3. Tek 7 series LCD (2019)
Tektronix 7000 series analogue scopes are in some ways superior to modern DSOs but where they are inferior is in lacking a digitised display that can be manipulated for measurements and saved. Even the digitised 7854 suffers from this due to the crudeness of its cursor design. Of course I would like it to work at the highest bandwidth, since this is needed for the LHC project.
Tek did produce the C1001 NTSC digital video camera assembly that fitted many of its oscilloscopes including the 7000 series, but although they crop up on eBay from time to time I have never seen any hardware or software associated with them that could perform this function:
http://w140.com/tekwiki/wiki/C1001
At best, I've seen ad hoc mechanical structures added to support modern digital cameras, but no attempt to go further, and certainly not in real time:
Point and shoot digital camera on a blackened cookie tin shade attached to a modern DSO
https://hackaday.com/2013/08/06/bench-equipment-tip-screenshot-of-old-oscilloscopes/
DSLR on a simple open aluminium bracket located in a standard Tek camera mount:
https://w140.com/tekwiki/wiki/519#/media/File:Timkoeth_519_TIM_4443.jpg
DSLR mounted inside a custom shroud made from bare pcb sheets (but later based on a Tek C-53 camera)
https://www.amplifier.cd/Technische_Berichte/Photographie_Oszilloskop/Foto_Halterung.html
All of these are large assemblies that clip-on at best. The original Tek C-5x series cameras that fit my 7 series oscilloscopes are more compact, designed to swing out on their bracket hinge, and quite cheap these days. I bought a C-53 with the intention of attaching a digital camera to its back.
Later I returned to www.amplifier.cd and saw he too had decided a Tek C-53 was the way to go however he he added a huge huge shroud between it and the CRT:
https://www.amplifier.cd/Technische_Berichte/Photographie_Oszilloskop/Foto_Halterung.html
I wanted to see if instead I could utilise the original lens in the C-53 with its mechanical assembly otherwise gutted, and hopefully not require that ungainly shroud, as well as find a way to add measurement capability to the signal; IOW, an easy means of digitising an analogue scope.
AWAITING PURCHASE OF A SUITABLE CAMERA TO TRY OUT THIS IDEA
A CHEAP GHz DSO
The 7104 does all of the high speed stuff. I just need to digitize its CRT, which is essentially what the Tek 7912AD does using a scan converter, see [References: Oscilloscopes].
The 400MHz 7B85 and 1GHz 7B15 timebases have on-CRT delta measurements using brightened lines, but these are undesirable on the 7104 as this will wear out the MCP.
Could I capture the CRT onto a camera and then process it to a form that can be displayed on an LCD, and add cursor measurements and storage?
PC BASED DIGIIISATION
My first thought was to simply capture the CRT to the PC and use the customise skin feature in the excellent Screen Calipers app [S1] to create on-screen line measurement cursors, perhaps running multiple copies of it for both time and amplitude
However having played in the distant past with Propellerhead's Re-Birth RB-338 sound loop synthesiser
[G13], I remembered I hated the virtual pot controls which I found clumsy to operate. Real controls cost money and wear out, so it is hardly surprising manufacturers of modern instruments have minimised the use of simple switches, and I like the rotary encoders that have come in their place.
Whilst Screen Calipers is a great idea, I didn't relish have to use a mouse to control everything, particularly for fine alignment, and it would be far more practical if the display was actually on the oscilloscope than on my PC.
PIC µC BASED DIGITISATION
My next idea was to use a PIC µC to process a digital camera video output and convert the CRT trace to a precision thin line that could be displayed on an LCD together with cursors and measurements, in the same manner as on a modern DSO. The LCD is mounted at the back of the camera mounted on and facing the CRT.
It so happens I already had a few rotary encoders, and in the dim past, I had made an analogue TV advert detector (to put the kettle on) based on a PIC, so I thought I could build on this idea.
I bought an old Tek C-52 hinged camera for $30 and removed the Polaroid film back.
http://w140.com/tekwiki/wiki/C-50_series_cameras
IMPLEMENTATION
Digital video camera to capture CRT
PIC to continually read video, reduce it to a thin line, store it
PIC to determine trace, timebase & amplitude from stored CRT video image
3 Rotary encoders to select functions (Cur X1, X2; Cur Y1, Y2; Select cursors, Store, Retrieve, etc.)
'OK' push button to confirm function selection
PIC to to overlay cursors onto stored video
PIC to write stored video to analogue VGA LCD mounted on back of camera
PIC to mimic video camera so PC can view / save modified image
Optional serial interface using USB port to transfer modified image to/from PC
Optional SD storage
Another option - synthesise the CRT signal & feed out as a slow signal to a cheap tablet DSO:
The FNIRSI 1013D is a particularly attractive model, but at 17:48 the SA plot in this video reveals this '100MHz' scope is actually only a 25MHz one! https://www.youtube.com/watch?v=kBH5WO9IOMU )
David Jones' review is more scathing:
https://www.eevblog.com/2020/07/05/eevblog-1317-140-2ch-100mhz-fnirsi-tablet-oscilloscope-review/
David seems confused by the name FNIRSI but I think he missed the spelling and logo font are suspiciously similar to high quality manufacturer ANRITSU, just as Siglent is I think, mimicking HP's Agilent brand.
A nice idea, but CRT analysis is difficult to implement. The TV advert detector only had to look for the chevrons in the top right corner and verify the screen and audio were blank. This system OTOH would have to analyse the entire screen for every instance of the CRT trace and decide where the centre was (effectively what the 7912AD does and we know how big that is) before displaying it on an LCD, and then it would have to add multiple moveable cursors together with their measurements. In order to allow live video, this would require a huge amount of high speed processing, not to mention a frame store larger than the amount of RAM in a PIC. I'd need a DSP or gate array with external RAM, and all that entails.
A SIMPLER ALTERNATIVE
A simpler approach is needed. One way is to dedicate a scope channel to synthesise the X & Y measurement cursors with a pulsed waveform. On a Tek 7k series there are no dual plugins faster than the 400MHz 7A24 which is itself a compromise: 5mV-1.0V/div (8 divs = 8V), 50R, dc only.
For X (horizontal) cursor, a variable voltage pulse is required.
For Y (vertical) cursor, a PWM square wave is required (use de-icing motor controller PWM SW).
A combined X & Y cursor arrangement might be achievable using a variable amplitude PWM square wave but given we have the option of a 2-channel plugin for the cursors, and its timebase would need to be triggered to the cursor signal, it would be of little use as a signal input. Therefore there seems little point complicating matters: separate X and Y cursor inputs is easiest to implement.
I could put a 1GHz 1-channel 7A29 or 2-channel 7A24 in slot 1 for the signal, and another 7A24 or slower 2-channel plugin in slot 2 for separate X and Y cursors. I'd use separate signal & cursor timebases.
This approach would still need a camera, but this time it would simply feed into an LCD screen at the back of the camera mount with no PIC analysis of the video signal. A potential (but still difficult) option is the PIC reads just the Tek characters on the CRT, as they are in fixed positions.
If a modern HAD camera is used and the CRT intensity turned down, the CRT trace should be very thin and therefore easier for the PIC to make out. However light intensity level will affect the ability to display the pseudo cursors.
Could an analogue VGA LCD screen take the video output from a camera without further processing, and does one exist that would fit onto the back of the camera?
Rotary encoders are nice, but timer / processor intensive beyond more than one or two, and it isn't necessary to have an endless track pot. PICs have multiplexed ADC inputs and it would be easier to have 8 cheap dc pots. X-axis pots can adjust the spare channel from say 0 to 5V on a 500mV/div setting (8 divs = 4V) so they have the option of going off-screen when only the Y cursors are selected (the normal scope position control would be used to pull the trace below when at 0V). When only the X cursors are selected, a dc signal is needed. Using separate dedicated pots means they can be left in their positions between use, making it easier to re-adjust the cursors.
CONTROLS
P1 = left X cursor <
P2 = right X cursor >
P3 = up Y cursor ^
P4 = down Y cursor v
P5 = select timebase: the widest Tek range
P6 = select amplitude: the widest Tek range
P7 = select measurement type similar to Rigol choice
Pushbutton 1 to select choice / store
Pushbutton 2 to delete choice / delete
Pushbutton 3 to hold snapshot / release to live video
P8 = select function: X cursors, Y cursors, X & Y cursors, store/delete, MATH functions for channel 3
CURSOR DISPLAY OPTIONS
The easiest option is to have a string of character LEDs showing XX & YY cursor measurements.
It may be possible to project bright red reversed-character LED matrix displays onto a frosted screen in front of the CRT face via the camera hood window, in which case the camera will see this too. The down side is any frosted screen will defocus the CRT trace.
A small backlit LCD could be run along the top of the scope, at the front of the box that contains the PIC and electronics.
The PIC could overlay the measurements onto the video screen but that would probably be too complex.
Automated measurement. This would necessitate the PIC having access to the video signal from the camera to the screen LCD and would remove the need for P5 & P6 . However instead of trying to recognise all signals on the CRT, it need only recognise the characters displayed in their fixed locations on the CRT. The PIC can then calculate cursor measurements. The only really hard part would be add these results to the LCD video stream.
SIMPLIFIED OPTIONS
2 dedicated push buttons to 'calibrate' XX & YY cursors placed at the graticule limits of the CRT
2 dedicated pots or rotary encoders to select V/div & timebase (NB CH-1 & ch-2 signal to be identical)
PASC to capture stills / videos; separate video camera for PC capture to storage if needed.
PIC to continually read controls and generate X & Y cursor digital PWM pulses as required
PIC controlled DAC to vary PWM voltage level
PIC to calculate measurements and display (in order of difficulty) on:
a) string of 7-seg or 16-seg LEDs mounted front bottom of camera side for my X & Y cursor differences
b) small LCD on back besides video display
c) matrix LEDs to project onto screen in front of CRT
d) overlay cursor measurements onto live video, or USB video dongle to capture video to PC for storage
Analogue video camera to capture CRT - will it be able to discern my cursor LEDs?
Harder - analogue VGA LCD to be driven directly from camera
EASIEST IMPLEMENTATION
H1 7A29 1-ch or 7A24 2-ch for signal(s)
H2 7A16/24/26 2-ch for x,y cursors (the slowest 2-ch I have is a 7A18)
Separate V1 & V2 7B10 x2 timebases for H1 & H2
2 push buttons to 'calibrate' XX & YY cursors aligned when the graticule limits of the CRT
2 pots or rotary encoders to select V/div & timebase (NB CH-1 & ch-2 signal to be identical)
2 pots for cursor xx & yy movement
PIC ADC pot to select Tek V/div
PIC ADC pot to select Tek timebase
PIC 10Hz int reads controls & generates H2-1 X & H2-2 Y cursor digital PWM pulses as required
PIC PWM to generate (fixed amplitude off-screen) H2-1 X cursor
PIC controlled DAC to vary voltage level to H2-2 Y cursor
PIC to calculate & display XX/YY difference on 7-seg LEDs front bottom of Tek hinged camera mount
A modern PASC (point & shoot camera) at rear of the Tek camera to capture stills & display on its LCD.
Even better would be a PASC with video output. If not:
It may be possible to simultaneously mount a video camera in the top viewing port on the Tek camera.
Analogue video camera captures CRT & LEDs to video output.
Video o/p optionally displayed on a small LCD monitor at back of video camera on top of Tek camera hood.
Analogue video output captured to PC for storage.
EXPERIMENTS
I don't have an AWG but the 10MHz Picoscope 2204A-D2 has a 1-ch (100kHz max) AWG o/p. It should reveal if my pulse idea will work. The 7104 needs >=43Hz to get a stable twin timebase display, see: [Projects: Tek 109 1-Shot & LF Mod].
Does my Panasonic FT5 have a live video output?
If not, is there a cheap PASC that does?
See if a monochrome PAL video camera can make out my cursor LEDs.
Can video captured to PC also see the LEDs?
NEW THOUGHTS
7M13 to display dynamic values
Cursor calculations of voltage and delay delta measurements can be displayed on the CRT via 7M13 dual display readout plugin if its switches can be bypassed and instead display V and T. x,Y Cursor controls would have to be a signal on one channel of a dual channel plugin with a dedicated TB, leaving just the other channel for the signal and its own TB.
Export mimicked signalAn improvement on the original idea of capturing the CRT to a camera and converting this to a display on an LCD would be to instead output it as a slow version of the signal to my cheap 10MHz Picoscope 2204A-D2 which can apply its full math capability, cursors and storage. The only downside is cursors would measure the slower signal, but it should be simple to manually multiply them by the original time base. This approach allows multiple traces. Surely someone somewhere in the World has done this already? It may be possible to use some kind of recursive coding to accommodate multiple traces captured by the video snapshot.
New text box
µ Ω ± ° ⌠ ⌡ ∫ │ ─ √ φ θ Θ ∂ δ ζ ξ ς λ ψ ω τ µ Ω ∆ Δ ∑ ∏ π Ξ ○ ≠ ³ ² ±