Surface mount and creepage

Sylvia Else wrote in news: snipped-for-privacy@mid.individual.net:

We put 'slots' in our PCB HV multipliers between HV diode nodes.

We were told by *our* pcb house that the smallest ream (read vertical mill) they use was 35 mils, and that would be the smallest slot they could mill. They also said that they could mill specific layers, so you could apply a 'trench' effect.

You would have to get with your decided upon PCB house to see what is possible with their processes.

I'm using PcbWay, and their minimum is just slightly less, at 0.8mm, but even that won't fit between those pads.

I'll use the through-hole version instead - I can modify the pad shapes a bit, and fit 0.8mm slots between them - but I wish manufacturers would think about these issues.

Even the creepage between the pins where they enter the component body looks marginal at best.

Sylvia.

It is just functional isolation

You could maybe trim the suggested layout a bit?

Cheers

Klaus

Klaus Kragelund wrote in news:27ff7571-2002- snipped-for-privacy@googlegroups.com:

That and things like HV diodes can also be "bump soldered", which is a hand applied application of solder to the pads then attach the part "elevated" or "floated" above the actual rest point reflow produces. This leaves air space under the part between the nodes.

A lot of times a failure can occur in that location as the crud of the flux or cleaning bath gets trapped under the part making for an arc creepage pathway.

I belatedly realised that using this zero crossing triac driver wasn't going to work anyway, because the thing it's controlling (electric blanket) also probably contains a triac for heat control (it's continuously variable) and they wouldn't play well together.

Rethinking the circuit got rid of the optocoupler.

Sylvia.

Who says you have to place the unused middle NC pad? Leave it out and place a 2mm slot in the 3.3mm space.

We do that with SOT89 HV transistors--just an L-shaped slot round pins 1 and 3, with pin 2 deleted and the collector connected via the pad.

Cheers

Phil Hobbs

Good idea, but if the pin is there still, it just jumps using that pin (but it may then be easier to make a hole in the first place that fits to the capabilities of the PCB manufactorer

Cheers

Klaus

I usually snip the pin off the IC, but that may be inconvenient in automated assembly. An unplated 2.5 or 3mm dia hole may be easier for the PCB house. And two series 1.4mm air gaps from the 0.5mm pin tip to the two active pins is better than nothing.

Nice. I'm a big fan of the SOT-89 package, get more power in a small space. But it's hard to find good parts in that package, especially high-voltage types.

There's the VN2460N8 600V, VN2540N8 500V, and TN2540N8 400V FETs, and the VN2460N8 500V p-channel FET. The TN5325N8 and a smattering of other 250V FETs. And the DN3535N8 350V depletion FET. The CXTA42 and CXTA92 300-volt 1.2-watt npn and pnp BJTs, by Central Semi.

Not a lot to chose from, have you found better ones?

Aha, not to forget the LR8N8, a 450V, 1.6W regulator in sot-89, $0.47 each. And LR8K4, 2.5 watts in DPak.

This was for an APD supply, 0 to -300V at 3 mA, and the SOT89s were a couple of parallelled LND150s for the current limit.

Cheers

Phil Hobbs

This is probably something one gets away with most of the time, but I always wonder what the resulting shock wave does to the innards of the device.

Sylvia.

Aha, I forgot about the SOT-89 version of the LND150. The LND150 was a favorite back in the TO-92 days, but their SOT-23 version footprint doesn't match the rest of the industry, so it fell out of favor.

I used the program "Everything" to search for SOT-89 in my computer, missing SOT89. That also turned up the 300-volt BF620 and BF621 npn and pnp SOT-89 types.

What do you replace the LND150 with? AFAIK it's a unique part.

Cheers

Phil Hobbs

For SOT-23 parts I prefer the BSS126. Its higher Idss results in a more predictable Vgs operating condition. 600 vs 500V. 1.6 vs 2 pF Coss. 7mA vs 1mA Idss(min).

As far as a SOT-89 footprint, maybe a VN or TN version. But don't get me wrong, we still love the LND150.

If one needs a more narrow slot, use a laser...

...are you talking about "no clean" residues?

Robert Baer wrote in news:y0XJF.379071$ snipped-for-privacy@fx47.iad:

Don't leave behind any carbonized surfaces.