Scope attenuators 1,2,5, why bother?

There was never any need: 360 degrees == 400 grads. ;-)

Cheers! Rich

What grads? College grads? ;-)

Hey, this is mutating into a serious case of feature creep ;-)

Some days you'll be celebrating your 95th birthday and it still ain't done. Oh, by the way, any chance to integrate a spectrum analyzer?

On a sunny day (Fri, 18 Jan 2008 00:07:43 GMT) it happened Joerg wrote in :

As to the spectrum analyser, well, if it was a 'PC scope' it is easy in software (you have a simple one here)

The issue is that in the PIC case I will have to program in ASM (PIC ASM at that), and the code space is now full :-) That leaves a bigger PIC, or an ARM, or all that + FPGA or FPGA with processor core. It is not a hurry project, there is one being done before this, good, that allows me to work out the hardware in detail.

software

that),

But I don't need it to 16kHz, I need it to 1GHz ...

I've yet to see any cheap 'PC scopes' that anywhere *near* the dynamic range of a decent 20-year-old analog spectrum analyzer. Granted, not every problem needs 100+dB dynamic range, but it sure is nice when you have it. :-)

We have one of those spendy Agilent EXA spectrum analyzers with something like >120dB dynamic range with "reasonably small" bandwidths... it's really quite impressive. The best ADC-based spec an we have is a Tek RSA series unit that only gets you ~70dB dynamic range, in many typical usage cases... although of case the fact that it's looking at the entire IF bandwidth at once is pretty darned cool and lets it do many things a traditional spec an never can.

On a sunny day (Fri, 18 Jan 2008 09:37:25 -0800) it happened Joerg wrote in :

software

that),

Well, it is going a bit of topic, but anyways, 'heterodyne' no?

The limitation is really this: with 16k samples / second the max bandwidth you can unravel would be about 8 kHz.

So we simply mix down your 1GHz with a dual gate MOSFET, a varicap tuned VCO, to say a few kHz (old TV tuner with some modifications). IIRC that was an old amateur radio project, spectrum analyser made from TV tuner.

And the mixing saves you a 1G samples / sec ADC, MOSFET costs 50 cents or less.

What doya sink?

software

that),

tuner.

I've seen those but TV tuners are IMHO the pits. You'd really need triple-conversion and much better linearity.

Methinks it could be done with some BF998, synthesizer chip and some filters if it has to be cheap. Something like TV-audio carrier filter ->

I've used a R&S portable lately. Nice, but expensive and it did fall off the cliff a few times since this EMI work had to happen right next to a busy runway.

The ones that come closest are from Aaronia but they don't make one that's suitable for pre-compliance work. Then there is a nice Thurlby-Thandar unit (Newark has it) but it's PDA based and to me that has a taste of yesteryear.

I have used software that does this, and it is way beyond confusing, more like tantrum-inducing!

Cheers Terry

piece

I am planning.

Think of the 5, 50, etc. as 1/2 (times 10, 100, etc.). Now even you binary buffs can be happy. :-)

Mark

A4 piece

scope I am planning.

Yep, but forget binary. 2, 5 (because it's half of 10), and 10 are easy to multiply without using a calculator, something kids likely can't get "their head around".

Aww come on, 0.00101 * 1.01 = 0.0011001. What could be easier? ;-)

Tim

-- Deep Fryer: A very philosophical monk. Website @

Grrr! I still can't find a binary calculator that knows what a binary point is or where to put one.

A4 piece

scope I am planning.

The PIC can be programmed in C quite easily too, didn't you know? Only masochists write entire programs in assembler these days.

Dave.

A4 piece

scope I am planning.

You'll get no argument from me. And multiplying & dividing by 3 is not that much harder. Most multiplications (in the head) then become some combination of operations with 2, 3 and 10. It might take a few steps but definitely doable.

Mark

EasyCalc for PalmOS handles arguments and results with a radix point for bin, octal, and hex. One small infelicity is that to enter a radix point while in "base" mode you need to switch to the standard keypad and then back.

It's GPL, available at

On a similar subject, I wanted to know what numbers look like in weird bases like base 5 or base pi, so I wrote this:

-=-=-

DECLARE FUNCTION Character$ (Value%)

CONST Pi# = 3.141592653589793# CONST e# = 2.718281828459045#

' Obtain inputs INPUT "Enter number to convert: ", Number# BadBase: INPUT "Enter base to convert from decimal to (enter Pi or e for the constants, or sqr to enter a root): ", BaseString$ BaseString$ = UCASE$(BaseString$) IF BaseString$ = "PI" THEN BaseNumber# = Pi# ELSEIF BaseString$ = "E" THEN BaseNumber# = e# ELSEIF BaseString$ = "SQR" THEN INPUT "Enter number to take the square root of: ", Root# BaseNumber# = SQR(ABS(Root#)) 'Neener, no imaginary base for you. ELSE BaseNumber# = VAL(BaseString$) END IF IF BaseNumber# = 62 THEN PRINT "Base must be lesser than or equal to 62.": GOTO BadBase

' Find the largest power in the input LargestPower% = FIX(LOG(Number#) / LOG(BaseNumber#)) IF Number# < 1 THEN LargestPower% = LargestPower% - 1

' Double precision is only 16 digits, so it makes sense to carry this ' to only as many decimals, plus room for an exponent. DIM Result AS STRING * 30 FOR i% = 0 TO 15 ' Pare down the input iteratively, counting the multiple each time ' and appending (or insert) it to the output string. TestPower# = BaseNumber# ^ (LargestPower% - i%) j% = 0 WHILE Number# - TestPower# >= 0 Number# = Number# - TestPower# j% = j% + 1 WEND IF i% = 0 THEN ' Set the decimal point in scientific notation. MID$(Result, i% + 1, 2) = Character$(j%) + "." ELSE MID$(Result, i% + 2, 1) = Character$(j%) END IF NEXT IF Decimal = 0 THEN MID$(Result, i% + 1) = "E" + LTRIM$(RTRIM$(STR$(LargestPower%))) PRINT Result

' Returns an ASCII character corresponding to Value. ' 0 through 9: as usual. ' 10 through 35: capital letters (the first five, having values 10, 11, ' 12, 13, 14 and 15 come through as A, B, C, D, E and F for hexadecimal; ' the rest of the alphabet follows similarly). ' 36 through 61: running out of choices, so going with lower case letters. ' I don't know of any standard numerals for values this large. FUNCTION Character$ (Value%) SELECT CASE Value% CASE 0 TO 9 Character$ = LTRIM$(RTRIM$(STR$(Value%))) CASE 10 TO 35 ' Capital letters, starting at index 65 (10 + 55) Character$ = CHR$(Value% + 55) CASE 36 TO 61 ' Lowercase letters, starting at index 97 (36 + 61) Character$ = CHR$(Value% + 61) CASE ELSE ' Should never come up, but just in case. Character$ = CHR$(32) END SELECT END FUNCTION

-=-=-

Probably quite slow (precision exponents? eww...), but it's not much and probably completes in a fraction of a milisecond, so who cares.

Come to think of it, the real datatypes are binary, so you could also print the first 48 (IIRC) bits of the DOUBLE to get its representation in binary, plus whatever exponent in the following word.

Tim

-- Deep Fryer: A very philosophical monk. Website @

Thanks. Hmm, looks nice but it's algebraic. Ouch. Were it RPN I'd buy something Palm to go with it. As is, I'll have to think hard. My other option was to learn how to write an Excel function. The drawback there is that I don't have an Excel license (that still works, anyway) and OO Calc is too different.