Advice Needed for 3500W Electronic Load Design

[Manvinder Bhullar]

Hi all,
I am writing this in the hope of getting some advice on the choice of
power devices for an electronic load I am going to build for testing a
48V DC power supply

The basic specs for the load are as follows:
Input Voltage :3 to 100 VDC
Max Current :120Amps
Max Overall Power Limited To:3500W

For the power stage, I am thinking of building 7 modules of 500W each
and paralleling them, for ease of construction and maintenance. Mostly
it is to get the heatsinks into manageable sizes.

Not having done such high powers before, I would appreciate advice on
1.What type of devices to employ (IGBTs or MOSFETs) and what
particular part numbers are readily available.
2. Where to get appropriate heatsinks.
3. Any other practical aspects which are not readily obvious when
building such equipment.

Thanks
Manvinder

 

[Tony Williams]

Hi all, I am writing this in the hope of getting some advice on
the choice of power devices for an electronic load I am going to
build for testing a 48V DC power supply

The basic specs for the load are as follows:
Input Voltage :3 to 100 VDC
Max Current :120Amps
Max Overall Power Limited To:3500W

Those numbers are not consistent, but assume that
the 3500W is the correct starting point.

I think you could get away with 10-off or 12-off
power modules, each with about 350W capability.

Each module would consist of an IXYS IXFN120N20
power MOSFET (SOT-227B package, 600W when mounted
on an infinite heatsink at 25C), mounted on an
IMI-Marston Coldfin heatsink, CF1-1250-0515-1250BA,
force-cooled with a 120x120 fan.

Each module would be an independant 0-12A current
sink, with current-shunt, opamp, TVS, etc. Maybe
also have a local analogue multiplier (or whatever)
to limit the demanded VI-product to 350W. SOT-227 is
handy here, because the control pcb can be bolted
directly onto the MOSFET terminals. And, of course,
a last-ditch thermal switch on the heatsink.

Verify the total current being sunk by passing all
10(12) output Drain wires through the aperture of
a LEM Hall Effect current sensor..... as they each
go to the final output bus-bar.

Hmmm.... tasty little job.

 

[Winfield Hill]

Tony Williams wrote...

Hmmm.... tasty little job.

Yes, and at 3500W, but with 12kW of dynamic range, with lots
of potential pitfalls for inexperienced engineers to fall into.
Manvinder would be wise to hire Tony for the job. :>) Ahhm,
assuming he's available of course.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Tony Williams]

Tony Williams wrote...

Yes, and at 3500W, but with 12kW of dynamic range, with lots
of potential pitfalls for inexperienced engineers to fall into.
Manvinder would be wise to hire Tony for the job. :>) Ahhm,
assuming he's available of course.

Unfortunately he's working 24/7 (at least it feels
like it atm) for the next few months.

Some nice little problems associated with it though,
especially a control scheme to reconcile that possible
12KW demand with the 3.5KW maximum VI-product. Some
interesting running analogue calculations and clamps
to be done.

I can't decide whether all 10(12) modules should be
brought up in parallel, or should a current-demand be
met by bringing modules up sequentially, each up to the
limit (of 12A or 350W, whichever is lower) until the
sum of the currents is equal to the demand (or is at
the maximum 3.5KW).

The latter method looks to have the advantage of probably
being more accurate at low currents, but requiring some
nifty footwork in the control scheme.

 

[R.Legg]

Hi all,
I am writing this in the hope of getting some advice on the choice of
power devices for an electronic load I am going to build for testing a
48V DC power supply

Are you sure this is the most sensible solution for your test
requirements?
Usually you wouldn't consider it unless:
- You anticipate it will be needed to perform similar tests on many
devices of a similar rating over a longer time period.
- there are no products on the market that fit your test budget.
- You have as much (or more) resources budgeted to design test
equipment as you have to design product.

The basic specs for the load are as follows:
Input Voltage :3 to 100 VDC
Max Current :120Amps
Max Overall Power Limited To:3500W

Either you are not just testing 48V devices, or you're wandering from
budgeted activity.

Are you familiar with the dimensions of similarly rated devics
currently on the market? Are you sure you want to burn 3500W when
testing a 3500W device?

For the power stage, I am thinking of building 7 modules of 500W each
and paralleling them, for ease of construction and maintenance. Mostly
it is to get the heatsinks into manageable sizes.

Not having done such high powers before, I would appreciate advice on
1.What type of devices to employ (IGBTs or MOSFETs) and what
particular part numbers are readily available.
2. Where to get appropriate heatsinks.
3. Any other practical aspects which are not readily obvious when
building such equipment.

When you wander from your actual goals, you run the risk of loosing
sight of what they are and of wasting resources by producing something
that is unsuitable.

RL

 

[Fred Bloggs]

Unfortunately he's working 24/7 (at least it feels
like it atm) for the next few months.

Some nice little problems associated with it though,
especially a control scheme to reconcile that possible
12KW demand with the 3.5KW maximum VI-product. Some
interesting running analogue calculations and clamps
to be done.

I can't decide whether all 10(12) modules should be
brought up in parallel, or should a current-demand be
met by bringing modules up sequentially, each up to the
limit (of 12A or 350W, whichever is lower) until the
sum of the currents is equal to the demand (or is at
the maximum 3.5KW).

The latter method looks to have the advantage of probably
being more accurate at low currents, but requiring some
nifty footwork in the control scheme.

I think it is the perfect application for a GMR bridge measurement of
total power dissipation going into your module array. Fix one module as
master and slave the the remaining to it- feedback error is on the total
array measurement and individual modules track with calibrated loop
precision . As usual the description is hopelessly underspecified as to
current mode, voltage mode , or power mode control, but, whatever, it
all boils down to a power VCCS on your individual modules.

 

[Don Lancaster]

Hi all,
I am writing this in the hope of getting some advice on the choice of
power devices for an electronic load I am going to build for testing a
48V DC power supply

The basic specs for the load are as follows:
Input Voltage :3 to 100 VDC
Max Current :120Amps
Max Overall Power Limited To:3500W

For the power stage, I am thinking of building 7 modules of 500W each
and paralleling them, for ease of construction and maintenance. Mostly
it is to get the heatsinks into manageable sizes.

Not having done such high powers before, I would appreciate advice on
1.What type of devices to employ (IGBTs or MOSFETs) and what
particular part numbers are readily available.
2. Where to get appropriate heatsinks.
3. Any other practical aspects which are not readily obvious when
building such equipment.

Thanks
Manvinder

And the problem with strip heaters is?

You can buy military load banks these days for pennies on the dollar.
--
Many thanks,

Don Lancaster
Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552
voice: (928)428-4073 email: [email protected]

Please visit my GURU's LAIR web site at http://www.tinaja.com

 

[Manvinder Bhullar]

Hi all,
I am writing this in the hope of getting some advice on the choice of
power devices for an electronic load I am going to build for testing a
48V DC power supply

The basic specs for the load are as follows:
Input Voltage :3 to 100 VDC
Max Current :120Amps
Max Overall Power Limited To:3500W

For the power stage, I am thinking of building 7 modules of 500W each
and paralleling them, for ease of construction and maintenance. Mostly
it is to get the heatsinks into manageable sizes.

Not having done such high powers before, I would appreciate advice on
1.What type of devices to employ (IGBTs or MOSFETs) and what
particular part numbers are readily available.
2. Where to get appropriate heatsinks.
3. Any other practical aspects which are not readily obvious when
building such equipment.

Thanks
Manvinder

Thanks for scrutinising the specs.

I admit that on the surface the specs dont look right. Customer
provided them and probably has some other reasons for the way they
are.

Here are some possible reasons I can think of:
a. 120 Amps is for over-current shutdown testing and providing some
margin beyond the OCP point.
b. There will be product of lower output voltages in future.
c. 100 V is to provide enough margin for OVP Testing and defective
products.

Its midnight here and I will get more clarification Monday.
Thanks
Manvinder