DIY thermal RF power sensor?

[rf-fil]

I'm attempting to source / build my own RF power meter based on heating of an RF load. I want this as a sort of a transfer standard so I can use it to calibrate other gear like log detectors, spectrum analysers, etc. The idea is that, if done properly, this type of sensor can be calibrated with DC voltage, and will be very broadband, well into GHz range. This type of gear from Keysight, R&S, etc is orders of magnitude above my budget, LOL.

I wonder if anyone has built something like that, so that I don't end up reinventing the wheel? I don't need a huge dynamic range or a short time constant. I just need RF CW average power measurement that is traceable to DC voltage. If there's not much out there, I might try to build something and document it here.

[dietert1]

Many years ago i made that device using a small NTC glued to a smd 50R resistor as transmission line terminator, mounted inside a small precision oven (+/- 1 mK or so). One needs a circuit to run the NTC at constant temperature. Then you measure it's self heating and it will mirror the RF power. The power range was like 1 mW. Higher power can be measured using an attenuator.
I even posted an image before in this forum, but it just shows the oven.

Regards, Dieter

[Gyro]

A couple of relevant threads...

https://www.eevblog.com/forum/metrology/rf-small-power-thermal-sensor/

https://www.eevblog.com/forum/metrology/diy-precision-ac-rms-to-dc-transfer-standard/

[Roehrenonkel]

Hi rf-fil,
 
in my stock i have some thermistors that might be just the right thing.
Thermal cross is probably the wrong translation for the german "Thermokreuz"
so i stick with the labeling on the box: Thermistor BS1380/90 by Stantel.

Never found any data/documentation about them. Can anybody help me out?

Although a Thermokreuz should generate a dc-voltage depending on the heat of the sensor
somebody before me measured them for resistance (+6dB 37Ohm).

Ciao4now

[dobsonr741]

Why thermal? Analog Devices has a huge portfolio of power measurement from DC to 70GHz. https://www.analog.com/en/product-category/rf-power-detectors.html

My fav is ADL5511: https://www.analog.com/media/en/technical-documentation/data-sheets/adl5511.pdf

[rf-fil]

Thanks, good info from everyone.

dobsonr741, these are great and I use them. The challenge is that they all need calibration with a signal generator. which itself needs calibration. For my use case, I need sub-1dB accurate power measurement with reasonable traceability (long story). With a thermal based method, the design only needs to be checked for VSWR flatness at its input. After that, calibration can be always be done with DC voltage. Basically, I am after flat response from DC to 3GHz.

[Odysseus]

I attempted something very similar, with the same goal of simplifying calibration requirements. I avoided an oven by using a separate RF and DC 50ohm terminations, and then balancing two thermistors in a bridge with a PC sound card control loop. I used some surface mount platinum RTDs for temperature measurement. I seem to recall the measurement sensitivity was on the micro-Kelvin scale. Despite that, I was never able to get the error drift below 100uW or so. As a measurement method, I think it was sound, but it was plagued by "common mode" temperature sensitivity and long lasting thermal drifts.

My suggestion is to use a sensor that inherently detects temperature difference, i.e. a thermocouple, thermopile, or miniature thermoelectric cooler.

[dobsonr741]

Got it. Thought about renting or finding someone who has access to one in a pro lab?

Or the other way around, to build a stable power reference oscillator? And assume to get the flatness of the silicon power detectors after calibrating on one frequency? Like http://www.w1ghz.org/small_proj/RF_power_ref.zip or closer to home: https://www.eevblog.com/forum/rf-microwave/10dbm-rf-power-reference-source-output-filter-needs-improving/

[Gyro]

... I am after flat response from DC to 3GHz.

I think you're going to have to modify your expectations there. The chances of maintaining a clean impedance match from connector, through the thin wires (needed for reasonable thermal isolation) and the heater resistor itself are minimal to non existent. 3GHz is definitely in stripline territory.

[dietert1]

That's why at the time i made an oven for it. The terminator/sensor sits at the end of the coax cable visible in the image. Nowadays we have thermal imaging cameras and it should be pretty easy to arrive at a working setup.

https://www.eevblog.com/forum/metrology/diy-precision-ac-rms-to-dc-transfer-standard/msg2622633/#msg2622633

Regards, Dieter

[G0HZU]

For many years I've been using an old HP432A power meter and HP478A thermistor sensor for stuff like this.
The HP432A has a DC substitution mode where the accuracy of the measurement transfers across to the accuracy of a 5 or 6 digit bench DMM. The DMM measures the equivalent DC required to balance the bridge system.

I use it to measure (and adjust?) the 1mW 50MHz reference in another power meter. Every year for many years the 1mW reference is within 0.02dB of 0dBm using this method. Calibration houses use a similar method. The only unknown is the efficiency of the 478A sensor but this should be stable for decades once known. It's usually marked on the sensor with a graph of efficiency vs frequency.

The other way I cross check this is with a home made low barrier diode detector. This is designed to work down to about 100kHz so I can calibrate the diode sensor at 250kHz using a Keithley 2015 or 2000 bench multimeter. These meters are very accurate for AC at 250kHz. This method isn't as good as the DC substitution but the diode sensor has a very flat response up to 500MHz and it still quite good to 1GHz. The input return loss is better than 30dB at 1GHz.

I also have a HP 8473C LB diode sensor that works to 26GHz as another cross check but it doesn't work down to 250kHz.

The diode sensors are temperature sensitive so it's best to calibrate the homebrew one at LF using the Keithley meter. The RF test source should also have low harmonic content. Ideally better than -40dBc.

Across all these methods I can gain good confidence in my test gear up to many GHz. It won't be adequate for some people, but in terms of accuracy, I would expect it to match or beat test gear that is simply sent for an annual cal check (with no cal chart or adjustment as part of the annual fee)

[rhb]

I urge you to join the amaterradiobuilders@groups.io mailing list.

The list owner has reported as much as 20% discrepancies between instruments that were sent out for a cal before use.  He has raised the topic of a group effort to design a reliably accurate reference.  Bill Schmidt has PhDs in EE and chemical engineering and owns an electronics company.

He insists that the only accurate measurements of RF power he has made were done with a load, styrofoam cup of water and platinum RTD in grad school.

This arose in part from my quest to calibrate my attenuators and discussion of a 20 A HF PA Bill designed that a list member recently completed.

I'm kicking around a non-inductive load in an aluminum cylinder with an RTD sensor wrapped around it and then placed in a well insulated enclosure.  By recording both the increase and decrease in temperature vs time the heat flow through the insulation can be precisely determined.  The cylindrical shape makes the equations reasonably tractable.

I want to calibrate a Weinschel 0.1 dB step attenuator so I need *at least* 0.01 dB accuracy.  And 0.001 dB to cal my HP 438A, 8481D & 8482A power meter and sensors.

Have Fun!
Reg

[LaserSteve]

I find it difficult to believe that there is 20% error between cal labs if the detector used is in a commercial instrument based on dual bridge,  closed loop, DC or AC   substitution made by a reputable vendor. (HP)

OK, find  NIST Technical Note 1374,  NIST Technical Note 1357,  NIST Technical Note 1511  See  Things I have attached...  Read the  journal references in the tech notes and app notes

see www.tegam.com  Makes calibrators / has interesting app notes.

HP application notes  64-1 and 64-1A, Multiple editions of the HP journal on diode and bolometer power meters and readouts... Start with HP journal vol 1974-09.pdf

Plenty of docs out there on HP 478a, a few episodes of Signal Path...

Steve

[G0HZU]

Yes, the DC substitution method should be very accurate and repeatable.
I've kept records for about 7 years using this method (with an old HP 432A power meter) and the biggest change I've seen is from -0.005dBm to +0.011dBm when I measure the 0dBm (50MHz) reference from an old Anritsu power meter. This reference seems to have aged to be very stable every year. It really is quite a boring task to measure it every year or so.

Before this, I used the same 478A sensor with an HP 431C power meter and used a DVM on the recorder output.

I also measure the 'recorder' output from the HP 432A meter every time I check the Anritsu sensor and it seems to be almost as stable as doing it via DC substitution. The biggest error I saw (as in disagreement with DC substitution) from the recorder output was +0.016dB one year.

[G0HZU]

Over the weekend I'll try and demonstrate the diode detector method. I've used this up to many GHz with very good results. It can't compete with DC substitution but it still gives results good enough for most applications. It's a cheap and effective way to gain confidence in the frequency response of test gear across a huge frequency range. It does require a signal generator that has low harmonic content.

It also requires two diode detectors. One is optimised for used from a few kHz through to about 150MHz. The other works to UHF (or even higher if I use the HP 8473C detector).

I haven't done this for a while but I use a classic levelling system to compare the diode detector against a thermocouple power meter (or the 478A thermistor sensor).

 I could try and design a diode detector that works from (say) 300MHz to 3GHz but this won't be easy. Getting a low input VSWR and a flat frequency response will be a challenge I think. I've made a few detectors that are good to 1GHz but the performance declines fairly rapidly above about 1.5GHz. I'm lucky in that I have an HP 8473C detector here for stuff like this.

[rhb]

I find it difficult to believe that there is 20% error between cal labs if the detector used is in a commercial instrument based on dual bridge,  closed loop, DC or AC   substitution made by a reputable vendor. (HP)

OK, find  NIST Technical Note 1374,  NIST Technical Note 1357,  NIST Technical Note 1511  See  Things I have attached...  Read the  journal references in the tech notes and app notes

see www.tegam.com  Makes calibrators / has interesting app notes.

HP application notes  64-1 and 64-1A, Multiple editions of the HP journal on diode and bolometer power meters and readouts... Start with HP journal vol 1974-09.pdf

Plenty of docs out there on HP 478a, a few episodes of Signal Path...

Steve

c.f.  https://groups.io/g/AmateurRadioBuilders/message/2252

I think Bill Schmidt just might know what he's talking about.

Reg

[szoftveres]

If I would DIY a thermal power sensor, I'd probably follow a similar philosophy to the HP432A and other HP true RMS meters. In the below circuit, I'd glue the 50 ohm smd RF load resistor (or two 100 ohm resistors) to one of the 2n3904 transistors and wrap them (together with the lead-in coax cable) in cotton wool and heat shrink tubing, and leave the other 2n3904 at a distance at ambient temperature. The circuit measures the voltage difference between the two bridges - one bridge is at ambient temperature, while the other one also gets the heat from the RF load resistor. The voltage on each bridge amplifier output is proportional to the Vbe of the corresponding 2N3904 transistor, which is largely temperature dependent (assuming currents are equal on both transistors; note that R5 and R2 are supposed to be of equal value, I just made one much larger in order to simulate the different Vbe of the two bridges). U3 is configured to provide 10x the difference voltage at its output.

[switchabl]

The list owner has reported as much as 20% discrepancies between instruments that were sent out for a cal before use.  He has raised the topic of a group effort to design a reliably accurate reference.  Bill Schmidt has PhDs in EE and chemical engineering and owns an electronics company.

He insists that the only accurate measurements of RF power he has made were done with a load, styrofoam cup of water and platinum RTD in grad school.

I have only ever seen errors like this with a broken power sensor that has been overloaded (assuming we are still talking about CW, modulated/pulsed power measurement can be more tricky).  I think you'd have to completely mess up the calibration otherwise, like an error in the algorithm, a wrong connection.

That being said, calorimeters are still the gold standard and the only proper way to get a traceable calibration from DC voltage. There are good reasons that people generally don't use them for anything else. First, they are not easy to get right, in particular you need a feedline that has both low thermal conductivity and good RF properties. But most importantly, they are very slow. If it takes 15 minutes to reach equilibrium that isn't just cumbersome, it also adds uncertainty through drift elsewhere in the system.

Of course, I definitely don't want to put you off trying to build one. I'm sure it would be a very interesting project. If you want a good starting point, you can try to find a copy of "Radio frequency & microwave power measurement" by Alan Fantom. It isn't exactly up-to-date (late 80s?) but the basics are all still valid and it has a lot of references to classic papers which may be more useful for DIY than the latest advances in mm-wave metrology.

I want to calibrate a Weinschel 0.1 dB step attenuator so I need *at least* 0.01 dB accuracy.  And 0.001 dB to cal my HP 438A, 8481D & 8482A power meter and sensors.

You may want to do some preliminary uncertainty budgets before you jump in.

The attenuator I guess could be doable (or close enough, maybe drop the "at least") since it is a relative measurement. I'd probably start by looking at (temperature) stability and repeatability first to make sure it is even worth the effort. Anyway, it doesn't require accurate power measurement at all, just good linearity. Thermistor sensors with dc substitution (like old 432A + 478A + 6.5 digit multimeter as suggested above) are also very linear and you don't even need the correction factors in this case, so no need for tracable calibration or a calorimeter. It still needs a lot of care. If you don't have a very stable source, you probably need a second reference power meter to track the incident power. At <0.1dB accuracy, mismatch uncertainty is likely a major limiting factor as well. You might have to characterise the S-parameters of your DUT and all components of your measurement system with a VNA first.

The second one... I am pretty sure it is way beyond the capabilities of even major metrology institutes. 0.001 dB works out to about 0.02%. The 438A manual calls for "only" 0.5%. I think NIST or PTB will do thermistor calibration against their microcalorimeters to 0.3% (maybe 0.2%).

[G0HZU]

For home use, it should be OK to simply measure the 0.1dB attenuator steps using a couple of modern (level stable) signal generators and a diode ring mixer (to make a simple downconverter) and then use a PC soundcard as the log detector at maybe 15kHz.

This allows the 0.1dB steps to be measured at various RF frequencies and there should be good log accuracy in the soundcard ADC. I'd recommend using some decent attenuators either side of the step attenuator to define a good source and load match for the step attenuator. I've used a PC soundcard to do this many times. It's important to keep well within the linear range of the diode mixer but this should be easy to achieve.

Modern spectrum analysers have a decent digital IF to replace the old school log amps and this should also give excellent linearity. I've used this method a few times too. For home use, there's probably not much point trying for an evaluation system that is better than this unless you are doing this as part of some sort of nerdy exercise in metrology.

I've got various decent RF step attenuators here (only one has 0.1dB steps) and I check them like this every so often. However, I only do this when there is a need to prove they still work as expected and this doesn't happen very often.

[David Hess]

I think I remember an alternative way where a single thermister is used as the 50 ohm load, so the RF is directly coupled into the thermister through a bias-T, and DC is applied to maintain the thermister at a constant temperature and 50 ohms.  As the RF signal is lowered, more DC has to be added to maintain the same temperature and 50 ohm load, so the RF power is a constant minus the DC power.  At higher frequencies the thermister is mounted in a transmission line environment.

[G0HZU]

I dug out my old homebrew 1GHz diode detector and tested it this evening. To minimise uncertainty, it's best to use it in a levelling system as in the diagram below. This system uses ALC (feedback) to level the signal generator to the accuracy of the power meter in the ALC feedback path. The source VSWR is defined by the quality of the 50R resistors in the splitter.

This is the classic method used by calibration houses to test the frequency response of various devices as it minimises uncertainty.

In the test I've just done. I put the diode detector in the ALC path (in place of the power meter) so the system now levels the sig gen to the frequency response of the diode detector. A sig gen with low harmonic content is needed here or uncertainty can become significant. However, the OP only needs to prove +/- 1dB accuracy.

I've used a thermocouple power meter as the DUT in the setup below so the frequency response plot below is my Anristsu thermocouple power meter measuring the levelled output. It therefore measures the frequency response of the diode detector.

I know that this detector is only good to about 1GHz. Above this frequency, the internal parasitic package inductance and capacitance cause the diode detector to over read the power level. The input return loss of the diode sensor is better than 30dB below 1GHz. Below 500MHz it's much better than 30dB return loss. You can see that the levelling performance degrades markedly above 1GHz. By 1.5GHz there is a 0.5dB error for example.

Whatever ripple is present on the plot below is partly due to the frequency response of the diode detector and from the influence of harmonics from the test generator (on the diode detector) and from the fact that the thermocouple sensor efficiency isn't perfectly flat vs frequency. The software knows the cal factor vs frequency for the sensor (as stamped in a table on the thermocouple sensor body) but in my experience, a typical thermocouple sensor will still have subtle efficiency variations between the cal points stamped on the sensor. So there will be various things contributing to the fact that the plot below isn't perfectly flat.

The performance is good when you consider the diode in the detector costs less than about £0.50.

Hardly anyone needs performance better than this. In a real system, the wiggles in my plot will be lost in the inevitable mismatch uncertainty between the stages of real world circuit designs. So there's not much point trying for better performance than this unless you are in the business of designing very exotic and expensive RF power detectors.

I'll have a look at extending the frequency range of the diode detector but I doubt I'll be able to get it to work to 3GHz. My recommendation would be to buy an old 26GHz HP 8473C diode detector for higher frequencies like this. The diode detector is good for measuring flatness but it will need to be referenced in some way if you want to make absolute power measurements.

I usually do this at a low frequency and then rely on the frequency response of the diode detector to verify the power at other frequencies. This diode detector method would be a good way to cross check the performance of any DIY thermal power sensor.

[rhb]

@switchabl

I spent an hour on a reply last night and instead of posting it, my iPad ate it :-(

I'm not going to rewrite that so I'll just summarize:

p 1-7 of the 438A manual states:

0.001 dB resolution. +/- 0.02 dB accuracy on the cal range and +/- .04% on other ranges.

The 0.5% is the zero settability which is irrelevant for relative measurements.

My lab doesn't currently have adequate temperature control for such work, so all I'm doing so far is error budget and similar preliminary work while I build a new lab.

I'd be very interested in the details of your diode detector as I have a Millivac with a blown sensor.  It only goes to ~1.2 GHz, so your design would be a huge improvement over not working or spending ~$400 for a repair by Millivac.

I have an 8481D 10 MHz to 18 GHz sensor for my 438A which is close enough to my general limit of 20 GHz to suit me.

Over the past 5 years or so I have accumulated a suite of lab gear with a 1990s list price of in excess of $600k for 1-2 cents on the dollar.  This poses a significant cal issue as I cannot afford to send that stuff out for cal. This includes things like L&N resistance standards, a General Radio precision pot.  No inductance or capacitance standards yet, but in time.  A long term project is to set up an automatic cal system that checks all the instruments againt the other instruments using a relay panel and GPIB control.  As long as everything is accurate relative to other instruments I really have no reason to care about absolute accuracy as much as relative accuracy between instruments.

However, I do want to have some reasonably accurate references to keep the absolute errors under control.

I'm also a general metrology nerd with gauge blocks, LVDTs accurate to millionths of an inch and am trying to acquire a Kern DKM1 autocollimating 5 second theodolite.  So I am clearly beyond hope of redemption.

Reg

[G0HZU]

Generally speaking, (across all makes/models) a healthy power meter with a thermocouple sensor can typically deliver 2-3% accuracy if you test the meter plus the sensor across its power levels/ranges and across frequency. It's generally best to avoid using it below -20dBm and above about +15dBm if accuracy is important.

It's unlikely the overall uncertainty would exceed +/-0.25dB and usually it will be somewhere in the window of +/- 0.1dB as long as you avoid the extremes of its frequency and level ranges. It's best to not expect much more than this unless you send the meter + sensor out together for a custom calibration for a specific frequency and power level.

That's my experience based on a decades long career working in RF labs with access to lots of power meters like the 438A, E441x and various modern USB alternatives from R&S.

[G0HZU]

If you are interested in metrology, then the thermocouple sensor based meters (like the HP 438A) are a non-starter as they operate in open loop. They absolutely 'need' the onboard 50MHz 0dBm source to provide the offset to the slope of the thermocouple sensor.

The thermocouple sensor usually has a very linear slope but it can't provide the 'c' in the classic y = mx +c equation for a straight line. So it is classed as an open loop meter.

If you want a power meter for metrology, then the classic HP432A and HP478A thermistor sensor is a much better choice as it operates in closed loop. That's how the 478A thermistor sensor can be used as a transfer standard. I don't think you can do this with the 438A meter + sensor.

The 432A + 478A sensor doesn't need the 50MHz 0dBm reference to 'calibrate' itself. Even though the 478A sensor dates back to the early 1960s it is still the favoured choice as a transfer standard for many calibration applications. As long as the efficiency of the thermistor is known, then the DC substitution method can be used to get accuracy in the ballpark of +/- 0.5% year after year. This is equivalent to about +/- 0.02dB.

With a 438A meter you won't know how stable the sensor is against temperature and time and only the accuracy of the 50MHz oscillator can correct for this. So the meter always relies on the accuracy and stability of the onboard 50MHz reference.

[switchabl]

p 1-7 of the 438A manual states:

0.001 dB resolution. +/- 0.02 dB accuracy on the cal range and +/- .04% on other ranges.

The 0.5% is the zero settability which is irrelevant for relative measurements.

I'm sorry, I should have been more specific. The 438A manual calls for a 0.5% thermistor reference (with very good return loss @50 MHz) for checking/adjusting the reference power level. The range calibration is done at DC with a 11683A or equivalent (basically a voltage divider along with a sampling circuit that mimics the one used inside the power sensors). The stated accuracies do not include the power sensor itself. In particular, the correction factors can have uncertainties of several % at high frequencies.

I think I remember an alternative way where a single thermister is used as the 50 ohm load, so the RF is directly coupled into the thermister through a bias-T, and DC is applied to maintain the thermister at a constant temperature and 50 ohms.  As the RF signal is lowered, more DC has to be added to maintain the same temperature and 50 ohm load, so the RF power is a constant minus the DC power.  At higher frequencies the thermister is mounted in a transmission line environment.

Yes, this used to be very common and basically all metrology-grade transfer standards use this principle even today. If you use two thermistors instead of one you can avoid using a proper bias-T (and non-ideal inductors). There is a smart arrangement where the thermistors appear in series at DC and in parallel at RF (HP AN-64-1 has a schematic). That's why you will often read about "200 ohm" thermistor mounts (two thermistors at 100 ohms each, so 50 ohms on the RF side). You still need some kind of temperature compensation, usually using a second thermistor bridge, potentially an ovenized enclousure.

If I was going to try and build my own thermal power meter, I think I would go for something along these lines. The additional circuitry is somewhat more complex but you don't have to worry about thermal contact between load and the sensor and not needing a power reference is another plus (you would still need a way to determine the efficiency, i.e. the power lost outside the thermistor, to get decent accuracy at higher frequencies). Finding suitable off-the-shelf thermistors may take some experimentation.

[rf-fil]

@switchabl

Over the past 5 years or so I have accumulated a suite of lab gear with a 1990s list price of in excess of $600k for 1-2 cents on the dollar.  This poses a significant cal issue as I cannot afford to send that stuff out for cal.

That's me as well. And now I'm in need to prove a radio RF power output complies to a local standard, without relying on my employer's gear.

I've run a very crude test just now. I strapped a thermocouple to a 20dB 50ohm attenuator and held it in front of a 12V fan with an alligator clip "third hand" fixture. Then I fed DC current into one end of the attenuator:



The results are surprisingly good:



The time constant was quite long. It took a minute or two to settle.  When I do linear regression on this, I get only 0.03dB of error.  The power range of a few watts might not work for everyone, but I'd be OK with this for my use case. One thing that works is that this uses a nearly perfect RF load. This attenuator is specified out to a few GHz. Using a small 50 ohm termination might work better.
Anyway, good discussion and input from everyone so far. I'll setup a pair of these sensors next, and see if I can keep the heat flow identical, so that the setup can then do real time DC substitution / comparison.

[dietert1]

Yes, this realizes the DC to RF transfer. It's fun and there is something to learn.
Don't you want to reduce the thermal mass of your sensor in order to get a one second or so time constant? Other people have used multiple SMD resistors to make a GHz terminator at the end of a coax cable, like 3x 150R. Or get a G150N50 at ebay (good up to 3 GHz). As soon as you have the time constant down you eventually find that a small oven with some thermal mass is more suitable than a fan. And running the sensor in compensation mode, i mean at constant temperature also helps with speed.

Regards, Dieter

[rhb]

For a precision reference a long time constant improves the accuracy.  The heat equation is an infinite sum of exponentials.  A long temperature vs time history will greatly improve the resolution of the coefficients, especially the constant in the exponent.

I am both surprised and encouraged by getting such results from such a crude device.

I was trying to reply last night but my ISP has borked their DNS & router tables in a manner that I couldn't even refresh my yahoo mail website Inbox.

I am queuing up some quality time with Carslaw and Jaeger's classic, "The Conduction of Heat in Solids".  Flipping through the pages I found analytic expressions for the temperature of a cylinder in an infinite medium.  I've not studied it yet.  What I saw was graphs of the first approximation.  The deviation from linearity is heat loss to the surrounding medium. 

Thanks for sharing your work.  I have a good (Clausing 4902) metal lathe, so making a bespoke thermal mass is not a problem.  Once we work out the details I'll be very pleased to make one and put it in the post to you.

The high specific heat of aluminum makes a long time constant especially easy to achieve.

I am insanely busy with ongoing projects which are higher priority, but I'm sure I can squeeze in a few evenings studying the mathematics this week. 

If you are impatient and want a hollow cylinder quickly, chucking  a 13 mm piece of aluminum in a drill press chuck and feeding it down on a drill bit so that the aluminum is rotating and the drill is stationary will quite precisely align the hole with the outside.

Have Fun!
Reg

[rhb]

Would you please repeat the experiment with a couple of inches of loose cotton or polyester batting wrapped around the sensor?

If there is no heat loss, the graph should be perfectly linear.  This would give an idea of the magnitude of the effect of insulation on the linearity and how much insulation is needed.

Very nice work.  Thanks for sharing it.

Have Fun!
Reg

[rf-fil]

Next set of experiments - I've spent a bit of time figuring out the best RF load for this. I wanted something with smaller thermal mass and as broadband as reasonably achievable. Luckily, I have access to a good VNA at work.

First, I tried an SMT RF load, DigiKey part # 4244-A100N50X4A-ND. The load had a decent match up to 3.5GHz, even though it was just soldered directly onto RG405 semi-rigid cable. Not bad. But after that, I also found that, as per suggestions above (thanks!), just simply soldering an 0603 51-ohm resistor onto a diagonally cut RG405 semi-rigid cable works even better. The return loss plot shows that this will work up to 8GHz, depending on desired accuracy:



Translated into mismatch loss, this gives about 0.1dB accuracy up to about 3.5GHz.

I've now got some tiny RTD temperature sensors and associated interface boards. The plan is to glue the RTD sensor onto the RF load and go from there..

[rhb]

Those ripples in frequency are reflections between impedance mismatches at the ends of the cable as you are probably aware.  The magnitude of the mismatches will vary by part, but the timing won't as that's controlled by the physical geometry.

This implies the possibility of correcting them in software and controlling their spacing.  The longer the cable the closer the spacing of the ripples.  By my calculation your cable is about 10 cm.  Is this correct? If it is, I'd suggest trying the same but using a shorter cable or a bare SMA connector.

I am currently working on repairing a Wiltron 560-7A50 sensor which uses a diode and thermistor and is supposed to be good from 10 MHz to 20 GHz.  The capsule fell apart and the two wire bonded parts are visible.  It will require a shrink fit to reassemble which has me somewhere between scared and terrified.  I've attached photos of the sensor just in case it sparks some random inspiration for this project.

The sensor head had gotten wet which resulted in corrosion at the steel aluminum threads contact between the APC-7 and the aluminum PCB housing.  A suggestion by a member of the Anritsu-Wiltron list to use a dilute acid solution completely removed all the corrosion (50-50 vinegar & distilled water in my case) after a short time in the ultrasonic bath, but when I pulled the brass capsule from the APC-7 it fell apart onto my bench top.

If you don't have access to femptosecond TDR I'd like to examine the device from this post on my Tek 11801/SD-24 sampling scope when you are finished with your testing of it.  I'll cover mailing cost both ways wherever you are.

Have Fun!
Reg