HP 5335A Timer / Counter - Anything i should know?

[max_torque]

After messing around with various projects that require accurate (or at least repeatable) timing, i've finally bitten the bullet and brought a "proper"  timer counter, the classic (very classic.... ;-) HP 5335A.  Unfortunately it doesn't have the high precision oscillator option, but i was wondering if there are any tweaks / tips / or things i should be aware of with this meter?

It seems quite a popular meter, so i immagine there are quite a few in hobbiest / Amature hands etc!



It ran out of calibration at the end of last year, but i guess now that it's nearly 28 years old it'll be fairly stable??

[Vgkid]

Nothing really bad with them, unless you fry the input assy. The only issue (mine will refuse to read input if chanels are switched during use,yet passes self check for both ). The opt 10 can be found on ebay if you want one.
Also you have the front display cover plastic grrrr. Mine doesnt.

[babysitter]

Having serviced its little brother 5334A just today:

Look for burned resistors at the input, that is the main thing. Mine had a burned 50 Ohm Termination on Channel A, which I didnt realize that the only time I used the internal termination I was using channel B. Also, the first series Resistor in the Arming path was burned, too, but it works again.

The termination was fixed before, the original ckt is 2 100 Ohm in parallel which were dark burned, and another pair of new 100 Ohm resistors was stacked on top without removing the first ones ,I fixed it by replacing the stacks with a 56 Ohm and a 402 Ohm O0 The Arm path was a 619 Ohm resistor, replaced with a 620 Ohm type measuring 619 - not that I expect any effect. All working well now. 

Mine was without Option 010 also, now wonder as the price was below a 10811 which is the thing you might want for the 5335 also. But a cheap OCXO from ebay went in with ugly connections, doing a good job, too.

[max_torque]

Yeah, i noticed the 5v max input limit in the manual when 50Ohm termination is selected, irrespective of frequency or attenuation.  Could imagine that it would be pretty easy to accidentally smoke that, especially as otherwise you've got 250V to play with!!

Mines Ex-MIL, must have been de-mobbed quite recently as the cal sticker says DEC13 for recal, so only about 6 months ago.  Will probably pop the lid off for a look around (check those 50Ohmers!) and perhaps any caps etc that are looking crispy (build date on mine is WK26 1986, so it's getting on a bit).  I suspect it's been rack mounted in a fixed system assy, because it's in excellent condition, with few scratches, all it's buttons, and ALL the BNC's are undamaged with little wear. THE HPIB connector also looks immaculate and dust/dirt free, so i suspect it's been remote controlled for a lot of it's life (which would also explain why it's only just been de-mobbed if it's been part of an active system, and hence difficult to just 'upgrade' unlike a unit used for bench test etc.

Will have to keep an eye out for the Oven Oscillator on ebay, and i'd love to find a cheap HPIB -> USB adaptor (or knock one up with a suitable micro controller ;-)

[babysitter]

I employ a Prologix USB Adapter for my GPIB Stuff.

Into my counter went a C-MAC STP2390C OCXO, free-wired to the Option 010 Connector (Not 10811 Layout); looking ugly but seems to work fine, tonight it is busy with a TIC Measurement Rb against its own reference output. The 5334 both A and B have a quite unsatisfying internal oscillator. Counter plus OCXO were ebay for less than a single 10811 or 10544

(Which i think will fit yours directly. Take care, several different versions of the 10811 exist, some with relaxed specs.)

Next intermediate goal: Hack together a C channel option...

[Dr. Frank]

After messing around with various projects that require accurate (or at least repeatable) timing, i've finally bitten the bullet and brought a "proper"  timer counter, the classic (very classic.... ;-) HP 5335A.  Unfortunately it doesn't have the high precision oscillator option, but i was wondering if there are any tweaks / tips / or things i should be aware of with this meter?

It seems quite a popular meter, so i immagine there are quite a few in hobbiest / Amature hands etc!



It ran out of calibration at the end of last year, but i guess now that it's nearly 28 years old it'll be fairly stable??

Nope, the standard XTAL is crappy and will stay crappy forever. (crappy means unstable)
The calibration therefore is near to meaningless.

An ordinary XTAL is good for 1e-6 stability over a day or so , short term stability also not very good.
Temperature coefficient is also bad.

Contrasting that, this instrument resolves about 1ns or 9 digits/sec, therefore an OCXO is required.

Try to get an option 010 =  10811-60111, that is the cheapest one of that series, but after 20+ years, rock stable, i.e. < 1e-10/ day / month.
Should sell for less than 100$. Or get an external GPSDO.

Well, a picture of the interior of your counter, and most important the revision numbers of the PCBs and the complete ser-No prefix of your counter would be great (see below).
Do you already have the opt. 040, that's the GPIB steered front end?
And what about opt. 030, C-Input?

There exist mainly two versions, the elder one covered by manual 05335-90005 (1980, ser prefix  #2024) or  05335-90021 (1983, #2224), both online, and the newer version covered by 05335-90044 (1994, #3145A) (available by Artek, I think).
I think, they made severe changes in between, regarding the counting chains. Not sure, anyhow.

Oh yes, and get HP journal 1980-09.

Mine is 2820Axxxx, from 1988, and it contains newer motherboard and opt. 010, 030 (new), 040.

Well I tweaked  one electrolytic, which controls the manual gate time, so that I could measure up to 12.5secs, (instead of only 1s), to get 10 digits from the front panel.
Otherwise, GPIB control would do the job, up to 11 digits.

Anyhow, it's a very nice, versatile  instrument, but you also need now a GPSDO to enter time-nuts world,as mentioned before
Thunderbolt and Lady Heather make a nice couple..

Frank

[babysitter]

My intention changed over time as I have quite good external 10 MHz available:

I first was thinking about ovenizing it with a QH40A from Kuhne elektronik which is a heater module for (HC49) crystals.
Then I thought about replacing the original ckt with a TCXO, but finally i have embedded this OCXO in it.

Consider using the ext. Ref Port: Either you could feed a good external 10 MHz source into the device, or put your best source inside and get it out of the Ref Port for other purposes. Put some planning into it.

Glad to hear that the 5335 also has a crappy timebase

[Dr. Frank]

Glad to hear that the 5335 also has a crappy timebase

No, my friend! 
Mine already came with the 10811 

[max_torque]

Mines 2626Axxxxx, and only has the 040 'extended HPIB' option fitted (no high speed channel C, although that doesn't matter for what i am going to use it for).  I'll pop the hood and take some pics inside tommorrow.

Funnily enough a GPSDO is something i'd started, but not finished a while back:





;-)

[Vgkid]

It would be interesting to know why they didn't fit the 10811A spec'd oscillators to these units, as opposed to the 10811-6011 oscillators.

[Dr. Frank]

It would be interesting to know why they didn't fit the 10811A spec'd oscillators to these units, as opposed to the 10811-6011 oscillators.

Hu, waddaya mean??

The 10811 OCXO exists in many different versions, i.e. HP10811A/B/D-xxxxx, differently specified (Allan statistics), different pinning, or in a doube oven.

The 5335A, opt. 010 and the 5370B both have the 10811-60111 version, which already is very good... after all those years.

If I find the datasheet, that specifies several of the versions, I'll upload it here.

Here's the document about the various versions:
http://www.leapsecond.com/museum/10811a/90027-1.pdf

Frank

[babysitter]

Glad to hear that the 5335 also has a crappy timebase

No, my friend! 
Mine already came with the 10811 

Well, yours... your equipment I suppose is one of the best cal labs around. And with competent end enthusiastic personnel, not just soe demotivated guy who needs to do it to bring home food 

But I somewhat understand the reasoning behind HPs decision for those designs.

[max_torque]

Here's a few pics of the internals.  Looks like it must have been built some time late in 1987, as most of the iC date codes are mid '87:






















I'm definitely going to be looking for an 10811a to fill the empty slot ;-)

[Wim13]

I got mine 5335a very cheap, because it was totally dead, but cosmetic very good condition

After lot of research, all of the tantal C's on the input board were totally short circuit. The blue bulbs in the picture.
The power supply is short protected, so replaced all tantals, and the unit was up and running, and in good condition.

In my unit there was a 10544 ocxo, whixh is very good, compared to my Rubidium standard.
i also have a 10811, but that unit is worse compared to the 10544.

For precision work i use the rubidium clock, connected to the external clock input.

Also replaced the FAN for a silent computer fan, works very well no fan noise anymore.

[max_torque]

Interesting about the fan replacement, mine i extremely noisy, and seemingly over powerful for the heat generation capability of the unit!  I wonder if being "rack mountable" the fan is spec'd to get relatively "hot" air, rather than the typical 25-30degC air of a typical electronics lab??  A quiet fan would be nice, looks like mine is powered off the unregulated ~25vdc rectified bus.  Fan is 30Vdc rated

Regarding swapping in a OCXO, the manual suggests a jumper wire must be removed to disconnect the original "on pcb" clock source, but where is this wire? (maybe later units don't have it?)  I can find all the components in the original clock source, but it's not immediately obvious (to me ;-) where that jumper is.

Thirdly, are there any affordable GPIB -> USB adaptors, they all seem silly money for what is effectively a parallel to serial converter 

[Wim13]

Replacing the FAN....

Old FAN: Papst 8500N, 110 Volts 34 dB, 61 m3/hour

New FAN, Noctua, NF-R8 12 Volts 17 dB, 53 m3/hour

The unit stays very cool, no problems with temperture.
The uP was getting hot, also in the old situation, so placed a
cooling strip on th uP.

Fore the GPIB i use the Prologic, is also quite expensive, but works very good.
But was already in use for other equipment.
can make your own software on the easy way, as post somewhere above.
i also use Free pascal for that.

[ve7xen]

I have no experience with it, but I've heard this Agilent 82357 clone works reasonably well:

http://www.measuretechnic.com/?product=m82357-usbgpib-interface-alternative-to-agilent-82357b

It can be had for < $100, though still a bit expensive for what it is, it's not too bad a price for such a niche item, assuming it works as advertised.

[AllanMN]

I also just bought one of these too!

I'll post pictures in a day or two. Also, w/o the high stability time base I believe but with the 1.3 GHz option.

Considering a necessary timebase upgrade, for a cheap option (<<$100), the morion mv89a double oven OCXO's for $25-35 on ebay seemed like the best option for possibly (depending it's condition) 10-11/sec stability and even 10-8/year discussed here: http://www.ke5fx.com/Morion_MV89A_IEEE_IFCS_2002.pdf

10E-8 would be quite good enough for me.

The HP 10811 option, from the specs I saw did not seem to be nearly as good, and looked likely $100+.

[babysitter]

You rather have a look at the power supply, the Morion neets lot of juice on the 12V line where the 10811 is only supplying oscillator and oven controller because of its dedicated Heater connector.

[AllanMN]

Well, here's what I got off ebay, a serial prefix 2323A model with option 010, 030 and 040. Typical horror story, the seller shipped with the fan flopping around by the power leads smashing things up a bit. I collected 5 loose parts and identified a couple broken leads.

But, I got the 10811-60111 timebase unexpectedly.

I need to replace the broken off caps and transistors before I'll power on. It was shown working in the listing.

Allan

[max_torque]

Wow, nice to get the precision Osc & high speed channel C options!  Less good about the shoddy shipping and subsequent damage 


The more i see how clean and undamaged my unit is, the more i'm convinced it's lead a sheltered life in some air-conditioned and filtered test system somewhere!

[Thomas]

For a GPIB interface, you should consider this:
http://galvant.ca/shop/gpib-usb-adapter-rev3/
Low cost, $59 on sale now

I got one for my Racal Dana 1992 Universal Counter, haven't really gotten around to testing it properly.
But I get some response on the counter, the REM and ADDR leds light. Just played around in a terminal.

Shows up as a serial port on the computer.

Edit:
Oooh, almost forgot: nice instrument you got there!

[Vgkid]

Well I tweaked  one electrolytic, which controls the manual gate time, so that I could measure up to 12.5secs, (instead of only 1s), to get 10 digits from the front panel.
Otherwise, GPIB control would do the job, up to 11 digits.
Frank

Can you go into detail on this.

[Dr. Frank]

Well I tweaked  one electrolytic, which controls the manual gate time, so that I could measure up to 12.5secs, (instead of only 1s), to get 10 digits from the front panel.
Otherwise, GPIB control would do the job, up to 11 digits.
Frank

Can you go into detail on this.

Change A4C36 from 10µ/10V to 22..47µ/10V. Correct value depends on tolerances in the circuitry, and on the capacitors leakage current.
This will increase "Gate Mode NORM"  time from (20ms..4s) to ca. (50ms..12s), delivering max. 10 digits instead of max. 9 only.

Edit: I assembled a 22µF/10V Ta capacitor for 12.5s max.
In later versions, HP replaced all the tantalums they used before.

[Orange]

Hi,

Attached you will find V1.1 EPROM images. My HP5335 counter was equipped with V1.0 firmware in two ROM sockets, so I split-up the V1.1 from an 27128 into two MCM68764 EPROMS.
Main changes for this firmware are related to the HP-IB section.
This is also a safe upgrade to get rid of the MK36000 time bomb.

Works OK in an 'old' style 5335 with two ROM sockets.

Edit : Beaware of the Channel C prescaled factor; V1.0 has a division ratio of 20. The version here posted requires a div. by 64