Active Load for PSU Testing

[drewspriggs]

Howdy,

Looking for an active load for testing some of the switcher supplies I've been working with. I've been using combinations of 10w resistors at the moment, but would rather something a little more elegant that is less sketchy and easier to measure.

Heaps available for batteries that can do real high current but only 50-100v. I probably need a max of 500v and maybe 50w max.

Happy to build something if I need to. There's a few different designs I've seen that have been proposed but nothing I've seen people actually finish. I've even looked into high power rheostats and a shunt for current measurement, but I can't find anything that's above 1k resistance because they're mostly used with high current circuits too.

Thanks!

[R.G.]

power mosfet and heatsink

[drewspriggs]

power mosfet and heatsink

As in, just run the MOSFET in linear mode with the biggest heatsink I can find?

[nuke]

It's a chunk of change but here's 500V, 300-watt B&K Programmable electronic load.

https://www.tequipment.net/B&K-Precisio ... ds/?v=7423

Hunt around for East Tester brand (China). I have one, not high voltage, but it's been really good for its rated purpose and was quite reasonable.

[trobbins]

Perhaps elaborate on what you are hoping to 'see' by using an active load, and how you are hoping to measure any change (ie. what scope and sensors are you aiming to use).

The reference to 500V indicates this is for testing a B+ rail and its dynamic characteristics ?

Is the testing for just step load transient assessment (eg. for an amp transitioning from say idle to max rated sinewave output) ?

[drewspriggs]

Perhaps elaborate on what you are hoping to 'see' by using an active load, and how you are hoping to measure any change (ie. what scope and sensors are you aiming to use).

The reference to 500V indicates this is for testing a B+ rail and its dynamic characteristics ?

Is the testing for just step load transient assessment (eg. for an amp transitioning from say idle to max rated sinewave output) ?

I have modified/built a number of switchers for generating +B from smaller DC supplies, and wanted a better way to test ripple, voltage drop at idle/high load, heat buildup, etc.

I'm currently using a resistor array sandwiched in some aluminium bar, and patching them out to get the approximate load I want.

It's fine, but I'm much limited in what loads I can present. An active load would simplify that a bunch

[nuke]

Look around for an East Tester ET5411A+. 400 watts, 0-500V, 0-15A. They're $180 USD on Amazon. Insanely cheap.

[R.G.]

power mosfet and heatsink

As in, just run the MOSFET in linear mode with the biggest heatsink I can find?

I was in a dead rush when I wrote that. But yes, run the MOSFET(s) in linear mode and keep them cool enough not to die.

You can get MOSFETs with high enough voltage ratings to test tube power supply voltages, and the ability to dissipate significant amounts of heat per device. Unlike bipolars, MOSFETs can be selected that will current share, not current hog, so you can add enough MOSFETs in parallel to dissipate whatever power.
The real key is the heat sink. Unless you are OK with PWM loads, the MOSFETs have to run linear, so the MOSFETs themselves have to dissipate the heat. Many power MOSFETs can dissipate up to 100W or so with the proviso (in small print on the datasheet) that the case is held at 25C. In the real world, this can't happen for long.

Most MOSFETs have a junction temp maximum of 150C, some up to 175C. The heat goes into the surrounding air, perhaps at 30C. So the sum of the thermal resistances, junction to case (Rjc), case to sink (Rcs), and sink to air (Rsa) will have to carry the heat flow in watts, and the sum of those (Rjc+Rcs+Rsa) times the watts has to keep the total of ambient temperature and and their temp rise under the junction max temperture. In practice, the only thing you can control very much is the heat sink thermal resistance in degrees C per watt.

Without forced air fan cooling, about all you can practically get is 20-30W per device. Forced air lets you get to 40-50W per device without killing them with heat, and MOSFETs let you parallel up enough of them to get a lot of watts. But the heat sink has to keep them cool, and hundreds of watts needs fan cooling. A non-pwm'd active load is an exercise in thermal management.

If I sound like I designed active loads - I did.

[flemingmras]

I was in a dead rush when I wrote that. But yes, run the MOSFET(s) in linear mode and keep them cool enough not to die.

You can get MOSFETs with high enough voltage ratings to test tube power supply voltages, and the ability to dissipate significant amounts of heat per device. Unlike bipolars, MOSFETs can be selected that will current share, not current hog, so you can add enough MOSFETs in parallel to dissipate whatever power.
The real key is the heat sink. Unless you are OK with PWM loads, the MOSFETs have to run linear, so the MOSFETs themselves have to dissipate the heat. Many power MOSFETs can dissipate up to 100W or so with the proviso (in small print on the datasheet) that the case is held at 25C. In the real world, this can't happen for long.

Most MOSFETs have a junction temp maximum of 150C, some up to 175C. The heat goes into the surrounding air, perhaps at 30C. So the sum of the thermal resistances, junction to case (Rjc), case to sink (Rcs), and sink to air (Rsa) will have to carry the heat flow in watts, and the sum of those (Rjc+Rcs+Rsa) times the watts has to keep the total of ambient temperature and and their temp rise under the junction max temperture. In practice, the only thing you can control very much is the heat sink thermal resistance in degrees C per watt.

Without forced air fan cooling, about all you can practically get is 20-30W per device. Forced air lets you get to 40-50W per device without killing them with heat, and MOSFETs let you parallel up enough of them to get a lot of watts. But the heat sink has to keep them cool, and hundreds of watts needs fan cooling. A non-pwm'd active load is an exercise in thermal management.

If I sound like I designed active loads - I did.

Ever try soaking MOSFETs in a vat of oil for cooling?

[R.G.]

Ever try soaking MOSFETs in a vat of oil for cooling?

I never wanted to clean up the mess I would have made.

On a more serious note, even in oil, you would need some kind of heat spreader. The postage-stamp sized thermal pad on a plastic power package will have an easier time getting heat out to oil, but unless you flow the oil past it i.e. use a pump or stirrer), there is a limit to how much heat you can get out at low temperature rises. stagnant air or oil doen't conduct heat away very well. You have to move the heat away by moving the air or oil. We often rely on thermal convection (i.e. hot things rise) to do the moving, but oil is thicker than air, so its thermal convection is less than air's. It can support hot spots better.

Yo can spread heat out by bolting a good thermal conductor - aluminum, copper, diamond, etc. to move the heat away from the transistor, and to offer a much bigger surface to move heat into the surroundings. I think of this kind of thing as "mpedance matching" the high-heat, small area transistor to a low-heat, very spread out medium.

[printer2]

I needed a load to test a SMPS that used to be pretty cheap, decided to use an amplifier I had breadboarded (just a Champ). Used a 6N3C and with 390V pulled 104 mA netting 41W for a short period of time. I was learning what the HV module could do, blew up the diode a few times. To be fair, it did not like supplying above 20W with a low supply voltage of 12V. The picture has the amp dissipating 28W. The temperatures are ambient and on the heat sink. I gave it a larger one as well as one on the diode.

[trobbins]

I have modified/built a number of switchers for generating +B from smaller DC supplies, and wanted a better way to test ripple, voltage drop at idle/high load, heat buildup, etc.

I'm currently using a resistor array sandwiched in some aluminium bar, and patching them out to get the approximate load I want.

It's fine, but I'm much limited in what loads I can present. An active load would simplify that a bunch

DC/dc like the YH11068A that printer2 is using may show skipping/burst mode operation at some loads, and would need a scope to identify. That can make it messy to interpret measured ripple levels and voltage change as load levels change.

Nearly all such cheapo dc/dc modules show thermal concerns for the switchs, diodes and transformer. So yes some bench testing is needed to either stay well within individual part thermal stress limits, or manage temp rises using custom enhanced cooling.

Using resistor loads does avoid any unforeseen control/ripple interaction between an active load and the PSU dc/dc - and best to check for any such interaction before spending a lot of time believing an active load is behaving in a benign linear manner.

[printer2]

My intent was to show you can use a tube with adequate plate dissipation, adjust the current with changing the cathode resistor or if fixed bias twiddle the pot. I did use the HV module in a SE amp and the biggest thing I noticed is no hum. Next I want to use one in a P-P amp using 6K6's. I did want to use the module and a laptop supply as a cheap way for someone to get into making a tube amp. I did try some buck modules to get the voltage down for the heaters, some are pretty noisy and do not live up to their specs. Got sidetracked due to life, hope to get back to it yet.

[drewspriggs]

power mosfet and heatsink

As in, just run the MOSFET in linear mode with the biggest heatsink I can find?

I was in a dead rush when I wrote that. But yes, run the MOSFET(s) in linear mode and keep them cool enough not to die.

You can get MOSFETs with high enough voltage ratings to test tube power supply voltages, and the ability to dissipate significant amounts of heat per device. Unlike bipolars, MOSFETs can be selected that will current share, not current hog, so you can add enough MOSFETs in parallel to dissipate whatever power.
The real key is the heat sink. Unless you are OK with PWM loads, the MOSFETs have to run linear, so the MOSFETs themselves have to dissipate the heat. Many power MOSFETs can dissipate up to 100W or so with the proviso (in small print on the datasheet) that the case is held at 25C. In the real world, this can't happen for long.

Most MOSFETs have a junction temp maximum of 150C, some up to 175C. The heat goes into the surrounding air, perhaps at 30C. So the sum of the thermal resistances, junction to case (Rjc), case to sink (Rcs), and sink to air (Rsa) will have to carry the heat flow in watts, and the sum of those (Rjc+Rcs+Rsa) times the watts has to keep the total of ambient temperature and and their temp rise under the junction max temperture. In practice, the only thing you can control very much is the heat sink thermal resistance in degrees C per watt.

Without forced air fan cooling, about all you can practically get is 20-30W per device. Forced air lets you get to 40-50W per device without killing them with heat, and MOSFETs let you parallel up enough of them to get a lot of watts. But the heat sink has to keep them cool, and hundreds of watts needs fan cooling. A non-pwm'd active load is an exercise in thermal management.

If I sound like I designed active loads - I did.

Legend - thanks for the info! I've got a couple of big heatsinks I was playing around with for some PCM stuff a few years ago that I can use, and occasionally see a couple of beefy MOSFETS posted up on one of the local electronics groups I'm in. Should be able to figure something out from there.

[nuke]

Man, for as little as they are on Amazon, ($180 to $260) I'd just buy one:

East Tester ET5411A+

https://www.easttester-cn.com/et54xx-a- ... d-product/