Go Down

Topic: Making a high-amperage, high voltage DC to DC solid-state relay for Arduino  (Read 672 times) previous topic - next topic

JohnDeere630

Jan 23, 2020, 08:30 pm Last Edit: Jan 28, 2020, 09:23 pm by JohnDeere630 Reason: Changed topic header for clarity.
I'm in the process of constructing a DC to DC relay, controlled by an Arduino Mega, that can handle up to 90 amps DC @ up to 150 volts. I didn't have a clue when I made this first post, (now deleted)  but got a lot of very helpful advice in response. I've kept the responses, and feel I have a workable solution, with it's diagram below.

JohnDeere630

#1
Jan 25, 2020, 10:10 pm Last Edit: Jan 26, 2020, 10:15 pm by JohnDeere630
Can someone more knowledgeable than I take a look at this diagram and tell me if it will work? I have attached the datasheets for the VOM1271T Photovoltaic MOSFET driver and the STP75NF20  MOSFET. Thank you!

(EDIT) Changed R1 to 150 ohms, and added a 12V zener diode.


MarkT

Yes I think that will work - however the VOM1271T is only able to provide about 8V rather than 10V, which is a bit marginal.  Make sure you give it a healthy 20 or 25mA of drive current, so 150 ohm resistor rather than 220.

You can also add a 12V zener across gate/source of the MOSFET to protect it from dV/dt induced gate voltage spikes as its in a fairly high voltage circuit.  That MOSFET has a +/-20V max gate voltage spec which is more sensitive than the more standard +/-30V parts.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

JohnDeere630

Thank you Mark. I will do that, and also look for an optocoupler that will put out 10 volts...

JohnDeere630

Mark, I hope I have the correct zener diode, and I think that's how it goes, but I have never used a zener diode before.

Another question, a 150 ohm resistor would allow 33mA, do you think that would be pushing the output pin of the arduino too hard? Should I use the arduino signal to turn on a small transistor to supply the optocoupler with 30-35 mA? I saw it has a maximum of 50 mA...

Also, I can't seem to find a MOSFET-output optocoupler that puts out more than 8 volts; is there a better way to drive the MOSFET? It lists it's gate threshold voltage as 3 volts, I just noticed.

Thank you for your help!

MarkT

8 volts will switch it, but it won't be upto datasheet specifications.  8V would be plenty for a logic-level MOSFET of course.

The ATmega based Arduinos have output pin drivers that have about 40 ohms, so adding an external 150 ohms gives 190 ohms - drawing a max of 26mA even if the opto coupler had no forward voltage drop, so no problem there - in fact 120 ohms might be a better value to use.

People often expect the output impedance of a logic pin to be zero, but they are basically MOSFETs around the periphery of the chip that drive the pins, and MOSFETs have on-resistance of course, and small ones have more on-resistance.   40 ohms or so is actually a very low on-resistance for a logic chip - that would normally be called a line-driver output.  Some of the other Arduino versions use microcontrollers capable of only a few mA per pin, which would be a problem driving opto couplers.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

ballscrewbob

@JohnDeere630

TOPIC MERGED.

Could you take a few moments to Learn How To Use The Forum.
Other general help and troubleshooting advice can be found here.
It will help you get the best out of the forum.
It may not be the answer you were looking for but its the one I am giving based on either experience, educated guess, google or the fact that you gave nothing to go with in the first place so I used my wonky crystal ball.

MarkT

The IXTX90N25L2 cannot handle anything like 70A, its got a whole 36 milliohms of on-resistance.

For 70A you need either a big MOSFET or a parallel array.

You will no doubt say but the IXTX90N25L2 datasheet says 90A.

However that's entirely infeasible to use in practice except as a pulse rating, that package cannot sensibly dissipate the heat or anything like it at those currents.  Unless you are planning to use liquid cooling don't believe a power MOSFET's nomimal current rating, other than as a pulse rating.  Some manufacturers even quote currents that will fuse the bond-wires, together with a little footnote saying "current spec applies to bare die".  International Rectifier hang your heads in shame I say!


Lets assume you want at most 50W dissipated in your MOSFET at 70A.  I-squared-R then says the maximum
on-resistance you can afford to have is 10 milliohms.

However that's almost impossible to find at high voltage, the best I found was https://docs.rs-online.com/f05f/0900766b814b7fa1.pdf which is 14.5 milliohm, so will dissipate about 75 watts requiring a massive heatsink.

A good choice for high power is a screwterm style packaged MOSFET such as: https://docs.rs-online.com/365d/0900766b80a3b4bf.pdf

This will run cooler on a big heatsink and its leads really can handle 70A without stress, unlike the TO264 packaged devices.  Its slightly higher at 18 milliohms though.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

JohnDeere630

@JohnDeere630

TOPIC MERGED.

Could you take a few moments to Learn How To Use The Forum.
Other general help and troubleshooting advice can be found here.
It will help you get the best out of the forum.
Thank you Bob. I thought about that, but felt that since I had changed direction, that would have made it confusing. I wish the forum had a delete post button....and you certainly have my blessing to just delete the first post in this thread.


The IXTX90N25L2 cannot handle anything like 70A, its got a whole 36 milliohms of on-resistance.

For 70A you need either a big MOSFET or a parallel array.

You will no doubt say but the IXTX90N25L2 datasheet says 90A.

However that's entirely infeasible to use in practice except as a pulse rating, that package cannot sensibly dissipate the heat or anything like it at those currents.  Unless you are planning to use liquid cooling don't believe a power MOSFET's nomimal current rating, other than as a pulse rating.  Some manufacturers even quote currents that will fuse the bond-wires, together with a little footnote saying "current spec applies to bare die".  International Rectifier hang your heads in shame I say!


Lets assume you want at most 50W dissipated in your MOSFET at 70A.  I-squared-R then says the maximum
on-resistance you can afford to have is 10 milliohms.

However that's almost impossible to find at high voltage, the best I found was https://docs.rs-online.com/f05f/0900766b814b7fa1.pdf which is 14.5 milliohm, so will dissipate about 75 watts requiring a massive heatsink.

A good choice for high power is a screwterm style packaged MOSFET such as: https://docs.rs-online.com/365d/0900766b80a3b4bf.pdf

This will run cooler on a big heatsink and its leads really can handle 70A without stress, unlike the TO264 packaged devices.  Its slightly higher at 18 milliohms though.
Thank you Mark, I was strongly concerned about that; those legs look mighty small to carry that much current. I cut apart a Chinese relay, and can't for the life of me believe that relay can handle it's claimed 90 amps. 15, maybe.

Anyway, I like the IXFN 140N20P part a lot. I didn't know that existed, as it didn't show up in my google searches, being 115 amps.

Which, of course brings up another question, with a threshold of 5 volts, the VOM part is going to do an even poorer job of driving it, what do you recommend using to control that mosfet with an arduino?

I also didn't realize the arduino output pin had 40 ohms of impedance; I'll file that away for future reference, for sure, as well as the P=I*I*R equation.

I figure that at 70 amps, that mosfet will dissipate 88 watts, although 70 amps would be rare,40-50 is more realistic, but I'd size the heatsink for the largest load. I'll look for a calculator to see how much heatsink that would need.

Thanks again! I know a lot more than I did when I made the first post in this thread.

JohnDeere630

#9
Jan 26, 2020, 10:13 pm Last Edit: Jan 28, 2020, 09:33 pm by JohnDeere630 Reason: Changed R1 to 330 ohms
This is a proposed circuit using the FDA217S driver and the XFN 140N20P mosfet:

(Edit) Changed R1 to 330 ohms




WattsThat

I would raise R1 up to 330 ohms, the parts spec'ed don't need the higher current.

You'll have about 100 watts to dissipate, maximum ambient will determine the heatsink required along with the RthJC which is 0.22 C/W which is quite low - but that is a big hunk of metal the mosfet die is attached to!

Datasheets:

https://m.littelfuse.com/~/media/electronics/datasheets/discrete_mosfets/littelfuse_discrete_mosfets_n-channel_hiperfets_ixfn140n20p_datasheet.pdf.pdf

http://www.ixysic.com/home/pdfs.nsf/www/FDA217.pdf/$file/FDA217.pdf

Vacuum tube guy in a solid state world

MarkT

Anyway, I like the IXFN 140N20P part a lot. I didn't know that existed, as it didn't show up in my google searches, being 115 amps.
The way to find electronic parts is using parametric search at one of the electronics suppliers websites, places like Mouser, Farnell, Digikey, etc.  You can search simultaneously on many different parameters including whether parts are going to be supported for the future, and most importantly price and availability.
[ I will NOT respond to personal messages, I WILL delete them, use the forum please ]

JohnDeere630

The way to find electronic parts is using parametric search at one of the electronics suppliers websites, places like Mouser, Farnell, Digikey, etc.  You can search simultaneously on many different parameters including whether parts are going to be supported for the future, and most importantly price and availability.
Thank you Mark. I use that search tool on Mouser all the time when I need resistors, small transistors, capacitors, etc. when I know or have some idea what I need. I've never used a mosfet before, so other than the amps/volts, I didn't have a clue what to look for. Now, I know what Rds(on), gate threshold, and a bunch of other new stuff is and means.

Thank you and Wattsthat for your help. I will have the parts tomorrow, but it may be some time before I assemble it. When I do, and plug it in, and assuming I survive, I'll post the results.

JohnDeere630

I got it assembled well enough for some preliminary testing, and it seems to work great. I'm only passing a few amps through it, as I'm still waiting for the heatsink to arrive. At 10 amps, it doesn't even get warm. I posted a drawing of the circuit.

My analog multimeter measures about 7 volts between the gate and source, but my digital meter measures 12.2 volts. I assume the digital is more accurate, as I'm measuring about .01 ohm between the source and drain.

I will be running two different voltages across the 2 mosfets, so I was concerned about them sharing a heatsink, but there doesn't seem to be any connectivity between the backplate and the 4 terminals. The spec sheet also doesn't mention it. I know some of the smaller transistors have one of the legs connected to the tab, but apparently not these parts. I hope I'm correct.





Below shows a different indicator LED circuit, and my calculations for the R1 resistor value. I'd appreciate it if someone more knowledgeable than I could take a look and see if I have made any mistakes. Thanks!


Go Up