#1447

[@tomsherwood4650]

I always held the engineering types that could take the time and explain stuff like this understandibly, in the highest esteem, in my career.

[@bborkzilla]

We used to lock up those sensors in the RF lab. Only very few people (I was not one of them) had the key. They lasted a lot longer that way.

[@fmashockie]

this video crushed my hopes and dreams... of finding a power sensor on eBay for cheap to repair 😭 but also prevented me from making the mistake of buying one to repair! Thanks! Now do one on why the power sensor cables are so expensive! 😂😂

[@tomstrum6259]

Wow !! .....This brings back +70 yrs ago memories of Microwave rf power measuring using HP 478A Coaxial RF power Sensors working my 1st days with the Bell System....At the time, the relatively Small size & lightweight of the waveguide Adapter, the 478A Sensor + the HP 432A Power meter seemed the Least expen$ive of all the rf test equipment we used !.....I always used Precautions & never blew one (Faik) & seemed reliable life as I remember...Thx for the memories !!!

[@gixerags750]

Passed on a few Boonton Millivolt and Microvolt meters over the years as they did not have their matching sensors. Great video

[@Dennis-uc2gm]

Back when I ran acceptance test on a particular C130 Radar system we use to build for the X and Ka bands I'd use a couple of similar HP sensors. I'd forget to power down the Power Meter before unplugging the sensor and something would blow in it. Needless to say after the 2nd time that happened I got a good chew'n from my boss and never did that again. I'd even leave a sticky note by the connector . They'd send them back to HP at the time to get them repaired but I think it was $$$.

[@InssiAjaton]

Interesting topic! Around 7 minutes the sketch probably should swap the capacitor and the diode. At least that was how Heathkit RF probe was done. That arrangement made it easier to provide bias DC to the diode. It also in effect doubled the voltage within the series capacitor, also helping to extend the diode operating point towards lower amplitudes. As to thermal sensors, my father had a big book called something like "Electricity and its Use". It was printed before WW2, maybe even the late '20s. There was a picture and description of an AC or RF hot wire meter. The wire was mounted end to end on fine adjustable clamps. The current was run from clamp to clamp trough the measuring wire. From the center, there was a perpendicular second, spring loaded wire, running around a shaft that had a pointer attached to it. When the current carrying wire heated up, it increased in length and the spring loaded cross wire pulled the center of the hot wire sideways, enabling pointer movement. The meter could be calibrated with DC current and was ready to measure the RMS value of AC or even RF current. No Schottky diodes or other fancy later inventions!

[@IMSAIGuy]

more about thermal sensors: hparchive.com/Application_Notes/HP-AN-64-1.pdf

[@aduedc]

Do you mean Heathkit used a clamper circuit to meaure?

[@rayoflight62]

When I begun as an hobbyist at the time all I had was a bolometer and an (excellent) Grid-Dip Meter. Later, the transition to Distributed Constants Designs was driven by HP and their design software. When microwave design turned into black magic, I turned to digital design. Thank you for the video - it clicked some ancient but important memories...

[@JxH]

Worth mentioning more-explicitly for those following along at home: I believe that the EEPROM is connected just to the bench instrument, effectively having no significant interface to the rest of the Detector circuit. In other words, the EEPROM with the Cal Data is more-or-less like a memory stick (as it were), storing the Cal Data physically within the Detector (for convenience, and being a perfectly-logical design concept). I don't believe that the EEPROM has any direct internal adjustment effect on the rest of the Detector's circuit. It's just a data store on the end of the same cable. I hope that my recollection is correct...

[@IMSAIGuy]

you are right, I thought that was obvious

[@electronics.unmessed]

Many thanks for the video, I used to work with that kind of sensors a number of years ago. Eventually, I know now what is inside. You know, open these was an absolute no go 😇

[@Zandraccoon]

The best source of parts probably be from other broken units, but even if someone had the tools to replace them, I assume that there are other issues issues at play like diodes need to be from the same batch.

[@mechmania8450]

The hp 478 and 8478 sensors compensate for any change in resistance from incident power by using a feedback loop which re-balances a bridge circuit. It therefore stays at 50 ohm so as to stay matched to the source. Also the 8481 and 8482 thermo-couple sensors have a series capacitor which limits LF response.

[@normanczerski5221]

Diode power sensors are less sensitive than the thermistor or wire type. The chopper circuit you speak of is controlling capacitor coupled RF to the thermistor or wire to maintain it's resistance while the said thermistor/wire is directly a part of a bridge oscillator. When microwave power is introduced across the thermistor/wire it requires less RF to the to the thermistor/wire to balance the bridge. The drop in RF power to balance the bridge indicates microwave power. Ambient temperature and temperature between the sensor and heat sinc must be accounted for.

[@johnwest7993]

I spent 30 years looking for one I could afford. I had gathered a good, relatively modern Agilent control box and the cable over the years, and had them sitting around for a decade before I found the power sensing head with a 25 Watt attenuator for a semi-reasonable price at a poorly advertised auction for a failed start-up. It was sitting in the original box with the cal certificate. I bid on it and bought it immediately for more than I wanted to pay, but far less than it was worth. Now it's on my bench, and NO ONE is allowed to touch it but me. I usually make a power measurement using a cheaper meter first before I use it, just in case. It's the one piece of gear in my lab that I don't even plug in the AC cord until I need to use it.

[@NimbleJack3]

Where do you get it calibrated?

[@johnwest7993]

@@NimbleJack3, I haven't gotten it calibrated. I've only used it twice since opening the box, when I really needed to. I use much cheaper stuff for approximate measurements.

[@ninefox344]

This was super interesting. Thanks for the excellent video!

[@Spookieham]

I've been wondering why the HP Power Meters are so cheap but the sensors are silly money - many hundred $$$ or more. Great video thank you. I'm building one from an AD logarithmic chip with a Arduino that has some calibration capability so it will be good enough for my hobby RF Stuff.

[@willthecat3861]

Yes, you will never get the band-width, and dynamic range, of one of these Gigatronics (or HP, or Agilent, or Keysight) power sensors. That's what makes them desirable for RF metrology work... the range of frequencies, and power levels... and their accuracy and precision. You won't duplicate that for 100 times the cost of a new sensor from Keysight.

[@johnprouty6583]

All the RF products went to Keysight at the split. HP/ Agilent/Keysight fabricated the diodes for many of their sensors if not all.

[@Mark-hb5zf]

I was kind of surprised they used a diode detector. I figured they used a log detector to get the dynamic range and to make it simple. I've made diode detectors in the past (

[@jeffreyyoung4104]

Nice video! The old sensors and detectors we use to used are now considered obsolete now that we have Vector Network Analyzers and Spectrum Analyzers available. I have a bolometer, but the meter portion does not work! I used to use it to test the output of signal generators, but I have a pair of spectrum analyzers, and the bolometer has sat on the shelf gathering dust. It may be ready to go bye bye! The last time I used it was several years ago, and it, and many other devices, are getting ready to be sold for parts or repair!

[@rfengr00]

I still use them frequently, as absolute power measurement accuracy on a spectrum analyzer is poor, as is measuring true mean-square power.

[@zaprodk]

Amazing insight. Thanks for sharing!

[@craxd1]

The diode cartridge works similar to the older designs for RF voltmeter probes using two diodes. However, they didn't have a great frequency response, where a Millivac, like I use, starts to roll off after 10 MHz. I think that is the reason Drake recommended using an RF voltmeter to calibrate their wattmeters at 14 MHz, though 10 would be best at 30 meters. At that, you can only measure 3 VAC RMS maximum, thus, for high power, you need a large and accurate attenuator. In the Millivac probes, the diode section used two diodes and a few resistors and caps, which was fed to the meter that used a chopper. However, it also had three thermistors that led back to the circuitry as well. I think Boonton's was similar.

[@IMSAIGuy]

that sensor is good to 20GHz and we made 40GHz version as well

[@IMSAIGuy]

they also work from -70dBm to +20dBm. you are right about needing an attenuators to move into the watts region. We made those sensors too. you might be interested in the box: kzbin.info/www/bejne/bpqYXoKVbaaNr80si=1Qlc3vAzuvjYRMm2 these are not simple watt meters but NIST traceable power meters good to 0.01dBm

[@khimbittle7705]

Cool, I've always wondered why these were so expensive

[@wherami]

i find the most important things are quite expensive

[@BuckleighPR]

As someone who has had an Agilent power sensor fail, it is good to know why it is so expensive to have it repaired! To be fair to Agilent, their repair service is excellent, but the cost is as you say, nearly as much as a new sensor. Really interesting video.

[@NimbleJack3]

Almost the entire worth of the sensor is tied up in the input assembly. The rest of the thing is just a relatively simple ADC. Most repairs require replacing the input assembly, while it's very hard to blow up anything else.

[@brianclimbs1509]

Is there any way to get a commercially available RF detector diode that is good enough for the lower end of the microwave range? I have some 1S1588 diodes stockpiled for amateur radio use, but I would have thought technology would have improved enough to have a reasonably-priced microwave detector IC by now.

[@Miller7747]

there also techniques that measure a temperature rise across a resistor, and compute power from that. But that is not as simple as you think, you need to keep the resistor isolated from ambient temp.

[@IMSAIGuy]

called a Bolometer

[@Miller7747]

@@IMSAIGuy HP used to make a generator, which was custom made for the telecommunications industry. It was called the HP 3335A "Synthesizer/ Level Generator" This thing was a beast when it came to amplitude performance. It was "ruler flat" within +-0.02db form 200 HZ -> 80 MHZ !! (It gets very expensive to be that amplitude precise, most gens today are spec to +- 1.0 db) Then the telecommunications industry changed overnight, and HP decided to discontinue it (I guess they were the biggest customer) That Bolometer technique we spoke about was used to confirm amplitude flatness in this instrument. They used the phrase "Thermal Converter" and they used this HP part number for it: HP 11050A (for the 50 Ohm output), HP 11050A/H01 (75 Ohm) I own one, it works well but the output attenuator is damaged, and all signals are 20 dbm below what you set on the front panel. A video on the HP 3335A might be popular with your viewers.

[@NimbleJack3]

Why is there a single frame of a coaxial connector at 1:00?

[@IMSAIGuy]

bad video edit. or maybe a secret message to my spy friends

[@dennisqwertyuiop]

I have a Krytar power meter the sensor goes to 18 G ,,works great

[@byronwatkins2565]

Rectification works to several kHz. Schottkey diodes work up to 200 kHz or so. Above that, we must use its nonlinear (I, V) to mix the signal with itself and to extract the squared term. RC averages this to a voltage proportional to power. Device uniformity is a problem so that these must be individually calibrated and the calibration stored in ROM.

[@IMSAIGuy]

for those interested in this topic: hparchive.com/Application_Notes/HP-AN-64-1.pdf

[@djmips]

What do all the triangle symbols mean on the various chips on this PCB?

[@joaopaulocoelho5401]

No microcontroller? Then the cal parameters in the I2C e2prom are used by the software that reads the power from the power sensor?

[@davidv1289]

Correct, the calibration factors are specific to the sensor. So when a sensor is connected to the power meter the cal data is read from the sensor. This makes use of different sensors easier. Prior to that, power meters had tables of "cal factors" stored in them and the user had to remember which table went with which sensor and manually select the correct table. Those tables had to be manually entered into the power meter memory. And prior to that, power meters had switches or dials that had to be set to the cal factor for the correct frequency based on a printed table attached to the sensor. Regards, David

[@IMSAIGuy]

the sensor is used with a power meter: kzbin.info/www/bejne/iqmpmJeVZstkadU

[@AJMansfield1]

A diode-based sensor might be pretty impossible to build yourself, but a thermal RF power sensor sounds like it'd be pretty easy to DIY -- a pair of 50 ohm resistors you've picked out to match against each other, with a thermocouple junction bonded to each one; one resistor connected to the test load, the other connected to an op-amp that has the thermocouple junctions wired in series between the + and - inputs, to servo the current to keep the temperature matched. Wouldn't have a fast transient response or anything, but the cal curve doesn't seem like it'd be bad at all for a system like that.

[@IMSAIGuy]

hparchive.com/Application_Notes/HP-AN-64-1.pdf

[@xandear1]

Amazing video! Thank you so much to share your knowledge.

[@lmamakos]

You blow up something like that.. the shriek of the high-pressure magic smoke escaping must be something! As small as that is, probably only your dog could hear it.

[@willthecat3861]

for these power sensors (some measure pW) ...you don't see any evidence of anything happening... all you see is an overload indication on the power meter. Next time you try to zero, and cal it... you can't do it.

[@IMSAIGuy]

I will have ImsaiDog on alert 🐕

[@NimbleJack3]

willthecat is correct - the components are too small and the power levels involved are too low for there to be any sort of alarming noise. the input resistor just quietly burns out without any sort of fanfare. You'd have to connect the sensor to a BIG amplifier, and all you'd probably get is arcing between the inner and outer conductors.

[@lmamakos]

@@NimbleJack3 it's a" magic smoke" joke. You know, about how components stop working when the magic smoke gets released? The more expensive and fancy the component, the more magic smoke it contains.

[@chrisscott1547]

If one doesn't need the greatest sensitivity or operation above a GHz, seems like it could be built. With those conditions, would 1n5711s be better than an AD8307?

[@IMSAIGuy]

kzbin.info/www/bejne/mXWllaqCjJaksKM

[@willthecat3861]

If you can get good RF diodes (and I mean good) and you can get someone to cut the PCB, with the appropriate PCB material... and you can do the transmission line, and capacitor design (or get the appropriate capacitors) then you can get something working... which you would have to cal... so you probably need a working HP/Agilent/Keysight sensor, and power meter. Many have tried... many have failed. Much better luck at HF and a couple of GHz through.

[@Broken_Yugo]

Down at HF and a couple percent precision it's easy, difficulty ramps from there.

[@KimJones-xd5bt]

The gold part is a PIN or Schottky diode , that's all. ???

[@lincolnworsham1085]

You didn't have to buy one, i've got a couple you can take a crack at, lol

[@jafinch78]

Ah, yes, I see now. Never have seen a marketed diode made out of gold. Same goes with a capacitor. Amazing how the manufacturers make batches and the QC dept (or maybe another term used) determines what the model-part-chip number in regards to the solid state electronic parts.

[@absurdengineering]

I’d hope the company making the diodes doesn’t care whether you’re an individual or not, they care whether they can afford to spend time talking with you. I’m not sure how much such perfect wafers cost, but if someone had a spare $50k or $200k or whatever, the company would be plenty happy to get their money :) For $1M and good leads on used equipment you could probably get everything needed to repair those things, including a wafer of diodes. It still would be an expensive service just keeping all your gear in calibration :(. And the cost of breaking a few things along the way to learning how to do everything right would be there too. So I guess as a business opportunity it would work, but if the money is someone else’s, their dividends would make the cost to customer in the same ballpark as manufacturer repair. And once the costs are similar, people will go with the more trusted vendor, even if it costs a bit more. I’m pretty sure that if those repairs could be done for much less and still paid a living wage, someone would be doing it. Quite a few metrology people out there who would know how or have the ability to learn. But it just wouldn’t cost any less to do it in a “garage”, and that would be some spotless air-conditioned garage, too :) I think the biggest unsaid/implied nugget here is that as much as these things cost, the market for them is competitive. To make those sensors cost significantly less, they’d need to be something normal people buy at Walmart. Then economies of scale would kick in and you’d be buying those sensors from Apple or Samsung for a couple hundred bucks. As much a metrology nut in me would like there to be a consumer market for these things, that’s all but a fantasy and will be for a while :)

[@IMSAIGuy]

here is one that would be ok: www.digikey.com/en/products/detail/macom-technology-solutions/MSS20-145-B10D/10249942

[@aduedc]

You just need a wide band AM de-modulator. It is unnecessarily designed too complicated. The reason is that it has too wide frequency range. I would design it in several ranges of frequency to ease the design requirement, and as a result cost, and probably you could do a better job of detection. Say you could break it like this, and design one device for each range: 0 to 100MHz 100 to 500 MHz 500MHz to 1GHz 1GHz to 2GHz etc. Also I would reduce the maximum power range to 15dBm (i.e. 3.5566Vpp)and minimum power to 1dBm ( ie. 0.70963 Vpp so it easily turn on the diode), so you need to use either a LNA or Attenuator before the unit for out of range power. Note that when you want to use a RF Power sensor , you need to detect the RF power for limited range of frequencies anyway. So by dividing the frequency you could put a band pass filter in the input to limit other frequencies influence. Now in unlikely case that you want to measure all the RF frequencies from say from 0 to 18 GHz yo use several measurements.

[@NimbleJack3]

A demodulator or filter would present an unreasonably high VSWR to the device under test. The closer you can get your sensor input to a purely resistive 50 ohm impedance, the more accurate the power measurement will be.

[@huf67]

12:44 ... Stay gold pony boy !!

[@Chris_Grossman]

A guy at the W6TRW swap meet today was asking $400 for a gigatronics meter with one power sensor

[@tomsherwood4650]

I got an old General Microwave power meter analog type at Dayton one year with a 3W. sensor. I think it was like 40$. Seems to work just fine. Although at very low power levels below the max it is a bit drifty down there.

[@ruhnet]

You need a little needle for pointing when under the microscope haha.

[@ploegmma]

You have the RF Power Snitch, that’s a diode detector.

[@IMSAIGuy]

oh, there are plenty of bad diode detectors around

[@UpinkProduction]

I love it good

[@tomstrum6259]

Actually +50 yrs memories from early '70's I meant !!

[@xenoxaos1]

I thought bolometers measured the accuracy of your bolo tie...

[@OceanusHelios]

Novice here. I might be off my rocker or talking out of my pay grade but that sensor looks like a MEMS device.

[@IMSAIGuy]

nope

[@craigwebster3828]

H74xt quad amp

[@ovalwingnut]

I have a new found respect for these sensors... I shall no longer "rage against the machine". Q: Is testing "relative RF power measurements" using a inexpensive Power Meter and a home-brew sensor in the 2-5Ghz range a obtainable quest? I would like to get a handle on WiFi output and seat of the pants antenna performance. As in "better or worse" case scenarios. I ask this without any foundational knowledge. My apologies. Thank you. p.s. No Commercials? You RoCk!

[@IMSAIGuy]

you can get a sensor: kzbin.info/www/bejne/mXWllaqCjJaksKM but you still need to calibrate it. so you need a friend with a good meter. Thank you, I turn off in video ads and I turn down sponsorship offers. I hate them as much as anyone.

[@ovalwingnut]

@@IMSAIGuy Thank you. My knowledge on the subject has increased by 3db :)

[@mikebond6328]

bolgnameter?

[@IMSAIGuy]

en.wikipedia.org/wiki/Bolometer

[@__--JY-Moe--__]

power sensor! sensor 4 power! no wonder it's broken!! to many certifications on it! hell, it'll take a week just 2 peal off all those stupid stickers!! jeez man! no wonder it blew up!

[@agnelodsa788]

Keeps thinking while presenting g the video. He has to get the facts together vefore starting g.

[@IMSAIGuy]

if you are looking for polished videos, go elsewhere. If you want lots of content (1447 videos and counting) stick around or go watch older ones.

[@mohsinsiddique2486]

Please make n arduino based swr and power meter fm/vhf/uhf

[@heinrichhein2605]

thanks for the vid, i think you did not resolve why it has to be gold, Thanks

[@chrisscott1547]

Very conductive, corrosion resistant, and no mating dissimilar metals which can have galvanic activity.

[@IMSAIGuy]

lead-copper contact will generate 5uV/degC gold-gold is zero

[@heinrichhein2605]

@@IMSAIGuy but is copper to copper not the best?

[@heinrichhein2605]

@@IMSAIGuy I only know it from low EMF connection even the under layer of the gold can be a problem

[@NimbleJack3]

@@heinrichhein2605 Copper oxidises, gold does not.

[@Jkauppa]

because its both the seller and the buyer who are dum dum paying any prices for free stuff

[@Jkauppa]

yes the distributor thought it was wise to tie into the useless money-law system

[@stepannovotny4291]

It's over-engineered. Boonton had much simpler probes which anyone can repair. I would not pay money for this!

[@IMSAIGuy]

Mercedes vs Toyota

[@vf3378]

I have couple of Boonton power meters with different sensors. If you are not able to find pair of meter and sensor as was calibrated in the factory, it is practically impossible to calibrate it later by your own. There we always calibration tables (several pages) for each sensor and also for meter (usually hidden inside of meter enclosure). If you don’t have it, I haven’t found the way how to calibrate it by your own. Compared to HP/Agilent where you connect power sensor to internal meter 50MHz/0dBm source and after simple procedure you are ready for measurement.

[@gixerags750]

I have a Boonton signal generator and was advised that Boonton Millivolt and Microvolt meters are useless without their ORIGINAL sensors... Cheers

[@stepannovotny4291]

@@gixerags750 Smart man, a meter without a probe is indeed useless, and also with a probe if you don't calibrate it. The calibration is multi-point so that you can accommodate different segments of the curve of whichever diodes you are using. Different probes (for the same meter) use different diodes. See the probe schematics floating around on the web.

[@NimbleJack3]

And I bet you dollars to donuts that the Boonton probes were calibrated using precision power sensors. These are lab-grade instruments, not for lugging around on a worksite.

[@tahirsongelen7781]

mni tnx ur efford BR from istanbul de ta1bj