Some old school instruments showing how it's done (HP 3325A and Fluke 8506a)

[SilverSolder]

Just checked what 1V RMS from the 3325A looked like on the Fluke 8506a thermal RMS voltmeter.  No real reason to complain today!  (Other than using up my luck for the rest of the year...)

[shakalnokturn]

Now don't you regret not having a couple of extra digits on the HP3325A's display?

[bob91343]

Very gratifying.

[0culus]

The 3325 A and B don't have all the fancy features of modern DDS generators, but they make very high quality waveforms. I have a 3325B with oven oscillator and high voltage option. Comes in quite handy sometimes!

[SilverSolder]

The 3325 A and B don't have all the fancy features of modern DDS generators, but they make very high quality waveforms. I have a 3325B with oven oscillator and high voltage option. Comes in quite handy sometimes!

Yesterday, I compared the specs with the current generation of function generators from Keysight (the 33500B series).   The new ones have cleaner output and many more features, and are very cool and high performing instruments.  But the fundamental specifications of the old machines are definitely not embarrassing in comparison, even after all these years!        The A/C amplitude spec looks very comparable, for example.

[0culus]

The 3325 A and B don't have all the fancy features of modern DDS generators, but they make very high quality waveforms. I have a 3325B with oven oscillator and high voltage option. Comes in quite handy sometimes!

Yesterday, I compared the specs with the current generation of function generators from Keysight (the 33500B series).   The new ones have cleaner output and many more features, and are very cool and high performing instruments.  But the fundamental specifications of the old machines are definitely not embarrassing in comparison, even after all these years!        The A/C amplitude spec looks very comparable, for example.

I would imagine that the modern ones are way better for anything approaching arbitrary waveforms. But most certainly don't allow for such high amplitudes without an external amplifier! I also have another old unit from Analogic Data Precision, a 2045 Polynomial Waveform Synthesizer. It has a rather unique feature set that I'm not sure has ever been reproduced. It can either reproduce waveforms digitized by a Data Precision 6000 series digitizer mainframe or you can define waveforms in terms of polynomials right from the front panel.

[joeqsmith]

Just checked what 1V RMS from the 3325A looked like on the Fluke 8506a thermal RMS voltmeter.  No real reason to complain today!  (Other than using up my luck for the rest of the year...)

(Attachment Link)

Curious if you still own the old Fluke.  I recently saved one from scrap and would be interested in how stable yours is compared with the one I am looking at.     

[srb1954]

Just checked what 1V RMS from the 3325A looked like on the Fluke 8506a thermal RMS voltmeter.  No real reason to complain today!  (Other than using up my luck for the rest of the year...)

(Attachment Link)

Curious if you still own the old Fluke.  I recently saved one from scrap and would be interested in how stable yours is compared with the one I am looking at.   

I would be interested as well. My 8506A performs pretty within spec for inputs above 10mV RMS but exhibits strange behaviour for inputs below that.

For inputs from 0 to 800uV the display shows a constant 131uV. Increasing the input to 900uV the display suddenly jumps to 1.074mV and stays thereabouts until the input raises above 1.1mV after which the reading increases along with the input but with a significant error of around +10%. As the input voltage is further increased the reading gradually converges to be within spec at around 10mV input. From that point on the reading stays closely in spec until the input reaches 30V RMS where the error starts slowly creeping up again.

I am not sure if the performance I am seeing with the 8506A is normal (due to some sort of auto-zeroing function) or is abnormal. The Fluke manual is not informative as to the expected performance at such low input levels.

I have been using a Fluke 5200A as a signal source with a Gertsch RT-7 ratio transformer to accurately divide down the lower level signals.

[SilverSolder]

Just checked what 1V RMS from the 3325A looked like on the Fluke 8506a thermal RMS voltmeter.  No real reason to complain today!  (Other than using up my luck for the rest of the year...)

(Attachment Link)

Curious if you still own the old Fluke.  I recently saved one from scrap and would be interested in how stable yours is compared with the one I am looking at.   

What do you mean by "stable"?   The zero stability is specced for 5uV / 90 days, which my examples easily beat. 

The temperature coefficient spec is 2.5 counts per degree C around room temperature on the 10V DC scale.

[...]
I would be interested as well. My 8506A performs pretty within spec for inputs above 10mV RMS but exhibits strange behaviour for inputs below that.

For inputs from 0 to 800uV the display shows a constant 131uV. Increasing the input to 900uV the display suddenly jumps to 1.074mV and stays thereabouts until the input raises above 1.1mV after which the reading increases along with the input but with a significant error of around +10%. As the input voltage is further increased the reading gradually converges to be within spec at around 10mV input. From that point on the reading stays closely in spec until the input reaches 30V RMS where the error starts slowly creeping up again.

I am not sure if the performance I am seeing with the 8506A is normal (due to some sort of auto-zeroing function) or is abnormal. The Fluke manual is not informative as to the expected performance at such low input levels.

I have been using a Fluke 5200A as a signal source with a Gertsch RT-7 ratio transformer to accurately divide down the lower level signals.

Mine behave in a similar way.

In the manual, the Minimum Specified Level for the 100mV AC range is 12.5mV.   Accuracy falls off a cliff when you go below that - reach for an amplifier or another instrument for lower voltage measurements!

Note, the other ranges also have Minimum Specified Levels and they are quite high!  e.g. 1.25V on the 10V AC range.

[joeqsmith]

What do you mean by "stable"?   The zero stability is specced for 5uV / 90 days, which my examples easily beat. 

Do you have the resistance feature?

[SilverSolder]

What do you mean by "stable"?   The zero stability is specced for 5uV / 90 days, which my examples easily beat. 

Do you have the resistance feature?

Yep.

[joeqsmith]

What do you mean by "stable"?   The zero stability is specced for 5uV / 90 days, which my examples easily beat. 

Do you have the resistance feature?

Yep.

Great.  Using an economy grade 1 ohm 4-W standard with guard attached and averaging enabled,  this meter will wander around easily +/- 0.0005.  The manual for the meter that I have doesn't seem to cover the resistance.   I'm sure some of it is temperature but it changes faster than the room temperature and I suspect most of it is just noise.   I'm curious how another meter would behave. 

If you have other standards available, I could try to compare mine with what you have.   

[SilverSolder]

What do you mean by "stable"?   The zero stability is specced for 5uV / 90 days, which my examples easily beat. 

Do you have the resistance feature?

Yep.

Great.  Using an economy grade 1 ohm 4-W standard with guard attached and averaging enabled,  this meter will wander around easily +/- 0.0005.  The manual for the meter that I have doesn't seem to cover the resistance.   I'm sure some of it is temperature but it changes faster than the room temperature and I suspect most of it is just noise.   I'm curious how another meter would behave. 

If you have other standards available, I could try to compare mine with what you have.

Let me see if I can find something to test.

You can always reduce noise by increasing the number of samples per reading -  E.g. press  STORE - 11 - SAMPLE   

This will set the number of samples to 2^11  (you can go as high as 2^17 if you have all day) -  default is 2^7, I believe.

It is similar to NPLC in other meters, but I believe these things take 4 readings per mains cycle so 2^11 (2048 samples) is done in 8.5 seconds, which is only 510 PLCs.

[srb1954]

[...]
I would be interested as well. My 8506A performs pretty within spec for inputs above 10mV RMS but exhibits strange behaviour for inputs below that.

For inputs from 0 to 800uV the display shows a constant 131uV. Increasing the input to 900uV the display suddenly jumps to 1.074mV and stays thereabouts until the input raises above 1.1mV after which the reading increases along with the input but with a significant error of around +10%. As the input voltage is further increased the reading gradually converges to be within spec at around 10mV input. From that point on the reading stays closely in spec until the input reaches 30V RMS where the error starts slowly creeping up again.

I am not sure if the performance I am seeing with the 8506A is normal (due to some sort of auto-zeroing function) or is abnormal. The Fluke manual is not informative as to the expected performance at such low input levels.

I have been using a Fluke 5200A as a signal source with a Gertsch RT-7 ratio transformer to accurately divide down the lower level signals.

Mine behave in a similar way.

In the manual, the Minimum Specified Level for the 100mV AC range is 12.5mV.   Accuracy falls off a cliff when you go below that - reach for an amplifier or another instrument for lower voltage measurements!

Note, the other ranges also have Minimum Specified Levels and they are quite high!  e.g. 1.25V on the 10V AC range.

Thanks. This is good to know.

I got this 8506A relatively cheap because the vendor said the instrument wouldn't zero and I was concerned that I would have to go through the incredibly complex calibration process to eliminate something that might or might not be a fault.

I have several other DMMs that perform better than the 8506A at very low levels but none can match the excellent accuracy of the 8506A over its mid-voltage ranges and over wider frequency ranges.

[SilverSolder]

I have four of these units with the Ohms option.  (Two are 8505A, two are 8506A, but everything related to Ohms is the same between the two models.)

I tested all four in parallel by hooking up radial wirewound resistors on all four units, 5 ohm and  0.25%, like this:




It seems very thermally stable, I could blow on the resistors and the readings did not deviate.





Here are the results:

2 hour warm-up.  No averaging, all default settings.  Peak mode turned on at start of test period to measure deviation.

4 wire Ohms mode.

Test 1: Peak to peak deviation in 10 minutes:
Unit 1 -  0.0005 ohms
Unit 2 -  0.0005 ohms
Unit 3 -  0.0001 ohms
Unit 4 -  0.0033 ohms

Test 2: Peak to peak deviation in 10 minutes:
Unit 1 -  0.0007 ohms
Unit 2 -  0.0005 ohms
Unit 3 -  0.0001 ohms
Unit 4 -  0.0046 ohms


Conclusion - Unit 3 is almost perfect, unit 4 is broken, while units 1 and 2 are probably "normal".


I have read somewhere (perhaps on the EEVblog) that the current generator for the 4 wire Ohms can become noisy on these meters.  Maybe that is what is going on with my Unit 4.   Perhaps the time has come to look at how that circuit works, and see if it can be made nice and quiet...

[SilverSolder]

Just checked what 1V RMS from the 3325A looked like on the Fluke 8506a thermal RMS voltmeter.  No real reason to complain today!  (Other than using up my luck for the rest of the year...)

How do you know they are not both off track, but with the same amount?   

I very specifically did not say that the signal was 1V RMS, for exactly that reason! 

Both instruments are good, but neither of these instruments can be trusted to that many digits of precision.

[joeqsmith]

Thank you very much for checking this.   If you do decide to try and improve #4, please make a thread for it.  I would like to follow along.

***
People who complain about changing fuses on meters have never had one of these apart.   Crazy 4 HRC fuses and GDT buried deep in the bowels. 

[joeqsmith]

The meter appears in fairly decent shape mechanically for the age.   Trying a few different measurements, it's not even close.   Shown with 1V from the Fluke 731B and 1K ohm 0.1% resistor attached.  It looks like it would need a full alignment to be useful.   

Have you had to send one out for alignment?  Curious who you used and the price. 

[SilverSolder]

It looks in good shape, even has the bails!

Did you set the DC zero?  Shorting bar, wait a few seconds for the reading to stabilize, and press Zero (the Zero LED should be lit).   The zero drifts a lot with time and temperature and needs to be set every time you are looking for max accuracy.

Once the zero is set, it can drift up and down by a few counts even on a good unit.  If it can't hold zero within about 5 counts, the temperature in the area may not be stable (or there is a problem with the unit).

I've never sent one out to be calibrated, but I have performed both the hardware and software calibration methods on some of them, using a 10V reference.  It isn't that hard to do (especially software cal, no need to even open it up), the manual is very comprehensive.  (I only did cals on some ranges, as I don't have the references needed for all the ranges...)

[guenthert]

     I bought a few older DMMs and all drifted less (or rather the accumulated drift was less) than the original specification would have allowed for (one HP 34401A even seems to have been "spot-on" in the 10DCV range).  That might have been survivor's bias or an indication that the manufacturers specification are wider than necessary for instruments living an easy life.
     In any case, I haven't seen one, which drifted so far as yours.  I'd watch it for a while and observe the current drift, which might indicate some fault needing a repair before calibration.

[joeqsmith]

Looks like pretty much every range is out for every function.  I checked the main supply and started to read the calibration adjustments section.   First step, install PN# MIS-7013K Bus Interconnect and Monitor Board.   They offered an extender board as well which would be handy about now.   

If I power up the meter, it defaults to DCV.   If I  then short all four connectors and select zero, it the Zero LED becomes active and it displays 0.000.   So it can hold within a 1mV for several minutes.  If I select the auto range, it looks like it will drift around by 5uV  (0.0050 -3) or so.  If I enable the average and filter, it seem to hold within 1uV (saw a 700nV).   

Having some software to store the min/max.... would also be helpful about now.  Does it have the ability to record basic metrics like this?

Did you write the software to run it through the software cal or is this something you found from Fluke?

[SilverSolder]

Each range has its own zero - you need to set the range of interest manually, then set the zero. 

Min/Max is the PEAK button.  Press it, and it starts storing min/max from that point.

To see the values, press buttons RECALL-HI-PEAK and RECALL-LO-PEAK respectively.


The software cal is described in the manual, you don't need any additional software, just some reference voltages - the values are not important as long as you know what the values are!

The references just have to be anywhere from 60% of the top of the range or above, in order to do a calibration.   So if you have a 7.007365 volt reference, for example, you can use it directly to calibrate the 10V range without amplifying it up to 10V or anything like that....  pretty cool, as you can use the most stable "naked" LTZ1000 circuit (without 10 volt up-conversion) to calibrate with.

The calibration is done by throwing the CAL switch on the back (note: lots of dangers as protection circuits are disabled), then set a permanent zero and enter the reference voltage for each range.  RTFM!

Hardware cal is more work, but not too bad.

[SilverSolder]

     I bought a few older DMMs and all drifted less (or rather the accumulated drift was less) than the original specification would have allowed for (one HP 34401A even seems to have been "spot-on" in the 10DV range).  That might have been survivor's bias or an indication that the manufacturers specification are wider than necessary for instruments living an easy life.
     In any case, I haven't seen one, which drifted so far as yours.  I'd watch it for a while and observe the current drift, which might indicate some fault needing a repair before calibration.

All of the ones that I have were bought very cheaply - as junk, basically... -  and they were all still within spec, unless they were outright broken.  When fixed, they came back to spec.

[joeqsmith]

RTFM!

No problem.

[SilverSolder]

RTFM!

No problem.

I meant in the sense that it is possible to damage the meter in CAL mode, the manual explains how to avoid that.

I've never seen the printed version of the manual, BTW.  It does look rather intimidating on the picture...  I have it in PDF form (easier to search, and add notes).  Got it from Artek.