-
#100 Reply
Posted by
Dr. Frank
on 17 Feb, 2019 23:13
-
Dr. Frank? That's not what I am asking.... I was asking about the jumper to remove. I posted a photo around that area with what I believe to be a fairly clear description. You'll see that part is slightly different.
Can you help me with that?
You said W1 is CLEARLY visible. I think so too. When I asked for confirmation, you told me to be careful and refereed me to a manual. That's when I got concerned.
Instead of a description, you better would mark the part in question in your picture, then I could safely say "yes".
I simply found your description dangerous to answer, especially such a term like 'cream colored "thing" ' .
I hope, you can now identify W1 on your own, safely.
And the OCXO should simply resemble the pictures of any 10811-60xxx on the net..
-
#101 Reply
Posted by
tkamiya
on 17 Feb, 2019 23:20
-
Done.
I removed what I thought to be a W1. While there is no indication on the module, some online resources refer to my unit as the right one. So I installed it. Power on! It seem to be working. Dr. Frank, thanks for putting up with me.
Is there any indication on front panel to oven is cold?
Now, the only thing left to do is to replace this noisy fan.
-
#102 Reply
Posted by
Dr. Frank
on 17 Feb, 2019 23:32
-
Was a bit laborious, with both of us, but I'm happy that everything works, in the end.
In the 5335A manuals, there is no mention of such an indicator, and I don't remember that it has one.
Sorry, can't test that, as my OCXO is still running.
PS: pin 11 of the 10811 is the oven monitor, but is N.C. inside the 5335A. Therefore, definitely no indication possible.
It should run 48h before making any calibration.
And a picture of the faceplate would be nice, for my collection.
THX - Frank
-
#103 Reply
Posted by
tkamiya
on 18 Feb, 2019 04:36
-
I'll have to replace a fan before closing the case. Then, I'll take the vanity shots. Thanks for your help.
-
#104 Reply
Posted by
tkamiya
on 19 Feb, 2019 13:36
-
I have a HP5335A just upgraded with HP10811 oven oscillator. I have GPSDO going to a scope channel 1 and HP5335A clock output going into channel 2. Trigger is set on 1. 20 ns / div.
What's reasonable expectations? So far, I have not been able to completely stop the trace movement. The "slug" is extremely touchy and has some hysteresis. (move slug, remove tuning stick, it goes back a little)
-
#105 Reply
Posted by
Dr. Frank
on 19 Feb, 2019 16:07
-
I have a HP5335A just upgraded with HP10811 oven oscillator. I have GPSDO going to a scope channel 1 and HP5335A clock output going into channel 2. Trigger is set on 1. 20 ns / div.
What's reasonable expectations? So far, I have not been able to completely stop the trace movement. The "slug" is extremely touchy and has some hysteresis. (move slug, remove tuning stick, it goes back a little)
I don't know how long GPSDO machine and OCXO are already running.
Anyhow, you can easily trim the OCXO to < 10^-9, i.e. to exactly 10.000 000 00x MHz.
Below, the resolution of the trimmer is too bad.
The metal screw driver you're using also influences the result.
I would apply the 10MHz, or the 1pps from the GPSDO to the A input and simply measure FREQ or PER with longest possible Gate Time, that's about 5sec.
That will give enough resolution to trim to that mentioned limit. I also have described an easy modification of a capacitor to get > 10sec Gate Time, and 10 digits resolution.
The OCXO will usually be stable to around 1^-9 over a longer period of time.
Oh yes, the GPSDO signal itself is quite unstable on these short time scales. 1E-9 at 1sec might be achievable, otherwise use TimeLab and GPIB for better trimming over longer time scale.
Frank
-
#106 Reply
Posted by
tkamiya
on 19 Feb, 2019 16:37
-
I should stop here then. I have 10.000,000,00x MHz
My GPDSO has been running for over a month, perhaps few months. I have a redundant setup and results are the same using either or. I'm at a point where mere pressure will change the result.
-
#107 Reply
Posted by
tkamiya
on 20 Feb, 2019 01:38
-
This is my 5335A on internal OCXO displaying rubidium source.
A little concerning that OCXO, when measured via 12 digit counter has an upward trend. It has been on for 24 hours already.
-
#108 Reply
Posted by
0culus
on 20 Feb, 2019 02:16
-
Upward trend? I guess that's not the picture.
-
#109 Reply
Posted by
tkamiya
on 20 Feb, 2019 02:20
-
No, the picture is a "beauty shot"....
I'm not sure what it is, but when monitoring internal clock frequency using GPSDO trained 12 digit counter, I see a very slow upward trend on clock frequency. It's VERY SLOW. But nevertheless, it should not be happening. I'm going to let it run for 24 more hours and see if it continues.
-
#110 Reply
Posted by
tkamiya
on 20 Feb, 2019 03:22
-
I wonder if my oven module is damaged.... I expected outside metal to be warm to hot. It's not even luke warm. It's same temperature as other components.... The upward trend is still continuing.
-
#111 Reply
Posted by
0culus
on 20 Feb, 2019 04:17
-
I wonder if my oven module is damaged.... I expected outside metal to be warm to hot. It's not even luke warm. It's same temperature as other components.... The upward trend is still continuing.
Yeah, that's not right at all. All of the ones I have (5335A, 2x 8568B RF sections, 3325B function gen) all are quite warm to the touch when the oven has had wall power for a while.
-
#112 Reply
Posted by
tkamiya
on 20 Feb, 2019 04:31
-
That's what I thought. Most of them gets hot enough that it's uncomfortable to touch for long. Darn!
Now it explains why it's behaving oddly. Tomorrow, I'll check 24V first, then start looking.
-
#113 Reply
Posted by
0culus
on 20 Feb, 2019 06:06
-
Here's a little experiment I did that might give you a datapoint. The 3325B has been unplugged for a couple of days, so I plugged it in and started monitoring a 10 kHz 0 dBm sine with the 5335A (which has been on power for about a week now). Took about 1.5-2 hours but the 3325B oven settled out reasonably well after pretty considerable swings when it was first turned on. I believe it may take quite a bit longer than that to truly settle, but at the same time, once the oven gets to temp it should stabilize a lot.
-
#114 Reply
Posted by
Dr. Frank
on 20 Feb, 2019 07:02
-
The 10811 has to get warm, otherwise the oven is not working. Strange, that you can trim to zero anyhow.
So, what is the rate, anyhow, and how / what do you measure in practise?
The 10811 is stable only after 48h, as I wrote already, and will be stable then to about 1E-9, maybe more, if it has been off for a longer time.
To check the oven, you may apply 24V to the oven circuit externally, that should consume 500mA on power on, going down to around 2..4W after 15 min, or so..
If not, the internal thermal fuse may be defect, what often happens on vintage 10811.
It can be replaced by a short w/o problem.
Frank
-
#115 Reply
Posted by
tkamiya
on 20 Feb, 2019 17:07
-
It's odd. I tested the oscillator by itself. It did consume 0.42A and declined. It got slightly warm.
Cleaned contact and reinstalled. Now it's QUITE warm. So it might have been a case of oxidation. I didn't do anything at all. I'll have to keep an eye on it. It *may* have some intermittent problems.
-
#116 Reply
Posted by
Dr. Frank
on 20 Feb, 2019 17:58
-
all contacts are gold plated, so that's not that probable.
This thermo fuse might have a problem, as well as the oven regulator circuit.
I once repaired a 10811 inside a 5370B, which has an oven indicator. So I could quickly find out, that its NTC intermittently failing to open, which gives no heating.
Was a mechanical / vibrational error also.
-
#117 Reply
Posted by
tkamiya
on 20 Feb, 2019 18:10
-
Oxidation may be a wrong word. More like contamination. I had plenty of times gold plated sockets and connectors failed me.
I believe I'm having an intermittent issue. One time, it's quite warm to touch. Another time ambient.
-
#118 Reply
Posted by
tkamiya
on 21 Feb, 2019 03:01
-
I have some data now.
The frequency drift appears to be a result of cooling by fan. To reach this conclusion, I stuck a thermocouple on side of OCXO and charted time, temperature, and frequency.
With case top open, it measured 34C and last two digit of frequency was 98. (last digit is 10^9)
With case top open, 30 minutes later, the temp went up to 41C and kept going. At that time, frequency was 94.
Now, close the case temp start to goes down. 14 minutes later, back at 34C and frequency was 98.
So temperature co-effecient is somewhere near 1 x 10^9 / C
Rate of drift is 1 x 10^9 / C (obviously...)
This should stabilize at some point. Tricky part is adjustment. It may take few hours once case is opened and closed.
Once it reaches equilibrium, I expect drift to be less.
Looks like HP has designed a very efficient duct system. I looked at the temp fuse. It's socket-ed. So I cleaned both and reinserted. I think this is working. Perhaps when we tune, we leave the case open, so we have impression that OCXO runs that hot all the time? I don't think this is over-cooling because some chips run quite hot and they are closer to the fan.
Opinions, please.
-
#119 Reply
Posted by
Dr. Frank
on 21 Feb, 2019 07:18
-
The 10811As specified T.C. is < 2.5x10-9 over an environmental temperature range of 0..71°C. (see page 1-2 in manual of 10811A).
So your observations might give too high a T.C., but maybe can not be distinguished so well from other influences.
Also, the specification is given as a 'box method', not directly as ppm/°C.
Maybe the oven control is not ok. A monitoring of the oven power consumption over temperature may give more hints.
Its timely drift is specified < 5x10-10 /day. So your observations are inside specification.,
I suppose, that you let the counter plugged in all the time, so that the 10811A is continuously powered.
When I switch my 5335A on, I also observe a slight drift of <1x10-9 after a few hours of internal heating up.
I use the counter in a room at 21°C +/- 2°C over the whole year.
The timely drift of my 5335A is less than 1x10-9 over a year, meanwhile the 10811A is so old, that it has settled.
Again, whenever I unplug the 5335A, it takes > 48h to reach the old value within < 2X10-9.
I recommend to let the 5335A warm up for several hours, with the lid unscrewed, but closed.
By briefly shifting the lid open, trim the 10811A as close as possible to 1x10-9 deviation and observe again with closed lid, and repeat adjustment 1-2 times.
I think, my counter meanwhile displays about 2x10-9 deviation, several years after last adjustment.
That's totally fine, because that is really at the limit of what is possible with an OCXO.
You need a Rb reference to get even lower deviation, about 1/100 of that ballpark.
Frank
-
#120 Reply
Posted by
Dr. Frank
on 21 Feb, 2019 13:55
-
Ok, I fired up my 5335A @ 9:39h. Room temperature was about 21.1°C, Gate Time 12.5sec for 1x10-10 resolution.
Reference is the Trimble Thunderbolt, with the "good" Trimble OCXO inside. Samples vary 1 digit only, i.e. by +/- 1x10-10.
9:39 => 9.999 999 970 MHz
10:34 => 9.999 999 969 MHz
11:24 => 9.999 999 967 MHz
So the warm-up or retrace effect (oscillator was off before) is <= 3x10-10
I opened the lid, and attached a thermocouple thermometer on top of the case of the 10811.
12:00 => 9.999 999 970 MHz @ 31.6°C, just opened lid
12:45 => 9.999 999 970 MHz @ 38.0°C, open
13:00 => just closed lid, 32.0°C
13:17 => 9.999 999 969 MHz @ 29.4°C, closed
14:17 => 9.999 999 969 MHz @ 29.3°C, closed
So, the change over temperature is about/less than 1x10-10 over a 9°C change.
Your OCXO seems to be too sensitive against temperature changes, although it might still be in spec. of the 5335A manual, page 1-4, i.e. <7x10-9, 0 to 50°C.
My 5335A drifted by about -2.2x10-9 over at least 2 years, maybe longer, see Reply #27 here in this thread. OCXO was continuously running since that time.
Frank
-
#121 Reply
Posted by
tkamiya
on 21 Feb, 2019 15:04
-
I guess you have a particularly good OCXO. I know mine sat idle for years before I started playing with it. I'm going to keep it plugged in for a while and repeat the test.
Maybe all it needs is a cup of morning coffee.
In few days, I will have another 5335A, this time with factory option 10. So I will be able to perform a direct comparison.
I actually have many rubidium modules. The problem is power consumption. It takes 2.5A total per spec, 1.8A actual measurement. I actually tried to fit that all in. I couldn't find a good placement for the power supply to be not near input and not near Rb. So I am going with an external solution.
-
#122 Reply
Posted by
Dr. Frank
on 21 Feb, 2019 15:19
-
Maybe, but in my memory, the other (repaired) 10811 inside the 5370B performs similar.
I also switch off my FRS-C Rb standard, due to high power consumption.. as it's mandatory to mount them on a big heat sink / metal plate.
Several Rbs inside one case, with one common heat sink might be a problem.
I packed 2 power supplies, 1 Rb, and the GPSDO inside a slim 19" case.. everything fine so far, only the Rb affects the Trimble stability a little bit.
Frank
-
#123 Reply
Posted by
tkamiya
on 21 Feb, 2019 15:30
-
I might have exponent expression wrong. My measurement was : 10,000,000,00x where x changes. That's 10^-9? or 10^-10? Input is 10MHz. I always mess this up!
The GPS I have is PRS-10 and FRS-C. I intend to use former and slave it to GPS. To avoid any interactions, power supply will be linear and external. So it'll be shielded from temp change and magnetic field.
I am revamping my house standard.
I was also surprised how much Rb drifts. It took good 24 hours to completely stabilize.
-
#124 Reply
Posted by
Dr. Frank
on 21 Feb, 2019 15:51
-
I might have exponent expression wrong. My measurement was : 10,000,000,00x where x changes. That's 10^-9? or 10^-10? Input is 10MHz. I always mess this up!
The GPS I have is PRS-10 and FRS-C. I intend to use former and slave it to GPS. To avoid any interactions, power supply will be linear and external. So it'll be shielded from temp change and magnetic field.
I am revamping my house standard.
I was also surprised how much Rb drifts. It took good 24 hours to completely stabilize.
That change would be x * 10
-10.
The un-modified 5335A has 4s max. Gate Time, so only 10
-9 will be displayed/resolved.
So maybe you should have a 2nd look, what you really measured.
My FRS-C needs about one week, until it has stabilized to parts in 10
-11.. that drift-time always depends on the limit criterion used.
I have provided an external potentiometer to adjust the Rb to 2*10
-12, but drifts are on the order of 1*10
-11.