Oscilloscope tube/display

If it can't refresh the display at the same speed as the data capture (albeit with a slight lag), it will have to either concatenate several data points into one display frame or omit some of the data. I doubt if any of the cheaper displays would include information about that in the specification.

I am looking at the vectorial stylus movement of a stereo gramophone pickup. Rough particles generate infrequent transients at various angles, depending on the contact angle between the stylus and the groove wall. These crackle patterns also show whether the stylus fits the groove properly and which size of stylus fits best.

Certain types of recording artifacts produce looping patterns which depend on the frequency being recorded, they tell the transcription engineer which type of recording equipment was in use at the time and can be a great help in setting up the correct replay characteristics. The distortion they generate can be minimised by altering the playback geometry.

Damaged groove walls generate other characteristic patterns and the balance between the right and left channels may have to be skewed to minimise the unwanted noise and distortion

Because of this, it is critical to be able to judge the angle and shape of these random excursions so that the proper correction can be applied. A modified X-Y oscilloscope has proved to be the best tool for the job since the 1980s but I was hoping that there might now be something off-the-shelf that would be a drop-in replacement.

It looks as though the trend for digital 'extras' has compromised the basic functions, at least in the cheaper equipment but I might buy one and see if it will do the job in spite of the shortcomings.

No great problem if you are only building a one-off and have a shed full of old parts.

Any mounting hardware will have to be built up from alloy sections, so they will just have to be a bit bigger for a CRT. The amount of work will be almost the same.

...But will it do the job? The answer so far appear to be "No". As the price is so low, I might be tempted to buy one and try it.

Yes - if it can be made to work in the first place.

Generally speaking, yes - but this market hasn't changed and the old equipment has proved adaptable enough to cope with unfoseen changes. Other markets have opened up alongside this one but they have needed completely new equipment to be designed from scratch.

Those prices are completely beyond anything I could afford. All my equipment put together hasn't cost that much.

...and it only needed to do that, so what was the problem that needed to be solved?

None that I have identified (other than those I have already dealt with by small modifications). Nobody else has come up with any equipment that is even as good as this for this particular job and the only reason I am building new equipment is to be able to carry it around for occasional demonstrations.

I have suggested modifications to other people's equipment based on my experience but they have always 'known better', because they were more focussed on selling it than on getting it right.

It simply needs to display the vectors correctly so that I can try to work out what they mean. Every pressing of every record is different and the display is a combination of many factors, there is no way software could sort that out. i sometimes have to sit for hours trying to work out the conditions under which that particular record was made and how it affected the decisions made by the recording engineer - and then how to get the best out of the surviving results.

Only the intermediate and final replays. This process started in the days when the output was on tape and storage was an expensive business.

The customers wouldn't have a clue what I was showing them, they only care about the final version. Very rarely I record the two channels direct from the pickup pre-eamplifiers but the limited bandwidth of the recording system means that that recording doesn't contain all the information. Also, if the vectors show that I have got something mechanically wrong, it need to be put right on the spot before the source material has been returned to its owner.

I have bitten the bullet and tried to order a Mini Oscilloscope kit but the supplier's purchasing website won't accept a UK landline telephone number. I tried to order one from a Dutch supplier but they will only deliver to addresses in continental Europe.

Thanks.

I think the 'refresh' rate of the Wave2 depends on triggering, even in XY mode. This might work if you're only interested in infrequent transients. Otherwise it uses whatever 'auto' spits out.

Latest versions offer a 2.5MHz sampling rate for dual traces - so yeah, ~5KHz maximum report rate. Without persistence, triggering above a few hundred Hz is probably a waste of effort.

Bugs me that there's not a direct conversion between bits and pixels, or easily retrofitted panel address formats or interface connectors, but that's what's out there.

Magnifying glass enhancing display legibility is right out of Terry Gilliam's 'Brazil'.

I gave away the record collection in '77, after selling the hardware to finance travel. Stylus, tape transport and loudspeakers were the main sources of audible foolery. These days it's just aging hardware and (probably) ears.

You seem to be looking for sound recovery, rather than cutting masters. There's probably a program or app for that.

RL

A 2-channel 30 MHz oscilloscope used to cost as much as a Chevrolet.

Now one can set up a very decent electronics lab for ballpark $600.

Yes, the 'product' is what the tool can do (a 'service' I provide). In the case of the portable kit, it is mainly intended for demonstation purposes but could be taken to the cuatomer if the recordings are too valuable to be allowed off the customer's premises.

I'm not sure if that applies here - but it is an amusing concept.

Yes, it tells me if the analogue computer settings are well-matched to the signals coming off the record. From that I can work out which of many parameters need changing and which way to change them on-the-fly (...sometimes with the customer breathing down my neck).

There are things that software can do (mostly related to the time domain) but it cannot correct faulty playback geometry. At every step in the recording and palyback chain, errors are introduced; they have to be undone in the reverse order from that in which they occurred. It is no use throwing a grossly distorted waveform at software and expecting it to work out what it looked like before it was subjected to:

Frequency response distortion by the microphone. Frequency response alteration by pre-emphasis. Frequency response distortion by the cutterhead. Groove wall phase errors by a skewed cutter. Recording machinery noise. Pressing errors. Groove damage. Poor disc material. Incorrect replay stylus radius. Tracing distortion on playback. Azimuth errors on playback. Speed errors. Playback machinery noise. Replay characteristic errors.

...and in what order they occurred.

Most of the errors that require vector analysis are either unique or occur very infrequently indeed. (The software would have been replaced before the same error occured for a second time.)

Start with a piece of shellac-slate compound, laminated shellac, nitrate on glass, nitrate on aluminium, gelatine on glass, gelatine on cardboard, cellophane on cardboard, embossed solid aluminium, chalk-filled vinyl (horribly noisy) - and, very rarely, vinyl itself. The output is a computer file in AIFF which I record to a CDR for the customer. Increasingly, customers are asking for the files on a USB memory stick in various formats, presumably so they can erase them by accident and expect me to provide another copy a few months later.

I have also filled in the contact form explaining the telephone number problem to the suppliers, they haven't replied yet.

Yes, that is my profession. (...one of several.)

Nothing that works to anywhere near an acceptable level.

Getting the hardware right is the key to good sound recovery, the software is just there to record the results. The X-Y scope is an aid to identifying the hardware errors and correcting them.

That is what I have done with my Mac G3s that run my business - but buying oscilloscope tubes in quantity doesn't seem to be possible nowadays. I can pick up a few on the secondhand market, but they are all different.

It has taken me since my childhood to identify these faults and I am still learning. I have made disc recordings myself, so I know what was involved at the various recording dates of the discs that people bring me. it would take another lifetime to put all this down in any form, let alone one compatible with a computer program, and it still wouldn't be complete because new discoveries are still being made.

I can't imagine a computer 'listening' to a recording and saying "that noise was the recording engineer engaging the scrolling mechanism prior to cutting the runout groove". It took several transcription engineers listening to dozens of records over and over, then searching the archives for evidence of how that particular recording lathe had been adapted to produce scrolls when automatic stop mechanisms first became available on clockwork gramophones.

Another example: The X-Y display showed strange looping behaviour which varied according to stylus size - but only on certain recordings made between certain dates on one particular type of cutterhead. This was traced to the contact points of the rounded stylus in the 'V' groove (which was unique to this particular manufacturer at the time) encountering a phase difference between the two groove walls (a bit like azimuth error on a tape head). The larger the stylus, the higher it rode in the groove and the greater the phase error.

We traced the cause to the Blumlein cutterhead which had the cutter tip trailing on a bar mounted on a vertical-axis pivot with rather low restoring force. If the cutting facet was not exactly at right angles to the groove, there was a sideways thrust which drove the stylus bar further sideways and exacerbated the angular error because of the low restoring force. The cutting lines on two groove walls were displaced from their correct positions, so the two lots of modulation were out of phase with each other by an amount that depended on frequency and their height up the groove wall.

Also the suction used for swarf removal could not be very fierce, otherwise the draught of air past the not-very-stiff cutter produced a faint roaring background to the recordings - so the engineer always set the cutting facet at a slight angle to help throw the swarf sideways. The problem wasn't just caused by an isolated error, it was unwittingly imposed on every recording for reasons that made sense at the time.

This might seem trivial but when the record company started producing frequency test records, they used this cutterhead because it had the widest response. On the 8 Kc/s band, the phase error was nearly 45 degrees so all the calibration and research subsequently carried out using this 'definitive' recording was invalid. That discovery was the result of over 10 years of research.

There are hundreds of examples like this, each one unique.

...and by whom? The customer?

I have tried to teach several people over the years but they soon loose interest. Many of them don't have enough basic knowledge of physics/ electronics/ engineering to understand the problems and, as long as they can do a quick-fix based on hearsay, they don't want to learn. Nowadays most people believe that digital can fix everything.

It isn't just me on my own. I have bookshelves full of books and documents ranging from research treatise on vibrational analysis through chemistry. electronics and acoustics to reminisences of the early recording engineers and their personal papers. The computer is stuffed with BBC monographs and I have access online to Bell Labs Journal, Philips Techinical Review and hundreds of individual publications. In addition to all that, I make contact with a network of other transcription engineers and record collectors; we bounce problems back and forth, trying to find explanations for observed phenomena.

Computers help with the communication but they haven't shown any signs of doing the sort of thinking required for that kind of research.

The magic word is "DEEMED".

Lots of copyright material is stored illegally by any responsible recording engineer, it is deemed not to exist and could not be found by a casual search. If the circumstances change so that the recovery of a lost gem becomes more important than its copyright status, it can be 'discovered' in some non-attributable way.

That is how many 'lost' radio programmes have come to light.

Mainly heater burn-out, loss of emission and 'soft' vacuum from electrode de-gassing or leaking seals. In addition, screen-burn if the designer or user is careless.

There are many parameters: EHT voltage, heater voltage, different numbers of electrodes, different electrode voltages, different deflection sensitivities, different base connetors (made of unobtanium) and even different bulb and screen shapes and sizes if I have to use what I can get.

...but that would only solve the problems we already know about. You couldn't train it to identify problems that might be encountered in future and work out their causes, because you don't know what they might be.

Why do certain recordings have a muffled 'boing' sound about 2 revs before the music starts? Is it due to the electrodes in a valve in one of the amplifiers expanding? Is it because the mastering studio used a Ferrograph with a clutch plate that magnetically snapped onto the HT choke and the shock wave vibrated the EF86 in the head amplifier? Is it because there were springs in the scrolling mechanism that were shock-excited as the mechanism disengaged.? Was it the sound of the switch controlling the 'start' light in the studio? Was it some accoutrement on the performer's clothing as they took a deep breath to begin singing? Could it have been all of these on different recordings at different times?

You couldn't train AI to 'envisage' all those possibilities unless you could set up a series of recording studios exactly duplicating the equipment, people and knowledge in use at each date - right down to the sort of clothing and accessories a performer might wear. Is the question important enough to justify this? Perhaps it would be if you needed to verify the provenance of a potentially valuable historic recording that claims to have been recently discovered but might be a fake.

As far as I can see, it wouldn't do anything because it would take longer to build and train than a group of us took to solve the problem - along with many other problems during the same period (e.g. Is the locked groove always exactly the same diameter? Is the pitch of the scroll significant? What are we having for supper tonight?)

Send me a machine that has passed exams in physics, electronic, electrical and mechanical engineering, that can solder-up circuits to test ideas and can make replacements for missing bits of mechanism on a worn-out lathe and search-out and interpret historical information from inaccurate sources ...and I will try to train it.

There is a moral obligation as an historian.

There is also a practical reason: I go to a lot of trouble to 'rescue' a recording and then give the results of my labours to a customer who will just as likely accidentally delete the files in six months time. A second attempt to recover the sound from the originals will take a lot more effort because they deteriorate each time they are played - and that is if the customer hasn't thrown them away in the meantime. Doesn't it make sense to keep a copy?

To be fair, I would rate the chance of obtaining in 30 years' time an exact replacement of some specific LCD oscilloscope originally obtained in 2025 from aliexpress considerably lower than the chances of obtaining a spare tube for a 1960s CRT oscilloscope. Ditto the chances of fixing any fault in the support circuitry. If the LCD one is reliable enough to never ever fail, and its flash memory retains the firmware long enough, then it might not matter.

In 30 years time I shan't be here to worry about it but I agree that older general-purpose components may still be available long after 'trendy' specialist ones have slipped into oblivion.

The chances of finding anyone who can repair anything to component level will be much less in 30 years time than they already are. People who can redesign circuity to suit a different CRT will be as rare in those days as the people who can design steam engines are nowadays. (There are people with the machining skills to copy existing designs, but how many people actually understand the design process?)

Just a fyi fact we discovered on Friday: the $300 Siglent scope we recently bought for a road trip has XY mode.

Cheers

Phil Hobbs

I wonder what sample rate, and whether it displays all samples from the ADC. I suspect it might capture a buffer full of 2 channels, diplay that, then capture another buffer full, display that etc. and perhaps only be capturing a small percentage of the time.

That was exactly my worry. With the CRT X-Y oscilloscope I connected a differentiator and full wave rectifier to each axis, this is coupled to the Z modulation, so that the infrequent short sharp pulses are highlighted.

I have got a very cheap digital hand-held 2-channel oscilloscpe kit on order. It claims to be able to do X-Y display, so I shall be interested to see how it compares with the CRT variety. If it is successful, I can build it into the record playing equipment but if it doesn't work adequately I shall have to use an external unit with a CRT.

Oh yes, there are people who know how to make beautiful emulations of CRT displays:

Sony Watchman or its CRT?

Sampling at 1 MS/sec (should be enough for 100 KHz signal) and assuming decay in 1s gives 1 mln points for display. For each point one needs to apply decay factor and redistribute value betwen corresponding points, I would expect about 40 machine instructions per sample. One needs to do this for each frame, assuming 30 frames per second this gives 1200 mln instrucions per second. Quite doable on Raspbery Pi class board. This assumes doing computation via CPU. For video core in Raspbery Pi this should be very easy job. Smaller RPi class boards can fit behind 10cm by 10 cm screen, so size should not be a problem.

I think the point is that the mechanical arrangement needs to be interactively adjusted before a good quality capture can happen. Things like stylus dimensions, tracking force and arm geometry were mentioned earlier. John

I think that this effect is in the eye. My estimate includes updating adjacent point, to improve visible quality.

There are many tricks which one could try to speed up display. Details should be worked out when doing actual implementation. My point was that implementation is feasible on available machines.

For such level of visual detail one needs multiple buffered display, drawing to memory buffer. This may involve copy from computer RAM to video memory. While such copy have non-trival cost, needed time is still smaller than time for computation.

Well, if display can not update fast, then there is no point in higher fram rates, just lower frame rate to what display can do,

IIUC the point is that data captured with wrong settings is essentially useless. I do not know what Liz is exactly doing, but clearly want to monitor and adjust settings in real time.

There may be psychological effect: human brain is better at finding dynamic changes than detail in static image.

I do not condider problem of automating what Liz is doing, for some problem of related nature there are remarkable successes, but also many problems currently lack cost-effective solution.