Making do with what you got/ An HP 3458 Opt002 reference with no heater

[View[+]Finder]

Hello All,
ViewFinder here.

In another posting elsewhere on this forum, I discovered that the HP 3458 Opt002 Reference I was intending to repurpose as a voltage reference had no current through the heater--hence no temperature control. That reference is out of service, on the sick list, waiting for repair and has been replaced with one that is working as expected.

I welcome advice on how to proceed with the repair and parts are on order. TIN has set the task in motion with direction on points to measure. So, when the OP Amp gets a new socket, I'll get started.

Stand by . . .

[TiN]

Do not use socket of any kind on precision analog electronics. What is your heater resistance on LTZ chip?
I recently made a mistake and supplied opposite polarity to A9 board. That killed opamp, but LTZ chip survived to my amazement  .

P.S. There is no point to add "making do with what you got" in thread name, we all doing that already. 

[Andreas]

Hello,

the first thing that I would do is measuring all voltages on the reference
and putting the values into the cirquit diagram.
It is very unlikely that the heater is defect.

with best regards

Andreas

[dietert1]

Yes, TINs conclusion about a burnt LTZ1000 heater was a bit premature. And even if the analysis Andreas proposed confirms the heater is gone, still don't throw away that reference board but try to get/make an external oven for it. A good exercise.
Replacement LTZ1000 is a bit difficult to get right now. Analog Devices offers samples in batches of twenty (800 or 900 $ entry) and the buyer needs patience.

Regards, Dieter

[View[+]Finder]

The 3458 reference (from Keysight) is on the bench for repair and testing; should have some data this week.

[View[+]Finder]

Hello,

the first thing that I would do is measuring all voltages on the reference
and putting the values into the cirquit diagram.
It is very unlikely that the heater is defect.

with best regards

Andreas

In an ambient temperature of 17C, the LTZ1000 is a couple degrees warmer (infrared thermometer) and the voltage output into an HP3478 is steady at 7.1409. The heater current--also on HP3478--was under 1mA.
Measured Voltages:
J401
1) 14.6
2) 15.7
3) GND
4) 14.6
5) 16.4
J400
1) 0.447
2) GND
3) 7.14198
4) GND

[Kleinstein]

The signal at pins 1 shows the LT1013 is correctly calling for a lot of heater power, so the temperature measurement may be still working, at least it is not the reason the heater is off.
One could check if the temperature-sensor is still working by applying a little more current to J400.1 (e.g some 47 K to 7 V or 15 V). This should shift the set point to a much lower temperature, so that the OP should turn off the heater. If this still works as expected, the temperature sensor seems to be working, which would be a good thing.

Chances are one could than add an external heater and still use the internal sensor. It would very likely need a modified control network (e.g. considerably larger cap for C412) as there will be more delay.

The voltage at pin J401.4 should be negative, for a standalone test one can also just keep this pin open.

One should check the voltage at pins 1 and 2 of the LTZ - there is a slim chance the bad part is the diode or transistor of a bad solder joint (unlikely, but not impossible).

[View[+]Finder]

The signal at pins 1 shows the LT1013 is correctly calling for a lot of heater power, so the temperature measurement may be still working, at least it is not the reason the heater is off.
One could check if the temperature-sensor is still working by applying a little more current to J400.1 (e.g some 47 K to 7 V or 15 V). This should shift the set point to a much lower temperature, so that the OP should turn off the heater. If this still works as expected, the temperature sensor seems to be working, which would be a good thing.

Chances are one could than add an external heater and still use the internal sensor. It would very likely need a modified control network (e.g. considerably larger cap for C412) as there will be more delay.

The voltage at pin J401.4 should be negative, for a standalone test one can also just keep this pin open.

One should check the voltage at pins 1 and 2 of the LTZ - there is a slim chance the bad part is the diode or transistor of a bad solder joint (unlikely, but not impossible).

One thing: I noticed that the board has an LTZ1000ACH which has a different heater curve in the data sheet.

I reheated the solder on the LTZ and Q401. Pin 1 (emitter of Q401) 14.91V Pin 2 (at CR401) .17V Heater current is still < 1mA.

What is the function of J400(1)? I don't have it connected to anything, same with J401(1)

[Kleinstein]

The pins J400.1 and J401.1 are very likely there for test purpose during the burn in period. Pin1 or  J400 can be used to independently check the 2 most critical resistors (that set the temperature) and for the burn in period they could also use this pin to temporarily change the set temperature - e.g. to speed up initial aging by letting the ref run slightly hotter for some time.
The voltate reading at the heater suggest that the heater is indeed open. So there is no more internal heater.
An external heater is more tricky with the A version, as the thermal coupling is not as good.  It is probably still the best one could do to revive the reference. Trimming the TC with a resistor at pin 3 would only work for a limited temperature range and likely not very good.

An exernal heater would likely mean:
1) add a heater very close to the case (fast coupling could be a challange)
2) slow down to heater control by larger values for C412, R416,R420 
3) likely a reduced temperature set point, as the external heater would cause more heat loss and would thus have a hard time to reach the same high temperature.
4) if the temperature control is only partially working  a resistor at pin 3 could be a last resort do lower the demand on the oven part.
 
The tricky part would be the heater - the faster the better. So it should be low mass and with good coupling to the chip.
The heater could be a resistor (e.g. resistive wire around the case) or a transistor (would replace Q401).

One could still opt for replacing the LTZ chip on the original board ( needs a little patience) and try the external heater separately.

[View[+]Finder]

Mouser has LTZ1000CH in stock, so "making do" and building a heater doesn't look all that interesting, considering my other projects. I have a reasonably good solder station and  desoldering tool from Hakko and can handle the mechanics of removal and replacement. Certainly some forum contributors will come forward with temperature suggestions. I'll replace the LTZ1000, maybe this weekend.

{I also found that one can order Cd and Sn on Amazon (for your 'elements' collection) so with the addition of a little crucible (also from Amazon) a home-brew 70:30 CdSn solder could be created. Is it even worth the trouble?}

[antintedo]

{I also found that one can order Cd and Sn on Amazon (for your 'elements' collection) so with the addition of a little crucible (also from Amazon) a home-brew 70:30 CdSn solder could be created. Is it even worth the trouble?}

Ignoring significant health risks associated with melting cadmium, which has a relatively high vapor pressure and is readily absorbed through lungs, it's not useful in this application. LTZ has kovar leads with 40uV/K coefficients against copper, 3uV/K extra from regular solder is insignificant.

[Kleinstein]

There is no need for special low thermal EMF solder with the LTZ1000 - this is about the 1 µV and maybe 500 nV range, not the the single digit nV. There will be some thermal EMF from the kovar leads - though even this is not so bad as often claimed: the temperature gradiens are largely symmetric and stable. The TO99 type case is not that sensitive to heat and the old ref is not very good anyway.

With the non A version, one could consider lowering the the set temperature, e.g. with a resistor parallel to the 15 K. This reduces the power needed and could reduce long term drift a little. Only really hot environments would need the high oven temperature.

[View[+]Finder]

The HP3458 A9 board reference is repaired! The easiest solution was a new LTZ1000CH from Mouser for $65. Getting the old one out was a PITA--had to cut the leads and carefully solder-suck the remains.

Attached is a plot of the first test.

UPDATE:
Overnight test, 5000 readings at 5 second intervals.

[dietert1]

To demonstrate that the heater is working now, please look at time correlations including significant ambient temperature changes.

Regards, Dieter

[View[+]Finder]

To demonstrate that the heater is working now, please look at time correlations including significant ambient temperature changes.

Regards, Dieter

Attached is a plot of measured voltage (by 3458 DMM) at 5 second intervals over 7 hours last night from 18:40--01:40. Also attached is a plot of ambient temperature as captured by thermistor in the vicinity of the DUT and logged on 34465 DMM.

The DUT was on a metal box. Photo to follow . . .  The heater is connected to the 'green' window on the PSU.

[View[+]Finder]

Attached are plots showing "Serial Correlation of Voltage Readings from an LTZ1000 Reference (HP3458 A9)" as measured by an HP3458 and for comparison, a KS34465 with a genuine Keysight 34470 reference. The testing comprised about 14 hours of data collection in two blocks over Feb 3-4 2021.

The DUT was the 3458 A9 board that was the subject of this post, now back to work, but not yet fully checked out for active duty in a DMM. The suggestions and well-wishes for a speedy recovery were most helpful.

The DUT was place in a metal enclosure and power was supplied by two separate LiPo 2P4S batteries which were in turn being charged by a Keysight 36313 bench PSU as pictured above. The output of the DUT was connected to the 3458 DMM as shown using copper forked-spade crimped connectors. All leads were twisted. The second DMM (KS34465/70) was linked from the 3458 by short wires (sorry TIN, that was the best I could do in the time at hand). Finally, a thermistor was placed near the DT and yet another DMM was tasked with logging temperature.