Author Topic: Fluke PM6666 counter freezing after 3 seconds from turning on. (Read 10787 times)

thinkfat · « **Reply #75 on:** October 12, 2020, 07:37:42 pm »

You guys haven't been singing along. The IC2 on the GPIB card is pulling SCL low when it's being accessed. I2C slaves can do that, it's called "clock stretching". The bus master then must wait until the slave releases SCL (never happens) which is the very reason why the counter hangs. Proof: counter continues as soon as you disconnect the SCL line.

coromonadalix · « **Reply #76 on:** October 12, 2020, 08:47:05 pm »

got schematics for the 6666

Well the OP did say the gpib card whas working fine in other instrument, did i read it wrong ?

reply #19

Could it be simply incompatible with the 6666 ? if everything else works and the gpib works in a 6669 what can we do to solve this ?

Is the jp101 test norm jumper closed or open ?

khs · « **Reply #77 on:** October 12, 2020, 09:35:21 pm »

Quote from: thinkfat on October 12, 2020, 07:37:42 pm

You guys haven't been singing along. The IC2 on the GPIB card is pulling SCL low when it's being accessed. I2C slaves can do that, it's called "clock stretching". The bus master then must wait until the slave releases SCL (never happens) which is the very reason why the counter hangs. Proof: counter continues as soon as you disconnect the SCL line.

thinkfat · « **Reply #78 on:** October 13, 2020, 08:33:57 am »

I found a datasheet for the MAB84X1 series. The clock input is specified as HCMOS and TTL compatible so we're good here (though I think 3.8V still a bit low), but the maximum rise and fall time of the clock is <10ns. Can you check the clock edges meet that spec?

thinkfat · « **Reply #79 on:** October 13, 2020, 08:37:26 am »

Quote from: coromonadalix on October 12, 2020, 08:47:05 pm

got schematics for the 6666

Well the OP did say the gpib card whas working fine in other instrument, did i read it wrong ?

reply #19

Could it be simply incompatible with the 6666 ? if everything else works and the gpib works in a 6669 what can we do to solve this ?

Is the jp101 test norm jumper closed or open ?

You're right, the GPIB from the PM6666 is working in another counter, and the GPIB from that counter also fails in the PM6666. That's why I keep pressing on thoroughly checking the 5MHz

PS: if you have a function generator, try connecting a 5MHz square wave from the generator to the GPIB module, instead of sourcing the clock from the PM6666. If it works then you have found your culprit. Then you need to find out how to restore or improve the 5MHz clock. Maybe add a 74HC14?

pizzigri · « **Reply #80 on:** October 13, 2020, 01:05:06 pm »

OK...
I connected my crappy FY6900 to the 5mhz pin on the board, 4V 50% duty cycle, square wave. I checked it on the scope and sincerely the original square wave from the meter is nicer. But I digress, on how crappy the FY DDS is. Anyway, turned on and the meter never goes past the POST, all leds lit and all segments in the LCD on.
So I thought to use an external 10 MHz reference using the rear BNC, bypassing completely the internal MTCXO oscillator, and no joy. Now it works as before three secs after power on, it locks up.

Then, I imagined that the 10MHz internal ref and 5Mhz square were out of sync, yielding the POST lockup, since I was using the DDS for the external 5mhz clock. SO, I connected the 10 MHz to the rear connector and 5 mhz clock to the GPIB card, using both channels of the DDS, therefore in sync with each other - reasoning being, this way I'm bypassing everything on the board save for the 5mhz clock for the master cpu (but still in sync due to the external ref coming from the DDS). Again, no joy; locks up three seconds after post. It's getting real frustrating.

pizzigri · « **Reply #81 on:** October 13, 2020, 01:23:35 pm »

Quote from: thinkfat on October 13, 2020, 08:33:57 am

I found a datasheet for the MAB84X1 series. The clock input is specified as HCMOS and TTL compatible so we're good here (though I think 3.8V still a bit low), but the maximum rise and fall time of the clock is <10ns. Can you check the clock edges meet that spec?

No they dont. It looks like a sawtooth especially in the rise edge, and not a square wave. Rise time is about 60ns, fall is 20ns. Amplitude is actually 4.20V, if I measure directly the pin.
Could it be the reason? The square wave I believe is built by the custom ASIC, if this is the case, I'm done for....

coromonadalix · « **Reply #82 on:** October 13, 2020, 02:27:55 pm »

you could use logic device ic's to square up the signal ??

thinkfat · « **Reply #83 on:** October 13, 2020, 09:40:36 pm »

Quote from: coromonadalix on October 13, 2020, 02:27:55 pm

you could use logic device ic's to square up the signal ??

Yes, a buffer with Schmitt trigger input for example, 74HC7541.

But, please show screenshots of the waveforms of the 5MHz signal.

I hope you disconnected the counter-generated 5MHz from the GPIB card before you fed it from the function generator? I don't see how it could be looking worse than the 5MHz from the counter itself - that one is abysmal.

Also, please show screenshot of the 5MHz waveform without the GPIB card connected.

Yes, there's one ASIC (IC104) that creates among other signals the 5MHz for the internal CPU (IC105) of the counter and the GPIB interface (BU105, pin 3). From what I can see in the schematics, the 5MHz go nowhere else.

I figure either there's something wrong on the GPIB board that loads the 5MHz signal and distorts it (that's why I want to see the signal without the GPIB card attached) or there's something wrong with the clock signal output driver in IC104, or the clock input of IC105 has a problem.

pizzigri · « **Reply #84 on:** October 14, 2020, 07:35:47 am »

Hello Thinkfat!
OK so here are the screenshots.

First one, output from FY6900 at 5MHz, using bnc end to end.
Then, output of the 5MHz coming from the pin of the GPIB option connector on the meter mainboard.
I then tried to trace the output directly out of IC104, third image. The quality is worse because I had to use the smallest grabber I had, form the logic analizer and that has a long unshelded wire. Gnd is always on the frame of the counter (except for the first image obviously). I tried to hook up the 10 MHz but if I do so, somehow the meter doesnt start giving garbage on the screen.
What I find strange how ever is that the onboard CPU, that is exactly the same part mounted on the GPIB card, can use the sawtooth signal nicely and works, and the one on the card cannot. As I mentioned before, the counter works as a charm without any glitch that I can see if the card is disconnected.

Oops I mean here are the images....

thinkfat · « **Reply #85 on:** October 14, 2020, 07:43:25 am »

~~It seems attaching the screenshots didn't work, however,~~ I have a general advice on probing with an oscilloscope: Don't use a far-away ground point for probing. Don't use the frame as ground, in particular. Find a ground hook on the PCB or something similar. And of course, use the oscilloscope probe and nothing else.

EDIT: there's obviously a TP100 next to IC104. It has a pin labeled "GND", connect the oscilloscope probe's ground lead to this pin. Put the probe tip directly on the chip pin.

EDIT2: IF the waveform we see is real (which I doubt when I see the probing setup), it looks like the output driver on the clock output of IC104 is damaged. The high-side mosfet might be dead or deteriorated. It looks a bit like the recent "Intel Atom C2000 bug" that killed thousands of NAS, switches and other appliances. Maybe a similar "fix" could be used: You can try pulling the clock line up with a small resistor (200 $\$\Omega\$$ or something).

Dr. Frank · « **Reply #86 on:** October 14, 2020, 07:51:39 am »

Hello pizzigri,
maybe it's a silly question.. have you ever attached the GPIB to a PC interface and tried to steer the PM6666 from there?
Reading through this thread, I am under the impression, that you never tried that.

I'm only asking, because I used a PM6669 with GPIB many decades ago, and I'm not sure if that was required also.

Frank

pizzigri · « **Reply #87 on:** October 14, 2020, 08:19:00 am »

Hello Dr. Frank!
the device freezes whether I connect a GPIB cable or not. However, I also have other devices with GPIB, a Philips PM2534 DMM, a HP34401A and of course the smaller sibling to the PM6666, exactly a Philips PM6669 which never showed the problems this one has.

0xdeadbeef · « **Reply #88 on:** October 14, 2020, 08:19:57 am »

Mainly to get updates on this thread automatically: I also bought a PM6666 for cheap quite some time ago and it locks up (no reaction to key presses) occasionally directly after startup (probably three seconds, didn't measure that). Originally, it worked most of the time, lately, I needed to start it a few times to avoid it locking up. So this seems to be something becoming gradually worse. Never really looked into this as I upgraded my Tabor counter in the meantime and even if I needed the PM6666, I got it to work after a few trials.

pizzigri · « **Reply #89 on:** October 14, 2020, 08:36:31 am »

Ok, so here’s the updated measure. Did as told (thank you Thinkfat) and I do get a cleaner wave but it is still a sawtooth.

thinkfat · « **Reply #90 on:** October 14, 2020, 08:46:42 am »

Aand you forgot to switch the probe to 10X again

Really, make this a habit when you probe digital signals or anything above audio frequencies: Your probe in 1X mode has a bandwidth of maybe 5 MHz, which means you basically get to see the fundamental tone of the square wave and all higher order harmonic will be attenuated a lot. So you see a distorted sine wave at best, but not the true waveform.

pizzigri · « **Reply #91 on:** October 14, 2020, 09:11:27 am »

Oh my god! But you pegged me perfectly: I'm an audio enthusiast, and I mainly play with amplifiers etc
Give me a moment, I'll redo it...

pizzigri · « **Reply #92 on:** October 14, 2020, 09:29:51 am »

Here it same conditions

thinkfat · « **Reply #93 on:** October 14, 2020, 09:34:31 am »

OK. Is that with or without the GPIB card connected? I'd like to see both.

pizzigri · « **Reply #94 on:** October 14, 2020, 09:39:24 am »

Without, this one is with. The shape doesn’t change much but amplitude is less

thinkfat · « **Reply #95 on:** October 14, 2020, 12:13:01 pm »

Well yeah, that doesn't look OK. The signal is out of spec already without the GPIB card, with the additional load of the card it's not improving of course. I suspect that there's something wrong with the output drive of IC104, it just cannot source enough current to drive the line high fast enough. The counter itself seems to be happy, though.

The fall time of the signal seems to be OK'ish. If you want to try a small experiment, get a resistor of maybe 500 Ohms, connect it between the clock signal and +5V, see if this improves the situation. If you see a difference, try a smaller resistor, maybe down to 200 Ohms.

However, lifting the clock output pin of IC104 and feeding it into a Schmitt trigger buffer and from there back into the board would be a better idea.

Maybe, just maybe, touch up the solder joints of IC104. If you have a cold joint, that could explain it, too.

shakalnokturn · « **Reply #96 on:** October 14, 2020, 12:18:30 pm »

Agreed, that doesn't look fantastic.
What does that 5MHz look like on the PM6669 with GPIB installed?

pizzigri · « **Reply #97 on:** October 14, 2020, 12:42:58 pm »

Well, IC 105 and 106 are mounted on a socket so maybe I could lift them out and clean the contacts, ow that I think about it it was a standard protocol for any problems on Macs with the ram. Will try that. And OK for the resistor, that’s next.
I don’t have the 6669 any more- waiting on an Hp 5385a with which I replaced it.

pizzigri · « **Reply #98 on:** October 14, 2020, 01:51:35 pm »

Wow what a difference, here’s the result of a R500 between +5v and clut on the signal:

...however it still locks up after 3 secs. I'll try now a R300

pizzigri · « **Reply #99 on:** October 14, 2020, 02:15:11 pm »

so taking +5v from pin 22 of IC104, closer than the 5v pin on the oscillator connector... going with a R280 tht's what I have now.
However, the meter doesnt boot, if I feed it this way


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

Author Topic: Fluke PM6666 counter freezing after 3 seconds from turning on. (Read 10787 times)

Share me