Repair of HP 5245L Nixie Frequency Counter - V1 finished , V2 in progress

[Cubdriver]

Will do!   

-Pat

[VK5RC]

An interesting comment re glass enveloped crystals I received from a fellow ham whose technical knowledge I would rate very highly, the 'glass' crystals pick up contaminants through the glass and this eventually can make the crystal off frequency HOWEVER after a few MONTHS of being left ON and HOT the contaminants will often evaporate and return the crystal to frequency. I am not planning to try this as my lab is not close to the house and I leave it unpowered except for 1 or 2 very reliable components. I thought very interesting. it might also be why HP wanted the oven/crystal left powered ON.

[Kleinstein]

The glass is usually quite tight, unless it is somewhere cracked. So nothing serious coming in from the outside, except maybe small amounts of oxygen, hydrogen and helium. The bigger trouble is volatile stuff moving from the walls to the crystal and back - this can happen during heating and cooling.

Having the oven hot will also speed up aging - so long term drift will be higher with a running oven.
Turning the oven off and on may give some variable degree of hysteresis, but I am not sure this old counter will resolve / note it, as most of the time the turn on sequence will be rather similar.

[VK5RC]

In buying a few plug ins for the 5245L I found an "As Is " unit for the price of the plug in.
The 5245L looked a bit rough -  see photo 1 - but not much to lose.
Opened her up - no nasty surprises - all the bits there - no bulging caps. Ohms test of the mains plug gave about the right figures. Fired her up with an AC lab supply slowly over 5-8mins or so. Drawing about 0.6A (with 230V AC).
Looked at the back of the unit - switched the frequency source to its internal unit! A counter always works better with a frequency source!!!!
She works, I had a few close fit knobs, I don't have a "double" knob (I don't know the correct terminology) so put a cheapie on and drilled hole for the small central pot control to stick through. Gave it a 15min clean with IPA.
Looks Good (to me)
Final photo is it counting a GPSDO, early after switch on. It got to 9.999 985 - no adjustment.

[Cubdriver]

Wow, Rob - that cleaned up quite nicely!

I'd be inclined to swap the sample rate knob with one of the double barred ones on the back panel to add a bit more uniformity to the front.  I'm curious to see how accurate relative to the GPSDO it winds up after it's been on for a while.  Congrats on getting the old beast back up and running - looks like it's a 1967-ish one.

-Pat

[bitseeker]

Wow, amazing what a frequency reference can do. All the Nixies look good too.

[VK5RC]

Thanks guys, I am amazed some eBay sellers put so little into selling the item, the clean up took me 20mins, he/she could have asked for another $100-150.
I have some original HP knobs coming but they are a little bigger, I think I will put them on the back, bring the back ones to the front. For the dual knob, I have a spare  little Tek red knob and have tried (successfully) drilling a hole in the sensitivity knob to make a hybrid dual knob.
Rob

[bitseeker]

Knob hacking! I like it.

[VK5RC]

I like the term "Knob hacking" , first coined by bitseeker
I used a cheap knob first,  seen here on the sensitivity control, I used a 6mm drill down the collett to find centre, then drilled a 3mm, then a 6mm holes from behind. The red little Tek knob is 10.3mm in diameter, so I drilled a 10.5mm hole from the front but used a drill press at quite high speed, with a low feed rate. I used a 6mm drill to hold the knob (held on by its own grub screw)
 I think I will try one of the bigger spare HP knobs next.
Below is what the counter settled down to after about 1 1/2 hrs with a GPSDO input.
I have been battling which plug ins can be used in which 'serial numbers' 5245Ls, as there was several changes throughout production , once I think I have it correct I will reproduce it here.

[bitseeker]

Rob, if you hadn't said anything, I'd have thought that concentric knob hack was possibly original. Well done!

[VK5RC]

I am reasonably happy with the hack. See below.

[bitseeker]

Yep, good stuff. Such clean knobs, too!

[factory]

Lots of cursing etc.
Rebuilt the oven using some stripped wire wrap wire in place of the fine 'not tin-able' wire, joining the crystal to the outside world. All back in the chassis.

The heater control circuit is a most unusual design in that it is not a DC sensing voltage but a 300kHz signal, derived from Q1 on A25. The wheatstone bridge type sensing circuit has resistive and capacitive elements.
I have tested all of the components around Q1 ( the oscillator) and Q2 the buffer/driver and can find no fault (pulling out both Q1 and Q2 and all of the caps as well). It is NOT oscillating. This absence of oscillation allows the heater circuit to switch on fully and stay switched on, certainly getting over 70C.

snip...

Looking at the manual it states the frequency of the oven controller oscillator is actually 3Kc (kHz) not 300kHz.
I'm having similar problems with a 5243L of similar age to the 5245L that started this thread, it has failed in a different way as it doesn't turn the oven heater on at all. I've found the oscillator based on Q1 is not running at all, I've tried changing both Q1 & Q2 with alternates but they don't work either. I found this out after fitting a fuse to protect the transformer in case the heater element fails S/C in the future. 

I also have an incomplete 5245L (missing all removable boards) that still has all the oven part intact, the 5243A-65T board has some component differences as it's slightly newer, testing that one the oven controller is running at 2.7kHz. I've now pulled both of the controller boards and tested them separately using some resistors & capacitors in place of the bridge circuit in the oven assembly and both the Q1 oscillator sections work fine, this makes me think the problem is actually within the oven assembly itself, either a component has failed in the bridge circuit or it needs adjustment, trouble is the adjustment resistor isn't accessible as far as I can tell.

David

[Cubdriver]

The yellowed electrolytic in the first board image certainly looks to be a bit past its prime...

-Pat

[factory]

It was knackered, the yellow stuff is just a sleeve, I've noticed the Sprague capacitors with the white rubber seal seem more prone to drying out, the blue Philips capacitor in the next picture replaced it.
There were two more on the regulator board that also needed replacing, the third had already been changed at some point in it's working life & was fine.

David

[Cubdriver]

Yeah, that particular series from Sprague does seem to dry out pretty much as a matter of course.  I suppose in their defense they probably didn't expect people to still be using whatever they were in 55+ years later (I think I see a 1963 date code there), so perhaps there should be some slack given in that regard.

-Pat

[factory]

To be fair those capacitors didn't stop the counter from functioning, just something I've noticed from working on other vintage HP TE, that those are usually bad and the others are mostly OK.

Yes this counter is from 1963 (serial prefix 328), the 5243L (20MHz) was introduced in 1962 and quickly became obsolete with the introduction of the 5245L (50MHz).

David

[factory]

Well it turns out the adjustment resistor (R4) for the oven temperature sensor oscillator is accessible, it's hidden under a cap on the side of the oven assembly, however this adjustment is completely undocumented, could I have the victim of the phantom twiddler?

Before I had a look at that I checked the cold resistance of the leads for the bridge components and compared these with the working one.

Resistance of wires in faulty 5243L;
Red to Blue = 9.83k
Blue to Grey = 862Ω
Grey to Orange= 1.215k

Resistance of wires in working 5245L   
Red to Blue = 10.40k
Blue to Grey = 1.323Ω
Grey to Orange = 1.214k

Ambient temperature was approx 15°C, the red/blue wires are the Thermistor (RT1), the blue/grey wires are the 1k pot + 442Ω and the grey/orange wires are the fixed 1.2k resistor.
After this I set the pot to match the working unit, reconnected the four bridge wires only and used an external -22V supply for testing.
Success it now oscillates, I noted that the adjustment will eventually stop the oscillator and it won't restart until the pot is turned back & the power supply cycled. I need to see if the other one does the same.
Also the amplitude differs between the two units, this maybe because the later one has a silicon transistor for Q2 & different resistor too, both are running about 1.3kHz without the rest of the circuit & heater connected, again need to do more testing.

Question is what to set the adjustment at, I would need to remove the crystal to find out the required temperature & fit a temperature sensor to monitor the oven temperature.

David

[Runco990]

Thanks guys, I am amazed some eBay sellers put so little into selling the item, the clean up took me 20mins, he/she could have asked for another $100-150.

Rob

When selling on ebay for profit, the whole idea is to go straight from landfill to buyer with minimal steps in between.  Just taking the photos is agonizing enough.   

It used to not be this way. 

Congrats on the amazing job restoring this!  I also have a very large collection of HP gear.  I restored and use every piece of it.  Sure don't make em like this any more. 

[Martin.M]

that beast makes me sick 

in normal counting mode without storage:  all ready counting the frequency.
in storage mode:  WRONG result.

the manual is here, also a nice programmable test generator from siglent.

PSU: done,

+13V exactly, no ripple.
+20V exactly, no ripple,
-15V exactly, no ripple.

the rectangle at #3 from A23 looks OK also.
the time base ist OK, 10/100/1k/ all ready

[factory]

There was a suggestion in another thread, that the neons in the display decoder age, the trigger voltage changes & they stop working correctly in storage mode.
In your counter, have you checked which display boards are giving the incorrect result?

David