HP Meter dial replacement

[Alan Bain]

It's a common story. Nice HP instrument with analogue meters, stored somewhere damp and the top layer of the dial starts to flake off and the flakes look horrible and get in the way of the pointer. I've tried various repairs with e.g. superglue, PVA which work if the damage isn't too severe, but this 3406A was in a bad way, and after my first glue fix another flake restricted the pointer's movement around the cal mark.

I've found a few articles around on these scales - as each meter was individually calibrated using a generic replacement scale is out.
https://www.hpl.hp.com/hpjournal/pdf...Fs/1961-03.pdf
None of the suggestions I saw handled the mirror portion well e.g.
https://www.eevblog.com/forum/testge...r-restoration/

Hence I tried a variant approach. I removed the meter scale, scanned it (greyscale) imported to photoshop converted to 1 bit colour (aka black and white) and cleaned up the lines (and also removed the scan of the mirror). I printed a copy onto clear acetate and made sure that I could print this to exactly the size of the original scale (see photo). I also marked the centre of the lower arc on the acetate (found by bisecting a chord).

I then printed this onto some QP self-adhesive photo paper found on ebay (see photo of pack). There are some other recommended products but they all appear to be US only e.g. Papillo photo film.

Now I used a "clock dial cutter" (from Cousins) basically a trammel with a bit of scalpel blade on one head to cut out the mirror portion. I located the centre using the pre-prepared acetate. First try was a fail. The blade was too blunt and the point insufficiently pointy. So put point in lathe and used a tungsten carbide graver to make it actually sharp, and sharpened the blade on an oil stone. This time it worked well. Getting the lower radius right is critical since it is just above a printed scale arc and if you get it wrong the difference in distance between middle and edges is very obvious. The second cut is just at a slightly larger radius. I couldn't figure out a reliable way to replicate the curved ends so I cut straight with a scalpel. Finally I cut round the outside of the scale accurately (the black from the scanner made this easy).

To remove the old scale I simply put it on a hotplate (aka the domestic cooker) and when warm the old scale peeled off. Clean up with IPA (more effective than acetone) and attach the dial (I didn't make a jig for this, I found I could do it accurately aligning the edges).

I thought the results were pretty good. Then was able to try a calibration (not having a thermal converter for DC transfer calibration I had to improvise with a power meter). It's often said that HP432s can do DC transfer calibration - but all the heads I have have a DC block capacitor (probably just as well or they would be very easily damaged!).

I thought this was a success -- although of course to actually do readings the best approach is to connect a digital true RMS meter to the BNC socket on the back!

[Radio Wrangler]

Originally Posted by Alan Bain

It's often said that HP432s can do DC transfer calibration - but all the heads I have have a DC block capacitor (probably just as well or they would be very easily damaged!).

This is often said, but it's not quite right.

The HP432 IS a dc transfer standard in and of itself.

The heas for it, like the 478A etc are bolometers. they contain two wheatstone bridge circuits. In each bridge one limb is a pair of thermistors. The imbalance voltage of the outputs of the bridges go to precision DC amplifiers and thence control the power supply energising that bridge. Two bridges, two amplifiers two power supplies. One of these systems is just used as a reference for comparison with the one that does the work.

Tht thermistors are nominally 100 Ohms at working temperature. The amplifier makes the bridge self-balancing and thus the series pair of thermistors is servo-controlled to match a precision 200 Ohm resistor.

The RF input to the head drives the midpoint of ther thermistors through the blocking capacitor you've mentioned. One thermistor goes to ground, the other one to the rest of the bridge, but with a decoupling capacitor to ground. So the RF input sees two 100 Ohm resistances in parallel to ground IE the head looks like a good 50 Ohm termination.

If I put some RF power in, the RF creates extra heat in the thermistors. They get hotter and this unbalances the bridge. The amplifier responds by turning down the DC power it feeds to the bridge. The clever bit is that the reduction in DC power getting to the thermistors is quite accurately equal to the RF power they are receiving.

This is where the second bridge comes in. It is sitting there as an example of what the DC voltage to the working bridge would have been without the RF being applied. It matches the working bridge in terms of effects due to ambient temperature etc.

So, you put RF into the 432A's head and on the back panel of the 432A are two BNCs giving the DC voltages of the two bridges. It transfers the true RMS voltage of your RF signal into the difference between these two DC voltages. It does it from first principles, thermally. It's not absolutely definitive, but it's very close. Only the 'Type IV' power meter designed by NIST in Boulder, Colorado is better. Put a pair of 6-Digit DVMs on those DC outputs and you have a standards lab grade RF power calibrator.

Those DVMs need calibrating They are your true measuring instruments. The 432 is just transferring an RF RMS voltage to DC for you. You can check that it's balanced properly ands not introducind anoffset but you don't need absolute standards to do this.

So you really cannot check a 432A by applying a calibrstion DC voltage to it. You can check it by applying a known RF power and check the DC outputs with some high-end DVMs. Play it the other way, have calibrated DVMs and check your 432A is properly balanced and you can calibrate your RF source!

One thing to watch out for is to use a network analyser to check the head is presenting a good 50 Ohm impedance. The transfer business will still be OK for RMS voltage, but your circuit loading could be off.

Typing fingers sore!

David

[Alan Bain]

Thanks. You've pointed out exactly what I had missed!

In fact I've recalibrated one of these HP432 meters and as you say all it needs is two voltage measurements - one differential and the other straightforward and with a decent 6.5 digit DVM like the 34401 no special magic is involved (other than a knowledge of the error in the DVM and I happen to know thanks to a helpful person that mine is within 70uV on the 1V DC range - or at least was on the day it was checked!).

Of course this recalibration is just about making the analogue meter show the right level, if you directly read the V_RF and V_COMP values you can compute the power accurately. There was a digital HP 432B (which I've never seen), must have a hunt for a service manual to see how it worked (i.e. does it do the computation digitally, or is it a digital voltmeter on the back of the analogue out of the 432!)

I tried a 478A on the VNA (taking care not to exceed the power rating) and at low frequencies S11 is a bit wobbly.

I'm sure my 432A isn't alone in getting most of its use calibrating the 0dBm reference on the more modern (open-loop) meters. I don't actually know why HP moved to the thermocouple design away from the closed loop thermistors, the only reason I use them is they have GPIB interfaces and digital readouts!

[factory]

That's an excellent job on replacing the dial scale, I have scanned in one or two scales in case they deteriorate further & become unrepairable.

There is a 432B & 432C that both use a digital panel meter in place of the analog meter, there is a 432B in the spare room here, can't remember if I mentioned it when you visited a while back, mine uses a bought in DPM, but later ones might use HP's own LED DPM.



David