what a horrible data sheet

You could ohm it out. Parts like this require a hands-on before you make the footprint anyway.

I don't think there is an eval board.

It looks ideal for what I plan to do, but I might just use a bunch of

The Diodes Inc web site invites me to ask for tech support, and provides no way to do that.

How can people publish stuff this bad? I sure hope the silicon is better than the data sheet.

Try snipped-for-privacy@diodes.com

Just guessing but it works for a lot of domains.

Just fill in the form on

We have very good support from Diodes

Cheers

Klaus

Hey - at least it's in English. I've been running into data sheets written in Chinese.

There is no form, mostly a blank page. I'm running Firefox.

I hate forms anyhow. What's wrong with email?

I've seen worse recently from an even-bigger manufacturer (Microchip). I decided to start a poll on Dave Jones's blog:

Spoiler: the majority vote appears to have been correct. It works OK with the pad grounded or not.

Another helpful tip they left off of the data sheet was the presence of protection diodes from the open-drain pins to Vdd. The specs say that the part will run from 1.8V and that the I/O pins will tolerate 5.5V, but they don't point out the headroom requirement that keeps the 'open' drains from being pulled up beyond about Vdd+1.2V. That one got me. :(

They're just plain wrong. Power vias between the capacitors and pins? This must be one of those cases where "everybody knows" you need solid planes for 6 different rails.

I'd be nervous about using this part without trying it out in person. I'm sure it's fine, but if one of the ADI/LTC clock parts did something similar, I'd choose it because of the data sheet alone.

-- john, KE5FX

That's what I did with the MCP23S18 I mentioned in the other message. It'd be clear enough if the resistance to ground were effectively infinite or in the Gohm range, or if it was obviously tied to the other ground pins. But what am I supposed to do if it's on the order of 1 Mohm? The pad is clearly connected to *something* internal, but they can't be bothered to elaborate.

I guess I'm supposed to make assumptions about the electrical properties of their (currently-used) substrate material.

-- john, KE5FX

It's all too common for data sheets to not mention the paddle, or hide it in some fine print somewhere. We always asign it a pin number and show it on the schematic decal.

Exactly. What would I ohm the pad to? Assumed ESD diodes?

No response from Diodes so far.

One bad data sheet can spin off hundreds of support requests, or hundreds of design failures. That's not smart.

I'm going to use MC10EP parts instead of the Pericom/Diodes thing. They are dual-sourced and have good data sheets. Slower, though.

They also show different packages, that have ground pins!

I really just want a big differential fanout, to drive a clock into a bunch of SFP modules. The Diodes part would be perfect if I could trust it. There are more weirdnesses than the missing ground.

I asked them if there was an eval board. No response.

I wouldn't expect an improvement, under new DI mgmnt. They don't actually develop product there - not even diodes.

RL

Am 25.06.19 um 21:26 schrieb John Miles, KE5FX:

Have you seen the case I opened on ADI E2E wrt the LTC6757 (???) sine to square converter? An open output with some cm of stripline features funny oscillations. OK, they die out and an open output is not badly needed, but that does not build confidence. Looks like sth. is going metastable.

Silently ignored.

cheers, Gerhard

No, I missed that somehow.

That's not a super fast comparator... 2.9 ns. Even the rise/fall are over 1 ns. But it wouldn't take much in the way of bouncies to make a comparator oscillate.

Source termination is good, to cut the rise/fall current spikes in half, or more.

I suspect that it's hard to keep good support engineers. Being a FAE is like having five job interviews a day. It's a great start for a recent grad, to see what's out there. Good FAEs seldom last; they get hired away.

The semi people also do everything possible to keep the rabble from disturbing the chip designers. I guess that I'd get real support if I bought 7 million chips a month.

But really, it would be a net benefit to everyone to publish good data sheets.

Sonebody could make a nice career being a free-lance data sheet editor.

Sometimes people make easy things hard. If the thermal pad is ground, which pin would it be connected to internally?

It's a trick question.

What would you ohm it to? Don't tell me ground. I'm looking for a pin number.

I don't get why you can't find the link for support on the web site. It's at the top of the page under "Contact Us", then "Technical Support". Could that be any more clear???

Some web sites do make it hard to contact support or any other part of the company, but not this one.

Close out Firefox and restart it. If that doesn't work, reboot your machine. It's fine here.

I get the same issue at times. Some webpages just don't work on some machines.

There is no form, on my machine at work or the one at home.

I did phone the SV office, left a voicemail, and eventually got an email from a guy in Plano TX who says that the paddle is ground. There are other hazards on the data sheet.

I can get to 3 GHz using EclipsPlus, which will also cost a bit more.