Fluke PM6666 counter freezing after 3 seconds from turning on.

[FransW]

I know and I am willing to do so.
The only thing is, I have to scan it.
And that will take quite some time or I have to rely
on a scanning service.

[pizzigri]

Wow, that is absolutely great!
Thank you FransW

As for what information is most important now, well, I imagine the part where the bus logic is explained, so I can try to understand where the whole process hitches and freezes the counter... I cannot believe the I2C logic is flawed, as the whole counter uses this bus and if a GPIB I/F is not connected to the counter mainboard, the thing works flawlessly. There must be something happening in the first 3 secs that hangs the machine. If I could pinpoint an IC, hopefully NOT the custom OQ0501 I, maybe I can replace it, or otherwise test it.
Of course I would be interested in the whole manual, once it is available!


all my best,

Franco

[pizzigri]

BTW, since the PM6666 is totally disassembled, I imagined I could take a few pics of the only option that I could not for the life of me find images on the web - the PM 9607 - 00 10Mhz high stability time base option, in other words, the MTCXO card. I took the copper shielding off to show the components on the board. It uses a Philips crystal that is actually encased in glass, I haven't seen many of these enclosures.
The values are stored in the 85c82 eprom; there'a also a hex ubuffered inverter MC74HCU04N and a LP2951 voltage regulator, plus an IC I cannot ID. The Fluke literature says that ..."The unique MTCXO principle gives a residual temperature stability of 2 x 10-7 over the temperature range"... and that it equals the stability of an oven XO, of which I have some doubts.
However here's the images.

[coromonadalix]

niiice a transparent crystal,   i dont catch why there is an eprom   85c82   IC403 if im not mistaken     what is the part number of ic402 ??

[pizzigri]

The EPROM stores the drift values recorded at the factory for this single crystal at a range of temperatures. As the sensor measures the temperature, these values are used to correct drift, hence “mathematically controlled, temperature controlled XO”...
I can’t read the markings on IC 402, and there is no schematics or pcb detail in the “old” manual for the MTCXO, only the standard oscillator. I suppose it’s a microcontroller. Maybe in FransW new manual there’s a schematic for it!

[pizzigri]

I have traced the I2C bus in the PM6666 and 69.
The whole device is based on this bus, it connects all subassemblies - display, oscillator, to BU105 connector that is used for both IEC (GPIB) bus and optional internal battery and it also goes to the undocumented connector on the rear. In the PM6666, the I2C bus also goes to the input amplifier, to the three PC8574 ICs (I2C bus expanders) connected to a PM7528 Dual DAC, all missing in the simpler, one input amp of the PM6669.

The I2C bus goes to both the CPU and custom IC, and I cannot understand why plugging the IEC - GPIB board freezes the machine. If the Custom IC was bad, or there was something wrong with the I2C bus, I would see malfunctions, and I dont. I can leave it on all day and it will work flawlessly.

So I looked at which other lines are connected to the IEC bus, and found that there is a "option" line that gets connected, going to pin 12 of the CPU, which is an external interrupt (labelled INT/TO), vs. the "ident" used by the prescaler and oscillator (these lines tell the CPU if a MTCXO is installed so that a single button press calibration is available, and if the prescaler installed; I did in fact install the prescaler and it works without a hitch.
But even if the option line int causes a crash, why does it happen three second after I turn on the machine? I would imagine it would be immediate.
An idea could be that there is some corruption in the programming of the CPU (stored in the internal 6K), but I would not know how either to test this or correct it.
I have an hobbyist oscilloscope (SDS1202x-e) and good multimeter, but that's all I have... is there something I can try to troubleshoot this thing?

[coromonadalix]

Nice detective work

Could it crash when enumerating the i2c peripherals ?

A data sniffer like bus pirate or salae logic probe / clones on Ebay  ??

[Ordinaryman1971]

Far fetched idea but... maybe your firmware is somehow corrupted. Let me see if I can burn the copy of of my counter.
Could you see the version you're running?

I know it's silly for me to ask... but did you inspect everything for possible shorts... the seller stated that the meter is working fine.. so, lets assume that he did in fact turn it on and it was working, then... something happen in transit.

What is the frequency of the prescaller you've had installed... can it go more than 1.1 GHz, that's what I have in mine. Actually I've tested mine some time ago and it goes all the way to 1.6GHz.... and I have standard time base btw..

[pizzigri]

No, I don’t think there’s any way to see the fw rev, however the board is rev 12, and the counter is branded Fluke.... built possibly in 1998 or 99 one of the last manufactured. The prescaler is the “B” version, it goes to 1.3ghz.
I took the whole thing apart, as I wanted to add lcd illumination - the screen is horribly hard to read, just like my old PM2534 - and I could not find any shorts, actually the counter seems brand new, clean and tidy.
The PM6669 I got as a great deal on ebay was used and abused, scratched, splashed, dirty as hell.... but works perfectly. Frustrating

[FransW]

In the men time I received this manual:
Timer/Counter PM 6665 and PM 6666, Service manual, 9499-465-01311, 90-10 Second edition, 132 pages.

It will take me quite some time to generate a complete pdf version. The manual contains fold-out A3 size diagrams in color.

To contribute something I refer to the attachment with an introduction to something called "Extended Test Program" which I gathered as important.

Attachment 1: Extended Test Program
Attachment 2: historic overview GPIB parts lists

Frans

[pizzigri]

Downloaded and looking at it! Thank you!

[FransW]

This time all the extended test pages from the manual, repair part.

[pizzigri]

OK, so now I'm going to pass the night working on this. Let's see what goes on! Thank you a LOT.

[FransW]

Rest of the repair section and a clearer  "fault finding tree".

Success,  Frans

[pizzigri]

So, I passed the last 24 hours going crazy around this. I tried following every step, but for starters, nowhere I found any freezing references, and all signals test good on my DSO. As soon as I mount the f**** GPIB card, the counter freezes, after three seconds. I can do anything in those three seconds, and the device will respond. Then it just locks up. Does not matter, either card I have gives the same result, and instead both test good on the 6669 I'm selling. No error codes on the display at startup (and in extended mode the screen is blank anyway).
However, in extended test mode, and with the card mounted, it does not freeze - it does the relay thing, and if I move the jumper as described, it stops doing that, but does not freeze. However, in this case something does not add up, I cant manage to get the signals described on the manual, on the referenced pins. Not sure I am doing anything correctly, the text here does reference other chapters of the manual, i.e. the I2C description chapter.
I also learned something new, the I2C bus includes the TXCO and prescaler, I did not see this on the circuit diagram I had. Possibly it uses the protocol on the"ident" lines I found? No SCL/SDA go to these options, unless I'm blind.
I admit to be a complete noob and hobbyist.
Ahhhh.... I'll just wait for the whole manual, and see if it could be a power issue in the meanwhile. Tomorrow I'll check up currents too from the MB to the GPIB card.

[FransW]

I will try to scan the I2C-bus pages tomorrow.
From the GPIB parts lists not much can be deducted.
They appear nearly exactly the same with minor variations.
Did you check the com port dataflow to your PC?
Is there data flow?
I can not check any GPIB irregularities since mine does not have that unit.
There seems to be something blocking the counter functionality when the
unit is installed. Waiting for (unknown command).
Since working with a GPIB unit is fairly common with the PM66xx counter/timers
I would not expect something like this would happen, given careful consideration
setting the command address switches.
Can you monitor the data exchange with your scope?
Bedtime now, 01:00 hrs here locally.
Regards, Frans

Scanning tomorrow. I will try to OCR these pages, given enough time.

[pizzigri]

OK thank you. I just trashed my SALEAE clone, it just died on me so I have now doubts on whatever i was looking at. I purchased another locally at double the cost (10 Euro)  but at least it should arrive tomorrow, hopefully it's going to be better to check the I2C bus.
I really appreciate your help and the manual excerpts!
Best, Franco

[FransW]

I2C, 3 pages.
Rest to follow.

[FransW]

Pages 27, 28 and 29

[pizzigri]

So I have been checking the I2C bus with the logic analyzer. I probed the SDA and SCL lines, using the rear diagnostic connector, see photo.
I am baffled, I can see that the bus at one point basically crashes but I cant understand why. The Saleae sw I am using shows me an interpretation of the I2C bus but I cannot correlate it to what I read in the manual excerpts, surely I am lacking in knowledge of the bus (dont know crap about it) but surely enough it has nothing to do with the little I do know about it from i.e. arduino.
That something wrong is going on I can tell by the last signals recoded when the machine crashes and locks up, all lines go down and they are supposed to stay up. See images and XLS files, the one with the crash is labeled "withgpib".
Can someone help me with this? Pleease! 

ETA: Image 3 is of the counter WITHOUT GPIB card connected, and 4 and 5 are WITH the card, up to the point immediately before crashing.

[pizzigri]

Far fetched idea but... maybe your firmware is somehow corrupted. Let me see if I can burn the copy of of my counter.
Could you see the version you're running?

I know it's silly for me to ask... but did you inspect everything for possible shorts... the seller stated that the meter is working fine.. so, lets assume that he did in fact turn it on and it was working, then... something happen in transit.

What is the frequency of the prescaller you've had installed... can it go more than 1.1 GHz, that's what I have in mine. Actually I've tested mine some time ago and it goes all the way to 1.6GHz.... and I have standard time base btw..

Ordinaryman1971....
do you have a PM6666?
I checked spec sheet of the MAB8461P cpu, it has Mask ROM therefore impossible to retrieve or program for that matter.
Can Mask ROM corrupt? I find it unlikely, but hey could be.
I still hope it is something else.

[pizzigri]

Additionally, I have run the Logic analizer on the surely working PM6669 counter, and I found that it is mostly silent: the bus stays completely quiet in high until a button is pressed or something else happens, and this, even including the GPIB board installed.
At startup it does a routine check of everything and then just stays listening.

The PM6666 is always "talking", but that could be because of the DAC? This component is not included in the PM6669.

[pizzigri]

coromonadalix, looking at these info obtained with the Saleae clone, do you think that what you suggested could be true?
All my best, Franco

[coromonadalix]

Strange bug i may say,   since i dont have the full schematic, i dont know if there are bus buffers, buffer with a latch / enable  etc ...


The i2c should be relatively easy to read,   i dont know wich software you use or witch version ??  normally you have the bus decoding option in it ?  i recall you have 2 compatible software who works with the clones ...

Could be the MAB8461P who's corrupt at one specific place when the gpib card is there, gpib protocol / enumeration, say a single bit

The best would be to have anoter 6666 and try this chip in the other ...

Or  the last option is to sell it "with the gpib not working warning" and try to find another 6666 ?


Your excel sheet  show  lots of "missing ack"   (acknowledge) ??

[Ordinaryman1971]

I do have two.. they work like  a charm. One with gpib and C channel 1.1GHz second without C channel, both with regular oscillator. Let me open it up and see.... I also have a spare gpib port from 6680B.
Here are pictures of both before and after backlight hack.
It’s not perfect but it will do...
I will open it up later today and get a look.
Odd question... what's the address on your GPIB card... try different setting.... like first or the last address.