Electronic load MOSFET replacement

[diegobarriga]

Hello All,

I'm trying to do an electronic load as the one shown in the video
http://www.eevblog.com/2010/08/01/eevblog-102-diy-constant-current-dummy-load-for-power-supply-and-battery-testing/

Can I replace the MTP3055 FET with one of the following ones that I have available?
IRFP260
IRFZ44N
IRFP150N

[Kleinstein]

The shown FETs are quite a bit larger. The IRFZ44 is more like 3 times the size - not sure if it is really well suited for linear operation. It might work at low voltage (e.g. up to 20 V).
The IRFP260 should be ok - though even bigger.

I would prefer something like an IRFP240 or IRFP250, so smaller brothers of the IRFP260, but still the large TO247 case.

The IRFP150 is not that good, as it is a lower voltage device.

[Benta]

Can I replace the MTP3055 FET with one of the following ones that I have available?
IRFP260
IRFZ44N
IRFP150N

No.
And you can't even use an MTP3055.

Somewhere in the video (as an aside), it's mentioned that a logic-level MTP3055 is used, which is something else.

What you can and should do, is use a rail-to-rail opamp instead of the LM324.

THEN you can use all of your different MOSFETs (and the standard MTP3055).

And before someone flames me: 5 V is sufficient gate voltage to let the MOSFET work in this configuration (linear).

Benta.

[diegobarriga]

Can I replace the MTP3055 FET with one of the following ones that I have available?
IRFP260
IRFZ44N
IRFP150N

No.
And you can't even use an MTP3055.

Somewhere in the video (as an aside), it's mentioned that a logic-level MTP3055 is used, which is something else.

What you can and should do, is use a rail-to-rail opamp instead of the LM324.

THEN you can use all of your different MOSFETs (and the standard MTP3055).

And before someone flames me: 5 V is sufficient gate voltage to let the MOSFET work in this configuration (linear).

Benta.

What would be your recommendation? What I'm planning is to make an electronic load that has one 12V battery as power supply

[Benta]

What I'm planning is to make an electronic load that has one 12V battery as power supply

If you can supply your LM324 from 12 V instead of 5 V, it's fine and will work, also with the MOSFETs you listed. This was not clear.

[texaspyro]

Very few MOSFETs are actually suitable for use as electronic loads.  Most FETs are designed for switching applications and not for operation outside of the hard ON/OFF states.  Using them at high currents in the linear region causes them, for various reasons, to die a premature death.  I've seen several commercial electronic loads fail because the designers selected inappropriate MOSFETs.

One failure mode is development of hotspots on the die.  Power MOSFETs are generally arrays of paralleled tiny MOSFETs.  Their individual gate threshold voltage / RDSon values in the linear region are not well matched.

Finding relevant information datasheets can be fruitless.  Start by looking at the DC safe-operating-area curves.

[Kleinstein]

Finding suitable power MOSFETs for linear operation can be a challenge. In principle the SOA diagram will show you, which current and voltage combinations are allowed. However with MOSFETs quite a few are missing the DC curve. If there is a SOA curve it is usually for 25 C case temperature - higher temperature shifts the curve to lower voltage.

Even worse there are quite some data-sheets that show a DC curve that is just reflecting the Ptot value and not the true SOA curve. They just ignored the possible thermal runaway that limits the use for high power linear operation. If there is no brake point, where the permissible current is dropping faster with increasing voltage, I would be careful unless there is note that explicitly explains this.

FETs from different manufacturers could be different in such details as the SOA curve. This is especially true for Chinese versions that may not have the correct chip inside at all.

Another point is that usually the MOSFETs are not tested for SOA performance. SOA tested MOSFETs get rather expensive. Not all samples are the same, so many units will be more robust than there SOA curve suggests, but a few can also fail early.

So there is a small chance the FET could blow on a first test - so make sure there is a fuse. Also a worst case test for each unit is a good idea.

[toxuin]

I am very sorry to hijack this thread, but I have a question regarding the topic.

I am building a scrapbin ghetto dummy load from the components I have on hand and some of the beefier mosfets I could salvage are Winsemi K2698. Their datasheet actually specifies the DC SOA curve and unless I am missing something really obvious, those should be good in their linear region for dummy load. Those aren't "logic-level", but I am driving the opamp with 12v which is well above their Vgs.

I am learning a lot with this project. I've let the smoke out of couple of things, but you have to bust some eggs to make an omelet

[Seekonk]

Speaking of Ptot, I was looking at a device today and a 220 FET device was rated at 45W, yet the the same device in D2PAK was 300W.  Specmanship at its best. Love to see them do that on a circuit board.

[Kleinstein]

The  2SK2698 fet has a good chance to work in linear mode, as it is specified for 500 V. So at least at low voltages (e.g. < 50 V) there are no instabilities expected. The gate capacity is rather high - so don't expect it to be fast.

[nctnico]

FETs which are intended for audio amplifiers will work well in a DC load but I prefer to use transistors in DC load design because these have an inherent current sink in their collector which makes sinking a constant current a matter of servoing the temperature effect out (which is a slow process). The NJW0281G is a nice beefy NPN transistor.

[Banjobeni]

Hi, I just registered to add my experience with dave's constant current load circuit.

I wanted to build a load capable of at least 2.5A at about 20V. I went through some component and what I found working was a NJW44H11G (80 V NPN 10 A Power Transistor, probably much oversized but it's what I could get my hands on easily) in combination with a MCP6002 (rail-to-rail opamp).

Before that, I burned through at least 5 MTP3055E (cheap ebay stuff, probably some other MOSFET relabeled) and one MTP3055VL (the logic level variant of the MTP3055E). What happened was that when stuff got hot (at about 2A, 40W) it burned through the MOSFET and as a consequence it started conducting for good, no regulation anymore. So my advice to someone trying to build the circuit is to go for a power transistor rather than a mosfet, and by all means use a rail-to-rail opamp. Hope that helps :-)

[magic]

You don't need rai-to-rail to drive an NPN. Only input range extending to ground is needed for current sensing and a fair bit of output current to drive the base.

I'm not sure if this heatsink is good for 50W, particularly without forced air cooling. What's the temperature of the transistor's metal plate? Longevity may be a problem if it's too high.

This circuit will probably work as-is without adventures because current draw is constant, but normally you should include some 100uF electrolytic at the input of every device powered through cables. Otherwise, sudden changes in current consumption cause ringing between cable inductance and the very low supply input capacitance of the device. Increasing supply capacitance reduces ringing frequency and amplitude, until it effectively goes away.

FETs which are intended for audio amplifiers will work well in a DC load but I prefer to use transistors in DC load design because these have an inherent current sink in their collector which makes sinking a constant current a matter of servoing the temperature effect out (which is a slow process).

You would think so but BJTs aren't always that great at low voltages and FETs too have channel saturation effect which makes them pretty much a CCS at sufficiently high drain voltages. The aforementioned MTP3055 is CC at currents up to 8A if Vds is above 3V. As usual, the datasheet of the specific device needs to be consulted.

Another advantage of MOSFETs is their usually lower thermal resistance. That being said, I've used a BJT too because they are slower and (I think) easier to stabilize. My sink has nothing between the opamp and pass transistor besides a base resistor, which is also needed as jumper. It's rock stable, at many combinations of currents and voltages up to 10A and 24V.