Researching Alternatives to Dim Bulb Circuit to Control AC Current While Testing

[Jwillis]

I have to ask, is that a doorbell at the fan location?  Evidently this is some new technology I am not up on.

 I think its an old style ceramic light switch .

[Berni]

There are some depletion N FET transistors out there that handle 400V and come in a large heat dissipating package. They act a lot like a JFET where they can work as a current source. A divider can be added to the gate in order to adjust its operating point to the desired current.

Hi, Berni. This sounds like a good idea. But, as I mentioned earlier, I don't have the skills to design something like this from scratch. So, can you give me the specifics, or tell me the name of such a circuit, so that I can google it?

A little googling actually results in finding this exact idea using a N depletion FET:
https://electronics.stackexchange.com/questions/218976/ac-constant-current-source-design


Now that blueskull mentioned it i also see you can get actual JFETs with high power handling capability such as this one (But depletion FETs are more still more common):
https://www.digikey.com/product-detail/en/unitedsic/UJ3N120070K3S/2312-UJ3N120070K3S-ND/12083301

Tho with the load being a dead short the transistor will indeed burn a lot of power on itself (Just like the light bulb), so the current must be set for something in the 100s of mA range or less in order to make sure the transistor can keep itself from overheating in all cases. Tho i suppose if you wanted a 1kW capable model you could still just put something like 20 of these current sources in parallel on a massive heatsink along with a bimetal thermal switch for protection.

[tautech]

What could possibly be easier to make from junk on hand ?



Incandescent bulbs will be around for many years yet in specialty shops.

I have to ask, is that a doorbell at the fan location?  Evidently this is some new technology I am not up on.

Old plastic surface mount mains switch. They were used on switchboards and wooden wall blocks on top of the end of steel conduit runs.
Just a simple sprung leaf switch and available in 2 way versions too.

More modern version of much older porcelain and Bakelite types.

Just junk I had around when I built that dim bulb tester more than a decade ago.

[t1d]

A little googling actually results in finding this exact idea using a N depletion FET:
https://electronics.stackexchange.com/questions/218976/ac-constant-current-source-design

Great find. I am sure it was no problem for you to find it. But, you have the advantage of knowing the proper terms to search by. That's why I always ask, specifically, for the proper terms... So, I can do the search myself. I hate to put the burden on someone else to do my "homework." It is totally reasonable for those that help to expect that those in need have done all that they can... first. I did do an extensive search, before posting. Thanks for the link.

Interestingly, as I was researching FETs, this is the exact series that I found best for the purpose. You beat me to proposing it. lol

Now that blueskull mentioned it i also see you can get actual JFETs with high power handling capability such as this one (But depletion FETs are more still more common):
https://www.digikey.com/product-detail/en/unitedsic/UJ3N120070K3S/2312-UJ3N120070K3S-ND/12083301

Tho with the load being a dead short the transistor will indeed burn a lot of power on itself (Just like the light bulb), so the current must be set for something in the 100s of mA range or less in order to make sure the transistor can keep itself from overheating in all cases. Tho i suppose if you wanted a 1kW capable model you could still just put something like 20 of these current sources in parallel on a massive heatsink along with a bimetal thermal switch for protection.

Yes, multiples FETs might be required. I did catch in the Data Sheet that the current rating may be wholly sufficient, but the watts rating is not. Tricky, tricky...

I surely do appreciate all the great help! Thank you.

[Berni]

No problem. These special types of FETs are actually not al that well known in electronics. Mostly because the common enhancement mode FETs are vastly more useful in most cases where you need a transistor, so they have become far more popular and so are the kind of FET everyone knows about. But this is one of the rare applications where a depletion FET is very handy. To be honest i didn't expect to find a write up on this online. I was just looking for a picture of a JFET current source to use as an example of how to do it but came across a schematic for the exact thing by accident (Too lazy to actually draw it)

In terms of power dissipation, yes you will run into this problem with all semiconductor based designs. The tiny silicon dies inside can only work up to about 150°C, so a lot of cooling is needed to remove this much heat at high powers. The reason lightbulbs are so much better at dissipating this excess power is that the filament inside can operate at 2500°C. This massive temperature rise gives entropy a ton of motivation to spread the heat out into the room temperature environment so a lot of heat energy is dissipated despite the small size.

But to be honest i don't really see any reason to build this circuit for supplying more than 50W. Such an amount of power is enough to start up pretty much any mains power supply. Once the PSU of the device being repaired is up and running then there is no reason to continue using the current limited supply. It will not protect anything downstream of the PSU since the PSU itself typically has a good bit of power stored in its capacitors, so despite a limited input power it is still capable of a plenty strong enough output surge to blow something else up. The only job of the current limited supply is to safely power up the mains side of the PSU in a way that doesn't cause a huge bang with sparks flying due to your wall outlet capable of supplying >300A for a short moment before the breakers trip. After the PSU it is the PSU deciding how much current it wants to provide.

[t1d]

To be honest i didn't expect to find a write up on this online. I was just looking for a picture of a JFET current source to use as an example of how to do it but came across a schematic for the exact thing by accident

After looking at the link, a bit, I realized that I had come across it. However, I did not consider it in depth, because it appeared to me that the remainder of the text indicated that it was not a good solution. Far more complicated circuits were then proposed.

But to be honest i don't really see any reason to build this circuit for supplying more than 50W. Such an amount of power is enough to start up pretty much any mains power supply. Once the PSU of the device being repaired is up and running then there is no reason to continue using the current limited supply. It will not protect anything downstream of the PSU since the PSU itself typically has a good bit of power stored in its capacitors, so despite a limited input power it is still capable of a plenty strong enough output surge to blow something else up. The only job of the current limited supply is to safely power up the mains side of the PSU in a way that doesn't cause a huge bang with sparks flying due to your wall outlet capable of supplying >300A for a short moment before the breakers trip. After the PSU it is the PSU deciding how much current it wants to provide.

From the very beginning of joining EEVBlog, I knew I had found a wellspring of wisdom and this proves it. I had not even thought about the capacitors building up energy. That's a great point!

The more we talk about this, the more I am convinced that the safest way to deal with the onboard PSU is to disconnect it from the rest of the circuit and, then, inject a controlled voltage/current source with a bench PSU. The effort to separate the two and the risk of rework damage is just part of the game, but well worth it. Of course, the onboard PSU would then be test-able, itself. I have also learned that I need to add a break-point, either jumpers or cables, between the PSU and the circuit, on all my PCB designs.

Do we need to consider this further, or should I mark the title as "Solved"?

[alsetalokin4017]

It's hard to beat a Variac and an inline ammeter for this task.  I usually just remove the fuse and clip the ammeter across the fuse holder, and slowly bring up the Variac until the indicated current approaches but never exceeds the fused value. Then look for hot spots in the DUT, using a q-tip and isopropyl alcohol. Or a thermal imaging camera, whatever you have on hand.

Unfortunately, some switching power supplies throw a temper tantrum when you feed them a low voltage from a variac.  At some voltage, some parts of the circuit starts to wake up, but other parts haven't started so things get locked into the wrong state, etc.  I've heard that this can fry a perfectly good power supply.  YMMV.  But to be on the safe side, when I'm working on a unit with a SMPS, I never use a variac - only a dim bulb tester.

Ed

Yes, you are right. I almost never work on things powered by switching power supplies so I didn't think of it, sorry.