HP8564E-How to test for faulty attenuator/s

[djsbriscoe]

How can I test the HP8564E I've just bought for any measurement errors? I only have a Marconi 2022E that has an RF output level range of -127dBm to +10dBm (don't know what this is in volts peak to peak, which I'm used to working in). The Marconi has NOT been calibrated, so I can only measure it against my Agilent MSO-X3054A and my Signal Hound SAA44B USB based spectrum analyser.
What should I be looking out for if there is a problem with the attenuator/s on this HP8564E? I don't REALLY even understand how attenuators work in the context of a spectrum analyser like the HP. I'm going to work through the user manual and quick start guide as soon as I get it. There are a few simple examples of how to use it in there.Also, has Anyone got any good links to demystifying all this dBm stuff for me? Thank you.

[loulou31]

Hi
Do you want to test the input attenuators of your HP8564E or level accuracy of your analyser?
Normally during auto cal the input test attenuator are tested and also level accuracy throught the internal calibrator source ( 300MHz usually).
If you want to calibrate level of your SA over all frequency range, you need high frequency synthesizer, a calibrated power meter, and calibrated attenuator.
For my opinion a SA is not an accurate absolute measurement instrument for signal levels. I worked and work today with SA and +-2dB absolute level accuracy is quite enought.
To understand how attenuator works in an SA, please read user manual and diagram of the instrument.

Jean-Louis

[Radio Wrangler]

Your spectrum analyser is essentially a superhet receiver.

The purpose of the step attenuator on the front end is to allow you to set the level of signals appearing at its first mixer so that you don't drive it into non-linearity (which can create spurious responses at all sorts of frequencies) while keeping the signal you are interested in as far out of the noise as you can, subject to the first limitation.

Consider your analyser in its base frequency range, zero to two and a bit GHz. There is only the attenuator and a 3GHz lowpass filter between the input socket and the mixer.

You do not want to run these mixers with an input of more than -20dBm AT THE MIXER for reasonable linearity. Occasionally you might find it OK to go up 10dB, but you need to understand what you're doing so you are not led up the garden path by spurs.

So, think of what level you are inputting. Now how many dB do you need to insert to bring it down to below -20dBm? That's your ideal attenuator value. But it only does 10dB steps so better a step more than the step too little.

DANGER!

With significant power input, you can damage your mixer (Ooooh $$$$$$$!) with more power you can burn out attenuators (also very expensive)

The attenuator is rated at 1W absolute maximum.

The mixer has no official rating but 100mW is a bad risk.

If testing transmitters, you need a serious external power attenuator.

Concealed danger:

The attenuator setting is latched and holds even with power off. With power off, nothing is on the screen to tell you where the attenuator setting is, and where it will be when you switch the analyser on. Apply NO signals with the analyser off just in case they get too big.

Even with the analyser on, there can be big signals outside your scanning range, so be aware of everything entering. Choose attenuation to suit the whole lot.

David

[loulou31]

Hi

Thank you RadioWrangler for the explanation.

What it is also important to understand in a spectrum analyser is that the RF step attenuator and the first mixer are directly connected to input connector. That means that if some high level power signal or DC supply can damage / distroy the RF attenuator and/or the mixer.
New SA are usullay AC coupled, but old and high frequency SA was very often DC coupled and very sensitive to DC current : e.g : 5V over 50 ohms is .5W that can damage mixer, 10V over 50 Ohms is 2W that can daamge the attenuator.....


Jean-Louis

[djsbriscoe]

So I should be getting on the HP exactly the same as I'm getting on my Signal hound. Just did an experiment where I set my 2022E to -10.5dBm at 10Mhz and fed this through a Minicircuits -20db attenuator and a DC blocker into the SA44B. The SA44B was reading around -29.98dBm. My scope was showing just over 200mV Pk-Pk when fed directly from the Sig Gen (into a 1M scope input impedance). Seems about right according to a conversion table I was looking at. Is there a quick rule of thumb to convert dBm to voltage?

[Radio Wrangler]

Watch out for the difference between EMF and PD. 1 meg won't load your sig gen much, so you'll get EMF.

Spectrum analysers are (almost all) 50 Ohm beasties and when you switch to linear detection, read RMS PD.

Marconi RF sig gens changed a bit over the yrars. Some have output scales in EMF, some in PD.

HP's mid-market 899x family of analysers have DC blocks and go down to 9kHz in most cases. The up-market 856x family have DC connected inputs and go down to 200Hz in several cases. The HP8566 was the true ancestor of yours. It's a two box rack sized job. THe later portabgle ones were designed to fit in a chosen portable size and to lose as little as possible of the 8566 performance and features. It was quite a remarkable achievement and I knew the people who did it. Oh, if you come across the HP 8568 analyser (1500MHz) while it shares the top box with the 8566, the RF section is a very different architecture.

David

[Cruisin Marine]

I would always leave your Spectrum Analyser with as much attenuation when you switch it off, they remember where you left them, start with more attenuation than you need and work up to avoid EXPENSIVE damage!