I used a cheap cheap clock

[poynton]

Originally posted by Ruach
Anyone use this clock upgrade?

Hi.

Jean-Paul's comment in the thread probably sums it up nicely :-

Cheap
Looks good
better than 99% of standard clocks


Andy

[pwillard]

George,

As I said, I'm no designer, but I dislike engineering based marketing pizzaz. I'm not saying this kit doesn't work great. It's a good price and it has what it needs... a quality crystal oscillator at a good DAC frequency.

Other people have tried to indicate that cascading regulator chips is bonus... it was TENT, I believe that originated this, but I'm guessing here.

What I see here is a nicely made circuit board, with some major marketing fluff and a tad bit more complicated than it needs to be to make people feel maybe it's somehow different or unique.

Why would you use LM317's, for example.? Lets see, In the Datasheet of the LM317 it reads "They are exceptionally easy to use and require only two external resistors to set the output voltage. Further, both line and load regulation are better than standard fixed regulators."

Well, now, that means this must be a SUPER regulator. Maybe we should realize that this part was designed in the 1970's and maybe, just maybe, a better 3-terminal regulator chip has been designed since then.

The reason I mentioned the Linear Technology part is because it's *is* a newer design and it has excellent specs. The line and load regulation specs for the LM317 and the LT1086 part nearly identical. So why would I prefer a Single regulator? KISS principle.

It's down to one 1 Part versus 6 parts and seems like a bargain to me if you get the same results. Pricewise, the LT1086-5 is $3.00 as compared to $0.75 for an LM317, so this is not about price, but do you really want more parts when less would do just as well?

Now, I'll admit that the benefit of using dual regulators exists. You see, each of the regulators only has to dissipate part of the difference between 12V rail and 5V output, unlike a single part. However, by using a LDO device like the LT1086 part with a 1V Drop Out, you can supply the circuit with 6.3VAC transformer instead of 12V, thus bringing your dissipation way down.

As I said before, the current draw of the circuit is about 100ma, which is well within the tolerances of the regulator and therefore it is not stressed. "Not stressed" means good regulation.


The Crystal Oscillator Can is where the accuracy is. I'm not trying to make trouble here by saying this, but... asking someone to make a custom frequency crystal oscillator package with less than +/- 50 PPM variance is as simple as "asking". Nearly all of the crystal oscillator manufacturers will or do make higher quality units. Citizens, for example, makes such a part for about $3.00.

While I could probably make a clock source myself for about $15 (and I probably will), I'm not disputing that the Cheap Clock Kit for $30 works well or not. I am also not saying that it is not a good bargain. It's super for anyone that doesn't like building from scratch and where things at $149.00 might be a bit too costly.

Me personally, I would initially start by replacing an on-board crystal with just an oscillator can with minimal "glue" components, using the available 5V supply for the existing logic and then compare that to a "kit" solution. We are talking a maximum of 30 additional milliamps here, which is about as much as adding Blue LED.

WHAT ME WORRY?

[fmak]

Now, I'll admit that the benefit of using dual regulators exists. You see, each of the regulators only has to dissipate part of the difference between 12V rail and 5V output, unlike a single part. However, by using a LDO device like the LT1086 part with a 1V Drop Out, you can supply the circuit with 6.3VAC transformer instead of 12V, thus bringing your dissipation way down.

-----------------------------------------------------------------------------------

The LT1086 is one noisy regulator 0.03%V if memory is correct. There are much better and newer LT and other regulators about.

It should not be used as a clock supply; even a selected 7805 with suitable output cap is better for noise. You can get 50uV in a good one; otherwise about 80 uV (1 MHz bandwidth)

[pwillard]

Ok, point taken... I actually haven't measured the noise on a LT1086.