HP 435B Teardown

[Electro Fan]

Just took a quick stroll through the 437B manual.

It has one really nice feature where you can store the cal factors for up to 10 sensors and then just enter frequency of interest.  You don't have to keep looking at the printed sensor cal table.

That IS nice ; after getting the first sensor there would be very little friction to inhibit buying about 9 more... 

Seriously, it is a good feature.

I just quickly looked through the 438A manual - it has support for two sensors at the same time but I didn't see the feature that would let you save the settings for up to 10 sensors (but I might have missed it).

[G0HZU]

My Anritsu power meter is a similar vintage and offers the same cal factor storage feature although I'm not sure if it supports multiple sensors in memory. It probably does.

But it is much easier to just use it manually because it has a rotary control to set the cal factor and it is a breeze to adjust it like this. By comparison, it isn't that intuitive how to enter all the cal data and it would take a fair while to do and you still have to dial in the frequency once completed. But again this is done with a rotary control so it's much easier than typing the number in.

When I did the above plots I did it remotely and stored all the cal factors in the PC and the program used interpolation between points to really get the best performance from it. The HP431C doesn't have GPIB so it was logged from the analogue recorder/DMM port at the back. The cal factor vs frequency on the HP478A sensor is fairly flat so I just used one fixed cal factor across the whole 4GHz range. I think it would have been slightly better if I had corrected it for cal factor but this would still have been dominated by the ripple caused by mismatch uncertainty in the 478A sensor head

[MarkL]

I just quickly looked through the 438A manual - it has support for two sensors at the same time but I didn't see the feature that would let you save the settings for up to 10 sensors (but I might have missed it).

No, you didn't miss it.  Sadly, the 438A doesn't have the cal table feature.

It was either the cal table or the dual channels.  And the 438A also had a metal case and a readily available service manual with full schematics.  So went with the 438A.

It also has an loud fan (same as the 8116A), so if noise bothers you don't get a 438A.  In my lab it's just one of many, so I'm used to it.

[Rupunzell]

Story behind the small wiggle meter on the hp 436A, 438A and others. When the hp 435 series of thermocouple based power meters were introduced the analog meter as typical and allowed observation in rates of change in power. Observing rates of change is poor to not possible with digital read out type displays. Here is a link to the hp journal introducing the 435 series of power meters. It is a LOT more complex and cleaver than it appears. An appreciation of how it worked needs to be studied to understand just how well the folks at hp addressed this design problem and solved it in a ver enduring way that lives on to this day.
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1974-09.pdf

RF power meters are much like a "Volt-Ohm-Amp" meter in the world of RF and microwave folks. Since most RF systems are 50 ohm based, this makes a plain and basic volt or current or current meter less useful as having a fixed impedance means power is more relevant than just volts, or current as either of those units are not relative to a impedance unless the impedance is known. This is why power based on a 50 ohm system is used as a normalization for measuring power in RF and power systems.

With the introduction of the hp 435 & hp 8481, power measurements were made easier. Before this system, thermistors/thermistor mounts and all the problems associated with that system of measurement was a source of never ending niggles. The analog meter work well, it was accurate, allowed rate of change observation and was reliable, durable and accurate.

In time the folks at hp decided to update aka "improve" the 435 series of meters by offering a digital readout version known as the 436A. During instrument trials, users trying to observe rate of change (in power) discovered it was near impossible with the digital display. To address this real world instrument requirement, the folks at hp added a small wiggle meter specifically designed and intended to display rate of change.

In the home lab, the 435A/B. 436A, 438A (fan sounds like a jet taking off) all live together as the complements to each other and allow great power measurement flexibility. The power meters are common, the power sensors and cables are precious as they are compatible over a broad range of hp power meters. These can be amazingly accurate for years and years. The power sensor heads can be fragile and easily damaged if excessive power is applied. They are tolerant to this kind of abuse to a limit. Go past that limit and the power sensor is dead.
Wiggle meters have their place and offer advantages digital displays cannot. Another case of knowing how to apply what work best for a given measurement requirement.

In the passing decades since the introduction of hp's thermocouple based power sensor, came the diode based power sensor and eventually other manufactures got into this game including Gigatronis (previously Systron Donner),  Wiltron / Anritsu and others including made in China knock-offs.

 
Bernice

[G0HZU]

Wiggle meters have their place and offer advantages digital displays cannot. Another case of knowing how to apply what work best for a given measurement requirement.

My Anritsu power meter has a digital display but also has a very similar type of 'wiggle' meter fitted as the HP436/7.

The wiggle meter looks like a direct copy 

However it compares really badly with my HP431C power meter dial in terms of 'wiggle' feedback in a visual sense. i.e. you have to hunch forwards and look very closely to use it and the little analogue meter is obviously a low cost thing on my Anritsu. By comparison the big old dial on my 431C can be used at a fair distance and gives good visual feedback.
it's not surprising that these tiny cheap wiggle meters were phased out pretty quickly.

I use mine mainly to let me know when the meter is near the low edge of a range and the measurement uncertainty is degrading. I suspect that other users will do the same.
If I want to use the (digital) Anritsu meter as an analogue 'wiggle' meter I can always connect the analogue recorder output to an analogue voltmeter. eg something like an AVO 8 and get a nice big wiggle meter. The AVO will probably load the output slightly (compared to a DMM) but I guess this doesn't matter for a wiggle meter. The alternative would be to log it over GPIB and have a peaking meter display on a PC/Laptop.

But usually I would rather work with a spectrum analyser with a 20dB attenuator selected if peaking something up

[Rupunzell]

That plastic wonder of what appears to be a wiggle meter has got to be one of the lowest cost wigglers they could have possibly used. It was all they had room for in the 436A as oops correction. That is why it looks less than ideal on the front panel layout.

Indeed, the 413/435 series have proper hp wiggle meters of proper size, mirrored scales and needle movement dynamics. These features are not often appreciated today as digital readout ease has replaced learning and understanding the advantages analog wigglers offer over a digital readout.

I do believe reading an analog meter properly is a learned thing that comes with time and practice and experience. Digits are direct reading like spoon feeding.

The set of power meters here are good to fractions of a db which is really good for a power meter. Still using an SA delivers a LOT more information about the DUT overall.


Bernice

My Anritsu power meter has a digital display but also has a very similar type of 'wiggle' meter fitted as the HP436/7.

The wiggle meter looks like a direct copy 

However it compares really badly with my HP431C power meter dial in terms of 'wiggle' feedback in a visual sense. i.e. you have to hunch forwards and look very closely to use it and the little analogue meter is obviously a low cost thing on my Anritsu. By comparison the big old dial on my 431C can be used at a fair distance and gives good visual feedback.
it's not surprising that these tiny cheap wiggle meters were phased out pretty quickly.

I use mine mainly to let me know when the meter is near the low edge of a range and the measurement uncertainty is degrading. I suspect that other users will do the same.
If I want to use the (digital) Anritsu meter as an analogue 'wiggle' meter I can always connect the analogue recorder output to an analogue voltmeter. eg something like an AVO 8 and get a nice big wiggle meter. The AVO will probably load the output slightly (compared to a DMM) but I guess this doesn't matter for a wiggle meter. The alternative would be to log it over GPIB and have a peaking meter display on a PC/Laptop.

But usually I would rather work with a spectrum analyser with a 20dB attenuator selected if peaking something up

[Electro Fan]

In watching this thread evolve it seems that there is a bit of a consensus developing that a Spectrum Analyzer not only delivers a lot more information than a Power Meter but that it might be the preferred tool for power measurements in general.  Or did I misinterpret that?

Some reasons to use a Power Meter:

1) It's less expensive than a SA and a SA might not be available
2) The less expensive PM might be a first line tool to protect a more valuable SA (in case something should glitch with a power reading)
3) The PM is smaller and maybe more versatile in not only size but with especially with some meters that can support 2 sensors at the same time
4) The PM could help confirm measurements from other equipment (SA, Sig Gen, etc.)
5) ? I'm out of ideas

So, is it possible that investing in a SA would largely preclude the need for a Power Meter?  (thereby freeing up funds toward a SA, or a better SA?)  Or is a Power Meter a basic tool like a screwdriver, wire stripper, etc.?

Thanks for all the insights.

[MarkL]

The SA is going to provide infinitely more information.  There's no question of an SA over a power meter if you only get to buy one of them.

The power meter is only going to tell you the average power of *everything* coming in, including all the harmonics and other distortions of a signal.

The primary reason I have a power meter is to calibrate my SAs.

[dom0]

The power meter is only going to tell you the average power of *everything* coming in, including all the harmonics and other distortions of a signal.

Doesn't that depend on the type of detector? Or do they all use thermal detectors?

[MarkL]

The power meter is only going to tell you the average power of *everything* coming in, including all the harmonics and other distortions of a signal.

Doesn't that depend on the type of detector? Or do they all use thermal detectors?

There are peak sensors too, but these legacy ones that we're talking about are only average power sensing.

Some sensors are diode based, where you can get much higher sensitivity.  Here's probably more than anyone would want to know on various sensor technology (part 2 of a 4 part series):

  (2) Sensors and Instrumentation: http://literature.cdn.keysight.com/litweb/pdf/5988-9214EN.pdf

The other parts are:

  (1) Intro and Standards:  http://literature.cdn.keysight.com/litweb/pdf/5988-9213EN.pdf

  (3) Measurement Uncertainty:  http://literature.cdn.keysight.com/litweb/pdf/5988-9215EN.pdf

  (4) Keysight Offerings:  http://literature.cdn.keysight.com/litweb/pdf/5988-9216EN.pdf


(Also, I'm not trying to be overly biased towards HP/Agilent/Keysight.  Other companies certainly sell power meters, but this is the only gear I'm familiar with.)

[G0HZU]

I use both a classic lab power meter and a spectrum  analyser to measure RF power. But usually the power meter is king if you want to measure power of a (dominant) signal with minimal measurement uncertainty.
However the latest generations of lab grade spectrum/signal analyser offer quite low uncertainty for power measurement.

Traditionally, the power meter was also king when measuring the total power of broadband noise but modern spectrum analysers (and some scopes) can do this remarkably well nowadays.
But you do need a sensor port with very low VSWR to get the uncertainty low.

The sensors for my Anritsu power meter maintain a typical port VSWR of 1.05:1 across a huge bandwidth and have minimal variation in cal factor across a huge BW. This makes them very useful for measuring broadband noise accurately.

Obviously, a spectrum analyser can see much smaller signals than a classic power meter so it has an edge here and you can also ask some of them to measure average power across a defined bandwidth/window or to compute average noise density across such a window. This can be very useful and a power meter can't compete here.

But a typical spectrum analyser made in the last 30 years will have so much overall measurement uncertainty that it can't be relied on to be within even +/-1dB. Some (many) will be much, much worse than this. Only the very best of the modern generation of $$$ signal/spectrum analysers can improve on this performance.

With careful management of overall measurement uncertainty a decent lab power meter should often get within +/- 0.15dB if used where its sensor VSWR is low and where the sensor range is at its most linear/accurate. But the overall uncertainty (worst case analysis) is much worse than this. Hence the need to try and manage/control uncertainty as much as possible.

[Rupunzell]

hp 438A example. Quick test using the cal output and with an attenuator spec'ed at 10db with label curve noted at 9.7 db at it's lowest range ( 0 Ghz) which is good enough for the cal signal of 50Mhz set to relative zero db on the 438A. Display on the 438A reads -9.63db, or if all is correct a difference of 0.07db.

438A, 0db ref set, cal factor 98% (from power sensor).



438A, display reads -9.63db. Attenuator is noted -9.70db at (0 Ghz ).



These power meters are accurate and does well at what they are designed to do when used properly.



Bernice

But usually the power meter is king if you want to measure power of a (dominant) signal with minimal measurement uncertainty.

Obviously, a spectrum analyser can see much smaller signals than a classic power meter so it has an edge here and you can also ask some of them to measure average power across a defined bandwidth/window or to compute average noise density across such a window. This can be very useful and a power meter can't compete here.

But a typical spectrum analyser made in the last 30 years will have so much overall measurement uncertainty that it can't be relied on to be within even +/-1dB. Some (many) will be much, much worse than this. Only the very best of the modern generation of $$$ signal/spectrum analysers can improve on this performance.

With careful management of overall measurement uncertainty a decent lab power meter should often get within +/- 0.15dB if used where its sensor VSWR is low and where the sensor range is at its most linear/accurate. But the overall uncertainty (worst case analysis) is much worse than this. Hence the need to try and manage/control uncertainty as much as possible.

[N8AUM]

Hello all,
just picked up my 1st power meter, HP-435B off of fleabay. Any ways, what are some of the symptoms of a bad sensor ? I cannot zero it and when I apply a small signal it reads more than 30db greater than it should be. It came with the 8484 sensor.

N8AUM

 

[MarkL]

Hello all,
just picked up my 1st power meter, HP-435B off of fleabay. Any ways, what are some of the symptoms of a bad sensor ? I cannot zero it and when I apply a small signal it reads more than 30db greater than it should be. It came with the 8484 sensor.

I'm assuming that would be a 8484A?

The 8484A sensor is a high-sensitivity diode based sensor, so by design it has a 30dB offset once calibrated.

Did you try to calibrate it using the 30db calibration attenuator (part #11708A)?  It's similar to the one Bernice shows plugged into his 438A a couple of posts back.

If you already did this, does the sensor calibrate and it just won't zero?

My reading of the schematic is that it could be either in the sensor or the meter.  There are some troubleshooting steps in both the 435B and the 8484A manuals, if you want to get into it.

But if you start troubleshooting, many sellers won't consider a return once the covers are off.  And if it's the sensor, you're going to be stuck with an unobtainium repair.


I don't have the 435B or that sensor, so maybe someone can comment who has direct knowledge if there's any quirks with that meter/sensor pair.

[N8AUM]

Thanks for the reply MarkL. I just noticed I made an error when I described the symtom. With a 30db attenuator on the cal output I can zero the meter but only if I set the meter to +10db and not 0db like it says to do in the manual.

[MarkL]

When you say "zero" do you mean while applying power input to the sensor?  To be clear on terminology, it's a two-point procedure. 

The first step is "zeroing" where no power is applied to the sensor input and the "Zero" switch is pushed on the front.  This automatically balances out any residual offset in the amplifier circuits.

The second is attaching the sensor to the front cal output and adjusting for a 0dBm (or 1mW) reading with Cal Adj.  In your case with the 8484A you need the 30dBm attenuator in there and you would still adjust for 0dBm (although the actual sensor input would be -30dBm).  I think on this meter the Cal Factor is always active, so you might also have to set this to the sensor's cal factor for 50MHz (usually 100%, unless printed otherwise on the sensor).

I think you're saying the second step is not behaving quite right?

[N8AUM]

I was referring to calibrating using the 1mW internal reference. Anyways, I just fingered it out, it has the wrong scale for this sensor    !
Damn, cant believe it took me this long to figure it out !   

[MarkL]

Great - glad you figured it out!

The manual mentions something about swapping range scale rings on page 2-1.  I guess that was it?

[N8AUM]

Great - glad you figured it out!

The manual mentions something about swapping range scale rings on page 2-1.  I guess that was it?

Wish it would have clicked when I read it but guess it didnt. Had to take a picture of the ring I need and paste it to the original ring that came with, not the prettiest but it will do.

[MarkL]

The manual says there should be two scale rings stacked behind the scale knob (with three different scales - I guess one is double-sided).  The knob unscrews so you can put the scale you want in the front.

Or maybe the ring you need is gone?  Such is the luck with used equipment.

[N8AUM]

The manual says there should be two scale rings stacked behind the scale knob (with three different scales - I guess one is double-sided).  The knob unscrews so you can put the scale you want in the front.

Or maybe the ring you need is gone?  Such is the luck with used equipment.

It just had one ring. I just purchased another 435A, hopefully it will have the ring I need.