Author Topic: Used Spectrum Analyzers for under $1k  (Read 71763 times)

0 Members and 1 Guest are viewing this topic.

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #125 on: March 19, 2015, 10:08:15 pm »
Probably the first thing you need to buy for your HP8566 is a decent N Type DC block. There are cheapish ones available that cover 10MHz to 18GHz or 100kHz to 3GHz.

I can't remember what make my N Type DC block is but it covers 100kHz to 18GHz with very low VSWR and through loss so there are decent ones out there that cover nearly the full range of the HP8566B.

Without this DC block you can easily fry the front end of the 8566 if you accidentally feed a DC voltage to it. It is only rated to 0V DC and can be (expensively) damaged by just one accidental feed of a DC voltage of just a few volts.  Especially if you select 0dB attenuation inside the analyser as there is a diode limiter in the 2.5GHz range that can be fried with a very low DC voltage.

At my place of work, more people damaged their analysers with DC than with RF overload so be very careful with that HP8566 !

I'd recommend you make or buy an RF limiter for it as well. Especially if you aren't very RF savvy yet :)
« Last Edit: March 19, 2015, 10:11:41 pm by G0HZU »
 

Offline TSL

  • Regular Contributor
  • *
  • Posts: 243
  • Country: au
Re: Used Spectrum Analyzers for under $1k
« Reply #126 on: March 19, 2015, 10:10:29 pm »
@Sailor

Glad you had a good experience with the guy, same as mine. $400 FEDEX is a bargain - I bought mine a few years ago and shipping was north of 700 then.

I hope you like the beast, and yes it has a loud fan as does most HP kit from that era. I have a HP54720A DSO and an ancient HP 8660C sign gen that compete with the 8566B for loudest fan. I think the 8660C wins!

I've also bought the 11970K external mixer for my 8566B to extend its operation to 26Ghz since I have some work to do at 24Ghz. The external mixers can cost a bomb but if you're prepared to wait a while you can pick them up for a good price.

Get yourself a PCI GPIB card or for USB a Prologix USB to GPIB converter and then go get the GPIB Toolkit from here...
http://www.ke5fx.com/gpib/readme.htm

This will allow you to capture your traces direct to your PC and there's other cool things such as, this...

"IMD_8566B_8568B.BAT uses TALK.EXE to transmit the downloadable program (DLP) T_HIRDIMOD to the nonvolatile user RAM on an HP 8566B or 8568B spectrum analyzer. This program performs basic third-order intermodulation distortion measurements on signals between 10 MHz and 500 MHz, and also calculates IP3. See HP product note 8566B/8568B-1 for more information. "

There is a couple of other add on programs to download to it too.

regards

Tim

VK2XAX :: QF56if :: BMARC :: WIA :: AMSATVK
 

Offline TSL

  • Regular Contributor
  • *
  • Posts: 243
  • Country: au
Re: Used Spectrum Analyzers for under $1k
« Reply #127 on: March 19, 2015, 10:23:14 pm »
Probably the first thing you need to buy for your HP8566 is a decent N Type DC block. There are cheapish ones available that cover 10MHz to 18GHz or 100kHz to 3GHz.


[snip]
I'd recommend you make or buy an RF limiter for it as well. Especially if you aren't very RF savvy yet :)
What he said !! :-+

I have a DC block plugged in all the time, you only need one faulty circuit to put DC on the input and poof!! there goes your front end. :scared:

I also have a selection of  attenuators from 3 to 40dB that go inline if I suspect something is going to output a sig higher than expected.  I also have a a handful of directional bridges with various dB coupling so the SA can be outside the signal path where needs be. Mostly when testing amplifiers above 100watts.

Don't be tempted to hook up an external reference to it either, the internal ref has exceptionally low noise figure and it would take an extremely good external reference to beat it.

regards

Tim
VK2XAX :: QF56if :: BMARC :: WIA :: AMSATVK
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #128 on: March 19, 2015, 11:34:51 pm »
The HP8566 is also a bit prone to developing minor niggly issues with the YIG/Control and at some point it may start throwing up onscreen BITE errors like YTO Unlock and also one or both of the little red STATUS LEDs may wink on or off intermittently after the initial bootup sequence. These LEDs usually indicate that a self test error has occured.

Usually this isn't serious and can be traced to a few dried out electrolytic capacitors in the RF unit. The HP8566B is quite easy to work on so the YTO Unlock fault could potentially be fixed in a couple of hours assuming you check lots of caps with an 'in circuit' ESR meter.

It's also worth checking it can pass its own internal error compensation program. This is a short calibration/correction routine that attempts to improve the factory calibration by measuring and compensating for ageing effects in the analyser RBW filters etc. You loop back the 100MHz cal signal to the RF input and then start the automated routine by pressing [SHIFT]  [FREQUENCY SPAN]. It takes about 1 minute to run.

Once it has completed the CRT will show a little 'Corrd' logo onscreen.

However, it will disappear if you hit PRESET or reboot the analyser.
You can recall this calibration (without rerunning it) after every reboot by pressing [SHIFT] [START FREQ].

or you can set it back to factory default with [SHIFT] [STOP FREQ]

It is definitely worth running this cal routine after the analyser is fully warmed up as it will improve the calibration across all the RBW settings compared to the factory default setting.
« Last Edit: March 19, 2015, 11:49:17 pm by G0HZU »
 

Offline Sailor

  • Regular Contributor
  • *
  • Posts: 170
Re: Used Spectrum Analyzers for under $1k
« Reply #129 on: March 20, 2015, 02:01:36 am »
@ everyone

Thanks for all the notes, and please, if any other odd bits of advice come to mind later, you could always just PM me. Good stuff like this comes from (hard) experience, not out of a manual.

I've been looking at the ke5fx site for three or four days now, lots of goodies there. I was going to ask about GPIB interfaces - I have several machines with ISA, PCI, or, of course, USB that I could use. I've got w2k, XP, and W7, so the question is - is it easier/harder/better with any of those systems? I vaguely remember reading of people having problems with some USB implementations, but I didn't take much notice at the time. Is there anything special about making a GPIB cable? Length, shielding, etc?

A DC block, and some form of limiting have been weighing on my mind for a while. I've always been paranoid about double-checking before turning on anything with significant powers involved, but this 0V DC and mWatts-max is going to be a whole new ballgame for me. Tim has a block installed permanently - I'll follow suit, but with some added measure to deter me from removing it inadvertently, like when my brain is five minutes ahead of my hands :P
BTW, whats with some of those Chinese devices - DC blocking (200V), but freq resp DC - xxGHz :-DD

Does the 8566 power-up with a default attenuator setting, or with the last-used setting?

... I also have a a handful of directional bridges with various dB coupling so the SA can be outside the signal path where needs be.
That sounds like a good idea.

That's good info about bad caps GOHZU. Last night I fed in 8566 to the search on the yahoo Agilent group, and was overwhelmed by the number of messages. I obviously couldn't even read a fraction of them, but I did notice that similar comments came up there also. I just hope I don't need to dive into this machine for a year or two. Much as I love exploring anything new, I know that I'm at the bottom of a steep learning curve right now, and I'd rather not run the risk of stuffing up something. Analogue, digital, power, I have no problem with, but I just know that I'm going to get caught out here more than once or twice...

Regarding the fan noise, has anyone looked at using different fans? I can remember some fan noise problems in the late '70s - '80s being mitigated somewhat with redesigned blades. I can understand the display generating a lot of heat, but the RF section?

Thanks again to everyone :-+

Sailor
 

Offline Mr Simpleton

  • Supporter
  • ****
  • Posts: 302
  • Country: se
  • Not the sharpest knife in the drawer
Re: Used Spectrum Analyzers for under $1k
« Reply #130 on: March 20, 2015, 08:41:24 am »
If I recall correctly the 8566 do have 10 dB attenuator switched in at start up and reference level is at 0 dBm (top of screen).

Assume it behaves like the 8568 which you cannot step the atteunuator to 0 dB, you have to punch in the digits. I see this as a safety measures.. :D

I am using a USB-HP/IB from National Instruments works like a charm with KE5FX programs. Have tested this with both 8560E and 8568B.
 

Offline l0wside

  • Regular Contributor
  • *
  • Posts: 118
  • Country: de
Re: Used Spectrum Analyzers for under $1k
« Reply #131 on: March 20, 2015, 10:23:46 am »
I am currently looking for an SA, mainly for EMC precompliance tests. The Rigol DS815(-TG) would most likely just be fine, but even if 1200 EUR is a bargain, it is a lot of money for a device which I will use only a couple of times a year.

I have been eying eBay for a while (and reading this thread, of course). Over here in Germany, the supply of used test gear is a lot worse than in the US, but considering shipping and tax, buying in the US does not make much sense.
There is currently an Advantest TR4131 on sale. According to my web research this is a late 80s device, 9kHz...3GHz, rather bulky and heavy, with some issues regarding frequency stability. For EMC, I feel this should not be too big an issue (but I am not an expert, EMC is just something I canĀ“t avoid). The unit in question has had its last calibration around 2002.

Does anyone know the device and can give me an idea how much this would be worth? The seller claims that the unit is in good state, I would have to rely on the self-test. Or would the unit make little sense for EMC testing?

Max
« Last Edit: March 20, 2015, 10:30:02 am by l0wside »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #132 on: March 21, 2015, 06:23:23 pm »
Quote
Regarding the fan noise, has anyone looked at using different fans? I can remember some fan noise problems in the late '70s - '80s being mitigated somewhat with redesigned blades. I can understand the display generating a lot of heat, but the RF section?
It may be possible to change the fans to an intelligent system where they only operate at speeds required for a certain temperature inside. Eg change the fan type and also use a cheap AFVR/PIC MCU and temp sensor and fan controller circuit. It might mean that the overall measurement uncertainty of the instrument degrades slightly (over time) but I think it would be worth it for many users.

On  the HP8566 the fans come on full blast from cold and make an awful lot of noise. These analysers are tolerable in a large open plan work lab because there are other sources of noise and the noise will travel across the lab. But in a typical hobby room/shack/shed the fan noise will dominate the room and will be difficult to live with unless the operator is insensitive to loud fan noise or there are other fan noises that are already as bad (unlikely!)

Quote
I can understand the display generating a lot of heat, but the RF section?
ermmm... have you seen the size of the fan on the 8566 RF section? It's a monster. Mine has a huge cover/filter on it and I think this is typical. I've not tried to see where the bulk of the noise comes from but my 8566 is definitely a bit louder than the 8568 so I assume the huge fan on the RF section on the 8566 is the reason.

I took a quick picture of the back of my 8566. See below. You can see the size of the RF fan (arrowed). Whilst it would be an exaggeration to say it sounds like an F16 fighter jet when it starts up, it does make a lot of fan noise especially if you operate it in a small room near a backwall.
« Last Edit: March 21, 2015, 06:45:14 pm by G0HZU »
 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2442
  • Country: gb
    • my website holding page
Re: Used Spectrum Analyzers for under $1k
« Reply #133 on: March 21, 2015, 07:19:16 pm »
So why does a high performance spec ana create so much heat?

One of the keys is dynamic range, so a cheap spec ana will be running level 13 mixers, so +13dBm as the LO drive level. The higher the level of mixer the higher the third order intercept point and the lower the spurious products. A lab grade analyzer will be running level 23 or 27 mixers, and the amplifiers driving those are dissipating at least a watt. The amplifiers in each IF stage will also need to be running at that level, so you end up with everything running far more power than on an economy unit. The 1st LO amp has to cover a wide frequency range, approaching an octave and you can't run it at the most optimum efficiency match so you've usually got several watts of heat being dissipated just from that one amplifier.

Offline Mr Simpleton

  • Supporter
  • ****
  • Posts: 302
  • Country: se
  • Not the sharpest knife in the drawer
Re: Used Spectrum Analyzers for under $1k
« Reply #134 on: March 21, 2015, 08:12:25 pm »
My take is that the logic is the main culprit of heat! This instrument was designed in late 70's early 80's and is probably chock-full with ECL and other power hungry semiconductors... Plus owens for ref oscillator and YIG...  A few mW amps/mixers would hardy be noticed  :-/O
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #135 on: March 21, 2015, 09:46:22 pm »
I think a lot has to do with the sheer complexity of the thing.

The swept LO system is a very complicated system with lots of active devices in it. Presumably all of this has to be carried out at a highish RF power level at each stage to keep the performance high and the noise levels low.

Then there's the signal path including the YIG filter and there's also the A15 controller unit. It all adds up. However, I just tried measuring both the Display and the RF unit for power consumption and it was lower than I expected!


My 8566 is an AB version that started out as an A but was converted to B with a factory upgrade. Also note that I only have a cheap and cheerful inline meter for this so it isn't going to be very meaningful/accurate...

I measured the RF unit from cold and also when the OCXO was warmed up:

HP8566AB

RF Unit:
Standby Cold:  24W (OCXO cold)
Standby Warm:  12W (OCXO warm)

Initial Power Up:   195W High band
Initial Power Up:   190W Low band

Warmed Up Power:   190W High band   (measures 244VA in VA mode)
Warmed Up Power:   186W Low band   (Measures 240VA in VA mode)

Display Unit:
Standby:      4W
Running:      138W  (Measures 168VA in VA mode)

Combined RF + Display Unit:

Standby: Warm = 16W
Running: = 328W   (Measures 404VA in VA mode)

I also experimented by placing my hand over each fan to baffle it and (surprisingly) the display unit fan noise is more annoying than the fan noise from the RF unit. But there's not much in it.
« Last Edit: March 21, 2015, 10:33:28 pm by G0HZU »
 

Offline TSL

  • Regular Contributor
  • *
  • Posts: 243
  • Country: au
Re: Used Spectrum Analyzers for under $1k
« Reply #136 on: March 22, 2015, 02:19:05 am »

I've been looking at the ke5fx site for three or four days now, lots of goodies there. I was going to ask about GPIB interfaces - I have several machines with ISA, PCI, or, of course, USB that I could use. I've got w2k, XP, and W7, so the question is - is it easier/harder/better with any of those systems? I vaguely remember reading of people having problems with some USB implementations, but I didn't take much notice at the time. Is there anything special about making a GPIB cable? Length, shielding, etc?

I use a National instruments PCI GPIB card in my Win7 PC. That works just fine with no issues. I also have an Agilent USB-GPIB adapter 82357B, plugged into my HP53181A frequency counter. I specifically bought the USB adapter for use with a Microsoft Excel plugin from Agilent that takes the counter output and dumps it straight into Excel. Great for measuring freq drift over time. The plugin would not work with my NI card :(

You can get clones of the Agilent USB-GPIB adaters,

http://www.ebay.com/itm/82357B-GPIB-USB-Interface-Compatible-with-AGILENT-82357B-/25164199097

I bought that one and it works just fine. If you want the genuine article keep an eye on eBay as they pop up for reasonable prices often.(not often enough when I wanted one)


BTW, whats with some of those Chinese devices - DC blocking (200V), but freq resp DC - xxGHz :-DD

Not sure what you mean here, but having a high voltage breakdown is good for working on RF stuff especially if you work in the lower frequencies. i.e. I'm working on a transmitter for MF about 475kHz. The antenna matching coil is around 1mH, and even thought the transmitter is powered by 12v and outputs only 10w, the resultant RF voltage on the top of the coil is around 1800v! :scared:

Needless to say I'm using a directional coupler to sample the RF :)

Likewise if you ever have to work on valve based RF amps, working voltages are often 200v to 4kv DC. Plenty of places to come unstuck if your not paying attention.

regards

Tim
VK2XAX :: QF56if :: BMARC :: WIA :: AMSATVK
 

Offline Rupunzell

  • Frequent Contributor
  • **
  • Posts: 349
Re: Used Spectrum Analyzers for under $1k
« Reply #137 on: March 22, 2015, 07:12:49 am »
Hewlett Packard Journal article about the 8566 from 1979.
http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1979-08.pdf

Keysight UXA is spec'ed with a 25W standby and 850Watt Max. Net weight spec'ed at 68 pounds.
http://literature.cdn.keysight.com/litweb/pdf/5992-0090EN.pdf


Bernice
 

Offline Andy2

  • Contributor
  • Posts: 35
  • Country: england
Re: Used Spectrum Analyzers for under $1k
« Reply #138 on: March 22, 2015, 12:28:47 pm »
I think Sailor is asking how a DC block can have a response from DC - (name any frequency). :-//
Andy.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #139 on: March 22, 2015, 06:47:22 pm »
If anyone is interested in the power consumption of the HP8568B and the Advantest TR4172 I measured them both today here in the UK (230V AC typical).


HP8568B

RF Unit:
Standby Cold:  24W (OCXO cold)
Standby Warm:  12W (OCXO warm)

Warmed Up Power:   104W   (measures 131VA in VA mode)

Display Unit:
Standby:      4W
Running:      136W  (Measures 164VA in VA mode)

HP8568B Combined RF + Display Unit:
Running: = 244W   (Measures 300VA in VA mode)


I also measured the complete TR4172
TR4172 Combined RF + Display Unit:
Standby (warm) 6W
Running 247W  (304VA in VA mode)

Note: It briefly takes about 270W at startup but falls to 247W in a few seconds.

According to the Advantest service manual it takes 'approx' 300VA and my meter appears to agree with this very well.


I'm not sure why my HP8566 and 8568 take much less than their HP datasheets says but maybe HP quote the worst case peak under the worst case line voltage and ambient temperature?

I looked up the specs for my cheapo inline meter and it claims to be within 1% accuracy for Watts and VA which seems a bit hard to believe...

However, the HP8568B and the TR4172 have the same measured W and VA according to my meter. Nothing in it!
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Used Spectrum Analyzers for under $1k
« Reply #140 on: March 22, 2015, 06:53:59 pm »
So according to my cheapo meter, the HP8566 RF unit is 190W vs 104W for the HP8568B RF unit. Which probably explains the big power bulge box in the back of the HP8566 RF unit that houses its huge cooling fan  :)

"We're going to need a bigger boat fan...."
 

Online Electro FanTopic starter

  • Super Contributor
  • ***
  • Posts: 3160
Re: Used Spectrum Analyzers for under $1k
« Reply #141 on: March 25, 2015, 11:44:56 pm »
Anyone ever use one of these?
http://www.tequipment.net/assets/1/26/Documents/Protek/7700.pdf
Any chance one of these could be worth $300?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf