Question for owners of HP53132A

[tkamiya]

If you just power on the unit, and connect 10MHz out from rear panel to front input A and B (use BNC Tee), how stable is the reading?  How wide does last digit flicker? 

(Please don't change gate time)

[Grandchuck]

10.000,000,00 with flicker to 01

[Grandchuck]

Sorry, just noticed mine is a 53131A.

[rf-messkopf]

With the default 0.1 sec gate time I get 10.000,000,000 MHz which often flickers to 9.999,999,999 MHz. The flicker is more pronounced when the oven is warming up. Directly after power up it flickers wildly between about 9.999,999,998 MHz and 10.000,000,002 MHz.

With a gate time of 10 sec I get a stable reading of 10.000,000,000,0 MHz.

This refers to a HP 53132A with the high stability time base and with input set to auto trigger at 50% AC coupled (the default after power up).

[tkamiya]

Mine at default gate time flicker between 10.002..... to 9.99....7. 

With gate time 1 second, #1 shows 9.999....9 to 10.00...1
#2 (I have two 53132A) shows 9.999....9 to 99.99997

Something is funny.  I've never seen counters act this way.

[jadew]

For what it's worth, on my two 53131A there's no jumping around, absolutely none, with the same default .1 s gate time and the default time base.

They just boot up and start displaying 10,000,000,00 MHz, and the 32A is supposed to be the more accurate.

Unless this is normal somehow, several possibilities come to mind:

1) You could be picking up noise from the environment. Wi-Fi, SMPSs, Cable TV, etc.
2) The noise is generated by the clock source. What time base are you using?
3) Something's broken - unstable power supply?

Also, do you have a 3rd channel option installed? If yes, maybe try disconnecting it?

[tkamiya]

The clock is THE clock internal to the counter.  I've been having noise issues.  So that's very probable. 

[jadew]

The clock is THE clock internal to the counter.  I've been having noise issues.  So that's very probable.

Yeah, but which one is it? The TCXO on the mother board, or an OCXO option?

[tkamiya]

It's the medium stability one.  53132A comes standard with that. 

[rf-messkopf]

You've got to remember that the 53132A is a reciprocal counter with interpolators. It does a lot more measuring and processing than just pulse counting when set to frequency mode. The interpolators are not perfect, for that reason I would not be concerned if there is a slight instability of the reading with short gate times in this test. Even with longer gate times an occasional +/-1 count ambiguity would not be of concern.

That is different from simple old fashioned non-reciprocal counters like the HP 5328A, which would always read exactly 10.000... MHz when counting its own time base, irrespective of the resolution set (this determines the gate time on these counters). Even though interpolation does get you more digits per second of gate time, it can have a side effect in this case. And I don't think it is a contradiction that the higher resolution 53132A is apparently more unstable in the last displayed digit then the 53131A in this test.

You can check if you have a noise issue in the input signal conditioning and Schmitt trigger circuits of the counter by feeding in a known number of pulses with controlled rise time and repetition rate from a pulse generator, and let the counter totalize the number of pulses.

I just did the same test with an older HP 5335A counter sitting on my bench, which I have just repaired (covers still open and the A2 board dangling). That's also a reciprocal counter with interpolation. With 0.1 sec gate time it reads 10.000,000,0 MHz, with an occasional flicker to 9.999... MHz.

[tkamiya]

Thanks.  You are right.  These are reciprocal counters.  Somehow though, I think 1 of 2 is moving too much.  I'll get it calibrated just to make sure.

Do you have any experience with HP5370A/B?  I have both, and they are both reading its internal clock (loop back test) 10......180 or something.  In other words, 3 digit off.  Obviously, they both failed calibration.

I'm going through my lab and testing ALL counters, as they are all the same type.  HP5335, I have 4 of them.  I'm very afraid now.  For remainder of the year, I am going to test everything and try to get them all calibrated. 

[notfaded1]

For what it's worth, on my two 53131A there's no jumping around, absolutely none, with the same default .1 s gate time and the default time base.

They just boot up and start displaying 10,000,000,00 MHz, and the 32A is supposed to be the more accurate.

On the 53132A and on my Pendulum CNT-90 you are going down into higher resolution so going beyond the capabilities of the 53131A.  Stability in -11, -12, -13 exponent you aren't going to see 100%.  This is why we use software like timelab and stable32... to test this stability.  One test to do on the counter is a noise floor test.  This will help test the counter itself and not the reference specifically.  You can do this with a BNC-T.

See chart for SI multiples of second(s)

Bill

[tkamiya]

Thanks.  I've done the noise floor test in the past on this counter.

It doesn't matter now because there is something going on on input channel.  Whatever result I get is invalid.

Very weird thing.....  I can use a certain coax between input 1 and 2 and it succeeds.  With any other coax it fails.  Try that with case off, it fails.  Sort of sounds like there may be a solder crack somewhere.  But I didn't see it.

This is going into service today. 

[notfaded1]

Thanks.  I've done the noise floor test in the past on this counter.

It doesn't matter now because there is something going on on input channel.  Whatever result I get is invalid.

Very weird thing.....  I can use a certain coax between input 1 and 2 and it succeeds.  With any other coax it fails.  Try that with case off, it fails.  Sort of sounds like there may be a solder crack somewhere.  But I didn't see it.

This is going into service today.

I've heard of people reflowing some of the solder joints to fix these HP counters before.

[tkamiya]

I thought of that, too.  The symptom is just too weird.

Works on some coax, not other.  Same coax works on another unit perfectly fine.  Jumper between coax connectors and it fails.  With case off, fails every time.  I saw some signs of previous work around input #1 area.  It may be related.

[jadew]

On the 53132A and on my Pendulum CNT-90 you are going down into higher resolution so going beyond the capabilities of the 53131A.

You're right. I got confused by reply #2, and assumed they default to showing the same number of digits when they power up.

Works on some coax, not other.  Same coax works on another unit perfectly fine.  Jumper between coax connectors and it fails.  With case off, fails every time.  I saw some signs of previous work around input #1 area.  It may be related.

A faulty connector could also cause this.

Looking forward to hear what it really was.

[tkamiya]

I just took it to a local service/cal place to get it looked at.

I will post what techs find.  I'm very curious myself!

[tkamiya]

The counter is still at the "sick bay", but I have a preliminary report.

The last digit fluctuating is NORMAL for this counter.

The Front End FAIL didn't reproduce but I gave the tech specific instruction to reproduce.  Still waiting to hear back.  What I found out so far is that there is some sensitivity to a cable between channels.  I have one cable that always work, and all other cables fails.  On different HP53132A, it works with any cable.  So something is going on. 

[tkamiya]

I have a solution.

My counter, HP53132A was failing on self-test "FAILED:  FRONT END".  It was done independently, and I did not run ALL.  There are differences between the two.  When you do ALL, there is no relay chattering, and it actually works without input connected to the reference out.  So my setup was, from 10MHz in rear to a tee, which plugged into input 1, then to input 2.  This always failed on this unit.  But always succeeded another unit.

I never noticed this, but THIS unit has a rear input BNC.  The way it's configured, the input amp connects to front input, then branches off to the rear.  So if I'm using front connectors, this wire to the rear acts like a stub.  It fails.  When we connected the same except using REAR input, it works every time.

According to HP's service manual, it says when rear terminal is installed, performance of front terminal is not guaranteed. 

And as to fluctuation on reading, according to the lead tech, it is normal for HP53132A.  He believes slight phase shift on internal oscillator is the cause of it, and it is perfectly fine as a unit.

Wow....

[tkamiya]

I am also a member of time-nuts mailing list.

I heard from someone who was a part of the original design team for the counter says, rear input socket was badly implemented after thought by less experienced engineer.  It was a demand from some large customer and decision made by marketing.  One line blurb about this in the manual was a simply CYA.  (Cover Your A**)

[rf-messkopf]

I never noticed this, but THIS unit has a rear input BNC.  The way it's configured, the input amp connects to front input, then branches off to the rear.  So if I'm using front connectors, this wire to the rear acts like a stub.  It fails.  When we connected the same except using REAR input, it works every time.

According to HP's service manual, it says when rear terminal is installed, performance of front terminal is not guaranteed. 

Interesting. Mine also has the rear panel input option, but it passes self-test and I never noticed anything out of the ordinary. When I got it I did a performance check according to the service manual, and as far as I recall, it passed.

[tkamiya]

Mine is an older model.  I wouldn't be surprised they revised a circuit a bit to compensate for this.

[tkamiya]

Also, this failure occurs ONLY if you are running front end diagnostic by itself.  If you choose to run all tests, it will not fail.

The difference is, in single test mode, it actually picks up 10MHz from back and switch in and out something at input.  I can hear relays chatter.  In ALL mode, no relay chatter and no cable connections.