HP 435B Teardown

[amc184]

I bought this HP 435B from eBay with the intention of reusing parts of it in a project.  It's a really nice, well built unit, but without the power sensor they're fairly worthless.  I got this one for about USD25, including shipping.  Physically it's pretty tidy, and it probably works (not that I care).





It's constructed of two cast aluminium side frames.  Folded aluminium braces sit between these, and painted aluminium panels screw to the outside. 









The main riser board, A4, contains most of the actual meter circuit, as well as the power supply.  There are five amplifier stages, as well as support circuitry for the sensor.  The grey box appears to be potted, and contains the auto zero circuit.  The power supply can work from mains or a 28.8V Ni-Cd battery, and includes charging circuitry.  Mine doesn't have the battery option fitted.



The small board in the can behind the panel meter, A3, provides the reference output for calibrating the sensor.



The only horizontal board, A5, connects the other boards together.  It also holds the two range switch attenuators, made up of two ADG201 CMOS switches and some precision resistors.  Each group of precision resistors is surrounded by a grounded guard trace.

[dave_k]

Very nice! What project did you have in mind .. ?

[sportq]

SInce it was made in the UK ('U' in the serial number) I was desparately hoping to see a date code in the 1987-1988 range as there's a good chance I might have done the factory verification and calibration on it. The chips I can see make it mid-1997, I can't believe they were still making them then. I thought it was being phased out when I worked on them. We also made the 436 and 437 meters on the same line too.

Pete

[wertyq]

very nice

[dom0]

Maybe it was a lower-cost alternative to the 436/437/438 meters.

By the way, does anyone know the exact date the 3400 was discontinued?

[amc184]

Rather unoriginal project, I'm making it into an alarm clock with a nixie display.  I wanted to make it out of an old piece of HP test gear in this upright form factor.  The 435B was cheap, and the controls suited what I wanted.  The power button will stay as is, the range switch will be the main UI control, I'll probably replace the 'calibrator %' switch with a potentiometer for volume, and the 'zero' switch will be the alarm cancel switch, as well as part of the UI.

This unit is the 2732U revision, made in at least as late as 1997.  I'm not exactly sure when these first came out, the service manual I have says at least as early as 1980.  Interestingly, it also says these were made in both the US and UK.  I'm not sure if it was kept around for its lower cost, it could have just been because it suited some applications.  An analogue meter can be really useful at times, especially for unstable signals.

I wanted to do a quick teardown before this thing comes apart.  This must have been a common unit, but I didn't see much info on it on the net.

[G0HZU]

Shame you are not in the UK because I would have made an offer on the analogue display meter part of that power meter if it was surplus to requirements...

I have the older HP431C meter and I 'think' your meter dial would fit it. Over time the microscopically thin 'transfer' layer that holds the text peels away from the white background and can jam the needle movement. Mine has done this several times now and this is a common problem on similar meters made in the 60s and 70s.

However, I usually use the 'recorder' output at the back into a DMM for accurate measurements so I can still use the meter.

[G0HZU]

This must have been a common unit, but I didn't see much info on it on the net.

An analogue meter can be really useful at times, especially for unstable signals.

The HP432 and HP435 meters were very popular when working up at microwave frequencies where something was being adjusted or 'peaked' in power. Obviously, the analogue display was very powerful here.

IMO the one to have was the older HP432A with the HP478A thermistor head. The HP435 supports the later (modern) thermocouple type head but the HP432A with HP478A head is the real star because it uses a closed loop system for calibration and can be very accurate and maintain this accuracy for very long periods. The HP435 needs its internal 0dBm reference because it uses an open loop cal system.

Most people either want the older HP432A or the HP power meters with a digital display and GPIB so the HP435 is kind of lost in the middle and isn't in much demand these days.

[amc184]

The analogue meter is yours for the cost of postage if you want it, I won't be reusing that part of the unit.  PM me if you're interested.

The HP432 and HP435 meters were very popular when working up at microwave frequencies where something was being adjusted or 'peaked' in power.

That's exactly the type of thing I was thinking of.  Thanks for the info.

[johnh]

So what happened to all those power sensors?

Blow up/lost or are they reusable with different model meters?

[Vgkid]

^^^ I've always been wondering the same thing, I'm guessing blown up.

[MarkL]

The 435B uses the 8480 series power sensors which continue to be supported to this day on the newest meters from Keysight.  They must be at least 30 years old by now and a set of them for different sensitivities and frequencies can represent quite an investment.

And they *are* easy to blow up.  Don't buy any used ones unless they are returnable.

The reference oscillators I've seen in these vintage meters have excellent stability.  I would at least salvage that.

EDIT: The 8480 sensor series started appearing in the 1973 catalog with the 8481A.  So it's at least 42 years old.  Now that's long term support.

[Electro Fan]

Nice looking meter and nice photos.

Slightly off topic....

Where on the continuum from older not so expensive power meters (such as the one in the OP) toward something newer and perhaps more expensive might be a sweet spot of value for an RF power meter (and sensor?) if someone wanted to measure/confirm power when using a 3GHz signal generator with a 3GHz spectrum analyzer?  Any favorite HP or other models?  Any preferences on digital readouts vs. analog meters?  Thx

[Electro Fan]

This must have been a common unit, but I didn't see much info on it on the net.

An analogue meter can be really useful at times, especially for unstable signals.

The HP432 and HP435 meters were very popular when working up at microwave frequencies where something was being adjusted or 'peaked' in power. Obviously, the analogue display was very powerful here.

IMO the one to have was the older HP432A with the HP478A thermistor head. The HP435 supports the later (modern) thermocouple type head but the HP432A with HP478A head is the real star because it uses a closed loop system for calibration and can be very accurate and maintain this accuracy for very long periods. The HP435 needs its internal 0dBm reference because it uses an open loop cal system.

Most people either want the older HP432A or the HP power meters with a digital display and GPIB so the HP435 is kind of lost in the middle and isn't in much demand these days.

Sorry, reading this thread backwards  ; maybe this is the answer or part of the answer to the question I posted above?  Thx

[MarkL]

Where on the continuum from older not so expensive power meters (such as the one in the OP) toward something newer and perhaps more expensive might be a sweet spot of value for an RF power meter (and sensor?) if someone wanted to measure/confirm power when using a 3GHz signal generator with a 3GHz spectrum analyzer?  Any favorite HP or other models?  Any preferences on digital readouts vs. analog meters?  Thx

Over analog meters, digital meters allow you to do easy relative measurements from a reference value, and they connect via GPIB for automated tasks.  They also have store/recall for multiple setups.

I happen to like the HP 438A.  Like G0HZU points out, this is an open loop meter.  It's not quite as old as the 435B.  It has dual inputs so you can do relative measurements between two sensors.  Option 002 adds a second reference oscillator which is nice to double check your calibration.  It can be had for $80 to $100.

8480 series sensors for the 438A (and the myriad of others) can run $200 to $400.  As I said previously, if you're buying used make sure it's returnable, and also make sure it has the calibration chart on the back (Freq vs. Cal Factor).  For some reason I've seen a lot of sensors on ebay with the chart ripped off or defaced.  If you can't see it in the photos, ask.  You need this data unless you want to have the sensor re-calibrated.

A more modern-ish looking unit is the HP 437B with an LCD display.  Same price range and also uses 8480 series sensors.  It's a single channel unit.

An improvement to the 8480 series sensors is the N8480 sensors which have the calibration table built into the sensor in an EEPROM.  Although they look the same and are similarly numbered, they are incompatible with these older meters.

More modern meters have sensors with better dynamic range, and there's also USB-based power sensors.  But for just an average power reading I find these old units work fine for a reasonable price.

[Electro Fan]

Where on the continuum from older not so expensive power meters (such as the one in the OP) toward something newer and perhaps more expensive might be a sweet spot of value for an RF power meter (and sensor?) if someone wanted to measure/confirm power when using a 3GHz signal generator with a 3GHz spectrum analyzer?  Any favorite HP or other models?  Any preferences on digital readouts vs. analog meters?  Thx

Over analog meters, digital meters allow you to do easy relative measurements from a reference value, and they connect via GPIB for automated tasks.  They also have store/recall for multiple setups.

I happen to like the HP 438A.  Like G0HZU points out, this is an open loop meter.  It's not quite as old as the 435B.  It has dual inputs so you can do relative measurements between two sensors.  Option 002 adds a second reference oscillator which is nice to double check your calibration.  It can be had for $80 to $100.

8480 series sensors for the 438A (and the myriad of others) can run $200 to $400.  As I said previously, if you're buying used make sure it's returnable, and also make sure it has the calibration chart on the back (Freq vs. Cal Factor).  For some reason I've seen a lot of sensors on ebay with the chart ripped off or defaced.  If you can't see it in the photos, ask.  You need this data unless you want to have the sensor re-calibrated.

A more modern-ish looking unit is the HP 437B with an LCD display.  Same price range and also uses 8480 series sensors.  It's a single channel unit.

An improvement to the 8480 series sensors is the N8480 sensors which have the calibration table built into the sensor in an EEPROM.  Although they look the same and are similarly numbered, they are incompatible with these older meters.

More modern meters have sensors with better dynamic range, and there's also USB-based power sensors.  But for just an average power reading I find these old units work fine for a reasonable price.

MarkL - Thanks.

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

Also, what's your preference on the digital readout vs. an analog meter, vs. the digital readout plus simple analog meter layout on a meter like the 437B?

Thx, EF

[G0HZU]

The analogue meter is yours for the cost of postage if you want it, I won't be reusing that part of the unit.  PM me if you're interested.

Thanks very much for the offer but, sadly, I think the postage costs would be too high to send it to the UK. But someone in your country will want this meter for the same reason as me so don't throw it away...

[G0HZU]

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

It's a fairly common type where the lower frequency limit is 10MHz and it probably gives best performance (in terms of minimising uncertainty) above about 25MHz.

You can also get the HP8482A power head that goes from 100kHz to about 4GHz but these may be harder to find at a cheap price. They are both very easy to damage though.

[MarkL]

MarkL - Thanks.

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

Also, what's your preference on the digital readout vs. an analog meter, vs. the digital readout plus simple analog meter layout on a meter like the 437B?

Thx, EF

The sensor looks like it's in pretty good shape, and the table is there.  I've seen some sensors that have a graph and not a table.  I prefer the table since it's easier read and type in the cal factor percentage for the frequency you're measuring.  The cal has expired, but it's very recent which is also nice.

It's shown connected to the cal output of the meter, which on HP meters is 0dBm (@ 50MHz), so the display is correct.

I have an 8481A and it's probably the sensor I use the most.  It can directly measure -30dBm to +20dBm, 10MHz to 18GHz.  If that's your desired power range I think this would be a good set.  Another good sensor to have is the 8481D which measures -70dBm to -20dBm.  The 8481D needs a 11708A calibration attenuator to connect to the meter cal output (added expense).

I've never found a need for the analog meter, but I can imagine some people like it for visual feedback while tweaking or maybe finding an intermittent.  I figured I could always connect an analog meter or a fast moving Fluke with a bar graph to the recorder output on the back if I needed something like that.

[G0HZU]

Also, what's your preference on the digital readout vs. an analog meter, vs. the digital readout plus simple analog meter layout on a meter like the 437B?

I don't think I've ever used a 437 meter but I would expect that the little analogue meter is there to tell you where the meter is (auto?) operating in terms of its amplitude ranges. A bit like a DVM operates in a number of voltage or current ranges.

The measurement uncertainty is different across each range for a typical power meter/head combo and the little analogue meter window probably tells you when you are nearing the lower edge of a range and the instrument uncertainty will be degrading.

If this happens then an advanced user might switch to manual ranging to try and keep the little analogue meter needle nearer to the upper end of its range.

[Electro Fan]

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

It's a fairly common type where the lower frequency limit is 10MHz and it probably gives best performance (in terms of minimising uncertainty) above about 25MHz.

You can also get the HP8482A power head that goes from 100kHz to about 4GHz but these may be harder to find at a cheap price. They are both very easy to damage though.

The reason that the 8481A and the 8482A are easy to damage is that they are thermocouple sensors instead of say the thermistor sensor used in the 478A?

[G0HZU]

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

It's a fairly common type where the lower frequency limit is 10MHz and it probably gives best performance (in terms of minimising uncertainty) above about 25MHz.

You can also get the HP8482A power head that goes from 100kHz to about 4GHz but these may be harder to find at a cheap price. They are both very easy to damage though.

The reason that the 8481A and the 8482A are easy to damage is that they are thermocouple sensors instead of say the thermistor sensor used in the 478A?

The thermistor sensor in the 478A is even more fragile     You are supposed to keep it below 30mW to avoid damage. I think the modern thermocouple sensors can handle 400mW briefly but you aren't supposed to go above 100mW. It's best to keep the input level lower than this for best accuracy and I rarely put more than 10mW into a thermocouple power head.

[G0HZU]

I think I've posted these plots up before but here's the datasheet/brochure spec for one of my sig gens in terms of accuracy at 0dBm up to 4GHz. It overlays the results for several different examples of this sig gen to show the spread. My particular sig gen had been calibrated at Trescal just before I bought it so it should offer similar performance hopefully

Then here's the result of me measuring it using my Anritsu (modern digital display and thermocouple sensor) power meter and also my 50 year old HP431C power meter using a HP478A thermistor sensor.

You can see they both agree very well but once the frequency gets bove about 2GHz the measurment uncertainty of the 478A head is easy to see in the form of ripple in the 431C plot. This is because the source impedance of the Agilent sig gen isn't an accurate 50 ohm and the Anritsu power head has much lower VSWR than the HP478A sensor and so it shows much lower ripple.

But both power meters are doing very well here IMO!

I've had the HP431C/478A for about 20 years and it's never been calibrated in all this time. But you can see how accurate this old meter still is! The HP432A is a newer version of this meter and is smaller and easier to use and has auto zeroing.

[Electro Fan]

Here is a photo of a 8481A (works with a 437B and possibly other meters); it has the cal data on it; any thoughts on this sensor? 

It's a fairly common type where the lower frequency limit is 10MHz and it probably gives best performance (in terms of minimising uncertainty) above about 25MHz.

You can also get the HP8482A power head that goes from 100kHz to about 4GHz but these may be harder to find at a cheap price. They are both very easy to damage though.

The reason that the 8481A and the 8482A are easy to damage is that they are thermocouple sensors instead of say the thermistor sensor used in the 478A?

The thermistor sensor in the 478A is even more fragile     You are supposed to keep it below 30mW to avoid damage. I think the modern thermocouple sensors can handle 400mW briefly but you aren't supposed to go above 100mW. It's best to keep the input level lower than this for best accuracy and I rarely put more than 10mW into a thermocouple power head.

Oh, other than exactly backwards, I had it right 
Thanks

[MarkL]

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.