Author Topic: Dynamic Electronic Load Project  (Read 144847 times)

0 Members and 1 Guest are viewing this topic.

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #75 on: August 20, 2018, 07:25:27 pm »
Hi group,

One question that comes up a lot in the discussion of electronic loads is:

Why is one op-amp per MOSFET better?



Most commercial loads, like the HP6060A/B and 6050x use one op-amp MOSFET.

The MOSFET datasheet has the answer:



This graph is from the obsolete On-Semi MTP3055V Datasheet. Other parts have similar characteristics.
If you fix the Vgs at 4V you will see that the drain current, Id, has a positive temperature coefficient. As the device gets hotter, there is more Id and the temperature rises. We have the conditions for thermal runaway.

Other hand if we have one op-amp MOSFET each of the drain currents is controlled separately and there is no possibility of thermal runaway.

Modelling



Here I have a load with two identical MOSFETs. There is no issue, the Drain currents are perfectly balanced.



If I unbalance the circuit by placing a 0.1V source in series with the MOSFET gate. This is the same as the MOSFET having a different transfer characteristic:



The currents are no longer balanced:



Separate Source resistors

One technique that is often employed is to use separate source resistors for each MOSFET. We can modify the load to this:



R7 and R8 were added so that the load regulates the average of the current in M4 and M5.

Results of separate source resistors.



Although there is an improvement, it is only slight. By inspection, for this to work well the voltage across R3 and R6 must be 'large' compared to differences in Vgs transfer characteristics. This means high power dissipation in these resistors.

Regards,

Jay_Diddy_B
« Last Edit: August 20, 2018, 07:32:53 pm by Jay_Diddy_B »
 
The following users thanked this post: BravoV, Kean, ogden

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #76 on: August 20, 2018, 10:59:51 pm »
Thermal runaway... Brilliant! Thanks, Jay. Excellent explanation, too. I suggest also posting this explanation, as its own thread, to make it easier to find the information.

Speaking of the MTP3055V, I have a supply of the MTP3055VL and AD8032ANZ Dual Op Amp. Would these work, for your single supply design?

I am building the single supply design schematic and PCB, in Kicad, in order to have the PCB manufactured. I will be glad to contribute the files, once they are finished. They will greatly reduce the work needed, to make personal customizations.

I will probably need help, with proofing the schematic and laying out the PCB (Though I am not bad, at PCB Layout. But, more eyes are better.) And, with multiplying the design... I am going for ~100 watts, to test PSUs, having 30v and 3a.

One issue, that I already see, is that the Op Amp schematic symbol dictates the PCB footprint pin assignments. So, the components will need to be placed, on the PCB, and, then, the Op Amp schematic symbols will need to be swapped around, to facilitate the best placement of the components, on the PCB.
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #77 on: August 21, 2018, 12:25:05 am »
Jay, as I am not a professional EE, I am not familiar, with this symbolism. I take it that the circles indicate complete nodes(correct term?) The one on the left being the 9 volt regulator, with all its supporting components, and the one on the right being the 2.5 volt reference, with all its supporting components? Correct?
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #78 on: August 21, 2018, 03:08:34 am »
Jay, as I am not a professional EE, I am not familiar, with this symbolism. I take it that the circles indicate complete nodes(correct term?) The one on the left being the 9 volt regulator, with all its supporting components, and the one on the right being the 2.5 volt reference, with all its supporting components? Correct?

These are symbols used in SPICE schematics. The circle with a V designation is a voltage source.



In this schematic:

V1 is a 9V power source, it could be a regulated power supply or simply a 9V battery.

V2 is the power supply being tested

V3 is a voltage source used to generate an offset.


In the other schematic there is a 2.5V voltage source, this could be a voltage output reference (example LTC6655CHMS8-2.5)  or a shunt regulator(example TL431) and its associated components.

It is a shortform, used in modelling.

Jay_Diddy_B
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #79 on: August 21, 2018, 04:57:24 am »
Background: KiCad components need to be shown, on the schematic, in order for the traces to be tied, pin-to-pin, on the board. This requirement dictates clarity, in the following matters. Meaning, I am not trying to be picky.

V1 is a 9V power source, it could be a regulated power supply or simply a 9V battery.
Here, I have added a L7809, with supporting caps.

V2 is the power supply being tested
V3 is a voltage source used to generate an offset.

V2 is not included, on the Single Supply schematic. I take it that, on the Single Supply schematic, R5, L1 and V3 should be removed and a DUT input connector (+ and -) should be added, in their place. Correct? Yes, I see that, on the “Real Schematic,” now. Added same.

In the other schematic there is a 2.5V voltage source, this could be a voltage output reference (example LTC6655CHMS8-2.5)  or a shunt regulator(example TL431) and its associated components.
Here, I have added the LT6656 voltage reference, with supporting caps.

I added the meter output terminating resistor, R19/51R, and its connector.

Please clarify the suffix, of the values, of these resistors. I could guess, but that is dangerous.  Such as R19 is a 51R resistor…

R1   100
R2   0.1
R20   100
R21   0.1
R23   3300
R4   3300
R5   0.1
R9   2.2

I am nearly finished. Do you use KiCad? I will provide a jpg, too, as the schematic will, now, be ready to use, without need of researching the items, above.

I am really enjoying this adventure. Thanks!
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #80 on: August 23, 2018, 03:22:38 am »
It is my goal to create a schematic, in KiCad, of the Single Supply Dynamic Load Circuit, to share and to use to produce a working board, for a finished device. Attached is my Rough Draft, of same, in pdf format. I will provide the KiCad files, after everything is finalized.

This schematic is a combination of the Single Supply Model and Jay’s Real Schematic. I have medical difficulties that make it nearly impossible to transpose visual information… (Flip a perf board over and solder it up… Forget-about-it…) So, I am not confident of my work.

Additional Questions, to my post #80:
#1) Red Block #1: The two pots (R21 and R22) and the three resistor divider network (R12, R13 and R14) were added, to the schematic, from JRS. I need someone to a) confirm the placements, within the circuit, and b) the values of the pots and resistors.
#2) Red Block #2: It is my guess that this resistor divider network is only for the purpose of modeling the circuit and should be removed, from the actual schematic. Correct?
#3) Some items, such as the voltage regulator, are substituted, for modeled items, from known, common applications.
#4) I need lots of folks, to proof the entire schematic, for errors, missing items, etc. The more eyes, the better.
#5) I would also like to know, if I can substitute AD8032ANZ’s, for the Op Amps, and MTP3055VL’s, for the MOSFETs. I plan to use five banks, to create ~100 watt capability.

I appreciate your help and look forward to finishing this up, with you.
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #81 on: August 23, 2018, 09:58:29 pm »
T1D and the group,

The circuit presented by T1d has some issues:



The op-amp U1B in configured as an inverting amplifier. It has positive inputs and a single, positive, supply. The output will try and go negative. Because there is only a positive supply, the output of U1B will always be 0V.

Remember, I have not built the single-supply version.

Regards,

Jay_Diddy_B
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #82 on: August 23, 2018, 10:22:44 pm »
Hi group,

Here is a little more modelling on the single-supply circuit that I proposed. I have modified the potentiometers to be voltage controlled. The control voltage is 0-100V 0V is fully ccw and 100V is fully cw.
These are representing normal, linear 10K, potentiometers, like those used when I built the dual supply version.

This is the new model:



The results are:





I have attached the model for those playing along at home.

Regards,

Jay_Diddy_B
« Last Edit: August 23, 2018, 10:24:39 pm by Jay_Diddy_B »
 
The following users thanked this post: HackedFridgeMagnet

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #83 on: August 24, 2018, 12:33:23 am »
The circuit presented by T1d has some issues:
Jay, please know that I was not intending to propose any engineering, of your circuit... I was just trying to turn the Single Supply model into a schematic. I was attempting to combine the SS model, with the pots, from the dual model. I really made a mess, of that. It makes perfect sense, now that you have labeled the pots.

The new pfd, of the Revised Single Supply schematic, is attached. Please proof it. If it is okay, I will post the KiCad files.

I left out some of the narrative text, that I thought was for modeling purposes. If that text should be on the schematic, please let me know.

Would it be okay to bread board the circuit? Or, will there be issues... Voltage, amperage, capacitance? I would think that I would not want to run it above 1v/1a. If I can Bbd it, I will order the exact parts, from Mouser.

Thanks, for all that you are doing, for us.
« Last Edit: August 24, 2018, 12:36:23 am by t1d »
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #84 on: August 24, 2018, 01:22:46 am »
At the risk of adding project creep, I have two questions:
1) If I wanted to add a switch, to bring in, or cut out, the oscillation section, without spinning the knob, where would it go?
2) If using an independent function generator would add functionality greater than the on-board oscillator, where, in the circuit, should it be injected? Just use the type jack that breaks the circuit? What issues might there be?

Non-creep question:
Would there be any great advantage to using ten-turn pots?

Thanks!
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #85 on: August 24, 2018, 04:41:43 am »
Hi T1D and the group,

It is so much easier to do this with a negative rail. I would go back to this circuit, from the first post in this thread:



I would sum the external signal into U1B by adding another 100K resistor to pin 6, the summing node.

I would convert it to single supply operation by adding a negative voltage generator. Here is an idea based on a 555 timer:



And the output of the circuit:



I have attached the model.

Regards,

Jay_Diddy_B
« Last Edit: August 24, 2018, 04:43:15 am by Jay_Diddy_B »
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #86 on: August 25, 2018, 06:11:09 am »
I have finished the rough draft, of the Dual Supply model. Here is the pfd.

I added some options, for the Vref, the V Supply and the number of MOSFETs, based on the parts I have in my stocks and what I intend to build. Take your pick and delete the remainder. Or, add your own twist.

I will post the KiCad files, as soon as I clean up a few details. To do that, I could use some help, with these (bumped) questions:
Please clarify the suffix, of the values, of these resistors. I could guess, but that is dangerous.  Such as R19 is a 51R resistor… These component reference numbers are for the Single Supply model, that I posted above, but they need to be updated in the Dual Model, too.

R1   100
R2   0.1
R20   100
R21   0.1
R23   3300
R4   3300
R5   0.1
R9   2.2

Would it be okay, to bread board the circuit? Or, will there be issues... Voltage, amperage, capacitance? I would think that I would not want to run it above 1v/1a. If I can Bbd it, I will order the exact parts, from Mouser.

At the risk of adding project creep, I have two questions:
1) If I wanted to add a switch, to bring in, or cut out, the oscillation section, without spinning the knob, where would it go?
2) If using an independent function generator would add functionality greater than the on-board oscillator, where, in the circuit, should it be injected? Just use the type jack that breaks the circuit? What issues might there be?
3) If two dual op amps are used, I would think that any unused units might be used to create a sign wave, too, in the oscillator section.

Non-creep question:
Would there be any great advantage to using ten-turn pots?

I still need lots of help with proofing, on both models. Please, please, oh please and thank you. <grin>
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1773
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: Dynamic Electronic Load Project
« Reply #87 on: August 25, 2018, 01:00:04 pm »
Hi,

you probably know that there are  specialized ICs to create negative voltages from positive ones with less effort (ICL7660).
Why not try one of those ?
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #88 on: August 26, 2018, 12:54:45 am »
Hi,

you probably know that there are  specialized ICs to create negative voltages from positive ones with less effort (ICL7660).
Why not try one of those ?

This is good advice. I suggested the CMOS 555 because they are widely available.

Regards,

Jay_Diddy_B
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #89 on: August 26, 2018, 12:58:03 am »
I have finished the rough draft, of the Dual Supply model. Here is the pfd.

Snip ...

I still need lots of help with proofing, on both models. Please, please, oh please and thank you. <grin>

There is a mistake in the VAC power supply option:



You can't use the LM7909 to generate a negative voltage from a positive input.

The rectifier section should look like this:



Two half-wave rectifiers, one is positive, the other negative.

Regards,

Jay_Diddy_B


« Last Edit: August 26, 2018, 12:59:42 am by Jay_Diddy_B »
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #90 on: August 26, 2018, 01:13:23 am »
There is a mistake in the VAC power supply option:
Well, I should be embarrassed... But, having no sleep, in 36 hours, I think I won't go to the bother.<grin>

I will get this cleaned up.

I have been using the time to work on my version of the project. I have the schematic and a rough draft, of the PCB. I will post it, just for fun, in a little bit...

Thanks!
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #91 on: August 26, 2018, 03:35:42 am »
I have the schematic and a rough draft, of the PCB. I will post it, just for fun, in a little bit...
Jay, thanks for proofing the schematic and catching the issue. Here is the corrected version, of the VAC Option. I will post the KiCad files, as soon as I hear back, on my other questions.

Posting my personal version will have to wait, I think...
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14082
  • Country: de
Re: Dynamic Electronic Load Project
« Reply #92 on: August 26, 2018, 07:20:54 am »
I have the schematic and a rough draft, of the PCB. I will post it, just for fun, in a little bit...
Jay, thanks for proofing the schematic and catching the issue. Here is the corrected version, of the VAC Option. I will post the KiCad files, as soon as I hear back, on my other questions.

Posting my personal version will have to wait, I think...

There is a wrong connection from output to shunt in the lower right corner of the circuit drawing.

For current measurement, one should average the voltage over the different parts, the current sharing may not be very accurate (good enough for sharing the heat, but not for measurement. So it would be 2 or 4 resistors to the measurement output.

The choice of OPs mentioned here is odd:
The LT1013 is rather slow. It can be an option for a more static load with single supply.
The TL084 has quite some offset, drift and noise. The offset can cause quite some trouble at the low end, when due to the offset only 1 channel might be active. So at least in this case one should have the option to disable all but one channel when using only a small current.

When making a board also be aware of the odd pin-out of the SO8 version of the LT1013.
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #93 on: August 26, 2018, 08:56:12 am »
There is a wrong connection from output to shunt in the lower right corner of the circuit drawing.


Here is a picture:



Remove the connection marked with the X.

Regards,

Jay_Diddy_B
« Last Edit: August 26, 2018, 09:27:20 am by Jay_Diddy_B »
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #94 on: August 26, 2018, 09:01:02 am »


For current measurement, one should average the voltage over the different parts, the current sharing may not be very accurate (good enough for sharing the heat, but not for measurement. So it would be 2 or 4 resistors to the measurement output.


The current monitor can be improved by averaging the voltage across the source resistors like this:




or (the same thing)



The resistor should be 50  \$\Omega\$ x number of output stages. With four output stages use 200 \$\Omega\$ resistors.

Regards,

Jay_Diddy_B
« Last Edit: August 26, 2018, 09:03:00 am by Jay_Diddy_B »
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #95 on: August 26, 2018, 09:05:03 am »
Hi,

There is still an error in the VAC power supply option:



Make the changes shown in red.

Regards,
Jay_Diddy_B
 

Offline Jay_Diddy_BTopic starter

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: Dynamic Electronic Load Project
« Reply #96 on: August 26, 2018, 09:18:50 am »


The choice of OPs mentioned here is odd:
The LT1013 is rather slow. It can be an option for a more static load with single supply.
The TL084 has quite some offset, drift and noise. The offset can cause quite some trouble at the low end, when due to the offset only 1 channel might be active. So at least in this case one should have the option to disable all but one channel when using only a small current.

When making a board also be aware of the odd pin-out of the SO8 version of the LT1013.

This is the most interesting part. The dual supply configuration allows a wide selection of op-amps to be used as it removes the need for rail-rail input or output.

I use the LT1013 for a lot of modelling because the LT1013 is very close to the LM324, except the LT1013 has better offset etc.

I would actually use the quad LT1014 if built the circuit. Standard quad op-amps will fit on the LT1014 footprint if pin 1 is lined up.




This model shows the same load circuit with the LT1013 (GBW 1MHz), the LT1801 (GBW 80 MHz) and voltage controlled voltage source (infinite GBW):



If I compare the LT1013 versus the LT1801:



The loop bandwidth is 40kHz for the LT1013 and 60 kHz for LT1801. In both cases the phase margin is greater than 80 degrees.

or the LT1013 versus voltage controlled current source:



The voltage controlled voltage source could be considered an 'ideal op-amp'. Again there is very little difference between the LT1013 and the ideal op-amp.

The circuit is dominated by the feedback network around the op-amp and the MOSFET input capacitance and the gate resistor. See earlier in this thread for more analysis.

I see that the op choice has little impact, as it should on the small signal behaviour.

The LT1013 (LT1014) is good enough and the offset performance is desirable.

Regards,

Jay_Diddy_B


« Last Edit: August 26, 2018, 09:26:02 am by Jay_Diddy_B »
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #97 on: August 26, 2018, 06:32:04 pm »
I would actually use the quad LT1014 if built the circuit. Standard quad op-amps will fit on the LT1014 footprint if pin 1 is lined up.
From when I was shopping for op amps, I looked at the LT1013/4 and TL074 data sheets. I remember seeing that, for some characteristic, that I now forget, there is some significant difference between the 13 and 14 specifications... Sorry to forget the exact item. But, if someone wants to use that brand, they should take a look... As I recall, the 13 had the better spec, for whatever it was...

I might take a look for it, later... But, I wanted to mention it, now, so, if someone was about to place an order, they could have this heads-up...

Thanks, soooo much, to everyone, for the notes, the schematic about... I am going to work on making those changes, now. I hope to post back, later in the day.
 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #98 on: August 26, 2018, 07:43:27 pm »
Here is the updated KiCad Schematic pfd.

Jay, I will leave it up to you (please,) to say when the schematic is complete. Once we do, you may want to add it to your very first post, for ease of finding it?

Jay, your making of handwritten notes, in red, on the schematic, is an excellent method of communication. Me grok. <grin> Please continue to include them...

Items from my prior posts, that we may (or may not?) want to consider, for finishing the schematic.
Please clarify the suffix, of the values, of these resistors. I could guess, but that is dangerous.  Such as R19 is a 51R resistor… These component reference numbers are for the Single Supply model, that I posted above, but they need to be updated in the Dual Model, too. EDIT: These resistors have been renumbered, in the Dual Supply Model, because of changes. Be sure to refer to the Single Supply Model pdf.

R1   100
R2   0.1
R20   100
R21   0.1
R23   3300
R4   3300
R5   0.1
R9   2.2

Would it be okay, to bread board the circuit? Or, will there be issues... Voltage, amperage, capacitance? I would think that I would not want to run it above 1v/1a. If I can Bbd it, I will order the exact parts, from Mouser.

At the risk of adding project creep, I have two questions:
1) If I wanted to add a switch, to bring in, or cut out, the oscillation section, without spinning the knob, where would it go?
2) If using an independent function generator would add functionality greater than the on-board oscillator, where, in the circuit, should it be injected? Just use the type jack that breaks the circuit? What issues might there be?
3) If two dual op amps are used, I would think that any unused units might be used to create a sign wave, too, in the oscillator section.

Non-creep question:
Would there be any great advantage to using ten-turn pots?

I still need lots of help with proofing, on both models. Please, please, oh please and thank you. <grin>

 

Offline t1d

  • Super Contributor
  • ***
  • Posts: 1212
  • Country: us
Re: Dynamic Electronic Load Project
« Reply #99 on: August 26, 2018, 09:29:45 pm »
I believe that the difference I noted, between Lt1013/14 models was this...
                            LT1013CDR         Other Models
Temp Max                 70*C                    125*C
Temp Min                   0*C                     -55*C
Input V Offset          0.3mV                   0.8mV
Given that either operational temperature range is acceptable, I take it that the lesser 0.3mV input voltage offset, of the LT1013CDR, would be preferable to the typical 0.8mV, of the other models. Correct?

How important would the better input voltage offset advantage be, in a cost/benefit analysis? LT1013CDR/Dual @ $1.91/unit needs more PCB real estate. LT1014/Quad @ $5.02; YMMV by footprint. For me, at 5 MOSFETS, 4 x $1.91 < $2 x 5.02 and real estate doesn't matter, because 100mm x 100mm is a flat price.
« Last Edit: August 26, 2018, 09:55:19 pm by t1d »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf