Thanks again,

I had the potentiometers in mind specially to control the gain and to achieve circuit stability.

I have studied the circuit you sent and I have some unclear points:
The OPA at the bottom Right is used to increase the Vsh level what I did not get is why it is provided through the resistors to the control OPA in the middle?

- why do we have the control signal connected to the resistor to the vp? (Is it a normal pull up resistor and would not this open the mosfet if the control signal is 0?)

-why do we have a voltage divider before and after the buffer OPA (control signal input and output)

I am so sorry for asking too many questions.

 

hi,

The middle OPA.
R6 and R7 set that LM324 middle OPA to a Gain of 1 for the va signal and a gain of 2 for Vcntrl, its function is to mix the Vctrl which is a high going voltage signal and va which requires a low going voltage.

So the Sum of the Vcntrl and va cancel when the two voltages are equal within the OPA.
R10 to plus Volts , supplies the +4Vg offset required to set Vg threshold to the MOS Gate.

R9 and R5 divider is so we can get a balanced signal voltages at the OPA input.

E

I have added a new image showing post and also a feedback OPA for use in the ESP32 ADC input so that you can actually monitor the Current.
Updated image.

 

Thanks again you are a great help.

I will try to digest all of this and will come back if I have further questions.
Thank you so much

 

hi,
This shows a very basic linearising circuit for the va signal.
E

 

Thank you, I really appreciate your help

 

hi Samy,
Checking thru the MOSFET datasheet and considering the temperature range over which the device is being used, the ADC feedback suggestion should be considered as essential.
Your present circuit has no feedback to the ESP32, so as the temperature increases so will the Drain current, which is not being monitored.

If you are open to alternative circuits, I have an idea that may improve the control.

E

 

Hi Eric,

in fact the shared circuit is just the part of the project that I am confused about however there will be a temp feedback for fan control through an IRFZ44N and current feedback through 16 bit ADC for measuring the running current and change control signal accordingly. also a voltmeter through a voltage divider and another port in the ADC.

Through SW (I already built the interface) I should be able to control CC or CV, CR or CP this is the easy part for me.
A contact relay (for high current) will be working as the OCP, OVP, OPP (This will be added later when I have a stable device)

having said that I am totally open for any suggestion and modification I built only the attached pictures the rest of the circuit is still under preparation so I can change everything if it will be better, more robust and/or easier to build.

Thank you so much for your help

 

hi Samy,
Looking at our original discussion circuits, you did not shown any form of feedback control for monitoring the actual MOSFET load current.
My concern is that is that as the temperature increase so does the Vg versus Id Gain in the MOS, which could cause thermal run away.

Your latest post explains that you have included current monitoring feedback..so you have covered that point.

Nice looking piece of kit.

E

 

Thanks again,
I am still having a lot to cover and I will share with you my final circuite to take your advice when it is done sure before that will come back with more questions.
Please if you have any other points that I should cover or think about it please share it I am sure it will help me.
Thank you so much

 

I've designed and am building a 1500W e-load (30v, 50A continuous for >10hr, 100A/3kW peak pulsed 0.1sec/sec) using a similar approach (and an ESP32! Great minds/fools etc... )

IMHO the IRFP260 is the wrong device to use. Its not characterised for stable linear operation at DC. The IXTK90N25L2 is a much better device as its designed and characterised for stable linear operation at DC and 75degC case temperature. Unless I missed it you don't say what your load/current capability is only voltage (150v). You may need multiple IXTK90N25L2 devices as you may still need to derate them to stay inside the SOA and appropriate heatsinking will be critical. Although the IXYS/LittelFuse device is better suited they recommend you still add some source resistance to protect against thermal runaway due to imbalance between devices - indeed their app note suggests separate parallel control paths and current sensing for each device at higher currents.

I'm using 8 x IXTX110N20L2 (the PLUS247 package is, I'm told, slightly better thermally than the TO264**) at 187.5W per MOSFET in 2 groups of 4 on a pair of force cooled 0.08degC/W heatsinks to run at a case temperature of 75degC and a junction of 137degC giving some overhead, that's the most cost effective solution with real-world air-cooled heatsinks. I have a Vishay 10mOhm 7W current sense resistor on each MOSFET giving 125mV at 12.5A (for 100A peak operation). The BOM cost of the multiple current sense/loop control is tiny compared to the heatsink/MOSFET cost. I did look at water-cooling but couldn't come up with an easy way to manufacture something with what's available to me.

**and smaller which makes it easier to fit multiple devices on the same heatsink. A PLUS264 package would be the best with 0.066degC/W, twic as efficient, but they don't offer the L2 devices in that package

 

Hi Thanks for your reply,
great that someone else has the same thinking, In fact I started with the interface and menus handling by using one rotary encoder which gave me some hard time however I manage to find a way, let us say I am dont with about 70% of the interface.

I am planning to share it publicly when its done and tested however I am willing to join efforts together to create two similar devices at the same time - I am good in programming however not so good in electronics. anyway we can think about it.

answering your questions: my plan is to reach 50 amp and 150v while maintain max power of 300w - 500w
the IRFP260 is just for testing purposes not to burn out an expensive IXTK90N25L2 during my trials.
I ordered too IXTK200N10L2 yesterday to achieve higher current rates or at least work safely at the 50 amp range with the minimum no of mosfets.

As for the heat management I got a snowman T6 cpu heatsink in case Iwill have one or two mosfets, If I will have to use several mosfets I have also 4 whole copper heatsink from HP servers each can serve for 2 mosfets and will add to them pwm controlled fans.

As for current sensing I am having a 50 amp 75 mv (1.5 m ohm) shunt resistors that I am planning to use it on the output of all mosfets combined where I will take the control signal to the ESP32 to manage the gate voltage.

From my point of view I think the L2 option is having the priority over the package heat dissipation as I understand it is designed to perform better in the linear range that means higher power tolerance and lower heat generation. ( maybe I am wrong) so I picked the L2 option. I got 5 of IXTK200N10L2 and 5 of IXTK90N25L2.

I saw a video for a guy tested two IXTK90N25L2 for up to 100 amps (50 each) and they stand the test very well

 

Hi Thanks for your reply,
great that someone else has the same thinking, In fact I started with the interface and menus handling by using one rotary encoder which gave me some hard time however I manage to find a way, let us say I am dont with about 70% of the interface.

I am planning to share it publicly when its done and tested however I am willing to join efforts together to create two similar devices at the same time - I am good in programming however not so good in electronics. anyway we can think about it.

answering your questions: my plan is to reach 50 amp and 150v while maintain max power of 300w - 500w
the IRFP260 is just for testing purposes not to burn out an expensive IXTK90N25L2 during my trials.
I ordered too IXTK200N10L2 yesterday to achieve higher current rates or at least work safely at the 50 amp range with the minimum no of mosfets.

As for the heat management I got a snowman T6 cpu heatsink in case Iwill have one or two mosfets, If I will have to use several mosfets I have also 4 whole copper heatsink from HP servers each can serve for 2 mosfets and will add to them pwm controlled fans.

As for current sensing I am having a 50 amp 75 mv (1.5 m ohm) shunt resistors that I am planning to use it on the output of all mosfets combined where I will take the control signal to the ESP32 to manage the gate voltage.

From my point of view I think the L2 option is having the priority over the package heat dissipation as I understand it is designed to perform better in the linear range that means higher power tolerance and lower heat generation. ( maybe I am wrong) so I picked the L2 option. I got 5 of IXTK200N10L2 and 5 of IXTK90N25L2.

I saw a video for a guy tested two IXTK90N25L2 for up to 100 amps (50 each) and they stand the test very well

The problem with CPU coolers is they actually don't need to do much cooling. The max output of a current i9 CPU is 135/140W over a 2 x 2cm contact area. That's one TO-264 or PLUS247 package.

A typical modern CPU air cooler has a thermal resistance of around 0.4degC/W. The TO-264/Plus247 package is 0.13degC/W and thermal paste adds another 0.25degC/W so assuming a max junction temperature of 130degC (to allow for various approximations) the max power dissipation is (130 - 25)/(0.4 + 0.12 + 0.25) = 135W, which gives a case temperature of 79degC, close enough to 75degC for the SOA chart to be meaningful. You might be able to push it to 150W, but the heatsink doesn't have enough thermal capacity to move more than that due to the small contact area.

So a single L2 MOSFET on a CPU cooler may do 50A but only to 3v! Or it'll do 150v @ 1A.

3 devices and coolers and control loops will get close to your top spec - of 500W - at 450W, which is 50A and 9v or 150v and 3A. Think of it as 3 x 150W loads in parallel - in fact many commercial e-loads are built modularly in that way. Indeed many of the 150W loads on eBay/Amazon/AliExpress are a single device on a modified CPU cooler.

The YouTube example you cite is Kerry Wong's 400W e-load on which I based my original thoughts. His approach to power calculation is not correct but he took an emprical approach by actually trying them out and yes he had 400W operation at 40A and pushed it breifly to 100A on 2 devices (both at 10v). But look at his absolutely massive heatsink with 3 or 4 120mm fans. That heatsink is around 0.05degC/W I'd estimate. Even then at 100A his case temperature was 95degC which means his junction temperature was probably over 150degC but the devices are robust enough to stand that for 5 secs. You don't want to go anywhere near that for reliability and you won't even get close with a CPU Cooler or a server heatsink (for the older Xeon processors which were lower TDP than the current i9s).

 

I See what you mean, So what is the best solution for the heat issue?

I thought about water cooling and another one built a modular e-load using water cooling solution but when I searched I found that the efficiency of the forced air heat sinks are higher than the water cooling.

The only Idea I have now is to finish the circuit and build it then replace the fans with stronger ones (It has two already) and see by experiment how much heat can be handled with it (anyway I am monitoring the temperature of the junction and case) to stop when the 75c is reached at the junction and that will be my absolute max.

please let me know if you have better idea.

by the way this is the heat sink that I have
https://www.ebay.com/itm/SNOWMAN-M-...Double-Fans-12cm-Cooling-Fa-R4F9/254624269970

 

The calculation for thermal resistance of the heatsink is nicely summed up here but even they recommend testing as its somewhat empirical as placement of devices on the heatsink, orientation and even the length of the pins have significant effects (the drain pin to PCB foil is a big contribution to cooling - short pin and 2oz copper is preferable to a long wire).

The heatsink I'm looking at is this one which is a reasonable balance of price v # of devices (8 devices, 1500w continuous load).

Adding more fans won't help with your heatsink, there's simply not enough metal to conduct the heat away from the contact area.

 

Thanks for sharing those information,

So how do you think the cheap chinese versions are getting away with their cheap cpu heatsink and single fan on a single mosfet (IRFP250N) for up to 180W.

Rigol DL3021 (200W) is using also IRFP250N but around 10 of them (I was thinking this is because they want to ensure smooth running for extended period of time)?

Anyway using multiple parallel paths is not very complex and easily done if you successfully manage to get a single path working correctly and accurately (This is what I am aiming at right now) after that some high power resistors and add more paths should be it. am I right?

 

Given IRFP250N is rated at 214W with a case temp of 25degC and derated at 1.4W/degC above that tells you something about the likely reliability of those loads. The original design for them about 10y back was 100W. Since then there's been a gradual arms race, but 180W might be possible with a better device.

The Rigol solution is engineered for reliability and 'low'cost but ultimately it's an old design as the IRFP250N is the wrong sort of MOSFET to use and 200W is easily doable with 2 devices now (or even just one on a suitable heatsink)

 

Thank you