Test Equipment Anonymous (TEA) : Discussion and Group Therapy Thread

[tggzzz]

Because of its immense strategic and economic importance, it's hard to see GNSS going away any time soon.

Except for localised intentional (and unintentional) jamming and jiggery-pokery.

There are advantages to an off-air standard, not taking time to lock being one, so not needing to be left on for long periods.

Hmm. There's warmup times, which are much longer than the lockin time of the 198kHz receiver. But the latter is much noisier; choose your poison.

However for practical purposes, calibrating SAs, frequency counters etc, cheap GPSDOs seem fine. Time nuts would have a different view. Industrial grade GPSDOs offer more but are substantially more expensive.

Here's a comparison of some cheap GPSDOs

https://reeve.com/Documents/Articles%20 ... DOComp.pdf

Useful; thanks.

[EC8010]

I used to calibrate my TTi PFM1300 frequency meter using PAL subcarrier, but it is no more. Now, I have one of Mr Bodnar's GPS things powered 24/7 to produce 10MHz (very sharp edges on the square wave) and 96kHz (or possibly 192kHz). I've yet to see a practical improvement in my measurements from having this, but it makes me feel better.

[tggzzz]

I've been playing around with Zenith's OCXO, comparing it with the 10811 in the Agilen53310. The prototype technique used is croc clips on flying leads, ahem.

It seems the control voltage is ~1ppm/V i.e. 1V->10Hz. It is easily adjustable to 10MHz, with Vcontrol 2.105V.

My bog standard TTI bench PSU has 100µV RMS noise, which equates to 1mHz. The measured frequency noise is 4mHz. That's too close for comfort.

My low noise LM399-based Power Designs 2020 bench PSU has 2µV noise, which equates to 20µHz. The measured frequency noise is 2.2mHz. Better.

Conclusion: need to ensure a stable and low-noise power source for the control voltage. That corresponds to what I've read elsewhere.

I wonder how much better it would be on a PSU in the same case as my HP10811.

My external 10811 has the LM723 PSU recommended in the datasheet (which was low noise for the time, 2.5µV). The control voltage is 0.1Hz/V (i.e. much less sensitive than the UUT) and recommends <100µV noise on the control voltage line.

I really ought to do all the arithmetic, and then use decent construction techniques.

[tggzzz]

I've yet to see a practical improvement in my measurements from having this, but it makes me feel better.

I can say that with many of my activities

[Zenith]

I've been playing around with Zenith's OCXO, comparing it with the 10811 in the Agilen53310. The prototype technique used is croc clips on flying leads, ahem.

It seems the control voltage is ~1ppm/V i.e. 1V->10Hz. It is easily adjustable to 10MHz, with Vcontrol 2.105V.

My bog standard TTI bench PSU has 100µV RMS noise, which equates to 1mHz. The measured frequency noise is 4mHz. That's too close for comfort.

My low noise LM399-based Power Designs 2020 bench PSU has 2µV noise, which equates to 20µHz. The measured frequency noise is 2.2mHz. Better.

Conclusion: need to ensure a stable and low-noise power source for the control voltage. That corresponds to what I've read elsewhere.

I wonder how much better it would be on a PSU in the same case as my HP10811.

My external 10811 has the LM723 PSU recommended in the datasheet (which was low noise for the time, 2.5µV). The control voltage is 0.1Hz/V (i.e. much less sensitive than the UUT) and recommends <100µV noise on the control voltage line.

I really ought to do all the arithmetic, and then use decent construction techniques.

I made a DIY PCB to put one of those Aliexpress units into a Racal 9904. The OCXO in the Racal really was in a bad state. It took hours to settle and settled at a different frequency each time. The control voltage for the new OCXO was taken from the 5V supply using a 20 turn trimpot. Checking with a scope, the 5V line is noisy as would be expected. The 9904 settles within five minutes and drifts within a few parts in 10^8 of the GPSDO. The noise on the 5V line is quite repetitive and I assume it averages out. There may well be frequency noise if it was looked at closely.

My experience of OCXOs in Marconi and Racal kit is mainly single overnised ones which take hours to stabilise properly and some never stabilise to the same point. Well they are about 40 years old.

There were several Racal frequency reference options.

A crystal oscillator, claimed to be good to about 2 in 10^6, intended for customers who intended to used the counter mainly with an external reference.

The single overnised one, which is the most common and about 2 in 10^7

The big double overnised one, which from memory is 2 or 3 parts in 10^8, although I recall mine does better.

A TCXO which I recall is about the same performance as the single overnised OCXO.

I mean to have a proper board made for the Aliexpress OCXOs so they can be a drop in replacement for the dodgy Racal ones. It makes sense to give more thought to the control voltage. It's very high impedance and something like a TL431 should be adequate and certainly an improvement over just using a potential divider on a power line.

[tggzzz]

All plausible. But since I'm "in there", I might as well do as good a job as I can.

I've just looked at the internals of my external 10811 oscillator. It seems like I ignored the possibility of fine control using the Vcontrol pin (-5V to +5V), and simply connected that to Gnd. I relied on the "coarse control" screwdriver adjustment to set the frequency. Definitely not timenut territory!

Hence I presume the 10811 is as noiseless as possible. Drift is another issue, and I'll ignore that drift is equivalent to low frequency noise.

[Zenith]

The double overnised OCXO on my Racal 1992 has coarse and fine controls. They disappear into the body of the OCXO and I have no intention of opening it up, so I have no idea what they used to stabilise the control voltage. I've never seen a cct diagram for those OCXOs, although someone has reverse engineered the Marconi equivalent and published the cct on the web. Their control voltage is supplied by an external arrangement. Anyway, for the Racal double overnised OCXO, doing the usual thing of comparing the OCXO to the GPSDO on a scope, and making the traces not move relative to each other, takes a very careful hand with the adjusters.

My view of making up a proper board for those Aliexpress OCXOs, is that it's received wisdom they should receive as stable a control voltage as possible, and TL471 (and better) are not expensive. From memory the input resistance of the control voltage pin is 100KR, so a voltage reference should easily be able to supply the current needed. It adds very little complexity and expense, so there's no reason not to do it.

When I tested those OCXOs I used a lash up with jumper leads, a bench power supply and a 20 turn trimpot. They all stabilised within about 5 minutes and their performance was impressive, they certainly seemed much better than the crappy old Marconi and Racal units I bought them to replace.

[tggzzz]

In the 10811, the screwdriver coarse control is a trimcap, the voltage control feeds a varicap.

Do you mean an OP471? I can't find a TL471.

I would expect the voltage reference to generate more noise than an opamp.

[Zenith]

My mistake. TL431.

It's a part that's been around a long time and they are cheap. I'm sure there are more modern ones which are better.

https://www.ti.com/lit/ds/symlink/tl431.pdf

The datasheet talks about noise. It looks a lot better than a voltage divider across the 5V line.

[Robert]

<SNIP>

I'm not sure what you mean when you say "GNSS availability". GNSS (Global Navigation Satellite System) is a general term for several satellite systems used for positioning and navigation. These systems are GPS, Galileo, GLONASS and others. GPSDOs use at least two. Because of its immense strategic and economic importance, it's hard to see GNSS going away any time soon.

There are advantages to an off-air standard, not taking time to lock being one, so not needing to be left on for long periods. However for practical purposes, calibrating SAs, frequency counters etc, cheap GPSDOs seem fine. Time nuts would have a different view. Industrial grade GPSDOs offer more but are substantially more expensive.

Here's a comparison of some cheap GPSDOs

https://reeve.com/Documents/Articles%20 ... DOComp.pdf

The Leo Bodnar unit seems one of the best and most versatile, although there isn't much in it.

It's sad that BBC Radio Four is finally going off the air after years of threatening to do it. From what I've read, transmissions will continue at reduced power for the Radio Teleswitch Service used by some electricity meters, but it wasn't entirely clear how this is to be implemented.

There's also Time from NPL - the Rugby Clock - on 60kHz. It's based on a caesium beam standard. It's used for clocks, even surprisingly cheap ones. I've never heard of an off-air standard using it.

GNSS availability is by no means assured. The are a number of areas around the would where it is jammed or spoofed. A large solar event could take out all the systems as could exoatmospheric nuclear devices.

Pub quiz question 1: How many ballistic missiles were launched during the Cuban Missile Crisis?
Pub quiz question 2: How many of the missiles in question 1 had nuclear warheads that detonated?

[EC8010]

My mistake. TL431.

It's a part that's been around a long time and they are cheap.

Agreed on both counts; I have a 1oz baccy tin full of TL1431 (very slightly improved version). However, they are quite noisy. Not as noisy as an LM317, but significantly noisier than an LT6655. However, despite all of Jim Williams's hype about how hard it was to measure LT6655 noise, it's actually quite noisy. Next down on noise is REF102. But a MAX6126 with a 100uF electrolytic capacitor hanging off the NR pin is even quieter. And if you put a foam house around the MAX6126 and another foam house on the underside of the board, the two will protect it against air currents and that's measurable too.

The last time I went to NPL, I bought one of their radio-controlled clocks. It runs off a single AA cell, does so for far longer than my other radio controlled clock that needs a C cell, and it works wherever I put it (the other one is fussy about positioning). It cost £25, and it looks nice, too.

[Zenith]

The question is, how much noise on the control line of the OCXO is tolerable? There's no point paying for something better than a TL431 if it makes no noticeable difference.

[Zenith]

GNSS availability is by no means assured. The are a number of areas around the would where it is jammed or spoofed. A large solar event could take out all the systems as could exoatmospheric nuclear devices.

We don't live in an area where GNSS is jammed or spoofed and if there was a Carrington Event, we'd have more pressing concerns than an amateur frequency reference not working.

[tggzzz]

Has something in the forum changed? There's no X allowing me to delete this post.

Ditto my previous posts in this thread, and other threads. But I can edit them. Strange.

1) Before I noticed this, I was quoting a post, submitted it and it required me to login (same as EC8010 recently?)
2) I logged in, recreated my edits, and successfully resubmitted the post.
3) I then noticed the post in (1) was there, but without my additions.
4) But, as you can see, I can still edit that post in (1); this is it.

[tggzzz]

GNSS availability is by no means assured. The are a number of areas around the would where it is jammed or spoofed. A large solar event could take out all the systems as could exoatmospheric nuclear devices.

We don't live in an area where GNSS is jammed or spoofed and if there was a Carrington Event, we'd have more pressing concerns than an amateur frequency reference not working.

There have been many NOTAMs saying don't rely on GPS near Aberystwyth. I know someone who claimed there was a 1 mile "discrepancy" near Boscombe Down. Then there are big events (e.g. Olympics), and truck drivers that don't want their location known.

A Carrington event would be more "interesting".

[EC8010]

The question is, how much noise on the control line of the OCXO is tolerable? There's no point paying for something better than a TL431 if it makes no noticeable difference.

Indeed. I leave it as an exercise for the student to determine maximum tolerable noise on the control line. Just be aware that a TL431 is quite noisy. A BZX55C5V6 is very quiet. However, (and it's an enormous however) its temperature stability is terrible and not easily compensated.

What's a Carrington Event when it's at home?

[Zenith]

Indeed. I leave it as an exercise for the student to determine maximum tolerable noise on the control line.

I've heard that one before, along with "it can easily be shown" which means someone on their way to a Fields medal could manage it in an hour. Lesser mortals might fall short.

Carrington event.

https://en.wikipedia.org/wiki/Carrington_Event

[tggzzz]

What's a Carrington Event when it's at home?

A repeat of the Carrington Event is one of those things that features in the UK's National Risk Register. Even smaller events have knocked out electricity grids over a wide area. Hence it keeps engineers responsible for the electricity grid awake at night.

The full effects on the regional electricity grid (and hence civilisation) are unclear, but transformers burning out is a real concern. It used to be thought that the best plan was to try to protect the distribution network components by disconnecting everything. I believe the plan is now to keep things connected in the hope that they can distribute and sink the currents somewhere or other.
https://en.wikipedia.org/wiki/Geomagnetic_storm#Impacts

Because of serious concerns that utilities have failed to set protection standards and are unprepared for a severe solar storm such as the 1859 Carrington Event, in 2013, the Federal Energy Regulatory Commission (FERC) ordered the North American Electric Reliability Corporation (NERC) to create standards that would require power grids to be somewhat protected from solar storms and equipment to be continuously tested for possible effects of solar storms.

Since it occurred before the extensive interconnectivity of electrical systems and the general electrical dependence of infrastructure in the developed world, its effect was restricted; however, its ground currents were up to an order of magnitude greater than those of the March 1989 geomagnetic storm which interrupted electrical service to large parts of northeastern North America.
...
Undersea telegraph cables were affected by the storm.
...

[tggzzz]

Indeed. I leave it as an exercise for the student to determine maximum tolerable noise on the control line.

I've heard that one before, along with "it can easily be shown" which means someone on their way to a Fields medal could manage it in an hour. Lesser mortals might fall short.

Start with the sensitivity to the control voltage, i.e. Hz/V. Determine the noise and stability of the control voltage. Multiply those numbers.

No doubt there will be subtleties, but that's a good starting point.

It is one reason why I prefer my 10811 to "your" base station OCXO: the 10811's voltage sensitivity is significantly lower.

[EC8010]

I've heard that one before, along with "it can easily be shown" which means someone on their way to a Fields medal could manage it in an hour. Lesser mortals might fall short.

Yes, I'm afraid my tongue was firmly in my cheek there. "It can be shown that" usually means, "I'm sure I read it somewhere but for the life of me can't remember how it's done." Gems like that are common in "A Random Walk In Science". Well worth a read.

I looked up the Carrington Event. I had heard of such things but didn't know that anyone had been immortalised for it. We have politicians who instead of doing something about ongoing man-made disasters continue squabbling for personal power (but without responsibility), so what likelihood is there of any serious precautions? It seems that technology is being used by those who don't understand it to make civilisation increasingly precarious. As an exam question would say: Discuss.

On the lighter side... I have been investigating insulators recently. Applying 100V to 1kV and measuring current with electrometer. Results have been somewhat variable and the next plan for the 485 electrometer is a temperature-controlled jig for the DUT. Yesterday, I speculated that a poor insulator might generate significant 1/f noise, so I took the pre-amplifier output from the 485 through a x500 amplifier and 1kHz low-pass filter to oscilloscope and applied FFT to produce a noise spectrum. Sure enough, the leakage between adjacent tags of a poor quality phenolic tag board generated a significant increase in 1/f noise compared to system noise. Similarly, surface leakage currents can be expected to produce 1/f noise. My point is that ensuring low frequency drift on an oscillator is not just about the components used (TL431 etc) but could also prove sensitive to construction quality and (especially) defluxing.

The National Risk Register is an alarming document. Not for what it says, but for what it doesn't say or glosses over. For example, experience has shown that governments respond too late to infectious diseases (COVID, AIDS, infected blood).

[tggzzz]

Gems like that are common in "A Random Walk In Science". Well worth a read.

Seconded.

Ditto the sequel "More Random Walks in Science" and Mackay's "The Harvest of a Quiet Mind".

The latter is available online: https://archive.org/details/B-001-014-4 ... 1/mode/2up

[Zenith]

Indeed. I leave it as an exercise for the student to determine maximum tolerable noise on the control line.

I've heard that one before, along with "it can easily be shown" which means someone on their way to a Fields medal could manage it in an hour. Lesser mortals might fall short.

Start with the sensitivity to the control voltage, i.e. Hz/V. Determine the noise and stability of the control voltage. Multiply those numbers.

No doubt there will be subtleties, but that's a good starting point.

It is one reason why I prefer my 10811 to "your" base station OCXO: the 10811's voltage sensitivity is significantly lower.

It's definitely your base station OCXO. I remember giving it to you wrapped in aluminium foil, and said not to open it in the middle of a rally because it was an ounce of dope. Your eyes lit up. Abandoning caution you feverishly ripped off the foil. Your expression dropped when you saw it was an OCXO, and then rose again slightly.

The first I heard of those Aliexpress OCXOs was from BD139. If you buy 5 a time they are < £3 each. Equivalent new OCXOs from Farnell are about £100. I have a few Racal and Marconi units with dodgy OCXOs and I've never seen NOS replacements. I thought these would do as the basis of substitutes. I bought 5 because I assumed at least a couple would be useless. They were all good. If they had a frequency stability of +/- 1 part in 10^7 and a reasonable warm up time of say 30 minutes they would be perfectly fine. My crude tests showed they had MUCH better frequency stability and warmed up in 5 minutes.

The HP 10811 is mentioned in TAOE as a good example of a high performance OCXO module. No doubt it cost a fortune when it was new. It's remarkable that a used OCXO costing about the price of a Sunday newspaper even bears comparison with a 10811. Judging from the one example of a double overnised Racal OCXO I have, they are also very good.

[Zenith]

I looked up the Carrington Event. I had heard of such things but didn't know that anyone had been immortalised for it. We have politicians who instead of doing something about ongoing man-made disasters continue squabbling for personal power (but without responsibility), so what likelihood is there of any serious precautions? It seems that technology is being used by those who don't understand it to make civilisation increasingly precarious. As an exam question would say: Discuss.

I don't hold politicians in high regard, but there are major trends which they have no control over. For instance we have become pretty much completely dependent on reliable mains electricity to the extent that most shops can't operate without it. Petrol pumps won't run without it.

Then there are possible natural disasters which either might or might not happen, or will happen eventually. For instance a volcanic eruption on La Palma could trigger a landslide which would cause a megatsunami in the Atlantic Ocean which would affect the East coast of the USA and other areas. How much planning should be put into preparing for that? There are many similar potential hazards.

On the lighter side... I have been investigating insulators recently. Applying 100V to 1kV and measuring current with electrometer. Results have been somewhat variable and the next plan for the 485 electrometer is a temperature-controlled jig for the DUT. Yesterday, I speculated that a poor insulator might generate significant 1/f noise, so I took the pre-amplifier output from the 485 through a x500 amplifier and 1kHz low-pass filter to oscilloscope and applied FFT to produce a noise spectrum. Sure enough, the leakage between adjacent tags of a poor quality phenolic tag board generated a significant increase in 1/f noise compared to system noise. Similarly, surface leakage currents can be expected to produce 1/f noise. My point is that ensuring low frequency drift on an oscillator is not just about the components used (TL431 etc) but could also prove sensitive to construction quality and (especially) defluxing.

Worth bearing in mind. Mostly we can get away without worrying about leakage paths across insulators and PCBs and inconsistencies in the material itself, but sometimes we can't. Electrometers are a different category. I did read about a valve circuit for a photoelectric cell. It said performance could be improved by increasing the load resistor but there was a limit of about 10M, because after that random leakage paths across the glass envelope became significant.

[tggzzz]

Then there are possible natural disasters which either might or might not happen, or will happen eventually. For instance a volcanic eruption on La Palma could trigger a landslide which would cause a megatsunami in the Atlantic Ocean which would affect the East coast of the USA and other areas. How much planning should be put into preparing for that? There are many similar potential hazards.

The UK's National Risk Register is a useful summary of such "eventualities". It assesses acute risks and explicitly ignores chronic risks. It is based on reasonable worst case scenarios and ignores highly unlikely ones.

"Space weather" including the Carrington event is rated as one of the more "interesting" possibilities: significant impact (range moderate to catastrophic) with a likelihood of 5%-25%.

Here are a few snippets: the definition of "likelihood" and "impact", the page on space weather, and how it compares with other eventualities.

Screenshot_2026-06-02_12-59-23.png

Screenshot_2026-06-02_12-52-29.png

Screenshot_2026-06-02_12-44-14.png

Screenshot_2026-06-02_12-45-40.png

Screenshot_2026-06-02_12-45-56.png

[EC8010]

For instance we have become pretty much completely dependent on reliable mains electricity to the extent that most shops can't operate without it. Petrol pumps won't run without it.

Indeed. Gone are the days of the "Open All Hours" mechanical cash register.