BBD milliseconds/clock hertz

[armdnrdy]

As some of you may know I'm working on a BBD tester design. I thought that I'd report that it's going well and getting close. As it turns out, this is no easy task!

Today I decided to figure out the relationship between BBD millisecond delay and clock frequency so that I may better understand what the LED display will end up reading.

I believe that fender3D stated that the readout will display the amount of stages. (I think he's correct)

The way that I see it, (please correct me if I'm wrong) it goes something like this:

Knowing that 1000 milliseconds equals 1 second, and 1 Hz equals 1 second, and using the MN3005 for example with the spec. sheet stating 204.8 milliseconds delay at 10kHz, I made the following calculations.

1000 ÷ 204.8 equals 4.8828125 (the delay of 204.8 is 4.8828125% of a second)
If the clock is running at 10kHz, and each complete clock cycle shifts the signal twice, then the signal would be pushed through the BBD at 20,000 times a second.
But remember that the delay only lasts 4.8828125% of a second. So armed with that information, we divide the 20,000Hz by 4.8828125.
The answer is 4096. The amount of stages of the MN3005.

This math works for all of the BBDs
MN3004.....1000 ÷ 25.6ms = 39.0625....20,000Hz ÷  39.0625 = 512
MN3007.....1000 ÷ 51.2ms = 19.53125..20,000Hz ÷  19.53125 = 1024
MN3008.....1000 ÷ 102.4ms = 9.765625..20,000Hz ÷ 9.765625 = 2048

This may seem trivial to some of you, but I've never had a reason to figure this out before.  

[PRR]

If a 1000 bucket brigade is passed at 1,000 buckets per hour, it takes one hour to get through.

If a 1000-stage delay is clocked at 1,000 ticks per second, it takes one second to get through.

Except that the electron buckets need two ticks. One to fill and one to empty.

So 1,000 ticks per second makes 500 complete passes per second. Then 1,000 stages clocked at 1,000 ticks per second takes 2 seconds to get through.

> 1000 ÷ 204.8 equals 4.8828125 (the delay of 204.8 is 4.8828125% of a second)

No. Your math is screwy here. Your result is right, but somehow you get it the wrong way.

204.8 milliSeconds is 0.2048 of a second. _Not_ 0.0488.. or 4.8%... that would be 48 milliSeconds.

BTW, using percents in the middle of engineering calculations is IMHO unwise. It begs for factor-of-100 mistakes.

Yes, yes, I know speaker efficiencies (not sensitivity!) can be quoted in Percent. (Not usually where buyers can see it.) Just like transformer efficiency can be cited in percent. The wise computer immediately converts "Efficiency = 1.4%" to 0.014. Then a 10 electrical Watt amp delivers 0.14 Acoustic Watts to the air.

[armdnrdy]

PRR,

You're right about a few things. I definitely made a mistake. The number I derived at by dividing the delay milliseconds by the amount of milliseconds in a second is not the percentage of a second. Wrong term!
But....even though my verbiage may have been off and my use of math may be frowned upon and considered unorthodox in engineering circles, I stand behind my equation as making sense and arriving at the correct answer. I came about the answer from a slightly different perspective. I saw a common denominator....milliseconds to seconds, and hertz to seconds. There are often different paths to arrive at the same destination.

With that being said, I looked through some of my BBD material and found an article that you are more than likely familiar with.

Analog Delay Lines by Ray Marston

http://www.aronnelson.com/gallery/main.php/v/diyuser/Analog+Delay+Lines.jpg.html

In the article he gives an equation for finding the delay time (milliseconds) when the delay stage and the clock frequency is known.

1024 / (2*10,000) =51.2ms

If you reverse the equation you can easily find the stage.

51.2 * (2*10,000) =1024

Good article!

[Fender3D]

I wouldn't rely on time measuring, unless you use a quartz clock generator (and I wouldn't trust it either...)
counting clock pulses (thus counting stages) will be safer.

BTW
You'll have to check what's the best clock freq. (signal detector might fail with a too much aliased signal or counting might be either too fast or too slow for the following counter/divider stages...)

[armdnrdy]

Hi Federico,

Nothing in this design has been set in stone. As I bread board sections, small issues keep arising and the design changes. Often info taken from spec sheets and snippets of circuits work on paper but when coupled with other circuit sections problems occur that need to be addressed for this design to be 100% functional.

I'm waiting for a HUGE bread board to arrive in the mail so I can bread board the whole circuit. I'm very close.

I went back and read a post where you relayed "getting an accurate measurement of the delay from a BBD would be difficult".

You were very correct my friend!!!

[armdnrdy]

I wouldn't rely on time measuring, unless you use a quartz clock generator (and I wouldn't trust it either...)
counting clock pulses (thus counting stages) will be safer.

BTW
You'll have to check what's the best clock freq. (signal detector might fail with a too much aliased signal or counting might be either too fast or too slow for the following counter/divider stages...)

Hey Federico,

I'm using the delay signal when it exits the BBD to stop and hold the counter/display ICs. The counter ICs are driven by one clock line at the start of the test using a momentary push button, 4013 flip flop, 4053, among other things! This circuit is basically like a runner and a glorified stopwatch.

I believe to achieve the BBD stage number on the LED display; I have to incorporate a frequency doubling circuit. (4011, 4069)

The counter/driver ICs I'm using are CD40110. The data sheet states no minimum frequency and maximum 2 MHz @ 10 volts, (I'm using 9 volts) so at 10 kHz it should be fine.

[PRR]

> I wouldn't rely on time measuring,

He's not trying to tell a 1024 from a 1025 or 1026.

VERY approximate "time" is good enough to tell 1024 from 2048.

Recirculating a thump at fixed clock rate, _any_ musician can count the tempo. Math, or comparing to BBDs of known length, tells the size ~~ 256 500 1K 2K 4K 8K.