Author Topic: Some old school instruments showing how it's done (HP 3325A and Fluke 8506a) (Read 48473 times)

joeqsmith · « **Reply #25 on:** January 24, 2021, 01:39:14 am »

I had removed the serial interface to change the cal switch. 4-68 Line 6, R64 will not come in. I pulled the covers from the Fast ADC and had a look. Problem appears to be the resistance module. Once removed, I was able to make it to 4-77. That's a job for another day.

Shown adjusting R64.

joeqsmith · « **Reply #26 on:** January 24, 2021, 04:21:13 am »

Looking at this Fluke 8505A, they have a good photo of the resistance board. It has a date of 1991 and uses a GAL. The schematic in my manual shows no GAL and a different layout. The board I have is dated 1983 and does not look like their photo or match the manuals schematic.

https://xdevs.com/article/f8505a/

Their photo appears to match the schematic in this manual:
http://ftb.ko4bb.com/getsimple/index.php?id=download&file=Fluke/Fluke_8505A/Fluke_8505A_6.5_Digit_Multimeter_Service_Manual.pdf

Mine appears to be something between the schematic in my manual and this one above. The one in the manual I have uses a single coil for K1. Where the board uses a dual coil.

SilverSolder · « **Reply #27 on:** January 24, 2021, 04:23:20 am »

Nice, she's looking happier! - The circuitry in these meters are full of JFETs acting as switches... they can go leaky and cause issues. I had a problem with a Filter module, where the switching between different filter sections was not working well - causing the Average reading to be significantly lower than the plain reading (because Average mode automatically turns on the "heavy" filter section). I can easily imagine a leaky FET in a module loading the analog bus and causing false readings, if the module cannot properly disconnect from the bus.

I made some progress with my Unit 4 with the noisy ohms: I looked at the sense current with another meter and an oscilloscope, comparing the noisy unit to a good one. The current turned out to be equally noisy on both the "near perfect" unit and the broken one - so there is something else wrong than the current generator circuit. More work for another day!
Modify message

SilverSolder · « **Reply #28 on:** January 24, 2021, 04:50:13 am »

Quote from: joeqsmith on January 24, 2021, 04:21:13 am

Looking at this Fluke 8505A, they have a good photo of the resistance board. It has a date of 1991 and uses a GAL. The schematic in my manual shows no GAL and a different layout. The board I have is dated 1983 and does not look like their photo or match the manuals schematic.

https://xdevs.com/article/f8505a/

Their photo appears to match the schematic in this manual:
http://ftb.ko4bb.com/getsimple/index.php?id=download&file=Fluke/Fluke_8505A/Fluke_8505A_6.5_Digit_Multimeter_Service_Manual.pdf

Mine appears to be something between the schematic in my manual and this one above. The one in the manual I have uses a single coil for K1. Where the board uses a dual coil.

There is a revision letter on each board, if you take off the cover and look at the PCB. I have encountered MANY different revisions of boards...

I have encountered revisions not covered by the above table too!

SilverSolder · « **Reply #29 on:** January 24, 2021, 04:55:36 am »

The newest manual I've seen is also on the Xdevs site:

https://xdevs.com/doc/Fluke/8505A/Manuals/fluke_8505a_multimeter_sm.pdf

guenthert · « **Reply #30 on:** January 24, 2021, 05:06:41 am »

Quote from: SilverSolder on January 23, 2021, 08:26:03 pm

Quote from: guenthert on January 23, 2021, 07:05:54 pm
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.

Well that might be the case, but joeqsmith, who asked about calibration, has a 8506A which is quite off.
(I should have been more clear whos "yours" is

SilverSolder · « **Reply #31 on:** January 24, 2021, 05:18:29 am »

Quote from: guenthert on January 24, 2021, 05:06:41 am

Quote from: SilverSolder on January 23, 2021, 08:26:03 pm
Quote from: guenthert on January 23, 2021, 07:05:54 pm
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.

Well that might be the case, but joeqsmith, who asked about calibration, has a 8506A which is quite off.
(I should have been more clear whos "yours" is

I was (badly worded) trying to agree with you that it is not normal that these meters have drifted as far as @Joeqsmith's (and it looks like the problem may indeed be a broken resistance module).

As an aside, I had not noticed the Xdevs article/teardown/repair of the Fluke 8506 (linked to by @Joeqsmith). I had been wondering why there haven't been any cheap ones on eBay for about the last year... I guess this is yet another case of prices shooting up when a previously "hidden gem" gets discovered by EEVblog and its friends!

joeqsmith · « **Reply #32 on:** January 24, 2021, 04:58:59 pm »

I finished up the DC alignment (through 4-85) without any issues. The Fluke pretty much matches my old HP34401A in all the DCV ranges. I wouldn't trust using the HP as a ACV reference so I'll leave that section alone.

An extender would be very helpful but for a one shot repair, time to add some test points and start working through it.

joeqsmith · « **Reply #33 on:** January 24, 2021, 06:16:40 pm »

The relay K1 on the resistance board will not change states. The interesting part is how this area of the design has at least three approaches. One design used 4 transistors to drive the two coils. My schematics show a single coil with a FET and two transistors. The board have uses 2 coils and three transistors. The software/logic to control the relay drive circuit seems the same. GAL of 4000 series, both use a single pin. My board uses the old 4042. Low-activates the resistance mode.

The relay checks out. I can change the state manually to disable the resistance mode and DCV works fine. They have this thing AC coupled to use the edges to fire the transistors. It's a bit Rube Goldberg. Maybe this section was assigned to the intern.

Time to reverse engineer the madness.

SilverSolder · « **Reply #34 on:** January 24, 2021, 06:42:09 pm »

The manual for the 8505A also works for 8506A for all the modules (except Thermal RMS).

The reason I say that, is that some of the 8505A manuals are from different years than the 8506A ones, so you get a broader range of circuit diagrams - one of which may fit!

I think the attached diagram may be for your board -

joeqsmith · « **Reply #35 on:** January 24, 2021, 07:01:00 pm »

I just finished tracing it out when I noticed your post. Indeed, it seems very close if not the correct one. I suspect I made a few mistakes tracing it out. I was jotting down the part values and part numbers as I was going through it and measuring them. Nothing seems out too far.

Thank you very much for finding this. Time to double check my work.

SilverSolder · « **Reply #36 on:** January 24, 2021, 07:34:07 pm »

The circuit for flipping relay K1 is very psychedelic - not sure I understand how it actually flips it, rather than just pull in both directions at the same time!

[Edit] Does it flip ON with a negative going edge, and OFF with a positive going one (from U3)?

joeqsmith · « **Reply #37 on:** January 24, 2021, 08:39:46 pm »

Quote from: SilverSolder on January 24, 2021, 07:34:07 pm

The circuit for flipping relay K1 is very psychedelic - not sure I understand how it actually flips it, rather than just pull in both directions at the same time!

[Edit] Does it flip ON with a negative going edge, and OFF with a positive going one (from U3)?

Matches. Maybe my hand drawn schematics will help.

joeqsmith · « **Reply #38 on:** January 24, 2021, 10:52:54 pm »

I wonder why they added the bootstrap for Q3 but nothing for Q2. The circuit all appears fine. Nothing damaged. It's almost like the slow CMOS switching times won't provide enough drive to the transistors to turn them on. Time for a little tweak to their design.

SilverSolder · « **Reply #39 on:** January 25, 2021, 12:12:07 am »

Quote from: joeqsmith on January 24, 2021, 10:52:54 pm

I wonder why they added the bootstrap for Q3 but nothing for Q2. The circuit all appears fine. Nothing damaged. It's almost like the slow CMOS switching times won't provide enough drive to the transistors to turn them on. Time for a little tweak to their design.

I take it that "Set" is working, but not "Reset"?

C24 losing capacitance?

Could the relay be getting "stiff", so the pulse isn't big enough to flip it over any longer (it needs a longer pulse)?

The relay may not completely symmetrical in how much pulse energy it takes to set vs reset, necessitating the difference in circuitry?

joeqsmith · « **Reply #40 on:** January 25, 2021, 12:38:03 am »

Both states wouldn't drive it. I increased both tants and it now will change states but I don't like that bootstrap circuit. I may limit the base drive of Q3 while I'm in there.

I haven't done anything with the alignment yet and will wait until I get the board back in its clam shell. Looking at the 10K in my test box.

***
Added a small resistor and ran it through the ranges before buttoning it up. Sadly, there appears to be at least one more problem. Looks like the current source. Fluke has a house number 261578. Appears to be a Siliconix logo, J2137. So far no luck finding a data sheet for it. The 1986 small signal databook I found did not list it.

SilverSolder · « **Reply #41 on:** January 25, 2021, 02:01:23 am »

The 10K resistor looks pretty good? What's wrong with the current source - noisy?

dietert1 · « **Reply #42 on:** January 25, 2021, 06:54:58 am »

Used our two old Fluke 8502A recently. One of them shifts 15 ppm in 10 V DC range when connecting Guard to Low, the other one doesn't. Any ideas?

Had a similar issue with the relays K1 and K2 on one DC input conditioner but it went after exercising them via GPIB for 100 or 200 cycles. The relays sound completely different between both units. I replaced all those orange tantals and small electrolytics by film caps, MLCCs and new electrolytics. All of them were still ok, though. Distributed about ten MLCCs onto each ADC digital board. Added a schottky diode over the +5 V adjustment pots for safety.

Otherwise those units are working - after the recommended service incl. reseating of all modules. I also replaced the 150K input protection resistors and some other carbon resistors that had drifted more than their 5 % tolerance. Day-to-night temperature changes in our lab are 2 or 3 °C and those units shift 0.7 ppm when logging an ovenized 10 V reference. I think they are very quiet and good for 2 uVpp.

Regards, Dieter

dietert1 · « **Reply #43 on:** January 25, 2021, 07:48:46 am »

About the relays: As far as i understand the 8502A schematic they tried to use the monostable input relays at constant current in order to reduce shifts from thermal EMF. So they are running the coils at a current that doesn't turn on a relay but holds a relay that has been turned on before. Then the action is by short pulses - either reducing current to zero or increasing it for a short time. Seems to work.

Regards, Dieter

joeqsmith · « **Reply #44 on:** January 25, 2021, 11:29:39 am »

With the later revision boards using a latching relay, there's no need to provide a hold current. Just a pulse.

Found the last problem with the current source and have the meter now working on all ranges. Time to find some low tempco parts and then it should be able to be reassembled and finish the alignment. I have some low tempco parts I had characterized with an 8.5 place meter that I will use to align it.

The meter uses RS232 rather than GPIB. I think the next step is to see if I can put some simple program together to tend the data.

SilverSolder · « **Reply #45 on:** January 25, 2021, 03:01:55 pm »

All mine have GPIB, I've never seen the serial option. You've got a rare one!

SilverSolder · « **Reply #46 on:** January 25, 2021, 03:08:10 pm »

Quote from: dietert1 on January 25, 2021, 06:54:58 am

Used our two old Fluke 8502A recently. One of them shifts 15 ppm in 10 V DC range when connecting Guard to Low, the other one doesn't. Any ideas?
[...]

One of mine had a problem where Guard was shorted to Ground, due to a too-long chassis screw touching the bottom of the PCB. It exhibited similar odd behaviour.

I had another weird problem, where the 10V range would shift by 10ppm when turning on the Filter. I'm pretty sure it is a leaky FET in the switching arrangement, but it made me think that since these things use JFETs as switches all over the place, any one of them could probably cause weird and subtle problems.... (I fixed mine by swapping the Filter module with a known good one).

The nice thing about these meters is that it is so easy to swap modules. Since you have two meters, you can try swapping modules until you see which one is causing the problem?

joeqsmith · « **Reply #47 on:** January 25, 2021, 07:00:57 pm »

New 2ppm parts installed and finished the alignment. Showing the Caddock TK series 10ppm 1.9Meg. It's about 10 counts lower from the 34401A but good enough. Checked out the serial board. Glad I did. That thing has jumpers that are just wires press fit in place. I'm surprised that they never fell out. I set it up for 9600 8N1 and stuck a null modem and my old RS-232 jumper board (a relic from the 80s). Fired right up. Shown tracking a 40ohm 0.005% 1ppm part. Top cover still isn't installed but with the clam shells, I doubt it will make much difference with the drift. I used some long lead to zero it and plugged the resistor directly into the unit. 2W mode. Not surprised it's reading a bit low.

The manual talks about sending a 0 to enable the average and O0 to disable it. I can't seem to get that to work but the rest of the commands seem fine.

I'll let it collect some data over the day and see how it behaves.

SilverSolder · « **Reply #48 on:** January 25, 2021, 07:48:14 pm »

I love your "draft excluder" box - what is it? Looks highly scientific!

Which parts did you replace? I've been thinking of swapping some of the op amps... there are modern ones that are orders of magnitude better.

I have never actually tried setting the zero over GPIB - in the middle of tidying up the work area, slow going... I'll have a go at it when I've set it all back up.

joeqsmith · « **Reply #49 on:** January 26, 2021, 02:11:06 am »

Quote from: SilverSolder on January 25, 2021, 07:48:14 pm

I love your "draft excluder" box - what is it? Looks highly scientific!

Which parts did you replace? I've been thinking of swapping some of the op amps... there are modern ones that are orders of magnitude better.

I have never actually tried setting the zero over GPIB - in the middle of tidying up the work area, slow going... I'll have a go at it when I've set it all back up.

The black box on top is just an old RS232 tester. There's no standard so these were common back in the day to sort out how to wire the cables. I have now fitted the correct connector and wiring so the meter just plugs directly into the PC.

Some parts were out of spec enough where it could not be adjusted. These were replaced with higher quality parts.

Where my manual show 0 as the commend to set the average and O0 to disable it, other manuals show it as O to enable the average. Explains why the command didn't work.

Showing the drift over five hours with the 40ohm part.

I then reinstalled the 1.9M as I had collected data from several meters using this part.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Some old school instruments showing how it's done (HP 3325A and Fluke 8506a) (Read 48473 times)

Share me