neat Hall sensor (2025 Update)

We have a LEM HMSR6-SMS instock... not sure why. There is a 30 amp version. It's more accurate than the Monolithic part and has nice chunky current leads, but it costs over 2x as much as the Mono part.

I'm designing a plugin to our modular power system, just a board with eight power relays. Then we figured, let's measure the voltage and current of each relay too. Featuritus follows. Why not make it a programmable circuit breaker? Why not a polyphase wattmeter? Somebody please stop me.

Demo board makes sense to me. Simplest thermal conductivity test.

Looks like ACS7xx from Allegro.

Rl

For sensing AC current, one could use a milliohms-range 1206 resistor and a signal transformer, a DRQ127 maybe. That all costs about a dollar. All surface mount.

They used 4 oz copper, hardly practical.

They mostly wasted the bottom layer thermally.

The pour geometry crowds the heat into two corners so adds thermal resistance.

The vias are too small and poorly located.

It's OK for maybe the 10 or 20 amp versions, but 50 will get nasty.

This is nice:

Why? The specialist printed circuit shops that make printed windings can cope with that and rather thicker layers of copper.

But you can't be bothered to tell us how they might have done better.

What's impractical about 4oz copper in a power circuit? Are you saying it's not enough?

I'd say it was more than could be expected from a simple demo board and is typical of higher current layout in this neck of the woods, where power and control circuits try not to share the same board or board layer.

Also, using a part with a 50A full scale in a 25-30A continuous rated circuit would be normal pactice, where duty cycle, crest factor or pulse/surge are anticipated under normal operating conditions.

There are 'better' SMD packages available, if that is your only concern. You seem to have found some.

RL

ACS711 has had commodity pricing for the last decade.

RL

I'm saying that you can't put any other reasonable parts on the board.

It would be interesting to get the eval board and actually push 50 amps through it.

I see a lot of data sheets with absurd power and current ratings, like a dpak rated for hundreds of amps and watts, with tiny footnote somewhere that cautions about package or lead limits. IR started the trend of specifying chip therortetical limits that can't be approached in real life, and then everybody had to tag along.

Exactly.

They could at least merged the high current pin pairs into big paddles.

It's a nice part. We're using the 10 amp version at up to 7.5 amps. The 50 amp version is silly.

It's pin compatible with the Microcircuits part if you ignore pin 6.

Ends up looking like an automotive fuse link.

Well, when you actually get around to using it (probably in a relatively remote location) be sure to R/C the power and data lines. R in series with source, C at receiving end.

RL

That Monolithic Hall sensor is analog output. Zero current outputs Vcc/2, and the sensitivity is ratiometric on Vcc. So one uses the same rail for sensor Vcc and for the ADC reference. We'll autozero it every powerup.

I'll RC filter a bit, right at the ADC input. Most ADCs kick out some charge when they sample, and that freaks out a lot of analog parts.

I'll have eight Hall sensors on a board, about like this one:

So, the sensors are very remote.

RL

No, surface-mount, on the board. Close to the ADCs.

Connector pins carry 8x 10A load current?

RL

I'm using 8 amp relays and rating things at 7.5 amps. We'll use Phoenix connectors rated at 12 amps per pin.

The relays we have are DPDT, so we can put two contacts in series to make a SPST closure and double the legal DC breaking voltage.

Here's what I have in mind:

The trick is to get the 2nd row of connectors onto the front panel. I'm thinking a baby board and ten 2-56 spacers to support it and bring the connections up.

Now I can contact some customers and see if anyone wants it.

you can get dual row connectors,

One problem with the Phoenix connectors is that they are big. Another issue is that you need to be Superman to un-mate them. The 5-pin pair that we use has ejectors so it's not so bad.