3 connectés     # # #

Advantest 3463 & 3465 tricks

english only

Advantest 3463

release: sept 1, 2005

Return to my ham page

Sold in May 2005 for purchase of an HP 8595E
This page is closed now


This note is only devoted to spectrum analyzers Advantest 3463 and 3465.

Next lines have absolutely no interest for you if you have not one of these devices in your laboratory.

Read complete specifications in this datasheet pdf document

Rear picture of my unit (click on the image) :

Date of construction is about 1998

The only purpose is to discuss about the dark points of the incomplete instruction manual and clarify how to use the complex menus.

I have a page with many screens and measurements situations here :
Analyses de spectres

Sorry, my English is poor! Don’t hesitate to send me bad sentences corrected. Thanks in advance.

Advantest R3463

 Haut de page


Using PCMCIA card to copy screen images to a PC

Without a GPIB link, a PCMCIA card is the only solution to get print screens. The RLE files are converted to GIF format on the PC. The RLE files are read and converted to GIF on my Sony laptop.

I have a Centennial 2 MB SRAM and a licensed version of Apsoft Cardware, v7 XP version, which works fine, but with a boring selection at dual boot.


Test period expired problem

The first one is very stupid. I first tried Apsoft Cardware in evaluation version only. I wait too much time, at the end of test period, of course it stops working. I got a new one with license, but it vas impossible to install it.
Message was, “an other version is installed”, with only one option “uninstall”, error and exit. No other way than a PC disk format and new Windows XP install. It’s a trap, but not a real problem.


Data lost problem

° If I write the card and put it quickly in the PC, all seems good.
° If I wait some time after writing (card not powered), all is lost…

Advantest R3463

I’ve understood the problem after reading the fatal sentence on the instruction manual, page 6-10: "The SRAM card contains a battery ", and on the datasheet from Centennial: "Up to 10 years rechargeable lithium battery ".
It is not too difficult to open the card it with a blade (see picture). Battery MnO2-Li is very thin and not easy to find. Soldering two wires solves the problem with any 3 V external battery. It is not very beautiful but it works...

Standby current is 4 µA. A CR2030 has a capacity of 220 mAh, it can maintain memory 55 k hours, more than 2000 days or 6 years. It is not necessary to remove it, if you don’t use the card during weeks.
Don’t spend $0.3 in a new battery, get it with the holder on an old PC board in the garbage tank.

Inside SRAM
Original battery

Ready to close !



It is not possible to format the SRAM card on the PC, in FAT32 or NTFS, the analyzer don’t accept it. Format is exotic on Advantest and it’s only valid if made by the analyzer.
The function “copy” writes an image of the screen in the SRAM card. You can read it as a simple bmp without problem on a PC.


These cards are obsolete and expensive, more than $50 for a 2 Mb with a dead battery! An HPIB link can be a better solution, but I have no experience. I’m waiting an adapter form GPIB to Ethernet or USB.
The old NI RS232 converter can’t be used, only drivers for DOS exists and it is absolutely incompatible with Windows.+

The perfect final solution

I tried some PCMCIA adapters for Compact Flash, flashdisk, and other cards that don’t works on analyser.
Frank Ingle sends me the very good information: Only one other memory card works well, without any special boot for cardware system!
The perfect solution is a card reader multi format, and a small MMC. It works perfectly, and the PC sees the adapter as an external drive. A 16 in 1 adapter plus 32 Mb card is under $50 on eBay.
Take the smallest one available, only first 2 Mb is usable on the Advantest.


bad and good


 Haut de page


Some questions

User manuals are not very explicit, so I will began to try to clarify some dark points.


Removing input capacitor

An other problem of the same type. There is a capacitor between attenuator and first mixer. This is a problem only on very low frequency signals for instance with video signal, under 10 MHz, measurement is made with a big value (about 470 microfarad) on the input, and it is necessary to by pass the small internal one.

How can it be do? It is possible on other HP models, but I don’t think it can be do on the Advantest range.


 Haut de page


Some answers

Setting relative marker

Exemple: test or attenuator or amplifier gain.
Place signal on screen (exemple 30 MHz, 10 dBm reference).
Marker > 1/3 Delta Marker > More 1/3 (to page 2/3)
> Fixed marker ON > 1/Delta Marker ON > Display level RELative
Value is now 0 dB : Place DUT, read new attenuation (gain ) in dB.


Removing 10 dB input attenuator

By security, a 10 dB attenuator is inserted between the mixer head and the input plug. For low level signals it is necessary (with prudence) to remove it. Of course, it is only for very low levels, if you are absolutely sure that there is no risk for input circuit, an error will be fatal...

The way to remove this attenuator is a bit strange (Page 7-13 in not clear) … (this note is from Miguel Martin)

1 Press ATT button
2 Set ATT Auto to MNL
3 Press Min ATT (On display)
4 Entry 0 dB (Keyboard)
5 Press ATT button again
6 Entry 0 dB (Keyboard)

That's all! To return to original state, only is needed to press ATT, MIN ATT, Set defaults or can be made with shift + preset (Panic button)


 Haut de page


Watching very low signals, under -90 dBm

This section was contributed by Frank Ingle, KG4CQK

In order to measure very weak signals it is necessary to eliminate as much noise as possible. We do this by:

Warming up and calibrating the equipment
starting with quality cables
ensuring they are connected tightly
reducing the SPAN and hence the RBW and VBW of the analyzer to limit the amount of noise power included in our measurement
disabling the internal attenuator
setting the signal under study at the reference level to take advantage of the internal amplifier's lowest noise gain settings, and using the "AVG A " function to take advantage of the statistical properties of noise to cancel itself.

This experiment will determine the lowest signal level which may be measured by the Spectrum Analyzer. This figure is known as the sensitivity.



You will need a fairly stable Signal Generator capable of adjusting its output continuously between -70 dBm and -120 dBm.

You may wish to use external attenuators if your generator cannot go this low. The signal generator in a 2-way-radio service-monitor would be suitable for this experiment.


1) Allow the Signal Generator (S.G.) and the Spectrum Analyzer ( S.A.) to warm up for 1 hour. Connect the internal reference signal and [SHIFT] [7] / Calibrate All on the S.A.

2) Connect the output of the S.G. to the input of the S.A. using a high-quality, low-loss coaxial cables with "N " connectors on both ends.
Be sure all connectors are connected firmly.

3) Set [FREQ] on the S.A. to match the output of the S.G. (e.g. 100 MHz)

4) Set [SPAN] to 1 MHz.

5) Visualize the signal on the display. Set marker to track input signal: Marker [ON] / Normal Mkr / Peak / Cont Peak ON

6) Reduce the input signal to < -60 dBm. Press [-> RL].

7) Disable the internal ATT in the S.A. (See the chapter on this page)

9) Reduce [SPAN] to 50 kHz. Press [-> CF] as needed to keep the signal centered on the display. (Although an even lower Span (and RBW and VBW) would theoretically provide an more accurate measurement, the improvement seems to be marginal while the measurement becomes more difficult and slower.)

10) Reduce input signal to < -90 dBm. Press [-> RL].

11) Press Format / Trace / Avg A / Avg A ON. (This will cause most of the noise to cancel out, leaving the signal in question visible.) The default value of 20 cycles will work well for this measurement.

12) Reduce the input signal to less than 10 dB above the noise level (shown away from center of display.)

13) Reduce the input signal very slowly until the red marker jumps away from the center. Increase it very slowly until the marker jumps back to center. The marker reading is your input sensitivity for that input frequency.

My reading was -113.4 dBm. The Advantest R3463 specifications indicate an optimum sensitivity of about -114.85 dBm based on the following formula: ADNL = -{115-1.55 x (.1 Ghz)} dBm = -{115 - .155} dBm = -114.85 dBm

 Haut de page

HPIB output

I have not a PC adapter, not checked at this moment.
To be continued...


On my unit :

R3463 : Revision
Main : K03 Nov 11 18:10 1988
Disp : A01 Jan 19 17:43 1995
TBC : A20 (DOS:A0)
DSP : CDM E03 98.05.08 (AD12)


Opt 15: Prog. Loader Nov 05 1997
Opt 09: CDMASG Jun 05 1997
OPT61: CDMA Jun 30 1998


First installation in 2003

Janury 2004 : Generator is now a Wiltron 6647a, 10MHz - 18 GHz

 Haut de page



Sold in May 2005 for purchase of an HP 8595E. This page is closed now


 Haut de page



Helpful Applications Notes from Agilent:

"Agilent Spectrum Analysis: Amplitude and Frequency Modulation, " Application Note 150-1 (28 pages). 
I think you will find this very helpful for testing Ham Equipment:

"8 Hints for Making Better Measurements Using Analog RF Signal Generators "
Application Note 1306-1, (12 pages):

"Agilent Spectrum Analysis Basics "  This is almost a textbook on the design of spectrum analyzers and the meaning of their specifications.
Application Note 150 (119 pages):

For more information on weak signal measurement, refer to "Hint 3: Measuring Low-Level Signals, " Page 5 in "8 Hints for Making Better Spectrum Analyzer Measurements
Agilent Application Note 1286-1, 12-page document : dead link :

* Verified links 22/10/09

 Haut de page

Voile Electronique Photographie Camping-car Divers


Liens matériels de mesure et TV ham

Autres pages sur la mesure et le radioamateurisme :

© Christian Couderc 1999-2024     Toute reproduction interdite sans mon autorisation

* Page vue   13667   fois       IP :

 Haut de page         Dernière retouche le 26 Mars 2019 à 12 h         Retour page précédente