HP 735A DC Transfer Standard

[gilbenl]

Another vintage piece of HP gear. I've been hunting one of these for a while and managed to get one off a member here. (THANKS!) Still looking for 3 more for a project, but wanted to share some pictures of one of these rare little boxes. Direct competitor with the Fluke 731, but lacks battery backup and 10V range. The Standard cell->1.018V output is 2ppm on both; the Fluke claims better performance on the standard cell->1V at 3ppm vs 10ppm on the HP.


P.S.:I followed the "Calibratory reference standard" thread and couldn't help but chuckle when I saw the "Stability" plot on this unit and then thought about the elaborate gilded lily test setups and parameters folks wanted. In 1969, you just left the sucker on for a month and had some poor sap plot the drift every week. Character.

[Vgkid]

Very cool, thanks for the upload. So are you going to tear down the oven?

[TimFox]

Any idea who manufactured the red precision resistors in the last photo?

[gilbenl]

I wish I could crack open the oven, but the manual indicates that it is not serviceable. It is replaced as an entire unit if something goes awry.

The parts list says HP was the manufacturer of all of the precision resistors.

One of the old caps is leaking so I'll be putting up some additional pictures when I get around to replacing the two.

[Vgkid]

I was curious to see if they built the oven in the same was as the HP105A,6012 crystal oscillator.

[gilbenl]

After leaving the unit on for about a week, it was clear that it was not meeting its stability spec and some repairs were needed.

First order of business was to repair the A1 power supply and preregulator board. I had noticed a bit of browning on initial inspection. Once removed, it was clear that the board had been exposed to excessive heat for quite some time. The epicenter was under the resistor that feeds into the preregulator for the reference circuit. I replaced both capacitors, the resistor and transistor. The transistor had been replaced before and the work wasn't quite up to par.

Desoldering was a bit tricky. The board did not tolerate heat very well at all, even with the iron set at 500. One pad began to lift in the process, but I was able to remove the heat before it was lost.


And with the new parts, which cost around $20 given that they were hard to find:


The next issue was that the unit had apparently been dropped at one point, causing the heavy transformer to deform the aluminum frame. As a result, the aluminum box had a gap that likely had an effect on temperature. This also caused the mains leads to be wedged between the transformer and outer case.


Despite the few dings and dents, I'm amazed at how new these parts looked. I have a few other HP devices from the late 60's and the transformers are all paper wrapped and come to me pretty rusty. This one looks brand new with no rust to be seen. Only sign of age is the "10 68" stamped on top. You can also see where some of the plastic had transferred from the mains wiring to the back case.


Very gently bent the case back into position without having to remove components. The unit was also humming rather loudly so I took the opportunity to crank down the transformer bolts with a ratchet. Silent ever since. I also removed Q1 to add some fresh thermal paste. Quite a hassle and probably wasn't necessary. Job done:


With the functional components repaired, I took a few minutes to give the exterior panels a facelift. Magic sponge+Goo Gone followed by IPA, then clean water and finally a light coat of McGuires Ultimate Black. The high gloss faded a bit after a few days and it looks pretty good, IMHO.



I've had it powered on for the last week or so and its remarkably stable. Maybe +/-3ppm noise. +/-1ppm transient drift as best as I can tell. My next project is to get my DMM's connected to a controller for data collection and will hopefully have some real numbers. Hope someone got some enjoyment out of this. I'd still like a couple more of these units!

[gridleak]

My loss ( oven failure )....group's gain ( access to construction of 735a oven)

My 735a oven lamp started to oscillate on/off every 10 seconds. That's a sure sign that either the thermistor had pulled away from the heater, or the thermistor failed. I had no option other than to crack open the oven and see what could be done. The insulation was old and very friable. Chopping at it with a screwdriver quickly separated the inner oven assembly from the outer can. The attached photos show that the heater (probably 28 awg flat nichrome) and the thermistor are all bonded together with some sort of enamel or epoxy. I did unsolder all the oven control precision resistors to check them out. They were still spot on.

The inner can must have first been coated with enamel, and then the bare flat nichrome precision wound (to avoid shorts) over the enameled can and another coat of enamel applied over the nichrome and the thermistor. Under a scope, it looked like the thermistor (sorry, no scope camera to show thermistor) had lost its glass bulb. The thermistor still had the proper resistance slope/beta. I placed a piece of aluminized mylar over the thermistor to retain heat, and then taped that down with kapton. I used fiberglass insulation as a temporary replacement for the original insulation, and reassembled the oven. No go, the oscillation remained. Since I can't bring a new thermistor out of the provided port to the outside without destroying the heater winding, my next move is to epoxy a new thermistor inside the can and hope that although there will be more thermal delay, there will eventually be a stable point reached.

Also note photos taken of the underside of the unit with three resistor adjustment ports on the smaller guard. Removing the guard reveals not only the three trimmers, but a set of plated holes labeled  A through G. These plated holes are joined to the leads of 6 manganin resistors. The original coarse pot was designed out and replaced with these "coarse" adjustment manganin resistors in later production runs in the early '70s. The circuit board in the oven is installed diagonally in the can. One side of the board is dedicated to the oven control circuit (right side), and the other side contains the reference circuitry (note metal can reference). Not having time to make a project out of the oven repair ( low probability of success) I bought another 735a off Ebay to replace my current unit. The Ebay unit was early production, and still had the coarse pot. I preferred the additional stability of the manganin jumper solution, so I cannibalized the recent arrival and put its oven into the original one. It's working fine now....As far as I know these are the first photos of the 735a oven internals to be posted anywhere.

[Vgkid]

Good luck with the repair. I guess i will need to tear down the 105a oscillator.

[gilbenl]

First, it makes me nauseous to see that an oven failed. I leave my 735A on 24/7 and as I write this, my eyes have oscillated to the oven lamp every 10 seconds. Those are probably the only pics of the oven guts on the net-probably anywhere. Thank you very much for posting!

The guts of the oven look like a mess, but as you said, the value comes from the precision resistors. Side note: The HP Journal from '66 announcing the 735a stated that the matched pair resistors were wound using wire taken from the same spool with each length cut consecutively...by nude virgins with grey beards and proper tongue angle.

The primary reference voltage out should be 5.91V (pin 7). The bridge puts out 35v, which is dropped to 12.7v which by A2R2. Why not swap out that resistor (A2R2) to give you 15V to the reference and throw in an LM399 or LTZ solution with the 7V output trimmed to 5.91? Stick that back in the can, use the pins you need and you'd have something equal if not better than the original.

If not simply age-related, my bet is that the failure was the bridge rect. or preregulator (what I replaced in mine). It was considerably overvolt before repair. For giggles, do you mind seeing what the voltages are feeding the reference and the heater circuits from the power supply board? If they're high, you might burn out the second board, too.

[gridleak]

When the temp. Controller circuit started oscillating months ago, I checked voltages and all were fine. BTW, there are no discrete transistor regulators in my version of the 735a.  It uses a 723 regulator instead.  Anyway, none of the temp control components on the oven board are fried. Even the thermistor looks okay using an ohm meter and a heat gun. I think that when the glass encapsulation came off the thermistor, it started losing too much heat....so that as soon as the target temp was reached the thermistor was now too cool. This resulted in the heater overcompensating causing the temp to overshoot resulting in heater oscillation. Thanks for the advise on turning the reference into lm399 or ltz. If the replacement oven fails, l'll go for that solution.

[Vgkid]

interesting that your 735 uses a lm723, I believe that the 10544 series oscillators used one in their oven controller. here is the schematic for the 105 oven.

[LazyJack]

Let me resurrect this thread a bit.
I recently scored a nice 735A, and the only problem was the oven lamp bulb was burnt out. Since i could not really find a good replacement, I went the LED way. However, simply a LED with a series resistor won't work, as the LED has a much smaller current and the oven activity is not really noticeable. So the solution was a bright orange LED, a 13V Zener and a 10k resistor. The bulb was smashed, and all these were soldered into the original lamp head. So, no changes to the instrument itself. The lamp nicely fits the holder and the orange LED gives a really fancy glow. With the above component values oven status is very well visible.

[pelule]

I got my HP735A in 2015 for just 15 EUR + 6 EUR shipping as defect/for parts only. I thought, for that price give em a try.
When it arrived, it was in a near excellent condition, just need some repair in the power supply (elocetrolyte), a new oven lamp and some calibration.

Mine was build in the 70s and the drift chart shows a much lower drift, seems HP improved the design.

In my experiance, it was quite easy to get all required parts from Farnell (see order codes on pictures).
As I was afraid regarding a potential lift-of of the solder pads on such an old PCB.
So I cut the wire and soldered the new part at the cutted.

I measuring the drift rate since that time, it is arround -2.2ppm/y (+/-1ppm with calibrated HP3458A)

Meanwhile I have added a 10V range (using a switch at the rear to switch between normal voltage output and 10V output.
Its actual in burn in - see chart

[LazyJack]

Yes, they have likely improved the design. Here is the chart of my instrument from 1974.

[blackdog]

Hi!

My HP735A is still going strong 

The drift with "normal" LAB temperature changes is less than 1PPM.
More than a week now the display shows this...


Its a nice little tranfer standard.

Kind regards,
Blackdog

[beanflying]

#1434 clocking in. 20 higher in build number than LazyJack's and on the other side of the planet 

So far after an initial tweak sitting very stably on my 34401a within a last digit on 6 1/2. micro volt had a small 3-4 microvolt offset to chase up other than that put the lid on and use it 

Interesting looking at how low the production numbers were over 4 years too.

[beanflying]

Apart from being sensitive to input loading this looks like being one of the most boring bits of test gear I own 

Min/Max Variation of 0.000001 Vdc with a 10+ degree temperature swing over 24 hours.

Mmmm now is the drift of the meter and reference excellent or are the reference and the meter just drifting together and really really terrible .....   

[BU508A]

Mmmm now is the drift of the meter and reference excellent or are the reference and the meter just drifting together and really really terrible .....   

You need more DMM (and perhaps more references) to verify that. 

[beanflying]

I have rack space available for the 3458a and a few LTZ's waiting for Vishay goodness

[jasonbrent]

These must be reasonably rare on fleabay. I didn't know they existed... would be a cool addition.

[ArthurDent]

Looks like I have the oldest 735A shown so far. Going by the serial numbers you can estimate how many of these HP units were made per year.  You can also see changes in the power supply board, the counter dial, and the oven board. It looks like it was a work in progress.

[LazyJack]

These must be reasonably rare on fleabay. I didn't know they existed... would be a cool addition.

Yeah, they seem to be quite rare. Ever since I got mine on a local auction site (for an embarrassingly low price), I was looking for an other one to have a comparison unit, I have seen maybe a couple for sale, either non working or for some crazy price.
 

[LazyJack]

Interesting to see that the chart scale was halved during the years. Shows that as they achieved better stability, it was not expected to go offscale on a tighter tolerance.

However, I do not understand how your unit can start from -21ppm deviation. Don't they set the deviation to 0 initially? Or was it not possible on the older models to set to this precision?

[ArthurDent]

LazyJack, I'm not sure if HP made great improvements in the 735A over the years to make the aging chart look better by half, or they just got better at selecting the zener so the spread was tighter, or marketing thought the +/-25 ppm chart looked better.  I assume they assembled the reference ovens to output about 5.9 VDC and if this output value was within that +/-50ppm range they could use the adjustments to trim the outputs; outside that range and you wouldn't have enough range on the adjustments to bring the outputs in. Other than selecting 2 series resistors before the oven was assembled, there are no adjustments on the oven. The main concern with the oven assembly was stability. The output series resistors in my oven are 200 ohms and 75 ohms.

My 735A oven output may have been 21ppm under the 5.9 VDC and they graphed the stability that indicated initially it was 21ppm low but still very stable around that value and within the range they could correct for with the adjustments. The later charts possibly ignored the initial absolute ppm value from the reference and just plotted the ppm difference from the initial value. Obviously the absolute value of the oven output is of no concern as long as it can be corrected for with the adjustments or my 735A would have never made it out the door.

One problem I had with my older oven and they may have made changes in the newer ovens was that there appeared to be corrosion where the leads were soldered into the holes on the oven board and the slightest touch would cause them to break off. I ended cutting a 1/4" off the ends of the teflon wires and stripping them to get to solid metal and resoldered all the oven leads. Here is a photo of my oven board that looks different than a previous posted photo of another oven board.   

[LazyJack]

Yeah, your explanation on the initial deviation does make sense indeed.