Can a TO-39 diode be replaced with a DO-41? HP428B Repair

[Kleinstein]

The proble uses a flux gate magnetometer to sense the rather weak magnetic flield of the DC current.  From the drawings it looks like parts of the probe are not in the 20 kHz AC flield and thus can hold quite a bit of residual magenitization.  Mechanical stress for example can also effect the magnetic properties and this way the residual field.
I have once build a flux gate sensor and a nasty source of changing magnetic field was turning the office type chair with some steel in in the back. As the sensor was very sensitive I could even see this some 3 m away. I would not expect the sensor here to be that sensitive, but 1 mA of DC current is still not a strong field.

There is a chance that there is still something wrong with the circuit / probe. E.g. dirt on the clamping part of the probe can make the proble more sensitive to external fields and possibly the clamping force.

For the adjustment and debugging it would really help to get rid of the feedback loop, as this can introduce quite odd effects and it makes it much more tricky to interprete the signals.
The wrong polarity would suggest that there is still some major problem in the closed loop case.

As an additional complication the zero adjust adds some 40 kHz signal to the probe.  In the open loop case one should be able to use the zero trim pots to counter the residual 40 kHz signal. The scope probe may be as senstivie as the tube base amplifier. So one should be able to directly see the probe signal.
If the drive signal is good (symmetric) and there is no residual magnetization, there should be rather little 40 kHz in the zero field case. As the probe can be quite sensitive even little residual magnetization may be visible.

[jeffjmr]

More progress. Alignment going well. Next step calls for minimum AC voltage by “padding” either of two 6800pF mica caps.

I can guess that “padding” means increasing capacitance. Seems an odd, very time consuming method of trying to get to a minimum voltage reading.

Can someone confirm or explain?

Update on L7. Removed and it was obvious someone else had been into it before. The ferrite was broken and part of it was no longer in the case. I found a place that claims to have them but they were closed by the time I called. We’ll see next week if the have them and the price. So I’m done for this week.

Thanks,
Jeff

[Kleinstein]

The adjustment with the mica capacitor looks like is for symmetry of the oscillator, mainly to get C53 and C54 with the same capacitance. Unless C53 or C54 have been changed, chances are that the balance is still OK. So this would likely be just a simple check.
The 2 equal resistors would check the the voltages are indeed opposite and same amplitude. 50 mV difference with some 18 V of amplitide is quite a tight tolerance.
There are kind of 2 resonators with C53 or C54 and the halfs of T5. So they are coupled but should still work best if symmetric.

For the L7 trim, there would be the alternative to adjust C27: the missing ferrite part would lower the inductance and would thus need a little extra capacitance in parallel to the existing C27. So one may get away without replacing the adjustable core part.

[floobydust]

[...] Update on L7. Removed and it was obvious someone else had been into it before. The ferrite was broken and part of it was no longer in the case.[...]

The slugs have standard sizes and threads, that can help track one down. I think the powdered iron slugs rust and jam in there.
Micrometals threaded cores
Surplus Sales of Nebraska

[jeffjmr]

That’s not how this one appears to be constructed.

The coil is fixed and has half a ferrite fixed around it. The upper half, shown on right in pic sits atop the bottom half and adjusts by rotation to align or misalign with gaps in the bottom half.  Again, the top half is the only part that moves, and there is no screwdriver slot or hex wrench slot to facilitate rotation. How that small change in alignment of the two halves can vary the voltage by any significant amount is a mystery to me. The whole assembly is only 16mm in diameter. There are six pins on the bottom. Those depressions in the center are part of the bottom half which does not move, not because it is jammed but because the coil wires protrude through the gaps. It is not meant to move.

Now that I know how it works I am going to reassemble hoping the small chip missing from the top half is inconsequential. The metal housing was crimped tightly preventing the top half from rotating. Now that I have corrected that I believe I will be able to adjust it with slight pressure from a pencil eraser.

If I measure C53 and C54 and find them matched, can I assume that will result in the minimum voltage asked for?

Jeff

[Kleinstein]

The adjustment would be with the small central part of the core of L7.  Rotating the halfs of the core again each other would also work, but would be a rather unusual way of adjustment, but could work. With the part missing the inductance likely is a little on the low side anyway. So likely keep them aligned to get the near maximum inductance.

L7 together with C27 forms a resonant circuit and it may only take a relatively small change (a few percent) to have have quite some effect on the resonant curve. R116 only gives a little extra damping.  If L7 is out too far one can just add to C27 if needed.


C53 and C54 are only part of the symmetry - though possibly a major source of tolerance. There is also a little effect of the tube. So it would not be about measureing C53 and 54 out of circuit, but testing in the circuit as described with the 2 additional matched resistors to make it a bridge. Ideally do the test twice with swapping the resistors and this way average out errors in the resistor matching. Still no need for overkill, as the following tubes can also have tolerances and there is an additional adjustment with R98.

For the drive adjustment it may help to have pin 3 and pin 4 of the sensor shorted or the link for the feedback loop open to avoid an asymmetry in the probe from current flow. Other initial adjustments for the amplifier would anyway work better with the feed-back link removed.

[jeffjmr]

So something is way off, as the voltage specified for that step is way off.

The page attached is from the wrong manual. My serial number starts with 0995A. HP’s serial numbering scheme back then leaves a lot to be desired. The manual from HP’s archive part number 00428-90003 dated 1970, says it is for serial numbers starting with 995 and above. I assumed this was applicable to my unit.

A version of the manual of which I have a very poor, mostly unreadable copy of is 00428-90005 dated 1983 and says it is for serial numbers 0995A and later. 0995A vs. 995?!?!?  Older units start with 131 or 601! That is the one I suppose is for my unit. Other discrepancies between instructions in the different manuals are some state disconnecting feedback while the other makes no mention of it or specifically says to reconnect it!!!!!!

I compared the alignment pages as best I can and they read practically the same, except the newer manual adds that the voltage expected for the test in paragraph 5-42 is 2.5-3.3v RMS. I am getting 102mv. Steps 5-36 to 5-40 are all good.

V5 tests fairly good though not super strong on my tube tester. The 12vdc feeding R49 is my measured rail of 11.7vdc. The 16.2 vdc shown for the resonant circuit is 15.8. But the waveform shown right above it which look like 10v p-p sine, is a jittery .5v p-p sine.

Again, the bottom half of the L7 core does not move. The coil is firmly attached to it and wires from the coil pass through the gap on the bottom half of the core. The top half is free to rotate when assembled but how they expected you to rotate it is a mystery. Either way, I don’t believe any setting of L7 is going to raise the voltage from 102mv to 2.5v.

Your knowledge of the circuitry leads me to believe you are already referencing the schematic, but if a full schematic is needed I’ll post it.

Thanks,
Jeff

[floobydust]

edit: with the powered instrument off, you can inject a sine-wave at the C28 0.01uF cap and use a scope to find the resonant frequency of the V5 input phase network at the grid, to see if it's alive.
 
It's a Ferroxcube pot core 1408PA250-3B7 material. I've never seen an adjustable one before. That would be 14mm OD, AL=250mH/1,000T per core pair.
source: old 1964 Ferroxcube databook then it became Epcos, TDK, Yageo P14 series. Philips did make them too.
Could not find 14mm parts here: Surplus Sales of Nebraska  Ferrite: Pot Cores - Slugs & Miscellaneous

[Kleinstein]

The adjustment of L7 would not be about rotating the whole lower or uper half but moving a small (e.g. some 4-5 mm diameter and maybe 8 mm long) screw like part in the center. The outer part is normally fixed and the parts aligned to match.

The correct / wrong trimming of L7 can have a quite large effect.  From a crude estimate of the resistance R116 relative to the impedance of C27 the Q of the resonator could be as high as 200 though more likely more like 100.  So with the right setting the amplitude could go up by up to a factor of 100 and the waveform would naturally get more sine like.

One could look for the resonance of L7 by some external excitation (e.g. to V7 cathode) with the meter not powered. With the broken core chances are the resonance would be a bit higher than the target 40 kHz.

Is the signal at C51 about correct - it looks like it should be some 0.5 Vpp ?

[jeffjmr]

“edit: with the powered instrument off, you can inject a sine-wave at the C28 0.01uF cap and use a scope to find the resonant frequency of the V5 input phase network at the grid, to see if it's alive.”

I assume I would sweep and look for max amplitude? Say 20-60mhz or so?

And I see now how that tiny center piece might be threaded to adjust L7. I’ll have to remove it and disassemble again to try to free it.

Jeff

[Kleinstein]

The frequency range of interrest would be some 20-60 kHz, not mHz or MHz.  The resonance would show a likely rather narrow maximum in the amplitude.

[floobydust]

If L7 is in bad shape, it will have low inductance or mangled windings as well. If it's still out of circuit you could try measure its inductance- goal is ~16mH end-end which gives around 40kHz with C27 1,000pF, for resonant frequency.
It's a bit odd V5 grid voltage is centered around +16.2V yet the cathode TP4 is less, run off +12V. So I would also check the +16.2V is good. With +ve grid voltage (wrt cathode) you could bias V5 on hard and draw grid current (seen as small voltage drop across R48) which would load the phase circuit tuning tank down a lot. That could be happening too.
Does that make sense? If V5 was biased wrong (grid current) it would load the phase tuning circuit down heavily and kill signal amplitude there. V5 plate would have low voltage then as well I think.

[jeffjmr]

Sorry. Hi meant khz but typed mhz.

There is a very obvious peak at 43.6khz. Falls off slowly below but sharply above.

Manipulating the top half as best I can only raised the peak to 45.2 which is where it is now. I can’t get it lower. Online calculator says I need 1277nf to get 40khz. But really need to pull the C27 to measure to calc the inductance now. Barring the ability to find a new L7 I guess I’ll try to get an appropriate cap.

I mostly understand your point about the grid voltage, Flooby. Still have a lot to learn about tubes, but I can check the volts and current after I get this resonant circuit squared away.

Kleinstein, I checked the C51 wave some time ago before the alignment attempt and it looked good though I can’t read the specified voltage. You mention .5vpp so I’ll check that and the other waveforms again also.

I’m taking up way too much of your attention but it is sincerely appreciated.

Jeff

[Michael Terrell]

I think that is a Stabistor, which is a special grade of reference diode. HP used them in their lab power supplies. They are temperature compensated, and are basically obsolete now that newer IC based references are available. They are built with a Zener and a normal doide in series to reduce most thermal drift