1. Home
  2. Electronics Design
  3. SG51K 3.6264Mhz clock chip: why are the outputs so flaky?

SG51K 3.6264Mhz clock chip: why are the outputs so flaky?

Yes, a UART is sensitive to frequency. Having as much as 1/2 bit of error at the end of a 10-bit character would be disaster. That's 5% variation.

There are gazillions of circuits out on the web. A 74LS04 is a good choice, as is a 74HCU04 (note the 'U'). A 74HC04 is _not_.

Define "garbage". As long as the overshoot isn't greater than 10% or so of the total then you should be fine, and depending on the shape even larger overshoots should be OK.

How are you measuring this waveform? It sounds like you're driving some pretty long pieces of wire to make things so sensitive to external influences, or you're powering the thing with long wire runs without any bypass caps. Have you looked at your VCC line to the chip? Is it bouncing around? Are you clipping your scope negative probe onto the chip's negative lead? Did you check the connection of your scope's negative probe to make sure it's still good?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Reply to
Tim Wescott
Loading thread data ...

Hi, I use these chips, I have had no problem with them. They can inject noise into the supply but a 50nf cap accross the supply pins kills it. Are you sure your using the right frequency? 3.6864 is more usual. If you live in the UK I'll drop a couple in the post for you.

Reply to
cbarn24050

Most of that "ringing" you see is the result of a lack of ground plane and bypassing, I bet.

And you blame the chip? Sheesh.

This is like complaining that my car's transmission must be damaged when in fact the car isn't moving very quick because it's underwater...

Tim.

Reply to
Tim Shoppa

At those frequencies you need to have a SHORT scope ground lead. Like to the chip's ground pin, ground lead no longer than 6 inches.

And what kind of scope probe are you using? It should be a proper scope probe, not just a coax cable with alligator clips.

Reply to
Ancient_Hacker

I bought a few of these SG51K' (3.6264Mhz) clock chips. You know the type ... 4 pins (1=OE, 7=GND, 8=ClkOut, 14=VCC).

Connected it up and the output is GARBAGE ... there is just no better way to describe it.

The output is full of overshoots and undershoots ("ringing"). And is so sensitive that all I have to do is pass my hand over the chip and it warps the output.

I checked for loose connections and what not ... thats not the case.

Unfortunately I need this blasted chip to work (I have 10 of them).

3.6264Mhz can be divided to give me the 9600x16 or 9600x64 I need to drive the 8250 UART i am using in my hobby project.

I see I have a few choices

1) use a 555 timer ... but I have told this isnt precise and the calculations dont take the RC tolerance into mind. But my question is: would this matter much to a UART? Because 555 give me the BEST squarewave output on my scope

2) use a 3.6264 Quartz xtal. Unfortunately I would need to devote considerable time studying feedback theory to be able to properly connect this ... maybe in rev 2 of my board ... but not now

My attempts at using an example I saw online with two inverters in a

74ls04 and a resistor (in parallel with the xtal) didnt produce the results I expected.

3) use the sg51k (confounded) clock chip. Ideally I would prefer to use this ... but its output is garbage ... on the three I tested. And its so sensitive that any device around it ... affects its output.

I am goign to have to read up on feedback/amplifier/oscillator theory sooner or later ...

Reply to
samIam

Figures

The one of two I saw didnt produce the expected output. But I am going to dig into it seriously tonight ...

I am going to get a picture put up later to show you (using 3.6264Mhz xtal).

But heres the output with the sg51k/sg531p clock chip:

formatting link

All I know is that ... the ringing on the outputs of that clock chip isnt allow my system to run ...

Now I have a function generator handy that I can tune to the same frequency and it drives my system without a problem.

Scope ... but I confess my experience is limited.

Well the clock portion of the system has been isolated and built on a solderless breadboard ... theres a few jumper wires about.

But my power lines are stable and I have bypass caps

No. Its pretty clean at 4.92v on my DMM. I also checked on the scope's DC selector ... good powersupply

not exactly ... I am clipping it on the power supply's negative line. Buts its all on a common ground

Yes ...

Reply to
samIam

Ill try that

You are right of course .. its 3.6864Mhz ... NOT 3.6264Mhz glitch in the matrix.

thanks. I am all the way in NY :(

Reply to
samIam

I posted a picture of some useful homemade wire probe holders that get rid of the garbage. Simple to make, just a couple minutes, once you have the proper tools assembled. See (abse) alt.binaries.schematics.electronics subject: Curley Q for noise reduction

Mike

Reply to
amdx

you got me!!! I am using the coax type with red and black alligator leads.

I have the better quality cable with 10x written on it. But I have never used it ... never needed to. and of course the BNC to BNC to test the output of my function generator.

Again I am surprised by all this because the SIMPLE 555 can be wired in mere seconds to provide me with a CLEAN clock (on the same solderless breadboard setup with jumper wires) ... although its not at exactly

3.6864Mhz ...
Reply to
samIam

Tim I cant make the leads too short I need to test this on a solderless breadboard before integrating it into my core system.

Same goes with the bypass caps ... theres just no way to make them shorter on a solderless breadboard.

LOL point taken! I got a lot to learn thats for sure

Reply to
samIam

with it's trimmer adjusted properly to avoid false under/overshoot.

Reply to
budgie

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.