Active loading for measuring high currents

[eblc1388]

Are there any models that have a low (0.5v) Vce_sat at a collector current of like 8-10A? All the common ones I've seen usually hover in the 1-2v range.

The common 2N3055 has a Vce_sat of 0.5V at 7A collector current, with base current at 0.7A.

 

[me again]

Active loading for measuring high currents.

All things being equal, a hall effect transducer would be the choice for picking off a current analog from a high current dc circuit.

Closer to earth, a 0.001 ohm, 0-100 amp, 0.001 volts per amp meter shunt is the ideal pick-off resistor for this application.

Jameco Electronics offers such a shunt for $24.95 as part#163209CB. This unit has a full scale accuracy of 0.8% +/- 0.2mV.

If you prefer an el-cheapo part, a pretty good shunt can be made from a 6" lienght of,(untempered), oil hardening drill rod. Drill and tap two two holes 4" apart for attaching the pick-off leads and insert into the circuit with clamp type lugs or however. Provide trimming in your circuit.

Woof

 

[Roff]

However, I see another problem. As I said, the batteries being tested are 1.2v NiMH/NiCD or 3.7v LiPo batteries. If testing a 1.2v battery, I'm sure the voltage will drop quite a bit at 100A. Assuming a worst case transistor scenario; if the shunt is dropping .1v, that means the tranistors will have to have a C-E saturation voltage of .7v or less (I figure I can stop the current test if the battery voltage drops below .8v since anything below that is unusable anyway). I can't seem to find a commonly available transistor that has a low sat voltage when passing any appreciable current - and the more collector current, the more the saturation voltage. So, I guess I'll have to use MOSFETs?

You have a valid concern in testing 1.2V battery. I think Ron would agree that the drain connection of the MOSFET driver has to be fed from 12V instead of the 1.2V battery under test. You can use more transistors to reduce the collector current of each and thus lowering the Vce_sat.

Good catch, LC. "Darlington" MOSFETs don't work!
I changed the schematic to correct this error.

 

[Roff]

Brian, why are MOSFETs out? I think this topology will work. I simulated it with the parts shown, but you could use a different op amp and different MOSFETs. You should probably get all the transistors from the same lot. I sim'ed it with a pulsed source in place of the control voltage to make sure it was stable (no oscillation) - at least in the simulation. :roll:
Keep in mind that a 1mv input offset in the op amp will cause a 1 amp error in the output.
I think Vishay has some resistors you can parallel to get the value you need. The problem with a homemade unit is calibration.

 

[eblc1388]

I changed the schematic to correct this error.

But now you have created another problem. The measured current value now includes the base current through all the transistors but not the load, which can be as high as 10A for a 100A load. The regulator will work but there is no easy way to know the actual load current.

The following suggestion is what I have experienced in building my 25A constant current source. I metered my constant current set value from the reference leg to tell me in advance what the current value will be so I can set the value up precisely without passing current through the load. Then I use a similar Opamp configuration as yours to regulate the current. As expected, the offset and drift of the error opamp produces error between the set and actual value of 5~10% especially at low current value.

So I would suggest in your present circuit, a X10 chopper stabilised Opamp be used to amplifiy the tiny current signal first. This would produce a drift-free output of 10mV/A of which the signal can also be metered by a DVM to indicate the actual current flow. With this voltage level and a reference of 1V instead of 0.1V, the circuit noise margin increases and the requirement on the drift and offset w.r.t. error opamp is greatly reduced. One problem though is the requirement of a -Ve power supply to power the chopper opamp.

But on second thoughts, if the OP is interested in tens of amperes, a few mV offset in the circuit would still be quite OK for his purposes.

 

[Roff]

I changed the schematic to correct this error.

But now you have created another problem.

Hey, it was your idea. I just documented it. :lol:
The problem with the original circuit was that Vce(sat) was basically identical to Vbe(sat), given a large MOSFET as the driver. Unfortunately, Vbe(sat) is around 1.5V for a 2N3055.

 

[Oznog]

Re: Active loading for measuring high currents.

If you prefer an el-cheapo part, a pretty good shunt can be made from a 6" lienght of,(untempered), oil hardening drill rod. Drill and tap two two holes 4" apart for attaching the pick-off leads and insert into the circuit with clamp type lugs or however. Provide trimming in your circuit.

This lacks several features needed for a good shunt:
1. A good way to make reliable high current electrical connections
2. A reasonably stable resistance over temperature
3. A reasonable amount of heat dissipation. This will not only keep the resistance stable but prevent permanent changes in the material's electrical properties from overheating.

 

[Oznog]

One problem though is the requirement of a -Ve power supply to power the chopper opamp.

Not a big problem since the current is very low. A switched cap voltage converter chip is cheap and will do the job nicely.

MAX1044 for example:
https://www.electro-tech-online.com/custompdfs/2006/02/ICL7660-MAX1044.pdf

Be sure you do not exceed the max input voltage of the MAX1044, the op amp, or the meter itself.

One problem you may notice is that a proper shunt has 4 wires, 2 to carry huge currents which includes significant errors from the interconnect resistance and 2 measuring wires which carry only minute currents. An op amp's normal configuration would make a signal differential versus the "-" shunt measuring wire. So if you do a 100x gain, on 50mV shunt voltage but there's a 200mV voltage between the shunt "-" measuring wire and the shunt "-" power wire- the op amp outputs 5v but that's versus the "-" measuring input, so it'll actually output 5.2V versus ground. All the rest of the circuitry is probably going to be referenced to ground so this may be an issue.

 

[eblc1388]

One problem you may notice is that a proper shunt has 4 wires, 2 to carry huge currents which includes significant errors from the interconnect resistance and 2 measuring wires which carry only minute currents. An op amp's normal configuration would make a signal differential versus the "-" shunt measuring wire. So if you do a 100x gain, on 50mV shunt voltage but there's a 200mV voltage between the shunt "-" measuring wire and the shunt "-" power wire- the op amp outputs 5v but that's versus the "-" measuring input, so it'll actually output 5.2V versus ground. All the rest of the circuitry is probably going to be referenced to ground so this may be an issue.

Good observation. That's why all things are referenced to the shunt voltage connection and not the high current battery (-) connection of the shunt. This applies also to the common of the reference voltage. The voltage between the voltage tap off point and the battery connection does not add to the error in this case.

If you looks again at Ron's circuit, you can see it is done this way already. The end of the 1K POT is NOT connected to ground but to the shunt terminal instead.

 

[BrianG]

A little update on this project.:

I decided to think a little outside the box by using a electronic RC speed controller! Castle Creations makes an ESC called the "Mamba Monster", which is a brushless ESC rated for 120A continuous. If configured in "brushed mode", all three outputs are tied together for a 360A theoretical rating. Why use this you ask? Because I can take advantage of the PWM output to regulate how much average voltage it outputs.

I also have a servo tester which outputs the PWM signal needed to control this ESC.

To measure the current and voltage, I opted to use an Eagletreesystems e-logger. A neat little device that measures voltage up to 70v, and uses a hall-effect sensor to measure currents up to 150A in 100ms samples. It also has other inputs such as rpm, temperature, etc. The neatest thing is that it can either record these values in stand-alone mode (to be uploaded to the computer later), or run in Live mode where I can see a graph real-time. Quite a useful device for under $70.

I then won an ebay auction for 220 3ohm 50w power resistors (aluminum cased). Putting 60 of these in parallel will get the overall resistance low enough to get the max current, while providing enough power dissipation (will also be heatsinked very well).

I still have some work to do for this, but at least the issue of regulating the amount of current is solved.

 

[bountyhunter]

Where are you getting a 1 milliohm, 10 watt resistor? :?

They sell wire for this puprose: it is called either "constantan" or "manganan" and it is formulated to have zero TC. You can get it in various thicknesses and you should select a wire gauge that gives you the desired resistance with workalble lengths like 4" or so.

 

[bountyhunter]

So, MOSFETs are out. Back to transistors then.

MOSFETS are not out, you can get FET's which still have gain with V(D-S) voltages of 1V, as long as the gate is pulled up to ten V or more. Their gain isn't great, but they do have gain and you can parallel them to share current. This one can do about 15A with V(D-S) of 1V:

http://www.datasheetcatalog.org/datasheet/motorola/MTB36N06E.pdf

I don't think testing batteries at high continuous currents is feasible because of power dissipation, you might do better with a pulse tester.

BTW: there are high current testers available that operate in both constant current and constant resistance load mode. I used to use them to test power supplies. You might want to check around before you spend a big chunk of your life trying to reinvent one.

 

[BrianG]

Well, my last post takes care of the FET/transistor decision, but I've also been doing some extra reading, and FETs are most useful when run in switched mode. Some are more suitable than others when run in linear mode, but tend to be expensive.

 

[bountyhunter]

Well, my last post takes care of the FET/transistor decision, but I've also been doing some extra reading, and FETs are most useful when run in switched mode. Some are more suitable than others when run in linear mode, but tend to be expensive.

I think you may not be able to use transistors. With a V(C-E) of only one Volt, any NPN transistor will be pushed deep into saturation where it's current gain is really low. To drive 100A current (even a pulsed current) you will have to have a current source to drive the base(s) of the transistors that can out put probably 20+ Amps.

The N-FET gives you the advantage of not needing to drive a ton of current into the gate to make it turn on fully.

I've also been doing some extra reading, and FETs are most useful when run in switched mode.

That simply is not true. MOSFETs are used frequently in linear mode and even in very high-end audio amplifiers. FET's (and any transistor) is more EFFICIENT in switched mode since it is either fully on or off, so it doesn't dissipate as much power. But FET's are designed to be used as linear amplifiers if needed.

FET's of a given current rating may be more expensive than bipolar transistors, but for your application the latter require so much drive current that the FET design would probably still be cheaper overall.

 

[BrianG]

Actually, the post I was referring to was the one where I said I would be using a brushless speed controller (which uses 30 FETs). So, that's pretty much taken care of. I totally understand the issue about deep saturating transistors to get minimal Vc-e voltage.

As far as MOSFETs used in linear mode: I was doing some research into another project (using FETs in linear regulators) and everything I've read leads to the same concusion: Yes, there are FETs that specifically designed for linear applications, but they seem to be relatively few and more expensive. I know some amps use FETs in the output stage, I have a few very old 80's-90's era Fosgate car amps that used them in this way. I just figured they used special FETs, or took extra precautions in the support circuitry. Either way, it was more work than I wanted to do.

As such, I've shifted focus a little from using FETs in their linear range, to using FETs in switching mode to achieve similar results. Now, I am looking at using a constant low-resistance static load, and relying on the FETs to limit the average voltage to the resistors to get the current load I'm looking for. Simpler and cheaper for what I'm after.

Just want to say thanks for all your input! I know it can be tiresome to explain fundamentals to pseudo electronics hobbyists, but it's been so long since I've done design work, that I need extra help. So, thanks for putting up with the questions.

 

[bountyhunter]

As far as MOSFETs used in linear mode: I was doing some research into another project (using FETs in linear regulators) and everything I've read leads to the same concusion: Yes, there are FETs that specifically designed for linear applications, but they seem to be relatively few and more expensive.

I can't figure out where you read that. FET's are FET's, the only real difference is in the voltage range they design the gate to fully enhance with. Logic level FET's turn on fully with realatively low gate voltage (like 3 - 5V) and "standard" FET's have a typical range of up to 10V gate voltage, but still turn on pretty well at 5V.

The functional difference between FET's and Bipos is that the former needs a larger drive voltage range and the latter uses a base current to drive with, voltage changes very little.

BTW, the semi company I worked at for the last 20 years is big in linear regulators and their high current (>2A) regs are nearly all FET types now, both P-FET and N-FET which requires a charge pump or external bias rail voltage..

I know some amps use FETs in the output stage, I have a few very old 80's-90's era Fosgate car amps that used them in this way. I just figured they used special FETs, or took extra precautions in the support circuitry. Either way, it was more work than I wanted to do.

It's not special circuitry, it's just a different design topology. They act more like vacuum tubes than bipo transistors.

As such, I've shifted focus a little from using FETs in their linear range, to using FETs in switching mode to achieve similar results.

You could put a power resistor in the drain to the +Batt and then switch the FET full on and off and vary the ON time duty cycle. The resistor burns most of the power that way.