HP Agilent 8648 Models - Opinions?

[Electro Fan]

Just checking to see if anyone has any experience or thoughts you can share regarding the 8648 (A,B,C,D) signal generators?  Strengths, Weaknesses? Different display types? Reliability? Accuracy? Control Layout/Ease of Use? Options? Other Considerations?  Other comparable models you prefer? Thanks

[1design]

I own the 8648B, it's a good unit and very easy to use, the front panel layout is better than on some modern parts. The display is the LCD one, no issues with it, it has a contrast knob on the back so you can set it any way you want, there is only one right way anyway . If the unit has been taken care of no issues with precision neither on frequency or amplitude. The PN is a bit high, but you cannot expect anything special from an economy unit that is 20 years old. I do prefer the R&S SMA100A and Keysight E8663D but I guess that's a different league

BR

[KJDS]

I've used dozens of these over the years. Spec wise they're not brilliant, but they always seem to work as they should.

[G0HZU]

At work we bought a batch of these about 12years ago or so. We bought the HP8648D version.

I assume the aim was to buy lots of sig gens at a low price because the company was expanding at a scary rate back then and test gear was in demand across the whole company as we took on loads of staff.

The performance limitations of the HP8648D were soon apparent and these generators were eventually seen as a last resort if a sig gen was needed for wideband amplifier testing or critical receiver testing. This was mainly because of the high noise floor when operating at power levels around -10dBm to -20dBm. Although the spec sheet implies they have good noise performance the wideband noise floor performance is markedly worse than most signal generators at these drive levels.

Whilst this might not be an issue for some users it was enough to cause a significant problem for anyone doing wideband and low spurious testing at my place of work. eg if you had a wideband amplifier that took in a drive level of -15dBm typical then the HP8648D would add lots of wideband noise and this was enough to mask spurious signals from the amplifier on typical spectrum analyser settings for something like a 500MHz span. Make a change to something like a Marconi 2024 or even a little Marconi 2022 and the problem went away.

However, I did like the user interface of the HP8648D a lot. I liked the two rotary controls. One for frequency and one for amplitude. This is a rare feature and could be very useful at times. Also, the interface is dead easy to use and the display is bright and clear if a little basic.

The internal modulation oscillator is very basic and limited in settings although you can apply external modulation.

Getting back to noise issues, the frequency plan for this sig gen is one of low cost and this means that the phase noise is quite poor across most of the range of the generator. i.e. the phase noise profile at 10MHz carrier freq won't be much different to that when tuned to 1GHz. This is because of the basic mixer system it uses to synthesise frequencies across most of its range.

By comparison, many other generators use division to get lower frequencies and so the phase noise profile tends to improve by 6dB with every halving of the carrier frequency.

The next/newer model up from the HP8648D is probably the HP/Agilent E443x range. These will cost more (used) today and will be bigger/heavier but the RF performance will be a bit better and the modulation options are markedly better. The phase noise won't be significantly better though although the E443x range don't suffer the same issue in terms of wideband noise floor although they are still noisy compared to the next class up.

As an example, if a ham/hobbyist used a HP8648D to critically test a decent HF/VHF receiver for various rejection or blocking tests then the poor noise performance of the HP8648D would dominate the test results making them meaningless. The E443x range of generators aren't that much better in this respect. But both the HP8648 and E443x will be very good for accurate testing of sensitivity and signal meter calibration. The HP8648D would be OK for doing modulation testing and the E443x range would excel here.

If you aren't bothered by wideband noise issues then the HP8648D is a nice, basic sig gen and they appear quite reliable. So if you are not interested in doing critical receiver or low level spurious tests then the HP8648D is a decent generator that is nice to use and is fairly light and portable compared to the usual competitors.

[Rupunzell]

If the upper frequency limit of 512Mhz is acceptable, consider the hewlett packard 8640 series of cavity tuned/ECL divider signal generators. Low noise, frequency stable, can be phase locked, AM-FM internal or external modulator input.

There is an external mixer to extend the upper limit to just past 1Ghz.
Common problems are dirty switches, broken plastic switch gears due to abuse or fried output amplifiers due to abuse.


Bernice

[Electro Fan]

Thanks for sharing the experience and thoughts.  The 8648 seems like a solid choice and a good value even if it's not a super higher performer.

Any reason to consider the 8664/8665 series as an alternative to the 8648 series?   (I'm betting this might be like comparing a 8560E series to the 8566/8568 series in spectrum analyzers - possibly more performance in return for a little more age and possible maintenance - just a guess....)

Thanks again

[G0HZU]

Any reason to consider the 8664/8665 series as an alternative to the 8648 series? 

The HP8665 is an old but high end lab grade sig gen. It will offer fabulous RF performance in terms of spectral purity and is designed for use in serious research labs but it will be big and heavy and much more complex inside. The HP8648 is an economy sig gen and aimed more towards service work. It is smaller, lighter and easier to use and repair.

The HP8665 is the type of instrument where the HP engineers were let off the leash and allowed to make something really special in terms of spectral purity (and very expensive to buy new)

If you don't need the performance of an HP8665 (few people do) then why have such a heavy beast on your bench?

I'd love to have one here just like I'd love to have an HP8642B sig gen but they are just too big and heavy and I don't really need the performance they offer. So the disadvantages outweigh the fabulous performance advantages in my case. It would make great eye candy and look very impressive and would be very rewarding to use but it is just too big and heavy 

[G0HZU]

To show you just how special the old HP8665 was, the modern equivalent to the HP8665 is the Agilent PSG generator.

http://www.keysight.com/en/pd-460865-pn-E8257D/psg-analog-signal-generator?cc=US&lc=eng

At work we have several of the basic 20GHz PSG models and the performance is extremely good but they are insanely expensive to buy new. I think they were over $60k each.

Note that the old HP8665 is much bigger and heavier than most modern sig gens and they weigh about 35kg 

[UA3BL]

I wonder if this topic is still interesting but I've bought 8648C three month ago and now I face a problem. At first it worked OK but after a month the sine signal went wrong and the output does not work properly. The Keysight experts say that the output stage A10 (08648-60199) is out of order. They don't repair it and I had to do it myself. I think the component 1GG34202 went wrong. I'm not sure if it is a transistor although it looks like one  because I can't find its datasheet. I think it got overheated because there is no any grease between it and the chase which works as a radiator. The question is if anyone had such a case and who can help me with the CLIP.

[cncjerry]

I have an 8648C.  I like it.  Very accurate, more so than the display allows.  Not the best phase noise.  I made my own table recently listing all the available generators and found you get what you pay for.  I would search craigs list before ebay.