UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Test Gear and Workshop Equipment

Notices

Vintage Test Gear and Workshop Equipment For discussions about vintage test gear and workshop equipment such as coil winders.

Closed Thread
 
Thread Tools
Old 12th Sep 2020, 3:26 pm   #141
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

I think the copyright on those manuals is actually owned by HP, passed to Agilent, passed to and currently owned by Keysight.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 12th Sep 2020, 9:11 pm   #142
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

Agreed but publishing links to the Artek files might cause some issues that are probably best avoided.

I therefore had a deeper search of the Keysight website and managed to find both parts of the manual although access to these may be limited to registered users.

I've uploaded these Keysight versions for the RF section to here:

https://www.qsl.net/g/g0hzu//RF%20Tr...90137_Vol1.pdf

https://www.qsl.net/g/g0hzu//RF%20Tr...80-90137V2.pdf

These should be fine to copy and distribute (I hope!)
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 12th Sep 2020, 9:30 pm   #143
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

Here's the service manual for the display section

https://www.qsl.net/g/g0hzu//RF%20Tr...5662-90085.pdf

I managed to find a list of about 1700 old HP manuals that have been archived on the Keysight website but I think you have to have a registered account to access them. To register with Keysight is easy, I've done it at work and here at home and the only minor baggage from registering is that I get the occasional marketing email from them.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 12th Sep 2020, 9:36 pm   #144
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: Replacement for HP8640B signal generator

Great, thanks. It looks like the calibrator is just an LM3146 4-transistor array plus a few discretes. At least it's available from UK specialist suppliers if a tad expensive.

Alan
Alan_G3XAQ is offline  
Old 12th Sep 2020, 10:01 pm   #145
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

If it helps, I've left my HP8568B disconnected from the AC mains for several days and I just powered it on. I'll have a look to see how much the calibrator output changes during warmup. The 20MHz calibrator hasn't been adjusted in at least 15 years because I think I've had this analyser for about 15 years. Prior to this I suspect it would have remained in spec for most if not all of its life.

My thermocouple power meter has been on for several hours and it is quite accurate so the results below should be quite representative of the calibrator performance over time.

From a cold start the calibrator showed -10.03dBm.
After about 4 minutes the level is at -10.04dBm.
After about 10 minutes the level is at -10.05dBm.
After about 20 minutes the level is at -10.05dBm.

The calibrator on my Advantest TR4172 analyser is also very accurate and consistent over time. I suspect it is slightly more stable than the HP calibrator but both are more than adequate for their intended use.

Note that the frequency response of the HP8568B analyser across LF through 1500MHz isn't spectacularly flat. It will probably show about +/-0.5dB ripple across its range. Back in the day at work the best way to calibrate the RF log scale (for a given spot frequency) was to feed a Marconi 2019 or an R&S SMH sig gen to a power meter via a 10dB attenuator. It is decades since I last did this but I recall that the idea was to adjust the sig gen such that the level on the power meter was exactly -10dBm and because there is a 10dB attenuator at the end of the coax cable the source impedance should have low mismatch uncertainty. This could then be fed to the HP8568B RF input port to see how much error there is on the analyser display. It is then possible to use the little amplitude cal trimpot on the front panel of the HP8568B to correct the error. This was typically done when trying to measure the spectral density of noise or for checking the power of a signal source on the analyser.

This correction only applies at the frequency under test of course...
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 12th Sep 2020, 11:13 pm   #146
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

Quote:
Originally Posted by G0HZU_JMR View Post
Back in the day at work the best way to calibrate the RF log scale (for a given spot frequency) was to feed a Marconi 2019 or an R&S SMH sig gen to a power meter via a 10dB attenuator. It is decades since I last did this but I recall that the idea was to adjust the sig gen such that the level on the power meter was exactly -10dBm and because there is a 10dB attenuator at the end of the coax cable the source impedance should have low mismatch uncertainty. This could then be fed to the HP8568B RF input port to see how much error there is on the analyser display. It is then possible to use the little amplitude cal trimpot on the front panel of the HP8568B to correct the error. This was typically done when trying to measure the spectral density of noise or for checking the power of a signal source on the analyser.

This correction only applies at the frequency under test of course...
We used a Wandel und Goltermann EPM-1 one-milliwatt meter as our level and flatness standard and got other levels using attenuators which were calibrated against an NPL waveguide-beyond-cutoff precision attenuator we had on the premises (used laser interferometer to measure piston position, the piston carried a 30MHz oscillator)

In the microwave division, we shifted to a 432 power meter with a pair of 6 digit DVMs following it and a 478 thermistor mount.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 12th Sep 2020, 11:54 pm   #147
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

At work we also found that with both the TR4172 and the HP8568B analyser it was possible to see subtly different results for things like noise spectral density (in dBm/Hz using the 1Hz noise marker) if the analyser was used on different spans with a different RBW setting. The most critical work was often done at just 70MHz so we ordered a couple of precision 70MHz bandpass filters that were calibrated for noise bandwidth at (I think) NPL.

The idea was to use a precision noise source and the 70MHz filter and a power meter to produce a repeatable result for the noise level from the noise source. The analyser was then checked against this. I can't remember now but we had a favoured analyser span and RBW to get the most consistent results for noise measurements using the 1Hz noise marker. Of course, these days modern analysers have a digital IF and a lot of these limitations are a thing of the past. The overall measurement uncertainty is a lot lower now and the results are repeatable with different RBW settings.

Alan: One way to get the noise marker to show up on the HP8568B is to press SHIFT and then the NORMAL marker button. This analyser does have quite a nice 1Hz noise marker feature as it uses sample data closely centred around the marker frequency to try and produce a smoothed reading. I'm not quite sure what methods it uses but the TR4172 noise marker is also very good in this respect. This makes it really good for measuring close to carrier phase noise for a typical VHF single loop synthesiser for example.

Note that I've now measured the HP8568B calibrator output after 1 hour and 2 hours and it hasn't changed at all. It is still at -10.05dBm. At each hour I checked both before and after an internal zero/cal check on the thermocouple power meter and the result was the same. This is a bit of a confidence check for the stability of the power meter over time.
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 13th Sep 2020 at 12:03 am.
G0HZU_JMR is offline  
Old 13th Sep 2020, 6:49 am   #148
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

There was an HP instrument which injected noise onto a clean signal so as to produce a controlled signal to noise ratio so that the bit error rate of digital microwave receivers could be explored across a range of S/N ratios. 3708A. It needed to have a group of filters of calibrated noise bandwidth (centred on 70MHz and 140MHz) as well as a fast thermal power sensor. I designed the filters and the work I did on early software for the bench lash-up version of it turned into an MSc project.

It takes time to measure noise of a given bandwidth to a given level of uncertainty. The 3708 could make noise of a level assessed long term by the thermal power meter inside it, and then modify it to track signal fluctuations with a fast-changing attenuator. The result being that it could produce rapidly changing levels of noise, so rapidly that you could mathematically prove that it was impossible to check that it was doing what it was supposed to be doing! An instrument to make your head hurt! We sold a lot of them and it kept the old Queensferry Telecom Division going in its latter years.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 1:37 pm   #149
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: Replacement for HP8640B signal generator

OK, I have bought and downloaded the Artek scans now. The 20MHz calibrator looks simple enough but I wonder why they specify things like a 681 ohm 1% resistor? Especially when it is in series with a 500 ohm pot!

I hadn't come across a noise marker function before. There's a brief app note on the Keysight site but at first glance it doesn't look as if it will be much use for the things that interest me. HF sources noisier than, say, -140 dBm/Hz at 1 or 2 KHz are too noisy for me. Did I misunderstand? Can it measure such sources?

As Jeremy sort of suggested, I used my 2019 signal generator but without any attempt at external calibration to set the AMPTD CAL adjuster to make the SA report -10dBm on the scale at 20MHz. I then looked for variations over a 0-70dB range and over a 1KHz to 1MHz RBW. The results look pretty consistent, with less than 0.5dB variation, except on 100KHz RBW where it seems to differ by 1.5dBm from all the other readings. Maybe I will look into that one day, or maybe I'll not bother.

If some knuckle dragger has transmitted into the CAL output then the little 1/8W resistors on the output attenuator may have smoked but my guess is the transistor array has died. I'm forbidden to work on this or anything else until the builders and decorators have been and gone but in the meantime I might order a new array chip.

And now I have to continue emptying my radio shack cum workshop into boxes to strip the room completely. I've been in here 30+ years so there's quite an accumulation of "valuable" belongings. I know for sure I'm going to lose of stuff but at least the HP gear is big enough and heavy enough not to be in danger of being mislaid.
Alan_G3XAQ is offline  
Old 13th Sep 2020, 4:04 pm   #150
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

Quote:
Originally Posted by Alan_G3XAQ View Post
but I wonder why they specify things like a 681 ohm 1% resistor? Especially when it is in series with a 500 ohm pot!
Simple!

The instrument divisions of HP decided to use 1% metal film resistors in 1/8 and 1/4 watt sizes as default choices unless somethng special was called for.

They are low noise and very stable (50 or 100 ppm/C) not a bad choice.

For 1% grade resistors, the natural series of values ie E96 meaning 96 logarithmically spaced values per decade. So take the 96th root of 10. Start at 1, for example and keep multiplying by that root result, then the value you see will step upwards. Round to three digit resolution, and you have the E96 series.

96 values per decade is a lot more than needed for everyday use. So we chose to concentrate usage on the 24 values closest to the familiar E24 ones.
That way we get better quantity discounts, buying fewer varieties in greater quantities each.

So your 681 Ohm resistor is the one we used where 680 ohms would have done.

Why not just buy a 680 Ohm 1% metal film resistor? Well, you can find them now at the usual distributors, but actually the E96 range is special for 1% resistors, it is calculated so that all the tolerance bands of each adjacent pair of values exactly abut. So if you made resistors randomly, you'd be able to flog them as one of the values. E96 1% leaves no gap, no overlap. So resistor manufacturers didn't used to make them rounded to E24 values very willingly and we got better price quotes... Also, if we ever had applications needing the intervening E96 values, we already had 24 values per decade set up with company part numbers in stock to go with them.

In special places we sometimes needed 0.1% jobs or better, in some places we needed custom values... precision attenuators for example. Yes we could have used multiple standard values to make what we needed, but often the stray L/Stray C spoiled the idea.

On the whole, it rationalised our stockkeeping, saved a bit of money and we often got to use good stable parts even in areas where it wasn't strictly necessary. We were paying 1 cent each for the 1/8 watt resistors.

Oh, 1/8 watt, but isn't that size 1/4 watt? Yes and no. Depends on the temperature range you specify it over. We called it 1/8 but in a less-derated lesser temperature range world it was 1/4 Watt.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 4:14 pm   #151
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

Quote:
Originally Posted by Alan_G3XAQ View Post
I hadn't come across a noise marker function before. There's a brief app note on the Keysight site but at first glance it doesn't look as if it will be much use for the things that interest me. HF sources noisier than, say, -140 dBm/Hz at 1 or 2 KHz are too noisy for me. Did I misunderstand? Can it measure such sources?
It's just the ordinary marker but selecting noise marker:

A) puts the detector board into sample detector mode rather than the peak detector This reduces an exaggeration effect on noise.

B) Applies calculated values for the noise bandwidth of whichever IF filter is in play. NB The IF filters are 5th order synchronously tuned, they are similar to Gaussian, but not exactly so.

C) And not a lot of people know this! It applies a correction factor for the detection process because it applied log conversion before averaging and this makes one-point-something dB difference, where the noise is Gaussian. It's one of those issues with not using true RMS detection when noise is involved.


One of the improvements I made in the noise figure analyser, compared to its revered predecessor the 8970, was to go to true RMS level detection. There were too many assumptions floating around that all noise creating mechanisms in devices under test would be Gaussian. This gave large errors in some cases, but smaller ones more often.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 4:16 pm   #152
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: Replacement for HP8640B signal generator

Sometimes you wish you'd never asked (about resistor values)...

Last edited by Alan_G3XAQ; 13th Sep 2020 at 4:18 pm. Reason: give context
Alan_G3XAQ is offline  
Old 13th Sep 2020, 6:39 pm   #153
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

Just trying to wean you off all these questions.......

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 7:28 pm   #154
Jon_G4MDC
Nonode
 
Jon_G4MDC's Avatar
 
Join Date: Nov 2015
Location: Welwyn Garden City, Hertfordshire, UK.
Posts: 2,013
Default Re: Replacement for HP8640B signal generator

They won't stop, since they like your answers.
NF calcs in my case, amongst many other things.

Which company made something like the 8970 but earlier. Was it Eaton?
Meh - off topic for signal generators - very sorry.
Jon_G4MDC is offline  
Old 13th Sep 2020, 7:55 pm   #155
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

Quote:
I hadn't come across a noise marker function before. There's a brief app note on the Keysight site but at first glance it doesn't look as if it will be much use for the things that interest me. HF sources noisier than, say, -140 dBm/Hz at 1 or 2 KHz are too noisy for me. Did I misunderstand? Can it measure such sources?
Not directly, as the HP8568B would typically have -112dBc/Hz phase noise at that offset. So it would be too noisy by about 30dB. It might be possible to do something to null the carrier with a narrow (500Hz?) crystal filter inline but this would be limited to the frequency of the crystal filter.

The other issue is that the HP8568B typically has a noise figure of about 24dB and this defaults to 34dB with the 10dB internal attenuator inline. So this means the analyser noise floor will typically be -174 + 34 = -140dBm/Hz with the internal 10dB attenuator selected and the noise marker enabled. By pressing [ATTEN] [0] [dB] it is possible to select 0dB attenuation but this isn't really recommended as it makes the analyser more vulnerable to damage and it also spoils the input VSWR of the analyser quite a bit. The noise floor would lower to about -150dBm/Hz. However, you may have already noticed that the user interface won't let you select 0dB attenuation using the up/down key. It will stop at 10dB and you have to deliberately ask it for 0dB with the button presses above.

The drive level to a crystal filter could be increased to overcome this but I'm not sure this would be a good thing for the filter.

The other limitation with these older type analysers is that to measure noise you really need to be measuring it about 10-14dB inside the genuine log range of the analyser. So don't try measuring noise with the noise marker near the bottom of the graticule unless you can accept some error or uncertainty.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 13th Sep 2020, 8:07 pm   #156
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

Quote:
Originally Posted by Jon_G4MDC View Post
Which company made something like the 8970 but earlier. Was it Eaton?
Meh - off topic for signal generators - very sorry.
Yes, but the names of the firms making it kept changing. It started as Eaton/Yale and ended up with Maury.

It was a good box, fairly closely equivalent to the HP8970, then Mury updated it somewhat and put in more processing power and I think it was a VGA interface so it could plot noise figure and gain from a sweep. This moved it ahead of the 8970 for a short while before the Agilent (It was HP's turn to change name!) NFA came out. Maury were showing their connectors, tuners and noise figure boxes at MTTS in Boston (2000? or 2001?) Agilent had the NFA in pride of place on their stand, and it wandered off with the prizes from the show. Shortly afterwards, Maury dropped their noise figure box from their catalogue. I was a bit saddened by this. THe NFA needed doing, the 8970 was full of life-bought bits which were running out of stock, and the new architecture seemed to me the sensible way forwards. Digitally defined NBW was a big improvement in stability, detector non-linearity was better than halved, and the measurement was now genuine RMS not diode.

Anyway, Maury make superb connectors and load pull tuners.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 8:13 pm   #157
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: Replacement for HP8640B signal generator

The spectrum analyser noise floor can be combated by adding suitable preamps. This is particularly effective with preamps tailored for limited frequency ranges. Phase noise can be combated by off-tuning the main analyser and using an offset IF filter. Some care is needed to not fall into the noisexnoise trap.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 13th Sep 2020, 8:32 pm   #158
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: Replacement for HP8640B signal generator

Sounds like in (my) real life I'm better off measuring receiver LO phase noise by tuning around a low noise signal (8568B, crystal oscillator, etc). Yes, you metrologists will savage me for not accurately knowing the noise bandwidth of my receiver and not characterising my signal sources in absolute terms, but on the other hand I'm more interested in practical questions like is the LO noise degrading the receiver noise floor enough to mask the IMD spurs.

I'll get my coat...
Alan_G3XAQ is offline  
Old 13th Sep 2020, 11:11 pm   #159
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

Have you tried running the internal calibration check/correction routine yet? This is supposed to require a connection to the 20MHz calibration port but you could try using the Mi 2019 set to deliver 20MHz and -10dBm at the main RF input as long as it is within about 1Hz of 20MHz. It should be if the internal OCXO is healthy. Otherwise clock it externally from the HP8568B 10MHz reference.

Then press (SHIFT] [FREQUENCY SPAN] to start the automated routine. It should check/correct for the attenuator and the RBW filters. It normally takes just over a minute to complete the routine and store the corrections.

Then the next time you run the analyser press [SHIFT] [START FREQ] to recall the corrections and the analyser should display CORR'D on the CRT display.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 17th Sep 2020, 9:29 pm   #160
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Replacement for HP8640B signal generator

If the HP8568B fails to complete this automated test or the corrected calibration appears worse than uncorrected then this may be due to intermittent connections in either the 0-50dB RF input attenuator or the input changeover relay right at the front end of the analyser.

The changeover relay swaps between the AC coupled N connector and the DC coupled BNC connector next to it. This clunky RF relay can be activated by the two little 'SIGNAL INPUT' buttons that sit between the two RF input connectors.

The input relay can become intermittent and this can cause the insertion loss of the changeover relay to be different each time the relay is swapped back and forth if it hasn't been used in a long time. I think the same thing can happen with the 0-50dB input attenuator.

Therefore, I would recommend you try exercising the input attenuator back and forth across its range a dozen times or so across 10-50dB and also try exercising the changeover relay a similar amount of times. In my experience you will sometimes see tiny (sub 1dB or even worse) changes due to the changeover relay if the analyser hasn't been used for months.

IMPORTANT: I don't recommend using the BNC RF input because this makes the analyser very prone to damage from external dc voltages. At a guess either the exotic input RF attenuator will get toasted or maybe the overload limiter diode stage will get damaged by the external dc. By all means make an external BNC (low frequency) dc block to attach here if you want to explore down to less than 100kHz but I really don't recommend making raw connections here.

At work this is by far the most common reason for a spectrum analyser to get damaged. The operator connects an active device to it and accidentally bridges out any dc block on their PCB due to clumsy probing near the output stage.
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 17th Sep 2020 at 9:37 pm.
G0HZU_JMR is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 8:10 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.