Purchasing HP 432A (was: eBay HP 478A thermistors. Any good?)

[Tony_G]

I did catch some 432 models have a sticker warning they are prepared for different, longer cables, and cable long must match what the sticker says for it to work OK. Mentioning it here, just in case.

There is a modification you can make to the meter to support a longer cable than normal. It's covered in the manual and can be reversed

TonyG

[tatel]

...the probability of some guy being here, having both 431 and 432, is probably next to zero.

I've also got a 431C power meter and two 432A power meters and two 478A sensors.  One of the 432A power meters is as new and unused and has the original HP seals and stickers on it. The 478A sensor is still factory (vacuum) sealed in its packaging and so is the sensor cable and the manuals. It's never been unpacked and used. It must have been stored from new and forgotten about.

You just achieved with me what Pavlov achieved with those dogs...

[G0HZU]

...the probability of some guy being here, having both 431 and 432, is probably next to zero.

I've also got a 431C power meter and two 432A power meters and two 478A sensors.  One of the 432A power meters is as new and unused and has the original HP seals and stickers on it. The 478A sensor is still factory (vacuum) sealed in its packaging and so is the sensor cable and the manuals. It's never been unpacked and used. It must have been stored from new and forgotten about.

You just achieved with me what Pavlov achieved with those dogs...

The unused 432A really does look like it's been in a time capsule as it is totally unmarked and as new. The gold contacts on the connectors look like brand new jewelry. The other 432A is fairly immaculate, but the unused one is really special, like it just left the factory yesterday.

Every year or so I use the other 432A + 478A to check the 50MHz 1mW source from my power meter using DC substitution. I've done this for quite a few years now. I can show you how I do it when you get your 432A up and running. I follow the factory instructions and created a spreadsheet to do the calculations.

[tatel]

Every year or so I use the other 432A + 478A to check the 50MHz 1mW source from my power meter using DC substitution. I've done this for quite a few years now. I can show you how I do it when you get your 432A up and running. I follow the factory instructions and created a spreadsheet to do the calculations.

I gladly take your offer.

[tatel]

I got the new cable and am able to coarse/fine zero, so I guess that's good.

Now I need to go to the other home where my gear is.

Thank you all

[Tony_G]

Nice to hear - Good luck with your use.

TonyG

[tverbeure]

This is for sale locally on Craigs List.  Though I have two 438a, I think that is the number, dual channel power meters, plus a module in my MMS, I've not run into the thermistor mount as he calls it, on this model.

This is the second 432A that this seller put on craigslist. I bought the first one.

If you're thinking about buying that one: go for it. He's a retired HP engineer who's downsizing his extensive home lab. I've bought a lot of gear from him. He always thoroughly tests things before he sells them.

https://sfbay.craigslist.org/sby/ele/d/los-altos-hp-432a-microwave-power-meter/7718776347.html

[TimYTTT]

G0HZU, could you please provide more info on how you use the 432A + 478A to check the 50MHz 1mW source using DC substitution. I do have similar needs and can probably benefit from your experiences. 

[tatel]

From https://xdevs.com/doc/HP_Agilent_Keysight/HP%20478A-H75%20Operating%20&%20Service.pdf I get there's a way to check if 478A is really in good order.

If I'm getting it right, one should make a circuit so:

a) pins 2, 4 and 6 of thermistor mount get connected to ground

b) a power supply giving 29 DC volts is put in series with a 2200R resistor

c) after the 2200 ohm resistor, a battery is placed against the voltage coming from the power supply, I guess it's to lower the 29V so the suggested DVM (HP3440) is able to get in a range where it can resolve 1 mW 1 mV

d) ground of that DMM is connected to power supply ground, like pins 2,4 and 6 of thermistor mount are, completing the circuit.

e) probing for milivolts against pins 1 and 3 of thermistor mount in some kind of voltage divider, connected between  resistor and bucking battery, should then show less than 30 mV difference. Otherwise thermistor mount isn't good. If that difference gets up to 150 mV, thermistor mount can't even be repaired.

Now my questions:

1) Is it worth the hassle if thermistor mount is able to be coarse/fine zeroed? I guess probably not.

2) I guess that 2200 resistor is there to limit current/voltage. So having a good modern power supply with CV/CC, one wouldn't need that resistor, provided the power supply is set to the correct V and I output --- or is it there to make sure there aren't any peaks able to damage thermistors?

3) Having not a 3440, but a 3456 6-1/2 digit DVM, surely that bucking battery wouldn't be really needed anymore?

[Tony_G]

Up front I have not tried what you are showing so my "guesses" may be way off - Before you go slamming V/I through the 478A I'd wait to get a consensus or commentary from someone who has done this.

c) after the 2200 ohm resistor, a battery is placed against the voltage coming from the power supply, I guess it's to lower the 29V so the suggested DVM (HP3440) is able to get in a range where it can resolve 1 mW

The battery was chosen to achieve a 1mV meter resolution - From the manual:

The value of this bucking voltage should be such that voltmeter resolution down to 0.001 volt is obtained.

Now onto your questions...

1) Is it worth the hassle if thermistor mount is able to be coarse/fine zeroed? I guess probably not.

2) I guess that 2200 resistor is there to limit current/voltage. So having a good modern power supply with CV/CC, one wouldn't need that resistor, provided the power supply is set to the correct V and I output --- or is it there to make sure there aren't any peaks able to damage thermistors?

3) Having not a 3040, but a 3456 6-1/2 digit DVM, surely that bucking battery wouldn't be really needed anymore?

#1 - No, if the mount zeros then it is pretty much working - TBH I've never seen one of these where it is working but 'almost failing', it seems that this is possible and may have been more common once given that the manual provides an explicit procedure to take to address but I think it is a pretty low chance case - YMMV as will others

#2 - +29V seems to be one of those numbers that are oddly specific - I would not be surprised to find out that the thermistors need to be seeing a certain amount of voltage/current so that the compensating thermistors react correctly to what would be expected from the thermistors exposed to the RF field. I would love it if someone had the real explanation for this and could share with us.

#3 - Most likely not given what the manual was saying earlier but you will still need to ensure that the V/I through the 478A is within spec - You could certainly math that from the manual but if the 478A was working then I lean towards this being a bit dodgy.

Sorry I couldn't provide more clarity - Interested in seeing what more experienced members say.

TonyG

[tatel]

Up front I have not tried what you are showing so my "guesses" may be way off - Before you go slamming V/I through the 478A I'd wait to get a consensus or commentary from someone who has done this.

Yep, I 100% agree with that

The battery was chosen to achieve a 1mV meter resolution - From the manual:

The value of this bucking voltage should be such that voltmeter resolution down to 0.001 volt is obtained.

Oops... typo. Yeah, 1 mV. Will edit my previous post

#1 - No, if the mount zeros then it is pretty much working - TBH I've never seen one of these where it is working but 'almost failing', it seems that this is possible and may have been more common once given that the manual provides an explicit procedure to take to address but I think it is a pretty low chance case - YMMV as will others

#2 - +29V seems to be one of those numbers that are oddly specific - I would not be surprised to find out that the thermistors need to be seeing a certain amount of voltage/current so that the compensating thermistors react correctly to what would be expected from the thermistors exposed to the RF field. I would love it if someone had the real explanation for this and could share with us.

#3 - Most likely not given what the manual was saying earlier but you will still need to ensure that the V/I through the 478A is within spec - You could certainly math that from the manual but if the 478A was working then I lean towards this being a bit dodgy.

Sorry I couldn't provide more clarity - Interested in seeing what more experienced members say.

Yep, again I agree... If it doesn't look broken don't touch it. Still interested in knowing more about that, this knowledge could be useful some day to any people looking at 478A.

[G0HZU]

G0HZU, could you please provide more info on how you use the 432A + 478A to check the 50MHz 1mW source using DC substitution. I do have similar needs and can probably benefit from your experiences. 

Sorry for any delay, sadly, I've been unwell recently but I'm a bit better now. I've had a go at putting the procedure into a word doc and it's probably best posted in the Metrology section. However, I've attached the first draft below.

It's been written in a hurry so it might not read that well....

[factory]

Yeah, it seems to be for 431.

Well, it seems I found the limits of the forum. Of course, the probability of some guy being here, having both 431 and 432, is probably next to zero. I have never seen the slightest mention of cables for 431 and 432 being different. So, here it remains this piece of information. Hope it will useful for somebody in the future.

I already ordered another cable, this one no only has the notch, but also has golden pins

Best wishes

The difference in the cables might have something to do with the later 8478B thermistor mount (first seen in 1968 catalog), it requires an adaptor to use with the older 431A/B (thermistor is unbalanced), but can be used directly with the 431C & 432A.

There is even an adaptor for the 430C to allow use of the higher frequency waveguide mounts, thermistor is unbalanced and there is no temperature compensation.

Apparently I have a 431C in storage, I also have several 430C, several 432A and one 432B & 432C, these last two both use a digital panel meter, in place on the analog panel meter. I might have a TEA problem. 

The 432 series may contain two Sprague 109D capacitors (or equivalent) one at 33uF 10V/15V* & one at 220uF 8V, these are wet tantalum and can leak corrosive sulphuric acid electrolyte, picture below is from my 1969 432A.
*voltage seems to vary depending on the age of the meter.



David

[TimYTTT]

I purchased recently one HP478A thermistor sensor without the calibration chart. The sensor is working properly and when I compare the results at 50 MHz with my Rigol DSA815 spectrum analyser, the measurements are close together. However, the question is if the results will be correct on the higher frequencies. Can someone kindly help me with suggestions/ideas on how to resolve this?

[Tony_G]

Not a direct answer and thus this is just a "for what its worth" post but here is the calibration report for one of my 478A's so that you can see how an in-specification one reacts over frequency.

Hope it helps in some small way.

TonyG

[TimYTTT]

Tony, thanks! Enclosed calibration report helps me to understand how the CF-table for correct working thermistor looks like. It tells me also that I can send my 478A for calibration to HP and get the missing CF data. What is unknown to me: What are the costs for such service? Could you please provide me some info?

[G0HZU]

I purchased recently one HP478A thermistor sensor without the calibration chart. The sensor is working properly and when I compare the results at 50 MHz with my Rigol DSA815 spectrum analyser, the measurements are close together. However, the question is if the results will be correct on the higher frequencies. Can someone kindly help me with suggestions/ideas on how to resolve this?

Probably the first thing to do would be to measure the return loss of the sensor across the frequency range you intend to use it over. This should provide a fairly good health check of the sensor.
At 50MHz it should typically be better than 25dB, but by 100MHz it should typically be better than 30dB and I'd expect it to maintain at least 30dB return loss to at least 1GHz.

However, it will degrade above about 1.5GHz. I don't recommend using the 478A above about 2GHz because the return loss isn't very good here. It is possible to fit an attenuator to the sensor to improve the return loss above 1500MHz, but you then need to know how accurate the attenuator is.

Obviously, it is also risky to use this power meter below about 30MHz because the return loss degrades fairly rapidly below about 50MHz. You can improve this with a (say) 10dB attenuator fitted to the sensor, but you need to know how accurate the attenuator is.

If the return loss is as good as I've listed above, then the sensor is almost certainly in good condition and the cal factor will typically be 99% +/- 1% across 50MHz to about 1500MHz.

For most home users, I'd recommend doing the return loss check on a decent VNA and then simply leave the cal factor set at 98% or 99%.

This ought to be good enough for all but the most critical applications. If the critical application is above 1500MHz, then I'd generally recommend using a modern thermocouple based power meter, unless the source impedance of the DUT has a very low VSWR.

I'd expect the cost of calibration to exceed the cost of buying another used and tested sensor, although the prices of these sensors seems to be climbing every year. Prices also seem to vary from £150 to over £1000 for a tested sensor.

[TimYTTT]

I purchased recently one HP478A thermistor sensor without the calibration chart. The sensor is working properly and when I compare the results at 50 MHz with my Rigol DSA815 spectrum analyser, the measurements are close together. However, the question is if the results will be correct on the higher frequencies. Can someone kindly help me with suggestions/ideas on how to resolve this?

Probably the first thing to do would be to measure the return loss of the sensor across the frequency range you intend to use it over. This should provide a fairly good health check of the sensor.
At 50MHz it should typically be better than 25dB, but by 100MHz it should typically be better than 30dB and I'd expect it to maintain at least 30dB return loss to at least 1GHz........

Thank you a lot for your valuable comments. I am going to check return loss of the sensor as you suggested.

[Tony_G]

Tony, thanks! Enclosed calibration report helps me to understand how the CF-table for correct working thermistor looks like. It tells me also that I can send my 478A for calibration to HP and get the missing CF data. What is unknown to me: What are the costs for such service? Could you please provide me some info?

These costs seem to be very country specific - If you go to the Keysight website and then into the support section, there are links to request calibration services - That is what I did here in the US - They are currently charging US$421 for a basic calibration.

TonyG