HP 3581 Wave Analyzer (and PC!)

[jackinnj]

For anyone interested, I posted most of the operational instructions for the HP 3581 a/c wave analyzer at
www.tech-diy.com/hp3581.htm Not the prettiest website, but most of it's there, at least the stuff you need to run the unit.

For those of you who are "scopeless", I have a Basic Stamp II routine which takes the 5 Volt X and Y outputs of the HP3581 and converts it to display on a computer screen via the RS2332 port. When I get time I'll put the code up on my website. Thus, the HP3581 becomes a very nice, high quality audio spectrum analyzer for around $100 as the unit has a tracking generator output (thus you don't need a sweep random noise generator.)

[tortello]

Well done jackinnj!
Your stuff looks very interesting.
Marcello

[jackinnj]

pretty interesting. we put the tracking generator on single frequency and measured the THD at various places -- around 0.02% -- not bad when you consider that it is a 1.0 to 1.5MHz sweep generator which is divided down by 10 and then mixed. Certainly good enough for testing the response of speakers, filters etc.

one suggestion I picked up from a ham operator is to remove the battery pack (if your unit came with one.) the battery was a $600 option when the units were new, but some were prone to leaking. the utility of the battery lies in the ability to completely isolate the HP3581 from ground loop. there is no reason that the battery pack can't be rebuilt with new NiCads, but it does add a lot of weight to the unit.

[jackinnj]

So if you get an HP 3581 here's a little Visual Basic/Softwire program module which will allow you to hook up the XY and Pen-Lift outputs on the back to a Measurement Computing USB Data Acquisition Card -- the data is stored in an array and passed to the XY Graph, when you are done you can dump the data into Excel.

[jackinnj]

I found 20 NiCad D-Cells for my HP 3581 and just wired them in -- the noise floor goes from 8.2nV/SQRT Hz (on A.C. Line) to 7.0nV/SQRT Hz with the battery pack.

[nimrod99]

We're using one of those old HP 3581A wave analyzers, mainly because of the sweep functionality they have.

My question is, isn't there some new (supported) fancy gadget like a techtronix super scope or something that can do the same things?
It'd be nice to have the same analog output via a BNC cable, but I can
deal without that if it can stream test results over a network and post process for a "near" real time test result.

Anyone ever get something like that?

[Conrad Hoffman]

I'm pretty sure the HP3581 selective voltmeter is identical to the HP low frequency spectrum analyzer of the same era. The only difference is it has a meter instead of a CRT, making it a lot cheaper. Displayed on a PC or a scope, it's a capable spectrum analyzer, but the big advantage of the PC is that you can see the rather slow sweeps necessary for taking advantage of the narrow bandwidth settings.

CH

[cliffyk]

The missing ingredient on most "affordable" digital instruments is the 3581A's 90+db dynamic range (properly adjusted the noise floor is -95db), that's why I hang on to mine (and a couple parts machines). I capture the output in two ways:

• Via feeding the Y output to a digital scope, matching up the sweep times, and using the external trigger option¹ to simultaneously trigger both the 'scope and the 3581A;
  
  These are captures of wave analysis of the 1kHz outputs of a Leader 1300s function generator, and a Kikusui ORC-11 audio generator.
  
  
  
  The vertical scale is 20dB/div (rescaled from the 0-5V output of the 3581A), the horizontal 1kHz/div (by setting the 3581A to a 200s 10kHz span, and the 'scope to 20s/div).
  
  
• Using a 10-bit ADC card to capture both the X and Y outputs and display same using a Windows application of my own design;
  
  Here is a capture of the Leader 1300s output using the 10-bit ADC (an AWC-GP3) and my app:
  
  

One comment: I had to fine tune the timing resistor (R45 on the A3 board) to get the sweep time sufficiently accurate to track the 'scope's sweep. The 3581A's sweep spec of +/- 5% did not cut it as compared to an average scope's horizontal accuracy of 10ppm (0.001%).

By tweaking--through trial and error--the value of R45 (it's mounted on posts, I used paralleled resistors to get the final value) I got the wave analyzer's sweep to +/- 0.01%--close enough.

-------------------------------------------------
¹ I used a simple circuit built in to a BNC connector to simultaneously trigger the 3581A and the scope:



It's just a 50Ω resistor from the center conductor in series with a NC momentary switch, to ground. This pulls the trigger input low (the specs say anything < 470Ω) inhibiting the sweep. Pressing the switch lets the line go high (5.0V) and the sweep begins.

Connecting this to a BNC T let's you trigger both instruments.

[jackinnj]

Originally Posted by cliffyk

It's just a 50Ω resistor from the center conductor in series with a NC momentary switch, to ground. This pulls the trigger input low (the specs say anything < 470Ω) inhibiting the sweep. Pressing the switch lets the line go high (5.0V) and the sweep begins.

Connecting this to a BNC T let's you trigger both instruments.

Necessity is the mother of invention. Mother is the invention of all necessity.

[ymwong]

Hi,

The original link to the operational instructions is broken. Would appreciate if it can be fixed.

thanks,
ym