The repair section is below, this first part covers some of the features of the meter.
About a year ago I picked up a Fluke 8506A on a whim off of Ebay. A 6.5 digit meter selling for less than a hundred dollars, I though why not. I already had A Keithley 2701 and needed a second bench top meter to simplify things when building circuits and trouble shooting but didn't want to spend an arm and a leg to get a newer meter.
Other than knowing that it was a "6.5 digit meter," that presumably did DC and AC volts (and maybe ohms or DC amps if your lucky enough to have one of them installed) I honestly didn't know much of anything about these meters when I bought it. After extensive use and comparison with my Keithley meter, I came to odly like this quirky multimeter. While rather large and slow behind the wheel, Fluke's 8506A DMMs are an interesting mix of 1980s metrology grade and general purpose features.
The DC voltage measurement accepts 100% over-range on the 100mV, 1V and 10V ranges, while the 100V and 1kV ranges are limited to a maximum input of 128V and 1.2kV, respectively. This feature gives a much appreciated resolution increase for readings below 20V, when compared to non over ranging DMMs. Further resolution can be had on the 10V range when averaging is enabled, which extends the 10V range to 7.5 digits. This is one of the reasons I decided to keep the 8506A for my work bench. Another neat feature is that you can independently zero correct each range of the DC volts and ohms functions and even save these corrections if you enable the cal switch on the back of the unit. Sounds boring, but it makes measurements that require changing between ranges more accurate.
Unfortunately, the DC amps, AC Volts and Ohms measurements are all 5.5 digits. However, with averaging enabled they extend to 6.5 digits but this slows the measurement speed down considerably, and this is especially true for hi-accuracy AC volts measurements which are already slow to begin with.
The AC voltage measurement uses thermal converting methods (presumably why it's so slow) and can measure up to 8:1 crest factor signals with full accuracy. 3rd dB bandwidth is excellent for a DMM with 3MHz on the 100mV range and 10MHz for the 300mV, 1V, 3V, and 10V ranges. Which is similar to the HP 3400A analog thermal converting voltmeter I also have on my bench that gives 3 digit readings but is capable of handling much larger crest factors.
Another neat feature is that you can do AC+DC TRMS volts measurements by pressing DC volts and one of the AC volts modes at the same time. My Keithley is only able to do DC or AC volts measurements, thus you would have to manually calculate the AC+DC TRMS value by hand when using it.
If installed, the DC amps module is quite handy with 100uA, 1mA, 10mA, 100mA and 1A ranges. It may not measure AC current or as high of DC current as most modern meters, but with small low power circuits it's more than adequate. I've frequently use the 8506A to measure currents smaller than 100mA, as my Keithley can't go that low.
The Repair:
A few months back I acquired three "dead" Fluke 8506A DMMs at my university's surplus sale and was curious if I could repair them and maybe pass them on to people who would actually use them, rather than let them become landfill fodder. With procrastination in full force, it wasn't until this afternoon that I finally had gotten around to giving them a rigorous tear down.
All three units turned on but gave different errors and symptoms. One read error "E" which means that an unallowed AC volts module is installed (i.e. average/RMS modules from the 8505A, the 8506A's little brother). It turns out the control module mistakenly identified the meter as a 8505A instead of a 8506A. Removing the AC volts module corrected this but then the meter latched on error 5, which corresponds to the recirculating remainder ADC module. Swapping it out with two other units returned the same error, so it's likely something else.
The second unit turned on with no error codes but gave spurious readings on the 1000V and 100V ranges (fluctuating around 600Vdc open circuit and 100V short circuit). The third meter latched on error code 6, which is a display overflow. Turns out meter two and three suffered from the same problem but with varying degrees of severity. Looking at the manual, you would believe that the DC signal conditioner, recirculating remainder DAC, or possibly the active filter were faulty. However, after randomly swapping U22 and U20 (both EEPROMs, but I didn't know that at the time) from meter one into meter three, I was able to get meter three to respond normally (no more overflow or spurious readings). At this point, I used meter three, to systematically test the modules and power supplies from all three units for faults.
Bizarrely enough, not a single module or power supply was determined to be faulty, each one measured and operated as intended. That is, except for the control modules. The control module for meter one was stuck thinking it was an 8505A. *Note that every module is interchangeable/shared between the 8506A and 8505A except for the AC volt modules.* Since it would be time consuming to figure out why this module thinks he's a dude who's playing a dude who thinks he's another dude (Tropic Thunder reference), I decided to work on why module two wasn't working.
Turns out module two is for the CT variant of the 8506A, a military oriented model. Further, it seems this variant is able to measure up to 1000V AC instead of the 500V AC of the non CT models and extends the 100mV AC reading to 6.5 digits when averaging is enabled. This is accompanied by a few changes to the input board, the rear input switch and rear input connector are deleted and in their place, two trimmer caps and associated compensation network were added to the input board that the non CT models don't have. Thus the chassis is slightly modified to accommodate adjusting these new trimmer caps and also means that the controller module for the CT is not compatible with the non CT models. Consequently, I wasn't able to place control module two into meter three for testing (it just latched on an error), instead I had to repopulate meter two.
On a whim I decided to exchange U20 and U22 with control module three (that I originally took from control module one) and like magic meter two was working A-okay!. Upon further inspection, only U20 needed to be exchanged for meter two to work. And with further testing I was able to repeatably recreate the original errors by swapping the faulty U20 and known good U20 into each respective meter with all their original modules installed. Thus it was a control module failure on meters two and three and not a DC signal conditioner problem that I had initially thought.
Now the damnedest thing about all of this is that U20 is a Xicor X2804AP-45 4kbit (512x8) EEPROM, why this would cause the 1000V and 100V ranges to have overflow/spurious readings but not the 100mV-10V ranges is beyond me. Truly some black magic that escapes my understanding. I would think that the meter would latch on an error for the control module, but instead it still turns on and attempts to work with the faulty EEPROM.
At any rate, I hope that anyone who is working on a 8506A or 8505A and observes spurious readings that would imply a bad DC signal conditioner comes across this and tries swapping control modules (if a second unit is available) or swap U20 on the control module before looking elsewhere. While I don't know how often this problem is statistically likely to happen, having got three meters each with faulty control modules seems to imply that the control module is a common weak point, and is likely a good place to look for faults when trouble shooting.
Since I feel reasonably accomplished at the moment, control module one will have to wait until I feel up to burning another day to discover why it has mistaken identity syndrome.
When I get a chance I'll upload pictures of the units and some of the more interesting boards. The DC signal conditioners and DAC's use custom made hermetically sealed resistors and later units like the CT model I have (chip date codes point to 1993 MFD), use laser etched resistors.
Home
Help
Search
Login
Register
