Autonerdz forums

Hi Matt,

The Pico 212/3 has one ADC and a 32k shared memory, so when using both channels, the speed is halved.  This is the unit that comes with the automotive kit and, for the moment, the only one compatable with the Atomotive software package.

The 212/3 can collect more information with a single processor on two channels, at common automotive timebases, than your Tek THS 720A with dual processors.  Your Tek has a 2500 point buffer.  So, 2500 samples on a pass.  The Pico 212/3 would still be able to collect 15,000 samples per channel even divided in half.

Vertically, the Tek is an 8 bit scope compared the Pico at 12 bits.  12 bit has 4096 voltage 'steps' on the screen and 8 bit 256.  So, the Pico can display small voltage changes with more resolution as well.

The Tek has other advantages though like pulse trigger that Pico does not.  It would also blow Pico away at the very fast time bases because it's a 100mhz scope.  We just don't use that speed in automotive application though.

You asked about zooming.  Glad you brought that up.  OK, so say we have collected 15,000 samples per channel.  That's way more than can be displayed on the PC screen.  To reveal detail we can zoom in up to 200x horizontally and scroll through all the data.  4096 is also more vertical than we can display, so we can zoom in vertically up to 20x.

You also asked about a Pico product with dual ADCs.

The ADC 212/50 has dual ADCs and separate memory for each ADC.  This means that in single OR dual channel mode its a 50MS/s 128K scope.  This is more than is needed, but it's sure fun to play with.  More expensive but perhaps a choice for the true power user 

Whoa, good questions, Matt.  Yes, record length is the number of samples that make up a trace.  You calculate the requested sample rate by dividing the number of samples by the time on the screen in miliseconds.  The result is the requested sample rate in KHZ. 

It's not quite that simple though.

There are other factors.  The ADC has available sample rates based on powers of two.  In the case of the 212/3, the fastest level is 3mhz, the next level is 1.5mhz and so on.  When a requested sample rate falls between available sample rates, the next lower speed is used.  We can't overfill the buffer. 

So with one second on the screen you will actually collect 23,437 samples.  Divided between the two channels. 

In the other direction we run into the ADC speed limits.  If we had 1ms/div that would be 32,000/10 for a 3.2mhz requested sample rate.  It doesn't go that fast.
It falls off at the very fast time bases.  With 200us on the screen you would collect 600 samples.

When capturing something as fast as a secondary firing line the 212/3 will begin to break down at 10ms/div.

Let's do the math for the 212/3 at 10ms/div:

32,000 samples / 100ms = 320khz requested sample rate

The scope then chooses 187.5khz since 320khz is between ranges.

187,500 samples x .00001 firing line duration = 1.9 samples during the 10us firing line.

This will show most of the high peaks. Longer time basses will not. You can overcome this by using the mix/max and current selection which will leave shadows of the minimum and maximum values for each trace while showing the current trace as it is updated. Just a few passes and you will reliably see the high spike shadows on the screen.

You have the idea though.  The power is in running long time bases and then zooming in on detail.  That's when Pico runs away from all the rest.  Everything has limits though   

Hi Matt,

Very good questions.  You must have watched that movie a couple of times 

Quote:


I worked the formula the other way.   Using your 200us screen time with the 212/3 st  at 32,000 samples the requested sample rate would be:

32,000/.2ms = 160,000khz or 160 million samples per second

That's how fast the processor would have to be, to fill that buffer.  DOH!  But it can only go to 3 million.  3mhz is 3000khz  So we work the formula with the known max speed and the screen time in ms to find the samples collected:

3000khz x .2ms = 600 samples on the screen

Actual samples is 602, so this gets us very, very close.

Quote:


Yes.  Very good.  At the very fast time bases the processor cannot go fast enough to fill the buffer.

Quote:


Fluke 97 collecting 500 points in .2ms requested sample rate:

500/.2 = 2500khz or 2,500,000 samples/sec

That's between available processor speeds.  Next lower available sample rate is 1,562,500  samples/sec

1562khz x .2 = 312 points on the screen

Pico is still faster.

Your Tek THS 720A (100mhz with 2500 point buffer):

2500/.2 = 12,500khz requested sample rate.  That happens to be one of the available processor speeds so you would fill the 2500 point buffer for 2500 points on the screen.  Kicks butt.     The processor had to be slowed down so as not to over fill the buffer.

BTW.  A note on the Pico ADC 212/50.  I stated that it has dual processors.  That is not correct.  It has a single 100mhz processor.  It's designed to run at 50mhz on one or both channels.

You made my head hurt again.  You know I'm math challenged.   

Hi Matt,

Hope you checked my math 

The 212/50 is one of the most powerful lab scopes ever applied to an automotive use.

Lets do the math for the 212/50 at your 200us screen time:

128,000/.2ms = 640,000 khz requested sample rate

OK, that's way faster than the processor can go so we use the max speed and screen time to get samples:

50,000 x .2 = 10,000 samples on the screen. 

4x faster than the Tek because there was no need to step down the processor to avoid buffer over run.

The 50 minute Flash movie class on the how and why of all this math was produced to empower techs to use the two parameters of buffer size and processor speed to compare DSO performance.  This is good to do before purchase, and also to understand the limits of something you are already using.  It's not always obvious that the tool is beyond it's limits during use and this can result in costly errors.  (don't ask me how I know that)   

Of course, it gets more complex when you consider more than one channel, but focusing on one channel use, when all the scope resources are available, helps compare apples to apples.  Some divide sample rate in half when using both channels, some scopes go into alternate sampling at certain time bases, etc.  We really don't go there in the movie.

In the case of the 212/50, the speed is the same using one or both channels.

The DSO Comparison Movie:

http://www.autonerdz.com/scopecomp.htm

Hi Matt,

Thanks for your support. 

I have created an Excel sheet DSO Sample Rate Calculator.  This will be included with each donation to watch the DSO Sample Rate Comparison movie. 

I sent you one.  Try to break it please

Like I said in the movie, Matt.  There are only two possible reasons why they would not tell you the buffer size.

1.  They don't know
2.  They are too ashamed to tell you

Neither is good. 

But without that information, you cannot compare.  Maybe they don't want you to compare.  I don't know anyone who has gotten a straight answer on that one.
If you can't compare, you have to gamble....

Do Ya Feel Lucky   ???

Hi Matt.

I doubt if you will ever find the specs that you are looking for in regards to the Modis. I have looked and looked to no avail.

The Modis is a really neat scope in it it compensates for any lack of record length by having a real cool feature. That feature is the ability for it to record continuous screens. Then you take the whole recording and zoom it out and see the whole recording if you like. That way you can still sample at the faster timebases but you can see long lengths of recording.

Think of it as the Pico in reverse.

Harvey.

Thanks for stopping by, Harv.  I learn as much from these forums as anyone.

You are right on the Modis.  While it can't hold a candle to the detail and resolution of the Pico, it can capture a longer continuous run of data than the 212/3.  Not as long as the 50 or 100 though.

At least the Modis screens seem to be continuous.  Have you been able to verify that there are no gaps?  I spent a little time with one but not enough to test it thoroughly.

Quote:

Hi Tom,
I have tried many times to find gaps but it seems to be continuous. The backwards zoom is pretty trick and probably needed with this unit as I can make it alias fairly easy. But the bacwards zoom seems to be the power of the unit. High frequency signals can be captured at a faster timebase then zoom out to look at many revolutions of the crank.
By the way, great Forum.