Strange behavior while calibrating a Fluke 8000A

[NoisyBoy]

Back in my college days, the Fluke 8000A had been one of the older bench meter I used in our lab.  I picked one up about 10 years ago and it had been in storage for the last eight years, in organizing my home test bench, I decided to bring it out of retirement.  Everything worked perfectly, so I decided to calibrate it for the first time.  I found it was mostly within spec and only very minor adjustment were needed - with the exception of AC V, which was out of spec.  I inspected all the caps, there was no leakage or bulging, so I decided to leave the hardware alone for now.

As one of the first step in the calibration, the Period Adjustment does not work as described in the manual.  I used a pair of freshly NIST calibrated and adjusted frequency counters to measure the period between TP4 and TP5, both counters came up with a reading around 154ms, the trim pot for period (R20) had very little effect on changing it, and it is far from the 100ms stated in the calibration procedure.  The same observation had been reported in a couple other older posts on this forum, so I am not alone.

On to measuring AC V, it was about 2-3 % high.  The calibration procedure suggested changing "HF ADJ" (C3), again, its effect on the readout was minimal and I could not bring it within spec. 

So I decided to experiment with R20 in AC V calibration (since it was ineffective in changing the period anyway), interestingly, I was able to bring AC V measurement to be spot on (where changing C3 had been a futile attempt).  I tried two AC source, one at 20kHz (as specified by the manual), the other at 60Hz, they both read spot on now. 

DC V and resistance calibration worked as described in the manual, I was able to bring them spot on with minute changes on the corresponding pots.  So I was very happy about how well it has maintained its accuracy as it has not been calibrated for a minimum of 10 years.

Has anyone had calibration experience with the 8000A?  It is ancient, and there is little information about its calibration other than the manual.  If you have done it, did you observe similar behavior on AC V calibration?

I have another 8000A with the battery option packed away in storage.  It never powered up when I got it, when I am able to dig it up, it will be my next repair project.

It is nice to get some red LED digital readout on the bench, a good reminder down the memory lane of my college days.

[windsmurf]

Dr. Joe Q. Smith is our foremost expert and ardent fan of the Fluke 8000A, and can build the whole thing from discrete transistors from memory if he had enough of them.
Perhaps he can weigh in on the beauty that is the Fluke 8000A.

[med6753]

I have experience with the 8000A although not with the ACV calibration since I don't have the proper standards to perform it. My experience comes from drift issues on DCV/Ohms due to the lousy power supply design. For some reason Fluke decided that a regulated 5VDC for the digital circuits wasn't necessary. I found with my particular 8000A that it had what I considered unacceptable levels of drift until I rebuilt the supply and included a regulated 5VDC. I also rebuilt the +/- 15VDC with modern regulators and eliminated the zener/pass transistor circuit.

Here is the original supply and the modified supply. The biggest issue was the original power transformer did not have enough headroom voltage to allow the 7805 to regulate. So an additional small transformer was installed in the case. This rebuilt supply made all the difference and cut the drift to almost nothing.



 

[coromonadalix]

Normally on the old zenner transistor psu's,  you add an capacitor in parallel of the zenner to help with regulation and noise.

And yes  for the 7805  you need at least 2-3 volts input higher than your output voltage for load and line regulation.

congrats for your mods   

[GregDunn]

Some good advice here!  I endorse med's power supply replacement.

I would add: pull the batteries out of that spare 8000 and throw them in the recycle bin immediately.     They probably have degraded, which may be why it won't power up; and will eventually (or may already have) leak and corrode the mainboard of the meter.  I have an 8600A option 01 (the battery version) which I fortunately got to before it self-destructed, and replacing the PSU brought it new life.  It is now essentially as good as the standard AC-only version in terms of stability.

[NoisyBoy]

Hey everyone, thanks for all your kind advices.  I went back over the 8000A today, and gave it plenty of time to warm up before a double check, so far everyone is in great working order.

Good point about the battery, I have no idea which box it is in storage, so it may be another few months before I can get to it.  But I will take the other unit apart and see if it can be repaired, otherwise it will just serve as the donor DMM for the working unit.

With regard to the power supply, I think I will leave it as is for now, as this is more of a nostalgic piece of equipment than something I rely on.  I have a couple Agilent/Keysight DMMs that I use for more precise work at home, those went back to the mothership to get annual calibration, I trust their reading a lot more than the 8000A.  But when my boys get old enough to experiment with electronics, they may find the old school 8000A interesting. 

When I started my EE in undergrad, most classmates were using the HP 3456A and 3478A.  The 8000A typically just sat around with zero wait, I ended up going to it whenever I need to measure something quick and if there's a line for the HP DMMs.  It is nice to bring one back to full working order, along with all its shortcomings. 

[joeqsmith]

Hey everyone, thanks for all your kind advices.  I went back over the 8000A today, and gave it plenty of time to warm up before a double check, so far everyone is in great working order.

.... this is more of a nostalgic piece of equipment than something I rely on.   

... It is nice to bring one back to full working order, along with all its shortcomings. 

Mine is not in the best of shape and certainly has little collectible value to anyone but me.   If it was not my first digital meter, I would have recycled it years ago.   After my recent sadistic relationship with it,  I decided to rebuild it.     

https://youtu.be/ObKomuLLqU8

[NoisyBoy]

Hi Joe, come to think of it, other than amongst one of the first DMM I used in college, the 8000A was also the first digital DMM I actually own (about 10 years ago).  Nice to see someone who share the same love for this old meter, there is nothing that speaks vintage equipment louder than full mechanical push buttons.

Have you had experience in playing with the period and HF ADJ calibration?  Did you observe the same thing I did?  I am leaning towards there is an error in the manual, which is hard to believe given it is an old Fluke.

[joeqsmith]

Hi Joe, come to think of it, other than amongst one of the first DMM I used in college, the 8000A was also the first digital DMM I actually own (about 10 years ago).  Nice to see someone who share the same love for this old meter, there is nothing that speaks vintage equipment louder than full mechanical push buttons.

I wouldn't say I have a love for the meter.  After it was damaged a second time and I could no longer obtain parts from Fluke to service it, I would say if anything I developed antipathy towards it and the company who produced it.  Mainly I know what damaged it and couldn't believe that a meter sold to the military would be such a princess.   I vowed never to buy another Fluke product.   

Then I conducted a pretty big test of several handheld meters to see how electrically robust they were.   Several of these were Fluke branded.  Some from Phillips, some Fluke, some from the USA and other from China.   I pushed all of the meters to their failure point using low energy transients, similar to what had damaged my 8000A.   Because of my disdain for Fluke,  I subjected the 8000A to further transients causing additional damage. 

I ran these tests for myself but made some videos for the people who wanted to follow along.   In the end I had to admit that Fluke had learned some very important lessons on how to design a robust meter.   They are no longer the little princesses they were. 

So lets say after all the abuse, it was time for some redemption. 

Have you had experience in playing with the period and HF ADJ calibration?  Did you observe the same thing I did?  I am leaning towards there is an error in the manual, which is hard to believe given it is an old Fluke.

Sure, I have some experience with it.   Not looking over your shoulder when you were working, I did not observe what you did.   That may sound obvious but I really have no idea what equipment you used, how you had it setup.

As one of the first step in the calibration, the Period Adjustment does not work as described in the manual.  I used a pair of freshly NIST calibrated and adjusted frequency counters to measure the period between TP4 and TP5, both counters came up with a reading around 154ms, the trim pot for period (R20) had very little effect on changing it, and it is far from the 100ms stated in the calibration procedure.  The same observation had been reported in a couple other older posts on this forum, so I am not alone.

Well, it's the internet and I can imagine the skill level of people working on them varies a fair amount.    What I can do is show you some data off of my 8000A.  From this, maybe you will see the problem.  I suspect you didn't follow the procedures. 

Let's start with R20.   From the manual, they state:

4-26.  Use the following procedure to adjust the 8000A measurement period:
a.  Connect the frequency counter between TP5 and TP4(common) and shown in Figure 4-1.   
b. Set the frequency counter to the time interval operating mode.
c. Using the appropriate adjusting tool, adjust R20 (PERIOD), shown in Figure 4-1 for a positive waveform portion of 100 ms +/- 5us.  Variations  of the time period should be <= +/-15us.

Seems clear to me but I can see someone jumping to the conclusion that they want you to measure the reciprocal of the frequency but that is not what they are saying.   

LeCroy149:   Showing the waveform in question.  The DSO is NOT the right tool to use and the manual is clear about that but I have shown it to make it clear what they have asked you to measure.  The "positive waveform portion of 100 ms +/- 5us",  not the period.

PIC1:  The manual is clear about using a frequency counter.  I have an old HP5328A that supports making this measurement.  Basically using two channels, one set to trigger in the rising edge, the other on the falling and measuring the time between these two events.  I have a GPS that is used to derive the reference clock for my equipment.  The timing error for what we are doing is negligible.

The procedure wants you to make sure the "Variations  of the time period should be <= +/-15us."  Here you can watch the counter to see the how stable my 8000A is.

https://youtu.be/qSqEX3nuF7c

Now, I suspect that should answer your question as I assume did not make the measurement  correctly but if not, lets move on.

On to measuring AC V, it was about 2-3 % high.  The calibration procedure suggested changing "HF ADJ" (C3), again, its effect on the readout was minimal and I could not bring it within spec. 

So I decided to experiment with R20 in AC V calibration (since it was ineffective in changing the period anyway), interestingly, I was able to bring AC V measurement to be spot on (where changing C3 had been a futile attempt).  I tried two AC source, one at 20kHz (as specified by the manual), the other at 60Hz, they both read spot on now. 

PIC2:   The procedure wants 19.0V at 20KHz to set C3.   In order to obtain the waveform, I use my Arb to drive a small signal step up transformer and my bench meter to monitor it's output.   

PIC3:   Here you can see my 8000A measuring the waveform.  The manual states adjust C3 for a display of 18.79 to 19.21.   

[NoisyBoy]

Hey Joe,

A big and sincerest thank you for your kind response, that is exactly what I am looking for, and I made the exact mistake you suspected. 

On my manual, under 4-26, section c, it simply state "Using an appropriate adjustment tool, adjust R20 (Period), shown in Fig 4-1, for a time period of 100 ms +/- 5 micro sec." There was no mention of only measuring the period of the positive waveform portion. 

When I went to Fluke's website and searched for the 8000A manual this evening, I noticed the updated version has changed to only measuring the +ve portion of the signal. 

Its time to go back to the bench to open it up and set up the counter with the right trigger.  It would also be interesting to see the waveform between TP4 and TP5 on the scope as you did as well.

It is the first time I see such detail info shared on the 8000A calibration, I hope it will also benefit all other 8000A owners who will be working on their meters. 

I think you are right, love is perhaps not the right term for the 8000A, but it certainly is nostalgic.  I don't do any high voltage work, that's why I have not run into the challenge you had with the meter - yet. 

Thank you so much.

[joeqsmith]

I went to BAMA and downloaded the manual they have and it is as you state.  I have attached the two sections.   Note that the manual that was supplied with my meter is Revision 1.  I can imagine this led to a few call early on.   Personally, I would have added a picture of the waveform to make it very clear.   

It's interesting to look at the note on the cover of the manual available from Fluke's site.   The claim is the manual has not been edited or updated since the revision date shown on the lower left hand corner of the first page.   

I never though I did much with high voltage as well.  Maybe 600V on old tube sets.  I never considered the act of sitting in the chair could create 10X more than that.   Not only has running all these tests shown that Fluke now makes some of the most electrically robust meters for techs and electricians, but also that UNI-T has no idea how to design a meter that will withstand ESD.    I've attached one of my favorite slides.   

Enjoy your old meter. 

[NoisyBoy]

Hi Joe,

Finally had a chance to revisit the 8000A and followed your instruction to only measure the duration of the +ve portion of the waveform with my counter using both channels and T.I. A-B, and I was able to achieve similar stability in what your showed in your video and keep the 8000A right within spec when it comes to period adjustment (that calibration pot was pretty sensitive though).

However, in calibrating the AC V using the 19V 20KHz signal, I found changing C3 has minimal effect on the measurement.  The result is within spec, but I was hoping I could get it to match against my other NIST cal DMMs despite 8000A's lower resolution.  Are you seeing the same in your 8000A?  If not, I was thinking about replacing C3 with a new 50pF variable capacitor to see if that's the problem.

[joeqsmith]

I wasn't paying much attention to how much adjustment I could make.   Specs for the meter are pretty wide.   I set it within a count of my bench meter as shown and called it good.  What are you using for a source?  What kind of distortion does it have?

[NoisyBoy]

Hi Joe,

Thanks for your kind advice.  I learned a lesson today that the order of the calibration steps definitely mattered.  When I started from the beginning again following the order Fluke highlighted, C3 worked properly and I was able to get it within spec now. 

It is nice to see the 8000A back to delivering accurate readings again.

[joeqsmith]

Hi Joe,

Thanks for your kind advice.  I learned a lesson today that the order of the calibration steps definitely mattered.  When I started from the beginning again following the order Fluke highlighted, C3 worked properly and I was able to get it within spec now. 

It is nice to see the 8000A back to delivering accurate readings again.

I thought it was in spec.

The result is within spec, but I was hoping I could get it to match against my other NIST cal DMMs despite 8000A's lower resolution.

Good to hear you have conquered your 8000A.  Someone had posted to my video a few days ago about still using an 8000A for their work in aircraft service.   Then again, people post about still using analog meters so I guess I shouldn't be too surprised.   

[NoisyBoy]

Hi Joe,

It was in spec before, but I was trying to get it to match my other DMMs that have been professionally calibrated, now it is "more" in spec 

I am not surprised that some are actually in active use, if you keep it calibrated, and do not require high resolution, or CAT ratings in high voltage work, then I can see how the 8000A can still meet the requirement.  Frankly, I think the majority of the work we encountered today can be done with a 8000A.  But with the abundance of handheld DMMs with equal resolution in the market today (especially Fluke's current offering), the 8000A just became a dinosaur. 

Have a great evening!  Again, thanks for all your help, I couldn't have done the adjustment without your insight.  Now I know what to do with my other 8000A tucked away in storage if I were ever able to find it.

[joeqsmith]

>>Frankly, I think the majority of the work we encountered today can be done with a 8000A.  But with the abundance of handheld DMMs with equal resolution in the market today (especially Fluke's current offering), the 8000A just became a dinosaur.

That's why it's hard to dump on the free Harbor Frieght meters too much.  I would imagine in many cases, they are good enough for what people use them for.   Similar to my comment on the analog meters.  Like any job, it comes down to what tools you need.     Of course, there people I see suggesting you can use an analog or free HF meter for everything and there really is no need for anything better.      I one person post the following, after I had modified my cheap DC current clamp:

What's the point of increasing the bandwidth if you are just going to be measuring 60 Hz current anyway? Do you have some other use for the meter?

We all work on different projects with different requirements for our tools. 

Good luck restoring your second Fluke 8000A.

[NoisyBoy]

Still haven't had a chance to look for the second 8000A in storage yet.  However, after the 8000A was adjusted to spec, I have included it in various measurements just to see how it behaves.  With only 3.5 digits to display and no auto-ranging, the meter is amazing fast.  I also found it to be very accurate when comparing with my other DMMs with fresh NIST cal.

The 8000a is dirt cheap on the used market, I will have no hesitation to use it as my bench meter if precision is not required.  It is more accurate and faster than any handheld DMM I have compared against it.  I thought it is an excellent candidate for those who wants their first bench DMM but running on a tight budget.

[maxwell3e10]

I recently got a Fluke 8600A with the battery option. To my surprise, the original batteries still work pretty well, giving 3-4 hours on a charge. I checked the meter calibration and its just a tiny bit out of spec on all the ranges.

One thing that is a problem is that it has a large offset on DCV/DCma when first turned on. It takes several hours for it to get all the way to zero, then it seems stable. Has someone else encountered this problem?

[NoisyBoy]

Beware of leaky battery destroying the meter, it is only one screw to open it up, check to make sure they are not leaking.  My other meter in storage is the battery model, it was supposed to be a working unit when I bought it, it did not turn on immediately when I plugged it in, and I just put it away thinking I would use it for parts.  I never gave it time to charge like you did, I thought the meter would power up regardless of the battery state.  That would be my next project when I am able to retrieve it from the storage locker.

I just checked my meter, the DCV/DCma both go straight to zero in about half a second after power up (except the 200 mV setting, which I got about 0.2mV).  I know for some production, there is a pot that adjust the zero reading, but it likely won't help if yours get there eventually after a warm up.  In my case, a warm up of 30-60 min makes the reading more stable and it is important to do that before adjustment/calibration, but the change is usually 1 digit, nothing big.

Perhaps someone else can chime in on your meter.

EDIT:  Sorry, just noticed yours is the 8600a, so the adjustment I spoke of may not apply. 

[joeqsmith]

Still haven't had a chance to look for the second 8000A in storage yet.  However, after the 8000A was adjusted to spec, I have included it in various measurements just to see how it behaves.  With only 3.5 digits to display and no auto-ranging, the meter is amazing fast.  I also found it to be very accurate when comparing with my other DMMs with fresh NIST cal.

The 8000a is dirt cheap on the used market, I will have no hesitation to use it as my bench meter if precision is not required.  It is more accurate and faster than any handheld DMM I have compared against it. I thought it is an excellent candidate for those who wants their first bench DMM but running on a tight budget.

When you make this statement, how are you comparing them?   Auto range vs manual, or are you running the handhelds in manual?   If auto vs manual, are you considering the time to select the range on the two meters?
 
For AC current they spec a response time of 3 seconds
ACV, 3 seconds
DCV, 500ms
DCA, 500ms
Resistance, 4 seconds

I would guess that is how long it takes to settle. 

The asking price for these old relics is in the $40 range now.  I personally would invest in something much better for that price.  Maybe a couple of low cost meters.