Hello,
Very informative analysis going on here. Thanks a lot for making it all available for us to see and learn.
I have been playing around with variations of the first time domain model in this thread, and I have a few questions from the perspective of a beginner in the subject.
I am looking into different variations of supply voltages, MOSFETS, number of MOSFET stages, and current limits. The main variables affecting the current drawn from the source are (as you of course know): the reference voltage (vref), the resistance of the current sense resistors, and the number of MOSFET stages. I understand the relationships between these, how they affect the current draw, and how they put constraints on the supply rails.
My questions are all related to which of these things can be changed without significantly affecting the stability analysis (I am still not understanding the subject well enough to trust my own analysis).
How sensitive is the system with regards to changes in:
1. Number of MOSFET stages. My uneducated guess is that this is not an important factor since they are all individual control loops.
2. Current sense resistor value. Again my guess here is that I can vary this within reasonable limits to adjust the current draw.
3. The reference voltage. My guess here is that this can also be changed somewhat without seriously affecting stability. For example, can a 1.25V reference be used instead of 2.5V? This would matter if a -5 rail is available instead of a -9V rail.
4. MOSFET choice. This is of course a major factor in the stability analysis. I can't source the IRFZ20 MOSFETs for a reasonable price, so I am looking for alternatives that don't require a completely new analysis. The MOSFET model used in the initial simulation model is the IRF530. I assume that this model will work instead of the IRFZ20.
In short:
I am thinking of implementing this circuit with 4xIRF530 to spread the heat over more devices. Thus, I would need to either change the reference voltage to 1.25V, or increase the current sense resistor to 0.2? in order to keep the same 5A ranges. My hope is that this won't negatively affect the stability analysis already made.
Thanks in advance for any further advice.
2) If you examine the voltage gain of the MOSFET stage from the gate of the MOSFET to the source of the MOSFET as shown in this LTspice model:
[pic]
The results are:
[pic]
So the maximum gain is 0dB if the Source resistance is high and the transconductance of the MOSFET is high. The gain is reduced if the transconductance is lowered or the source resistance is lowered.
...
When the control loops are designed there should be a least 6dB of gain margin. 6dB is factor of 2. so this allows for some significant changes while maintain stability.
Do not forget the RC damping network. This is critical in making a good load.
Good luck with your project, please share your results.
Thanks for your response!
snip ...
So, I have a couple more questions:
1. Would it be desirable to have faster rise times, or are they already fast enough for basic load transient testing?
2. How could the circuit be modified to get faster rise times, while maintaining stability?
Hi ,
Nice and Interesting work Jay_Diddy_B. well Done.
However, I am trying to build a dynamic electronic load with the following specs:
- frequency: 0.1Hz - 10kHz
- Voltage range: 0 - 40V
- current range of 0 - 30A
Please what's your advice and could this be modified to accommodate the requirements?
Thankyou.
Aroogz.
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Hi ,
Nice and Interesting work Jay_Diddy_B. well Done.
However, I am trying to build a dynamic electronic load with the following specs:
- frequency: 0.1Hz - 10kHz
- Voltage range: 0 - 40V
- current range of 0 - 30A
Please what's your advice and could this be modified to accommodate the requirements?
Thankyou.
Aroogz.
Thank you for the kind words.
You indicate that you need 0-40V and 0-30A, do you need 1200W?
Because the MOSFETs are being used in the Linear region, 40W per MOSFET is a sensible target. The HP 6060A/B uses 8 MOSFETs for 300W.
Each MOSFET needs it own op-amp to control the current.Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
This is great advice!!
My original circuits has two MOSFETs, but it can be extended to many.
The oscillator in the circuit shown is set to run at 330Hz. This is a good frequency for testing most power supplies. There is no reason why it can't be set higher.
To get high bandwidth in the load you need to use small MOSFETs and lots of them.
Good luck !!
Regards,
Jay_Diddy_B
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Thanks very much for your reply.
The 10kHz is a sine wave signal (of about 250mVp-p) which will ride ontop of a set DC current of the load.
Regards,
Aroogz.
The goal is to use this to perform a frequency sweep (0.1Hz - 10KHz) at set DC load point.
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Thanks very much for your reply.
The 10kHz is a sine wave signal (of about 250mVp-p) which will ride ontop of a set DC current of the load.
Regards,
Aroogz.
The goal is to use this to perform a frequency sweep (0.1Hz - 10KHz) at set DC load point.
250mVpk-pk.. this is a DC load - that doesn't make sense as an 'output' of the load.
Do you mean that you will adjust the load till you get 250mVpk-pk ripple?
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Thanks very much for your reply.
The 10kHz is a sine wave signal (of about 250mVp-p) which will ride ontop of a set DC current of the load.
Regards,
Aroogz.
The goal is to use this to perform a frequency sweep (0.1Hz - 10KHz) at set DC load point.
250mVpk-pk.. this is a DC load - that doesn't make sense as an 'output' of the load.
Do you mean that you will adjust the load till you get 250mVpk-pk ripple?
Quite ambitious.
Probably start with building a 1A 40v load and getting that working at 10KHz. Then build up from there.
Also, what is your 10KHz - a squarewave? sine? will make quite a difference to your acceptance criteria.
Thanks very much for your reply.
The 10kHz is a sine wave signal (of about 250mVp-p) which will ride ontop of a set DC current of the load.
Regards,
Aroogz.
The goal is to use this to perform a frequency sweep (0.1Hz - 10KHz) at set DC load point.
250mVpk-pk.. this is a DC load - that doesn't make sense as an 'output' of the load.
Do you mean that you will adjust the load till you get 250mVpk-pk ripple?
Hi,
first I want to thank Jay_Diddy_B for your effort on this project, Im looking into building a 2nd load since my first DIY one melted its plastic case...
Im considering ways to get away with a single supply. I see why you did this as a non-inverting amp in order to easily add up the voltages, but it makes you to use symmetric supply.
Could buffering the oscillator help with this and keep it single supply ?
Snip ..
Thank you very much! I will build it on the weekend. Is LM324 fine for this? I also have TL082/072.
Thanks again!
Hi, JDB, Thanks for referring me to this load project. Your work, on this, is wonderful, impressive and gracious.
You are helping me, on another thread, to troubleshoot a load. But, its design is not your favorite. So, I am considering your design, to replace it, once I have used the present fault, as a learning opportunity.
Firstly, I looked up the difference between a Constant Current and a Dynamic load. Okay, I get it... The dynamic load allows the current draw to be varied (at a known frequency), so that the characteristics of the supply can be observed.
Here is my first Noob question. Can this circuit be modified, to turn off the oscillation? In other words, have a Constant Current/Dynamic Current option? If it is not too much to ask, what would be that design? Is it as easy as adding a switch, to by pass the oscillator portion, of the circuit?
Thanks!
There are two controls, one sets the constant level, the other sets the pulsed level. To get constant current, you simply turn the pulsed control fully counter clockwise:
Using a pulsed load is used to look how a power supply responds to step changes in current.