Dynamic Electronic Load Project

[bicycleguy]

t1d,
Seems like 1 IXTN could replace maybe 3 to 4 typ mosfets and associated parts, depending on what V and A you were looking for.  I think it would be cheaper and easier.

edit: Also provides a way to mount your PWB 

[t1d]

Easier, yes... But, the typical MOSFET needed is cheap... About $1/ea... If I had the money, I would surely do as you suggest. I have even been trying to figure out how to do it, on a retirement income budget. I haven't figured that out, yet.<g> There are some less expensive models of this type... $25/ea. So, do some looking...

[Kleinstein]

The AD8032ANZ op amps and MTP3055VL are both poor choices:
the OP is very fast (which only makes life more difficult and is not needed) but has a rather large offset drift and like high LF noise. What is needed is more like a precision single supply OP with something like 1 MHz GBW, likely BJT based. So the LT1013 is a little on the slow side and still rather noisy and maybe on the low speed edge.  If there is a negative supply available it would be something like an OP27 or ADA4075 that would work well. A good single supply choice could be the OP113 - though a little expensive.  Depending on the frequency range one is interested in, one might also consider an AZ OP like AD8551 or MCP6V27. A low cost candidate might be TLC272 or TLV171, maybe even LM358.

The MTP3055 might have trouble with reliable operation due to SOA limitations  .
A better choice would be something like IRFP250. It is more like looking for 200 V - 800 V MOSFETs, even if only 30 V are planed to use. A lower number of larger case FETs also reduces the number of OPs needed so that better quality ones could be used.

[t1d]

Great information, Kleinstein. I will investigate these parts.

[t1d]

Hi, Kleinstein,

I have been giving your suggestions consideration. They have good merit. I think it would be a good idea, for you to start a separate thread discussing
- the op amp selection factors
- the MOSFET selection factors
- and, what changes are needed (broadly,) in the circuit, based on these selections.
I know that I would be interested and, certainly, many others.

I looked at the IRFP250. It appears to be a good, reasonably priced candidate. What do you think of the Microchip MTP6002 op amp, as a match, for it? It might be best to answer, in the new thread, to keep from highjacking Jay...

[t1d]

A circuit to add a Function Generator Auxiliary Input to Jay's design can be found here, at Post #42...

https://www.eevblog.com/forum/projects/hacking-a-dynamic-electronic-load-circuit-to-use-an-external-function-generator/new/#new

[t1d]

Here is another option, that I made, using TLV171 op amps and IRFP250 MOSFETS. The complete project release can be found here:
https://www.eevblog.com/forum/projects/electronic-load-project-ltv171-irfp250-with-kicad-files/msg1840739/#msg1840739

[jrsikken]

Hi there,
I have designed an Arduino compatible electronic load and sold it to a hunderd people in the past year. You may want to look at my latest deisgn.

https://github.com/jrsikken/ElectronicLoadR3

It uses the super robust BTS133 in TO-220 package. It has over voltage, over current, over power, ESD protection and thermal protection. It will save a lot of mosfet replacements. And a thermal safety control circuit.

In my circuit I had oscillation and so I had tuned the constant current circuit (opamp, mosfet, feedback circuit) to prevent oscillation but have fastest response. Now the current settles in 20us. Now it's so fast it supports pulses loads. For example the 550us wide 2A pulses from GSM can be simulated.

My design uses a DAC, the MCP4725 to generate an analog input for the constant current circuit. My embedded code can change DAC output voltage at max 5kHz. On this forum post http://www.stm32duino.com/viewtopic.php?t=1048 there is a guy who generated a 3kHz sine wave, with 60kHz sample rate. Basically it means you can program any waveshape from he MCU. 

[GigaJoe]

jrsikken,
I'm not certain C11 cap to load opamp are good idea, in addition to mosfet capacitance ...   and C9 ... ,  seems all to fight for oscillation suppression

[Jay_Diddy_B]

jrsikken,
I'm not certain C11 cap to load opamp are good idea, in addition to mosfet capacitance ...   and C9 ... ,  seems all to fight for oscillation suppression

I have had a look at the load designed by jrsikken. This is the circuit that GigaJoe is talking about:



This is fairly conventional feedback around the op-amp IC1A. The note is interesting about the tuning.

The BTS133 Smart Sipmos device has a lot of extra 'stuff' in it:



It has over current and over temperature protection which is good, but it also has circuits to limit dv/dt which is bad in this application.

The datasheet shows the dynamic performance as:



This is really slow, 40 to 100us to turn on, 70-170us to turn off, compared to a normal, unprotected, MOSFET.

This graph shows the transfer characteristics:



From this graph I can get the transconductance, there is a 20A change in Id for a 2V change in Vgs so the transconductance is 10.

This protected device may be good for static loads, or slowly changing loads. It is not well suited for a dynamic load.

Regards,

Jay_Diddy_B

[Mechatrommer]

This is really slow, 40 to 100us to turn on, 70-170us to turn off, compared to a normal, unprotected, MOSFET.

according to him, after tuning, the settling time is "fast" 20us (i'm not sure how he get 20us settling time from a 40us rise time device) but well, isnt that a good thing? slower settling/rise time means less overshoot, C11 probably helps in that respect as well. as OP circuit you made is also somewhere if not slower than 20us settling/rise time (not sure if you've made faster latest circuit in the middle of this thread), there is another load circuit, also reporting somewhere 20us settling time but faster ~2us rise time, but that has more overshoot than yours. i guess we only can pick 2 from those 3. anyway jrsikken sold his earlier unit (probably oscillated) to hundred of happy customers already

[jrsikken]

Hi, I quit selling my electronic load because last year sales dropped dramatically, I think because people prefer cheap chinese electronic load. For just $18 you get one with higher power rating, a display, a dial and with graphical userinterface on the pc. I tested one myself and it is pretty cool. The chinese electronic load does not support pulsed loads yet, but it won't take long. I  also quit selling because assembly and testing/calibrating took a lot of time. I think it was a wise decision.

[MarkF]

Has there been any issues with @jrsikken circuit spiking to max current when the Device Under Test (DUT) is turned on after the load power is applied? 

The basic issue:
   When the power is turned on and there is NO load attached there will be no voltage across the R_SENSE and any small voltage on the (+) opamp input (or opamp input offset voltage) will cause the output to the MOSFET gate to go to max.  Therefore when the DUT is turned on, max current will be pulled until the opamp recovers.

My question comes from this thread and how to address the issue:
https://www.eevblog.com/forum/projects/mosfet-and-opamp-load-tester-constant-current-load/msg2485161/#msg2485161

[Jay_Diddy_B]

Hi MarkF and the group,

I am more than happy to answer your question here 


I am going to illustrate the issue with a simplified schematic of the Dynamic load schematic I presented in this thread.

Model - Normal Operation

If the power source being tested is turned on at the same time or before the load is turned on the current in the load will not overshoot.





The waveform to look at is the output of the op-amp. The voltage slews to the operating point and the load current rises to the target value.


Model - Load on First






If the load is turned on first. The op-amp will be unable to increase the current and the output of the amp will saturate at the positive rail.
When the DUT voltage is applied, the current will overshoot and the op-amp will reduce the gate drive to bring the current to the target value.


Model - large time constants

The model above uses relatively fast loop. At of the load designs found on the internet, do not care about the loop response. They are not interested in measuring the step response of a power supply.
If I populate the circuit with 'slow' components, I get:






The overshoot is higher and takes longer to recover.

Commercial Load

Commercial loads generally have this feature. If I want to avoid a start up transient I use my HP load in the constant resistance mode. In this mode the current reference is obtained DUT voltage.


The model is attached.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,

If the Power supply being tested, DUT, is current limited it will take longer for the load current to come in to regulation:






Regards,
Jay_Diddy_B

[MarkF]

Thanks Jay.
My only interest in common with @jrsikken circuit and mine is the single power supply architecture.
https://www.eevblog.com/forum/beginners/a-load-off-my-mnd/msg2151265/#msg2151265

I'm looking at the power up sequence where the load is turned ON before the DUT is turned ON and the current overshoot.  I looks like both our circuits show the same issue.

I want to propose that a resistor be added between the R_SENSE high side and the positive op-amp supply rail.  This would always provide a minimal current across R_SENSE.  Therefore when the load is set a 0A, the op-amp would always try to turn OFF the MOSFET even if no load is attached.  This would prevent a turn ON overshoot.  I'm thinking a value between 10K and 100K to provide a minimal current.

   

[t1d]

My sincere apologies, Jay. I thought this was on my thread, as it looks very similar. I am embarrassed... So much for not being able to sleep and being on the web at 4 am.

[Jay_Diddy_B]

Thanks Jay.
My only interest in common with @jrsikken circuit and mine is the single power supply architecture.
https://www.eevblog.com/forum/beginners/a-load-off-my-mnd/msg2151265/#msg2151265

I'm looking at the power up sequence where the load is turned ON before the DUT is turned ON and the current overshoot.  I looks like both our circuits show the same issue.

I want to propose that a resistor be added between the R_SENSE high side and the positive op-amp supply rail.  This would always provide a minimal current across R_SENSE.  Therefore when the load is set a 0A, the op-amp would always try to turn OFF the MOSFET even if no load is attached.  This would prevent a turn ON overshoot.  I'm thinking a value between 10K and 100K to provide a minimal current.

   

Hi,

The current introduced by the 10K resistor, you have added, will not be enough to help.

Here is a concept that should work. It essentially modifies the load to be constant resistance at low input voltages. It requires a summing node.
The circuit works by amplifying the difference between the input voltage and a reference voltage and reducing the current sense point if the input voltage is low.

Model


Modelling Results




I have attached the LTspice model

Regards,

Jay_Diddy_B

[MarkF]

Thanks Jay for taking the time to look at this.
I am going to wire this up and see how it performs.
Having a constant resistance mode would be a nice to have option.

A couple questions: 
  - I assume the R9 value was chosen to lower the amount of constant resistance influence?
  - The R7:R4 ratio is suppose to be a 10:1 voltage divider?

[lordvader88]

 Ok I too shall try this as mine is dangerous on startup with a LM358 and IRF540. I haven't tried changing the op-amp, I have LT1013 too.

[MarkF]

I finally got around to working on this...
To review the issue:

Initial conditions:
  - With the DUT disconnected or powered off
  - The control voltage (V_control) set to 0V.
Resulting in:
  - The op-amp drives the MOSFET gate to 11V (max voltage)
  - The MOSFET is fully turned on
  - When DUT is connected and turned ON, maximum current is initially drawn

Solution:
  - Added R_min-set to supply a minimum current (10mA) through R_sense
    (My initial R_min-set of 10K was too high.  A value of 1.2K seems a good compromise)
  - This forces a minimum voltage across R_sense
  - This artificially raises V_control to start turning ON the MOSFET
  - The MOSFET gate voltage now goes to 0V for V_control less than approx. 10mV
Possible issues:
  - When using external voltage control, the actual desired load current will be slightly lower
than what is expected.

[MarkF]

Jay,
Maybe you don't see this issue since your control is via the (-) op-amp input.
Where my control is via the (+) input.

[Kleinstein]

Starting with the gate voltage at a high value, because the DUT can not deliver the commanded current (e.g. DUT open) is a known problem.
The main cause is that the control loop does not know about the limitation of the DUT and thus starts with a very high gate voltage it than need quite some time to recover, as the capacitor in the regulator part is charges quite high.

A solution to this is to limit the gate voltage to a low value, if the voltage at the MOSFET drain is too low (e.g. below some 100 mV or so).

A minimum solution is to limit the maximum gate voltage to a reasonable value, just high enough to deliver the maximum current. This would limit the initial current peak, though not fully prevent it.

Adding current to the shunt only adds a small offset. This may solve the problem of having an OP offset that does not allow to set the current all the way to zero. However it does not help if the initial current is set higher than zero when the DUT is connected.

[Jay_Diddy_B]

Hi group,

Let me start with a little benchmarking. I will display the test results for an HP 60501A (150W) load module in a 6050A mainframe. The load is set to the 3A range and 1A constant current mode.
The current was measured with a Tektronix TCP202A DC current probe.
The power supply used for the testing HP665A set for 5V and 2A limit.

Test 1



This is using the ON/OFF button on the power supply.

Test 2



This is 'hot-plugging' the power supply into the load.

Note: the current scaling was changed from 200mA to 500mA/div


It would be interesting to see the results for testing other popular electronic loads for example B&K, Rigol, Maynuo etc.. I don't have any of these.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

Let me do the same tests on the unit that I shared in the first post of this thread.

Link: https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/


Test 1



This is turning using the power supply on/off button.


Test 2




This is 'hot plugging'. Turning the supply on first before connecting the load.





zooming in, you can see the transient is essentially over in 22us.

You can decide if this a problem that needs fixing.

Regards,

Jay_Diddy_B