this frequency, to do general testing?

- The FG signal swings positive and negative. But, doesn’t the E-load circuit need the signal to just be positive? Meaning that I need a Positive, Unbiased Clamper?

- The voltage reference supplies 2.5v. So, doesn’t the clamper need to limit the signal voltage to 2.5v?

Thanks for your help, with these noob questions…

[MarkF]

I did something similar.  If you feed the set voltage into the "+" input instead, you don't need a negative supply.  You still would need to clamp the input to 0-5V.

   

[David Hess]

- I would base the calculations of the clamp on the fixed resistor value of 100 ohms, as this is the minimum resistance of the pot and resistor, together. But, wouldn’t this resulting variable resistance mess up the resistor to cap relationship?

A series coupling capacitor?  It would only change a little as shown.

- I need to calculate the cap value to accommodate the upper speed of any frequency that I would inject. My FG only goes to 20MHz, but I want to make the E-Load as versatile as possible. What would be a good number, for this frequency, to do general testing?

This design is never going to reproduce a 20 MHz signal.  Just achieving 1 MHz would be a challenge which is why I said power supply testing usually uses an open loop pulsed load.

- The FG signal swings positive and negative. But, doesn’t the E-load circuit need the signal to just be positive? Meaning that I need a Positive, Unbiased Clamper?

That is right.  Since this electronic load only supports 1 quadrant operation, the signal from the function generator should be clamped to prevent negative operation.

- The voltage reference supplies 2.5v. So, doesn’t the clamper need to limit the signal voltage to 2.5v?

That is how it would normally be done but it also depends on other details like gain of the separate circuit paths.

[t1d]

Thank you, David. Good information. I will have to process it, a bit...

[t1d]

David,

Here is the Zener, I am thinking of using.
Nexperia NZX2V4A,133
Zener Single 2.4V 4% 500mW 2-Pin

Here is the clamp design that I propose. A 100K resistor, already exists, at the injection point.

Is a capacitor needed? Most clamp designs seem to have it. If so, how do I calculate its value?

I would think that using a resistor divider network, to drop the input voltage, before the diode, might be a good thing. But, I don’t know what to expect, as the common auxiliary output voltage, of frequency generators. Do you have an estimate?

My Tenma FG’s output reads as
Output Amplitude
2Vpp to 20Vpp at open load
1Vpp  to 10Vpp at 50 ohm load

Would it be good to use a 51 ohm resistor, to terminate the e-load FG input? I may have not worded that correctly.

Thank you, for all your help.

[David Hess]

The frequency response is so low that a terminating resistor will not be needed for the function generator.

Why is the zener diode in series with the function generator input?  Clamping requires it to be in parallel with a series resistor to limit the current.

[t1d]

Why is the zener diode in series with the function generator input?

Because I didn't have it straight, in my mind... lol

Zener Capacitor Calculations
The capacitor must have enough charge, to last the entire period of discharge.

Here, we will assign the slowest frequency (thereby having the longest discharge period) that the function generator might provide, as 1 Hertz. The capacitor charge time is one-half of the complete wave cycle. Therefore, for this calculation, T = 1 and 2T = ½. A safety factor of 10 x T will be included. The resistor will be 100 ohm.

Resulting Formula
5RC=T/2
5(100)C = 1/2
500C = ½
C = 0.5/500
C = 0.001uF
C = 1nF, aka code 102

Do I have the math correctly? Thank you, so much.

The corrected schematic is attached.

[David Hess]

That is not going to work.  The inverting input of the operational amplifier is a virtual ground point (summing junction) so the voltage never changes there.  Directly connecting the coupling capacitor to that point will destabilized the operational amplifier.

Given your current design, I would expect a 100k resistor between the junction of the zener diode and coupling capacitor and the summing junction or maybe a 50k resistor on each side of the zener diode junction if you want to keep the coupling capacitor.

Note that such high impedances like 100k are going to limit frequency response because of stray capacitive coupling.

[t1d]

Thanks, David, I will keep at it.

[t1d]

That is not going to work.  The inverting input of the operational amplifier is a virtual ground point (summing junction) so the voltage never changes there.  Directly connecting the coupling capacitor to that point will destabilized the operational amplifier.

Given your current design, I would expect a 100k resistor between the junction of the zener diode and coupling capacitor and the summing junction or maybe a 50k resistor on each side of the zener diode junction if you want to keep the coupling capacitor.

Note that such high impedances like 100k are going to limit frequency response because of stray capacitive coupling.

I discovered the answer to my question, in Jay_Diddy_B's original thread:
"The U3 -input (on the LTspice model) is a summing node. You can add the output from your signal generator, through a 100K resistor, to this input. You can then use all the waveforms in signal generator."

David, I am a bit confused. I understand that you are suggesting changes, to the clamping circuit. But, are you also saying that I am not injecting the FG signal, into the circuit, at the correct place. If so, I think that I am at the location that Jay suggested. So, where is the best spot? After sorting this out, I will have questions, about the added components. Thank you!

EDIT: If the capacitor is not needed for biasing, then, without good benefit, I do not want to add it. Maybe a drawing would help and I could reverse engineer the understanding of the operations... Thanks, yet again...

[David Hess]

The summing node works fine for injecting the signal and the function generator schematics that I checked do it that way but you cannot clamp the input there because as a virtual ground, it essentially never changes voltage.

Technically you could clamp the signal in series by liming the current into the summing node from the auxiliary input but that gets away from simple zener clamping.  I would consider this I wanted a very fast and very accurate clamping level.

[t1d]

The summing node works fine for injecting the signal and the function generator schematics that I checked do it that way but you cannot clamp the input there because as a virtual ground, it essentially never changes voltage.

David, I moved the injection point to in front of the R21 100K resistor. I tied the sink, for the clamping diode and resistor, to -9v. Is that what you are explaining, in the quote?

Technically you could clamp the signal in series by liming the current into the summing node from the auxiliary input but that gets away from simple zener clamping.  I would consider this I wanted a very fast and very accurate clamping level.

I don't mind a more complicated design. Is there a name, for this technique/circuit? Or, could you point me at a link, so I can learn about it and get a design? Just a good old transistor current limiting circuit? Or, does it need to be faster?

[JS]

  I don't know the problem with the thing going low, certainly a zener from the negative rail won't help much. In the case the input goes negative the load will turn it self off and nothing really bad will happen. In the case you don't want it as it will disturb the dynamics of it you could clamp it from a different place. A good option could be adding a diode in parallel with each feedback cap of the driver amplifiers so the output can't go too negative and keeps the cap from charging too much and one in parallel with R16 so the output can't go too positive and they all recover faster. With this you could use a negative input to turn the load off but the recovery from that state will be much faster than letting all saturate and wait smacked up to the rail.

  With that I'd take out the coupling cap, just run from DC so you can control the load with an external reference and forget completely from the external reference, with this and a programable AWG you can set up any test for a wide range of conditions to run and go to take a coffee as the measurements are ready for evaluation.

  Also, I'd sum the signal to the proper point, so a resistor to the inverting input of the amplifier replicating R21 for the external input.

  Having some series resistor with the input and some attenuation doesn't seem like a bad idea to me either, any signal generator will go well over what you need for this load reference...

  So, my circuit would be a 50Ω load to ground, a 100k resistor to pin 13 of U1D and the diodes in the feedback of each opamp to provide fast recovery. If you want you could add an input attenuator (change the 100k resistor might not be practical as two decades up it's just too big to work properly).

JS

[t1d]

Good stuff, JS. I will be trying to draw your design, in schematic form. If you could provide a drawing, for me to work from, it would help, as I still have noob limitations... I really appreciate your participation.

[t1d]

So, my circuit would be a 50Ω load to ground, a 100k resistor to pin 13 of U1D and the diodes in the feedback of each opamp to provide fast recovery. If you want you could add an input attenuator (change the 100k resistor might not be practical as two decades up it's just too big to work properly).
JS

Well, I have to admit, just with a first look, I am confused. U1D is an individual MOSFET load, isn't it? How would it distribute the signal, to all the MOSFETs. I apologize, if this is a basic question. I am still learning how the circuit works.

And, it might be easier to communicate, using the schematic that I am adapting. I have attached it, here, in its entirety. I should have provided, the complete schematic, and not just the summing node, earlier. My apologies.

[JS]

I can't se te schematic, tapatalk Xiang Sant me to download pdfs... Will check it later on the pc.

Sorry if i messed the naming, I'm pretty sure I was talking about the general opaml, not a single driver to get the signal into.

JS

[t1d]

This is the only information that I found that is on-point, after hours of additional (frustrating) research. It is only general, in nature. It shows the Function Generator signal injected into the op amp, with no conditioning, between the positive [this being a non-inverting example] input and ground... Even though it is an ac circuit.

http://gpete.blogs.keysight.com/2012/03/building-electronic-load-with-general.html

[t1d]

Sorry if i messed the naming, I'm pretty sure I was talking about the general opaml, not a single driver to get the signal into.
JS

Oops. my mistake... Yes, the first post of the schematic lists the op amp as U1D... But, this schematic was, later in the thread, revised and renumbered. I'm sorry for the confusion.

[t1d]

  So, my circuit would be a 50Ω load to ground, a 100k resistor to pin 13 of U1D and the diodes in the feedback of each opamp to provide fast recovery. If you want you could add an input attenuator (change the 100k resistor might not be practical as two decades up it's just too big to work properly).
JS

I have been thinking through your solution. I believe I understand your instructions up to "and the diodes in the feedback of each opamp." It sounds as if you are suggesting diodes in the feedback of op amps other than U1D (now know as U2B)? If so, which ones and why?

For attenuation, how about injecting the signal, before Pot2? I would like adjustable attenuation, if this would work. If so, wouldn't the value of the 50 ohm resistor, R45, need to be adjusted? If so, to what value?

Here is a drawing, of my understanding, so far... Notice that I make use of the existing 100K resistor; R21.

Thanks, for getting me through this.