150w DC Load project (low cost, med difficulty)

[Cliff Matthews]

Our good friend at Scullcom Hobby Electronics is back with final refinements to his Nano-controlled DC-load. For something that looks like it could be biult for under $100 its got some nice software options. The PCB looks reasonable but he elected to limit the fuse to 5 amps (not sure about software yet..). I'm curious, for testing batteries etc.. <24v 10amp, can the fuse and dual diode be bypassed without effect on calibration?

[Cliff Matthews]

[AllTheGearNoIdea]

That looks about a hundred times better than the one I built. This guy has some really interesting projects on his channel and everything really well explained. Definitely 10 out of 10 and gets my vote. 

[Dubbie]

It’s enjoyable seeing the design process blow by blow, and also seeing the improvements he makes throughout the versions.

I can’t even imagine how excited I would have been to have resources like this when I was a kid. It’s better than university in its own limited scope.

[Inverted18650]

I built V7 and can tell you without a doubt, you can by-pass both the fuse and diodes. But why? The calibration of the unit is done by adjusting the voltage at the base of R6 and dialing in the "sense" voltage on the input

the cost was about $145 for 90W unit. Its very simple to recreate as he has literally done all the work.

I have since torn the unit back down address the issues I found. I am working to push the unit up to 500W. I have had two successful runs at 200W for three hours each (using standalone cooling) but need to finish the PCB layout. I moved the sense resistor, temp sensor, and changed out the fan drive tranny/diode combo.



I still have a couple of his original V7  PCBs, that I had made, if you want one. He says I should not have had them made and waited for his double-sided layout but wasn't upset, just didn't like the links in the layout and wanted only "DIY" types to make their own.. Since he didn't release anything but PDF's this time, I think he wants to "guard" it a bit more this go around. He is now more interested in selling them which is cool, he has earned it, it is a cool project.

[Cliff Matthews]

I built V7 and can tell you without a doubt, you can by-pass both the fuse and diodes. But why? The calibration of the unit is done by adjusting the voltage at the base of R6 and dialing in the "sense" voltage on the input

Thanks. BTW, where did you get your Mosfet's?? It seems NXP trashed most of their TO-220 TrenchMOS stuff and won't give replacements..
https://www.nexperia.com/products/mosfets/automotive-mosfets/BUK956R1-100E.html#discontinuation-information

I am Parametrically confused when trying to find equivalents on Digikey, but I'd like near triple Vds while getting similar gate characteristics..
Is this even possible (yet) in logic-level parts??

Edit: Wow, nice build and nice channel ! Have you thought about posting to get listed on EEVblog's big YouTubers list? 
https://www.eevblog.com/forum/other-blog-specific/a/

[ArdWar]

I am Parametrically confused when trying to find equivalents on Digikey, but I'd like near triple Vds while getting similar gate characteristics..
Is this even possible (yet) in logic-level parts??

It's hard because not that many manufacturers willing to claim DC FBSOA in their datasheet, much less parametric-ized. Even then some of the claims are just too optimistic.

The usual suspects for good DC FBSOA (IXYS) doesn't make logic level devices, but their devices maybe still usable if you only need <5A rating.  Closest direct drop-in replacement that's on stock I can find maybe Vishay SUP85N10. Maybe someone have alternative?

[Inverted18650]

Thank you for the compliments mate, greatly appreciated but Scullcom Hobby gets all the credit for the design on this project. I used these fets:

https://www.infineon.com/dgdl/irlb4030pbf.pdf?fileId=5546d462533600a4015356604640258d

Had no issues until I accidentally shorted test battery to the board while it was off and blew a fet. I ramped it up 20A's. But I have modified my unit a lot & right now its completely apart and I'm playing with putting the Load section on a separate PCB. If the dedicated INXY werent so expensive I would love to try them but I have also just swapped the sense resistor and used other logic-level fets with a bit higher RDS(ON).

 

[ArdWar]

R_DS(on) doesn't matter anyway for this application, unless you want to make a high current, very low voltage load.



Hmm, this FBSOA is actually believable. Maybe even already accounting for thermal instability region.

[Cliff Matthews]

I have 4 of these TO-264 bad boy's, they're not logic level, but could his circuit be tweaked to double Vgs without mucking with the DAC?

@Inverted18650 - I subscribed to your channel. Looking forward to Part 2 and maybe seeing how you increase the capacity..

[blackdog]

Hi,

I see several problems in his design, the AD8630 is a nice opamp, but like almost all other chopper or auto zero opamps, it is not suitable for driving capacitive loads...
Take the output of U7D, 100 Ohm and then 100nF to mass, you must be kidding ... 
Look in the datasheet on page 10 and figure 25, above 100pF it is inherent instable, 500pF load 90% overshoot TADA!!!

And also the output section 2K2 resistor for a gate input who is 14nF 
Probably the circuit is stable for DC use, say for testing batteries, but if you have a load that is not constand, the entire output section can not keep up with it.
In the Netherlands we say: as slow as a snail on a tar barrel.
The AC otput impedance of this current source is low, even in the low frequencies, and likely to get easily generated, during error conditions.

Kind regards,
Blackdog

[ArdWar]

Can anyone points me to the schematics, please?

[Cliff Matthews]

Hi,

I see several problems in his design, the AD8630 is a nice opamp, but like almost all other chopper or auto zero opamps, it is not suitable for driving capacitive loads...
Take the output of U7D, 100 Ohm and then 100nF to mass, you must be kidding ... 
Look in the datasheet on page 10 and figure 25, above 100pF it is inherent instable, 500pF load 90% overshoot TADA!!!

And also the output section 2K2 resistor for a gate input who is 14nF 
Probably the circuit is stable for DC use, say for testing batteries, but if you have a load that is not constand, the entire output section can not keep up with it.
In the Netherlands we say: as slow as a snail on a tar barrel.
The AC otput impedance of this current source is low, even in the low frequencies, and likely to get easily generated, during error conditions.

Kind regards,
Blackdog

Just measured my 4 IXFK90n20's at 24.5nF.. I think I'd need a costly PA to drive each one!! There's no free ticket is there?

[Cliff Matthews]

Can anyone points me to the schematics, please?

No problem..

[Mechatrommer]

you must be kidding ... 

iirc when following his venture with this dc load project sometime ago, with so many days, weeks, months of design iterations and revisions, he never got into assessing his load performance for stability, transient, pulse etc. the best i can see is he's probing the gate driver pulse. he's too busy drawing schematics, playing with keypads and firmware. not to screw him up but his load could work, but may as well it could not.

[blackdog]

Hi,

The way I test the output section of a power source, for example, is this:
I take a power supply that can provide sufficient voltage / power.

I then connect the output section directly via twisted connection cables to the power supply and measure then over the 0.1 Ohm resistor R17 how the signal looks.

Which signal? connect a function generator to pin-5 of U7B.
Set the DC shift on the function generator so that the desired test current is running.
Now also set the AC signal to the function generator, usually I start with 500Hz and 10% duty cycle block signal.

Even now I measure over R17 and the signal should not have any overshoot!
If the signal has overshoot, the compensation is not correct yet.
And now the difficult part. Now place a resistor in series with the + connection of the power supply.

Test with 1, 5, 10, 22, 100 Ohm, think about the current, at 100 Ohm you usually can not run 2 Ampere.
Make sure you stay in the linear area of the dummy load.

The higher the chosen resistance, the lower the bandwidth, especially with the proposed power mosfets.
This is due to the large feedback capacities of the power fets.
It is easier to get a larger bandwidth with fast audio transistors in the power section.

Designing a good dummy load is not easy and at least 75% of your time goes to getting the output section properly stable and bullit proof 

Down hear some links of some older design of mine, there are a lof of measurements, use Google translate becaue i'am Dutch, sorry fore that 

https://www.circuitsonline.net/forum/view/121338/4#highlight=load+dummy

And one with Audio power transistors
https://www.circuitsonline.net/forum/view/106770/1#highlight=load+dummy

Maybe my designs give you guys some inspiration...

Kind regards,
Blackdog

[ArdWar]

My first impression is that there's a lot of unnecessary part. I don't think U7A and U7C is needed.

Also, passively balancing MOSFET in linear operation is kind of risky. I dunno how much his balancing transistor can equalize, but I'd prefer individual  drive 

[Inverted18650]

R_DS(on) doesn't matter anyway for this application, unless you want to make a high current, very low voltage load.

I use it almost exclusively to test my DIY li-ion battery banks and like to run them from 4.2V down to 3.0V (2.75 depending on the manufacturer of the cells) @ 100 A's. I've spent many hours testing the unit and trying to find ways to tailor it to my application.

edit: I want to get it up to 100A's*

[Inverted18650]

Maybe my designs give you guys some inspiration...

For sure brotha, thanks for the link.

[Mechatrommer]

using 4X 120A mosfet in a 5A dummy load is kinda nuts imho, its kinda mosfet-toss-phobic design. there will be only 1.25A max running through each mosfet and the 0.1ohm power resistor. that is 0.125V across each of the R22-25, i wish i could see those nodes probed to be certain they are not oscillating during step respond and steady state.. build a 10X or 100X amplifier if DSO is noisy.

I built V7 and can tell you without a doubt, you can by-pass both the fuse and diodes.

until you plug the voltage source backward.

[ogden]

using 4X 120A mosfet in a 5A dummy load is kinda nuts imho, its kinda mosfet-toss-phobic design.

It looks so, but it is not. When you use switching mosfet in linear mode, nasty things (hot spots) can happen on it's die. If you pick smallish mosfet, most likely it will make just quiet puff. What's worse - they die short!

[edit] Reading: http://www.ixyspower.com/images/tech-papers/Article_Linear_Power_MOSFETs_2007.pdf

[alex-sh]

Hi,

I see several problems in his design, the AD8630 is a nice opamp, but like almost all other chopper or auto zero opamps, it is not suitable for driving capacitive loads...
Take the output of U7D, 100 Ohm and then 100nF to mass, you must be kidding ... 
Look in the datasheet on page 10 and figure 25, above 100pF it is inherent instable, 500pF load 90% overshoot TADA!!!

And also the output section 2K2 resistor for a gate input who is 14nF 
Probably the circuit is stable for DC use, say for testing batteries, but if you have a load that is not constand, the entire output section can not keep up with it.
In the Netherlands we say: as slow as a snail on a tar barrel.
The AC otput impedance of this current source is low, even in the low frequencies, and likely to get easily generated, during error conditions.

Kind regards,
Blackdog

Just measured my 4 IXFK90n20's at 24.5nF.. I think I'd need a costly PA to drive each one!! There's no free ticket is there?


Should not be that high. IXFK90N20Q-ND is rated 6800pF @ 25V
Did you manage to complete your DC load?