Pratical Hardware Questions for DIY Electronic Load

[t1d]

I am building my own version of Jay’s Dynamic Electronic Load. https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/  I have several practical hardware questions…

For the DUT Input and DMM Output wires, from the banana sockets, on the case, to the PCB, what type of wire should I use? Twisted pair? Coax?

I use a switch, to select between the Reference Voltage and Auxiliary Function Generator inputs. As this switch will pass the signal, itself, what type of wire does it require? And, should I use some type of switching circuit, on the PCB, instead of taking the signal off of the PCB, running it through wires and the switch? If a switching mechanism on the PCB is best, what type of switching circuits might I consider?

I know that I need to separate the power circuit (ground plane) from the signal board(s) (ground plane) and I plan to do that. But, should the voltage reference be on its own, separate board (ground plane), too?

This is a rough draft, of my circuit. I have not proofed it, yet, but you can get the idea.

Thank you, for your help.

[Kleinstein]

The switch for choosing between the internal reference and the external generator signal is not critical, as the signal level is relatively high there and the signal is low impedance. So those simple lever switches are OK.
One could consider a circuit that does not need the switch at all, if the internal and external signal are just added. So the internal reference would be still used to adjust the offset and leaving the input open is enough to disable an external signal.

There should be no special need to use separate ground planes. It is more like on should avoid a ground plane at all, especially with the high current paths at the output. This type of circuit is more suitable to use a star ground point. A ground plane is more for higher frequency and less suitable for precision.

A point too look for might be protection of the ground side of the F-gen input and the current monitoring output. These are connected to the power side and with wrong use could possibly carry high currents. So it might be a good idea to have something like a fuse or fusible resistor there.

Another point to consider would be range switching to allow better accuracy at low currents by only using one output stage. In the current circuit it gets possibly nonlinear below about 5 mV / 0.02 Ohms = 250 mA, as the OPs can have different offsets.

[schmitt trigger]

Very interesting concept to include an external waveform generator input, to provide a a variable load to -let's say- test the a PSU stability. 

The maximum frequency would, of course, be limited to the frequency response limited by the integrating capacitor in the feedback loop.

[t1d]

Kleinstein, I was hoping you would drop by... You always have the good information and suggestions. Thanks, for your help. It is much appreciated.

So those simple lever switches are OK.

Great, that's easy...

One could consider a circuit that does not need the switch at all, if the internal and external signal are just added. So the internal reference would be still used to adjust the offset and leaving the input open is enough to disable an external signal.

Good trick...

There should be no special need to use separate ground planes. It is more like one should avoid a ground plane at all, especially with the high current paths at the output.

I deleted the bottom ground plane, to lessen capacitance. It would be kind of difficult, to eliminate the top plane, because of the use of SMDs and dense populations. But, I will give it a look. I will use the bottom plane, for common feeds... power, load, etc. and separate them, with distance.

This type of circuit is more suitable to use a star ground point. A ground plane is more for higher frequency and less suitable for precision.

I had to look up "star point" ground. I have been using it, all along, but didn't know it had a name.. lol

A point too look for might be protection of the ground side of the F-gen input and the current monitoring output. These are connected to the power side and with wrong use could possibly carry high currents. So it might be a good idea to have something like a fuse or fusible resistor there.

Excellent idea... Maybe I will make their trace margins wider, too...

Another point to consider would be range switching to allow better accuracy at low currents by only using one output stage. In the current circuit it gets possibly nonlinear below about 5 mV / 0.02 Ohms = 250 mA, as the OPs can have different offsets.

I have been thinking on this one, but I could use help, with the switching design... What type of switch, where to place them, etc...

From the op amp and MOSFET suggestions that you made, on the other thread, I decided on the TVL171 and IRFP250... Does that sound good (enough)?

I will be posting, the final design... So, please watch for it. But, there is still lots to do, first.

[t1d]

Thanks, ST, for your help!

The maximum frequency would, of course, be limited to the frequency response limited by the integrating capacitor in the feedback loop.

I take it that you mean the cap in the feedback loop of the summing/driving op amp... Correct?

My FG only goes up to 20MHz. And, I really only intended to be using signals, to test power supplies. Do I just need to look on the internet, for a cap/frequency calculator? Or, is the concerns, for the needs of the summing/driving op amp circuit, too? Remember, I did not design the circuit... Jay_Diddy_B did...

[Kleinstein]

The capacitor at the OPs limit the frequency response, though there is a fast kind of feed forward part. One could speed up the response with a suitable series resistor to the cap, that would increase the fast part. Anyway it would get challenging to make the current regulator really fast. So it is a little questionable with there is enough higher frequency response to really test a good lab supply.

To really get a fast step in load current, it would likely be better to use a different design, with something like 2 current sinks: one for the fixed background and one that is switched on and off fast.

For the choice of OP it depends on the requirements. The TLV171 is a low cost single supply option, somewhat better than the TL07x (needs a negative supply) in most aspects. With a dual supply there might be other good options, like RC4558. It depends on what properties matter.

Usually one should have a fuse for the input, possibly separate for the MOSEFTs - this is especially important if testing powerful batteries. For such an application it might be a good idea to have a low voltage turn off option.

[t1d]

So it is a little questionable with there is enough higher frequency response to really test a good lab supply.

Okay, I will leave that design performance specification, to a future design.

For the choice of OP it depends on the requirements. The TLV171 is a low cost single supply option, somewhat better than the TL07x (needs a negative supply) in most aspects. With a dual supply there might be other good options, like RC4558. It depends on what properties matter.

Doh! I missed a step... I confirmed that the TLV171 is a dual supply, by looking at the max supply voltage rating; -20v to +20v. With that in mind, I did not register the single supply notification, in the text. Shoot! And, I just finished the PCB layout, based on a SOT-23-5 footprint. Many hours, of hard work... Good thing I think this stuff is fun...

Usually one should have a fuse for the input, possibly separate for the MOSEFTs - this is especially important if testing powerful batteries. For such an application it might be a good idea to have a low voltage turn off option.

Oops, I forgot the fuse... You had mentioned it, earlier... I will try to remember to add it.

EDIT: Can I stick with the present design, just by changing the power supply to a single source? That would be easy enough.
EDIT2: The TS321 is a LM358 compatible OA. It comes in a SOT-23-5 package and can operate on single, or dual, voltage supplies. Maybe it would do?

Standard
https://www.mouser.com/ProductDetail/STMicroelectronics/TS321ILT?qs=sGAEpiMZZMuejTFqD3Nt%252bJa7v1LNQLtS

Enhanced
https://www.mouser.com/ProductDetail/STMicroelectronics/TS321IYLT?qs=sGAEpiMZZMt9sXxwSQcXSUlXRjgf3oyO

[Kleinstein]

The TLV171 is also capable of working with a single supply. The TS321 would be an alternative - slower, but likely less low frequency noise.
The old TL074 is what needs the negative supply. The single supply OPs also work with a split (e.g. +-9 V) supply if the voltage fits.

The last circuit looks like it need a negative voltage anyway.

[brabus]

Just a small note on a topic that is often misunderstood: Opamps have no idea if the supply is single or dual and they don't care at all.

Take a loot at any opamp: there are only two supply terminals: if we supply +/-12V or just +24V, for the opamp this means shizzle.

What really matters is the input common mode voltage, regularly well specified in the datasheet.

Opamps advertised as "single supply" often accept a common mode voltage near 0V, but it still has to be checked and double checked.

[t1d]

Thanks, Kleinstein and brabus, for the good information. Phew... I feel much better, now. I had a lot of work, in on this one...

I am told that "The usual input bias cancelling resistors from the +inputs of TLV171 to GND are not needed when working with the TLV171, though, because the input bias currents are ultra low." by Kia, over at the TI forum. So, the +input can go straight to ground? That makes me a bit nervous... I had checked, over there, because the TI data sheet had me confused...

So, back to it! I'm excited...

I will be posting the complete project, soon... But, I think I will do that, on a separate thread.

Woot! As the cool kids say...

[brabus]

You can connect them to GND no problem whatsoever. I would in any case keep the resistors, you may need to patch some wire or try something exotic on your board, so having a clean way to break the ground connection can be useful.

[Mechatrommer]

Very interesting concept to include an external waveform generator input, to provide a a variable load to -let's say- test the a PSU stability. 

The maximum frequency would, of course, be limited to the frequency response limited by the integrating capacitor in the feedback loop.

the concept is more suited to 4 quadrant PSU/Load supply. with 1 quadrant load, you have to limit the FG signal to 1st quadrant region only. and you need to stick the V to A conversion formula in mind when adjusting FG magnitude so shunt wont go outside that region.

[t1d]

Thanks, guys, that's good information. I will give it some thought.

I have thought, a little, about the FG magnitude... I have looked at a few limiter and compressor circuits, but I have not designed anything, yet... They can be complex.

A 2.5v zener could be used, but wouldn't that clip the signal? Might that work as a cheap and easy warning and give some protection? You could see it on an oscilloscope...

Do you have any suggestions, for a circuit type name, that I could research?

[jrsikken]

For your inspiration please take a look at my design.

https://github.com/jrsikken/ElectronicLoadR3

It uses a DAC to set the current, the MCP4725. Pretty cheap and 12bit.

I have tuned the constant current circuit (opamp+mosfet+currentsensefeedback) for fastest response. It settles in 20us. So it can do pulsed load. Using my embedded code it can do 3kHz square wave, or pulses that are 156us wide.

Yesterday I have tried someone's sketch that overclocks the MCP4725 DAC and it generates a 3kHz sinus. Sample rate is 60kHz. The values for the sines are simply looked up from an array in program memory. I have modified the sketch to generate a 6kHz square wave, or pulses that are 78us wide.

What I am trying to say is that the MCU can also be a function generator that can generate any wave shape, and then you don't need external signal generator.

Regarding opamps, I started with MCP6002 and later I changed to MCP6072. I did not need to use input current bias resistor because current is extremely low. I changed to MCP6072 to reduce the offset error and to make the output voltage swing to GND as low as possible.

[ogden]

For your inspiration please take a look at my design.

https://github.com/jrsikken/ElectronicLoadR3

Please do not assume that everybody uses Eagle or have it installed on phones/tablets they often use to read news & forums. It is advised to provide information in more popular file formats like PDF, PNG, JPG. Could you show schematics as picture attached to forum post?

[t1d]

Very interesting concept to include an external waveform generator input, to provide a a variable load to -let's say- test the a PSU stability. 

The maximum frequency would, of course, be limited to the frequency response limited by the integrating capacitor in the feedback loop.

the concept is more suited to 4 quadrant PSU/Load supply. with 1 quadrant load, you have to limit the FG signal to 1st quadrant region only. and you need to stick the V to A conversion formula in mind when adjusting FG magnitude so shunt wont go outside that region.

Hi, Mechatrommer. I have been studying on the needed converter and have realized that I don't know enough, to narrow my search, to be on point... For the quadrant converter, is what I need called a "Four Quadrant Converter?" Do you have a favorite circuit, for this purpose?

I did breadboard just the driver circuit, with an LM741, to try the concept of reading the FG signal. That worked and all four quadrants, of the signal shape, did appear, on the oscilloscope, appropriately. But, you are saying a problem can arise, when you introduce the load, to the remainder of the circuit?

For clarification, the shunt, that you speak of, is the current sink shunt, correct?

Thank you, for the explanations.

[Mechatrommer]

some explanation...
http://www.epanorama.net/sff/Power%20Electronics/Power_Supplies-Linear/Four%20Quadrant%20Power%20Supply.pdf
https://www.accelinstruments.com/Applications/TS200/Four-Quadrant.html
but i think its not relevant to dynamic load circuit (1 quadrant aka +ve voltage and sink current).

[t1d]

some explanation...
http://www.epanorama.net/sff/Power%20Electronics/Power_Supplies-Linear/Four%20Quadrant%20Power%20Supply.pdf
https://www.accelinstruments.com/Applications/TS200/Four-Quadrant.html
but i think its not relevant to dynamic load circuit (1 quadrant aka +ve voltage and sink current).

Thank you, for your super fast reply.

I apologize, but I am not sure I understand... Are you saying that the information in the link is not relevant to fixing my circuit? Or, are you saying that the FG operation, of my circuit, will work, without any changes?

I appreciate your continued help.

[Mechatrommer]

Are you saying that the information in the link is not relevant to fixing my circuit?

my reply to scmitt trigger concerning using FG to control dynamic load. since dynamic load only works as current sink (not current supply as in PSU) and +ve voltage (Vds), hence 1 quadrant... and a FG can generally output +ve and -ve voltage, ie 2 quadrant in your application (by applying "power source" to the dynamic load input), so we need some care on using FG in your circuit. for example if we feed -ve signal, by definition your dynamic load should supply current to DUT (-ve shunt voltage), but in reality its not (signal or current clipped will occur).

Or, are you saying that the FG operation, of my circuit, will work, without any changes?

it should work, not looking closely on your circuit, but user need to know what he is doing when applying FG signal into your circuit. btw, one good application to this is applying a suitable +ve volt PWM signal to the dynamic load to analyze DUT power transient respond, (as i've implemented internally in my dynamic load project) ymmv.

[t1d]

Thank you, for straightening me out.

"but user need to know what he is doing when applying FG signal into your circuit"
Excellent advice. Obviously, I designed the circuit to work with my particular FG. But, for other folks' equipment...

About pulsing... My FG has this feature.

Good help... Thanks!