In the first post in the series, we looked at the Garmin 18x LVC “puck”. We talked about a particularly insidious issue that affected [Andrew] – both of his GPS units. And we saw that Oscilloclock owners really need to be able to update the firmware in these units.
In Part 2, we went through the design of an Oscilloclock Garmin 18x USB Adapter, that would allow the GPS to connect to a PC where the Garmin software runs to upgrade the firmware.
Now we conclude the series, with a treatise on the construction of the Adapter. Enjoy!
The final design
Here’s the design we arrived at in the last post. Let’s go through the steps to build it!
Fish out Fake Chips
The key component required is a decent TTL serial to USB adapter with programmable inversion on the signal lines. But here we have to careful: many low-cost adapters out there are built around fake FTDI chips!
As mentioned before, we at Oscilloclock are pacifists. But if we were to wage war against anything, it would be fake components. They are unsafe, unreliable, unworkable, and entirely unethical. You get what you pay for, if you pay the right people. The people who design, manufacture, and support the real McCoy.
Besides ethics and reliability, there is also a practical reason we must avoid adapters based on fake FTDI chips – often the fake chips are not programmable. A true no-no. So watch out.
Program the inversion
FTDI provide a nifty utility called FT_Prog. Below shows the utility running on a PC with the adapter connected, and configuring to invert the transmit (TXD) and receive (RXD) signals.
Dividing the input signal
We need to figure out the most elegant way to install the voltage divider – the two resistors we described earlier that reduce the impact of noise.
The cleanest way seemed to be to install the 1.2k shunt resistor directly across the receive and ground pins in the adapter itself, as below.
What about the 270 ohm series resistor on the RXD line? Well, installing this inside the adapter unit itself would require cutting tracks on the PCB. And that would compromise our effort, reliability, and aesthetics goals! So instead, we’ll insert this into the cable later on.
Cable Connector Conundrum
Recall that [Andrew] has two Garmin 18x units – one fitted with a small GPS connector and the other with a large connector. Wiring up two independent cables would have been natural. However, the TTL Serial to USB adapter came with only one cable pre-fitted with the necessary “DuPont” (a.k.a. Qi or 2550) connector.
What’s the big deal? Surely we can just attach a Qi connector to another cable?
Ha! Connector tech is never that easy! It turns out that to make a perfect connection with Qi connectors, you need a special crimping tool. The Oscilloclock Lab does not have this tool. And we do NOT compromise on perfection! Given that this adapter is not the best reason to invest in an incredibly expensive tool, we decided to use the single pre-fitted cable and split out to two GPS connectors, with the larger one serving as the split point.
(In hindsight, we could have separately purchased another quality cable that was pre-fitted with the connector. Next time, folks!)
Wire up the cable
Our beloved ultra-quality Hirose connectors are a joy to look at, and a joy to use. But wiring the tiny smaller units up with high precision doesn’t exactly “spark joy”. Still, we persevere…
Now we need to install the 270 ohm series resistor. We simply cut the wire and splice it in.
A bit more heatshrink applied, and we’re done!
Closure at last
Using the 18x USB adapter, [Andrew] is at last able to upgrade his pucks and enjoy his clocks in their full glory with GPS-synchronized time and date once again!
In the previous post, we looked at the Garmin 18x LVC “puck”. We talked about a particularly insidious issue that affected [Andrew] – both of his GPS units. And we saw that Oscilloclock owners really need to be able to update the firmware in these units.
We introduced the Oscilloclock Garmin 18x USB Adapter, that allows an Oscilloclock owner to connect their puck to a PC to enable the firmware upgrade.
In this post, we’ll take a look at the design of the Oscilloclock Garmin 18x USB Adapter. It wasn’t GPS satellite launcher (a.k.a. ‘rocket’) science, but it certainly wasn’t as straightforward as it might seem!
The Garmin 18x LVC electrical interface
Referencing the manual, the Garmin 18x series comes in 3 basic interface variations:
USB – USB 1.x interface, with a USB(-A) connector to plug into a PC
PC – RS-232 serial interface*, with a DB9 connector to plug into a PC, and a massive cigarette lighter adapter plug to obtain power
LVC – RS-232 serial interface*, with no connector – for wiring into a device
For our Oscilloclocks, we use the LVC variation and fit an attractive custom connector solution, avoiding the PC variation with its venerable, utilitarian, and aesthetically unpleasant DB-9 connector and cigarette lighter plug combo. (We may buck the trend one day and intentionally fit such sockets into that special retro clock build – who knows?!)
* Astute readers noticed the earlier asterisks. PC and LVC units are not quite true RS-232; their output voltage swings between 0V and +5V. Not so with devices having true RS-232 interfaces! A swing from -25V to +25V is legal and also lethal for any unsuspecting microcontroller. In the Oscilloclock design, we take advantage of Garmin’s voltage range cap to avoid having additional circuitry to adjust voltage levels.
Interfacing the 18x LVC to a PC
To upgrade the GPS firmware, the 18x LVC needs to connect nicely to a PC. But [Andrew] is an Oscilloclock Owner. He deserves more than just a good electrical connection. The interface also must be elegant and aesthetically pleasing, lightweight (for shipping), and easy to build. And – most of all – it has to be interesting enough to write a blog series about!
We can start with Figure 1 in the manual, which describes the most basic interface hook-up possible.
This interconnection option assumes two things: the PC has a DB-9 serial port, and there is a power source.
If we extend this option slightly, to take power from the PC’s USB port, we arrive at this:
A few months ago, [Andrew] – of Metropolis Clock fame – reached out for help. He had just pulled his lovely Oscilloclocks out of storage to put on display, when he observed odd behaviour in both units: the time was accurate, but the date was stuck – to some random date back in 2003!
What on earth was going on?
What’s going on was not “on Earth” after all! [Andrew]’s clocks synchronise time and date against satellites, using an external Garmin GPS unit. And this unit happened to have a serious flaw. In this series of three articles, we’ll look closer at this accessory, identify this issue, and see how we were able to resolve it. Enjoy!
Our longevity dream
We want your Oscilloclock up and running as long as you are – and even beyond! Our dream is to see these beloved devices inherited by loved ones, and even available on the second-hand market as antiques one day.
In an era of throw-away technology, we flaunt an unthinkable target: Decades of trouble-free* operation.
* Excluding the CRT itself – although we really try hard with that as well, as this post explains!
To maximise usable lifetime (and safety!), we construct Oscilloclock units from the finest materials and components available. As part of this, we also select manufacturers that guarantee their components and provide decent after-sales support.
And Garmin is one such manufacturer…
Welcome to the Garmin GPS ‘Puck’
All Oscilloclock models that synchronise time using an external GPS unit have so far been supplied with a Garmin 18x LVC GPS unit, colloquially known as a ‘puck‘. (Note: to extend the lifetime of the pucks, we do not recommend using them on the hockey court.)
Now, this is not the smallest external GPS unit on the market today. But it has been available from Garmin since 2007, and is even being manufactured today! It is one of the most sensitive, robust, and well-supported units out there.
(Of course, for every new Oscilloclock delivered we evaluate afresh based on the latest devices available.)
This puck has a special connector …
How many times have you relegated an expensive laptop, phone, or other random device to the trash just because the power socket or headphone jack failed? Some of the weakest components of any electrical device are its connectors – plugs and sockets.
To combat such failures, your puck is wired with an exceptionally high quality connector from Hirose. This connectivity solution is not only robust, it even feels good! There’s a lovely audible and tactile ‘click’ when you engage the plug, and it locks securely in place. And unlike cheap chrome-plated connectors, we’ve proven that these babies do NOT corrode, even after a decade.
Whether directly or indirectly, the pandemic seems to have slowed everything down: chip production; the global economy; and even Oscilloclock blog post publishing!
But perhaps most impacted of all is transport logistics. [Dante] in Brazil discovered this to his dismay in July 2020, when he purchased an Oscilloclock Bare unit. The P.O. had stopped all air service to Brazil just 3 weeks earlier – well after our discussions had started. Oh no!
[Dante] waited patiently for 6 months for the post office to resume accepting airmail service to Brazil. But they never did. And FedEx and DHL came at too hefty a price. In desperation, he authorized shipment by sea – and at last, in December 2020, his package was off!
Absence (of air mail service) makes the heart grow fonder...
After an agonizingly long wait, [Dante] finally received his unit 6 months later – in July 2021. He then spent the next 5 months completing his dream project!
[Dante]’s Dream: A Hewlett Packard retrofit
The Oscilloclock Bare is designed to be a no-frills controller assembly that highly knowledgeable folks can install into their own displays. [Dante]’s dream was to use this to convert his beloved HP 182T / HP 8755C unit into a living, breathing scope clock.
And convert he did!
Clearly, [Dante]’s 18 month end-to-end was worth the wait.
[Dante] was kind enough to supply a write-up of his project, including some clever solutions for pitfalls along the way. Let’s hear from him in (mostly) his own words!
The model HP 182T is an oscilloscope featuring a large CRT with a graticule of 8 x 10 major divisions and a display area of 133 cm2, coated with a P39 aluminized phosphor for high brightness and long persistence.
The HP 182T works as a display mainframe supporting other HP plug-in test equipment, such as the HP 8755C, a swept amplitude analyzer.
Both items are nowadays considered “vintage” test equipment. But with the Oscilloclock board installed, they have been transformed into a unique appliance with a natural appeal for practical use. Far better than the regular surplus market destinations, or — even worse — destructive disposal!
HP 8755C in short
This plug-in unit works primarily as a signal conditioner and a multiplexer for “almost dc levels” from three RF detector probes attached to three input independent channels. There are front panel adjustments for the scaling, gain and multiplexing controls that provide the appropriate Y-Axis composite signal for displaying by the HP 182T mainframe.
The Oscilloclock control board was elected to be installed inside this plug-in unit.
HP 182T in short
This oscilloscope is built around the CRT with its high voltage power supply.
The X-Axis signal from the Oscilloclock board is fed to the HP 182T’s chain of the horizontal pre-amp plus output amplifier, which drives the CRT horizontal deflection plates.
The internal wiring of the HP 182T connects the CRT’s vertical deflection plates directly to the plug-in cabinet of the display mainframe, so the Y-Axis signal from the Oscilloclock board is routed inside the HP 8755C itself.
The Z-Axis signal from the Oscilloclock board is fed to the HP 182T’s gate amplifier.
Contrary to any standard X-Y scope where the two input channels are always supposed to have electrically similar (if not identical) characteristics, the correct operation of the Oscilloclock board for the application here was shown to be not as seamless as first imagined. You have to face some details of these integrated “host” equipment (HP 182T + HP 8755C) to see why…
As described, there are distinct amplification chains accepting the Oscilloclock output signals. This presents specific challenges regarding (a) the differential gain for the X and Y signals, and (b) the differential time delay between any combination of the three X, Y, and Z signal outputs of the Oscilloclock board.
Before having the board at hand and expecting to make it work as soon it arrived (the shipping took longer than expected due to COVID restrictions), I first planned the signal flow and did the wiring. I had one eye on achieving a ‘clean packaging’ of the board inside the HP 8755C, and the other on ensuring compatibility between the Oscilloclock’s X-Y-Z output signals and their respective chains planned in the host equipment, considering signal amplitude and required frequency response.
The adaptations made at this time considered a minimally-invasive approach, where the criteria was to “make it simple”. This was limited to just opening or re-using connections and keeping the existing routing, in order to use the Oscilloclock’s X-Y-Z output signals in the most simplistic way possible.
Another necessary one-time adaptation was for the board’s power supply, and integration of its PSON output signal with the equipment’s hardware. This part of the design was successfully kept to the end of the project without any further modification.
First time installation of the oscilloclock board
Upon arrival and a bench test of the Oscilloclock board with a scope, I immediately figured out that the amplitude levels for the X and Y output signals were lower than expected (maybe due to my misinterpretation of the specs). I did the gain compensation corrections again and went thru the complete installation of the board inside the host equipment, anxious to see it working.
What a disappointment when instead, up came a completely distorted and elliptically shaped image, blurred with noise, and what looked like un-blanked retrace lines. Worse yet, mainly when alphabetic characters were displayed on the screen, none of the shapes were correctly formed.
Of course, that was time for a break — and a complete review of the job and the work done so far!
Chasing the problems
The Lissajous figures generated by the Oscilloclock board use an approximately 40 KHz sinusoidal signal, so I started to play with an external generator at the same frequency and amplitude for the X and Y signals (at about 1 Vpp) and trace it inside the HP 8755C and HP 182T.
At this time, I’d already exercised the Z-axis waveform from the Oscilloclock board and the expected processing through the HP 182T. There was no evidence of problems with this Z-axis signal chain, and I achieved a measured propagation delay of around 50 nS.
The minimalist approach mentioned earlier showed its consequences, when a propagation delay of an impressive 8 uS was measured at the vertical deflection plates, and around 1.5 uS at the horizontal deflection plates! It was time again for another break, to elaborate a new routing scheme for the X and Y signals.
From the previous analysis, I ended up with two different and both very large propagation delays for each of the X and Y signals (as compared with the measured 50 nS for the Z-axis). How to solve this? It did not seem to be only a routing problem.
I decided to investigate X-Y-Z signal propagation delays in the two units separately. After a thorough measurement of propagation delays inside the HP 182T itself, comparing with the HP 8755C plug-in itself (where the Oscilloclock board was installed), I concluded on two countermeasures:
1. The complete removal of the Processor board XA-6 from the HP 8755C. (This is where the Y-axis signal from the Oscilloclock board had initially been connected.) Instead, this routing was transferred directly into the Normalizer Interface board XA-11 (which interfaces with the HP 182T).
2. Also at the Normalizer Interface board XA-11 inside the HP 8755C, the substitution of two original op amps U9A and U9B (HP #1826-0092) by TL072 op amps, which are faster and have a higher slew rate.
These solutions were enough to align the signal propagation and complete my project!
Dante JS Conti, 8 November 2021
Like what you see?
We do! We love to hear back from Oscilloclock owners, to hear their stories.
Check out our previous posts and the Gallery for info on other unique creations!
These days, just about everyone has an old oscilloscope lying around. You know, an old, dusty, derelict scope handed down from Grandpa (or Grandma). Well, [Paul] had something even better – an old Tektronix 602 X-Y Monitor! Could an Oscilloclock Control Board drive this vintage beauty? Absolutely. Could I make an aesthetically pleasing case? Definitely. How about time sync via WiFi? Stock standard!
Presenting the Oscilloclock Connect:
Here’s what it looks like plugged in to my fabulous old Tektronix 620 monitor:
And why not have a pair of Connects drive a Tek 601 and 602?
The main component of the Connect is, of course, a standard Oscilloclock Control Board. As usual, all 121 parts on Paul’s board were individually mounted and soldered by hand. The board then was programmed and underwent rigorous inspection and testing. Finally, the board was cleaned to remove flux and renegade flecks of solder, and sprayed with HV coating for humidity protection and – arguably more importantly – to give it its glorious sheen.
The case was custom-made and professionally machined right here in Japan from 6mm-thick sheets of pure cast acrylic (not extruded). This is an extremely transparent, hard, high grade acrylic – and Oscilloclocks deserve nothing less!
The case was sprayed with a special acrylic cleaner and static protection solution, before fitting the various components. Naturally, every part was cherry-picked, right down to the three BNC connectors – they needed an aesthetically pleasing colour, but they also had to have a shaft long enough to mount through 6mm-thick acrylic!
Finally, the physical interface! The knob was chosen for its perfect finger-fit and delicate aluminium/black tones, which gently contrast with the rest of the unit.
The Compatibility Crisis
Over the years, many folks have observed that the scope at hand has an “X-Y mode”, and asked if they could just ‘plug in’ an Oscilloclock Control Board. “Is it compatible?” Unfortunately, the response has usually been disappointing.
You see, creating figures and characters with Circle Graphics relies on the scope’s ability to turn the beam on and off at split-second intervals. This feature is called a “Z-axis input”. While many scopes from the 80’s and beyond do sport such an input, there are two common limitations:
Limitation 1: AC-coupled Z-axis inputs
The input is connected to the CRT’s grid or cathode circuit via a capacitor. This is a low-cost, effective way to isolate the (usually) very high negative voltage of the grid circuit from the input.
The problem here is that the capacitor, by its very nature, removes the edges from the pulse. The controller is no longer able to control the beam on/off timing, and you end up with uneven blanking across the segments, as shown in the screenshot at right.
Depending on the values of the capacitor and the surrounding resistors, the symptoms may not be severe. However, the best way to resolve this problem (while still keeping the oscilloscope’s original circuit intact) is to insert an isolated DC blanking amplifier directly in series with the grid (or cathode). See the Kikusui 537 Oscilloclock for an example of this.
LIMITATION 2: INSUFFICIENT BLANKING AMPLIFICATION
Most oscilloscopes tend to require at least +5V on the Z-axis input to noticeably blank the beam. The Connect, however, is only capable of delivering +2.5V. It works just fine if you set the scope’s Intensity control very low, but as you increase intensity, the blanking quickly becomes ineffective.
Below we have a beautiful Japanese YEW (Yokogawa Electric Works) 3667 storage scope. The left shot is misleading due to the camera exposure; the displayed image is actually extremely dim. The right shot shows the same* image with the intensity control increased – the image is bright, but there is no blanking!
* Astute readers will observe that the time is significantly different between the two shots. This is a result of the WiFi NTP sync kicking in right in the middle! More (or less) astute readers may also notice that the scope’s trace rotation is not adjusted very well…
Of course, it would be a simple matter to incorporate a pre-amplifier for the Z-axis, which would solve this problem. This will be introduced with the next Control Board revision!
Like what you see?
Nothing brings more joy than connecting this bundle of usefulness into a woefully unused old oscilloscope or X-Y monitor. If this is of interest to you, visit the Availability page for more information, and of course see the Gallery for other unique creations!
The Oscilloclock Deflection Board currently assumes X and Y input ranges of 0-5V, centred on 2.5V. However, the customer was programming an Arduino-based controller board with analogue output from 0-3.3V. Applying this directly of course didn’t break anything, but sure did make it hard to centre on screen! Would there be a quick way to adjust voltage levels?
Another issue was that the gain in the current-revision Deflection Board is hard-wired, and the image was not the right scale to just fit the screen. The gain could be changed via a single resistor per channel, but would there be an easier, more flexible way?
Recently, I’ve seen quite a few search hits and even an enquiry regarding the 400-LED dual-trace oscilloscope that I briefly mentioned on my History page. With renewed enthusiasm therefore, let’s take a trip down history lane and see what I was doing back in 1990!
A compact dual-trace 1MHz DC scope – what more could a high school kid want?
Last month’s post about the Heathkit Oscilloclock generated tremendous interest, and I’ve heard from several folks keen to try their hand at preserving their own beloved instruments.
… so let’s take a brief look at what was involved in the Heathkit OR-1 conversion!
Approaches to conversion…
There are many approaches to retrofitting a scope into an Oscilloclock, but it really boils down to how much of the original circuit you want to re-use, vs. what you will bypass with Oscilloclock boards.
In Transformer Corner Part 3, I looked at how to choose materials for a custom HV transformer. One way was to pull stuff from the junk-box – I did this in my early Prototype. The much, much better way was to use an off-the-shelf core with documented specs.
Let’s look at winding up the transformer. It’s amazingly easy to get a workable result! Continue reading →
Now, let’s see how I could choose the materials and design the transformer – without any pesky mathematical formulae!
The end goal – a hand-wound HV transfomer!
Picking a core
The first challenge was to find a suitable core from my junk box. First off, recall from Part 1 that this couldn’t be iron (too ‘slow’ for 151 kHz), and it couldn’t be air (too ‘weak’ for 25mA). I suppose I could have tried plastic, milk, or even beer – but I knew better. I knew about a substance called Ferrite.