Flat Cap Electronics

JFETs Don't Leak

2023-06-13T03:40:21Z

Ok, to be fair JFETs do leak, just not very much. Also, they leak more with temperature, but all diodes do. Did you know JFETs are diodes?

Disclaimer: All prices are from time of writing. If you're reading this in the future, good luck!

Here's a picture of the process die of a JFET as an example. In an N channel JFET:

The gate is made of P doped silicon, colored dark red.
The channel is N doped silicon, colored grey.
The pads with the letters correspond to JFET terminals, colored gold. Bond wires connect these pads at one end and the lead frame pins at the other.

Note the pair of pads for the gate and the interchangeablity of the source and drain. Looks like a JFET, smells like a JFET. Still, there's a PN junction there between the gate and the source/drain. It's a diode junction just as much as any old two terminal diode, and despite the unintended nature of this arrangement it's perfectly serviceable.
The ratings must be respected of course, but in the end you have a diode good for a few tens of mA, a few tens of Vf, and potentially extremely low reverse leakage.

Once upon a magical time, back in early October of 1980, the Royals were just about beat the snot out of the Yankees for the American League pennant.
In the electronics world, Burr Brown had just published an application bulletin in EDN on the misuse of the above mentioned JFET diode. Their aim was to preserve the super high input impedance in their DiFET® precision opamps but still provide input protection. The DiFET® is a 'dielectrically isolated FET' that was used in the front end differental pair of opamps like the OPA124 and allowed for typical input bias currents of +/- 350fA with a max of 1pA.

Time to Obsess Over a Single Number on a Datasheet #

Let us compare some diodes and get a feel for what leakage currents look like:

Old germanium diodes like the 1N34 are legendary for leakage. Typical datasheet maximum leakage at 10V is 30μA, rising to 500μA at 50V.
A modern 1N4148 signal diode has a datasheet room temperature reverse leakage of about 7nA at 30V, a worst case of 25nA.
The FDH333 used in B&K's 2120 20MHz scope from the late 80s is listed as a max 3nA at it's full rated 125V, with it's fancier brother the FDH300 as max 1nA.
The 2SK30A maximum gate leakage is listed in Toshiba's 1997 datasheet as 1nA at 30V.
The Burr Brown app note's 2N4117A is 1000 times lower at 1pA max at 20V with a 'typical' of 200fA.

It's difficult to get apples to apples comparisons of the performance differences. JFETs aren't really characterised like a protection diode.

One thing that haunts all PN junctions is leakage scaling with temperature. It roughly doubles for every 10C rise, and this will be reflected in the specs for all the really good stuff down in the picoampere range.

At some point there was suitable desire in industry for ultra-low leakage diodes, so the PAD (Pico Amp Diode) series was created and is still sold by InterFET. They're available in rounded E3 steps (1, 2, 5) in three decades (x1, x10, or x100) of picoamperes. Just append your chosen rating after 'PAD' and check the package types as not all ratings are available in all packages. Prices hang around the $7US mark and range up to $20US for the good stuff. Oh, and you'll never guess how they're made . . .

For the SOT-23 package, it is up to the user to short the source and drain pins.

Time for the big reveal. Those PAD diodes use the same die as the InterFET 2N4117A. This is the very same example die I used in the beginning. I guess if it isn't broken you probably shouldn't fix it.

The utility of JFET and PAD protection diodes do have their limits. Another member of the DiFET® family was the legendary INA116, an instrumentation opamp with two guard pins per input and built in +/- 40V input protection. If you've never heard of instrumentation opamps before they're kind of like an opamp with opamps for claws. The main draw is that they usually have amazing specs and you can adjust their gain with a single resistor. These particular beasts were boasting typical input bias currents of +/- 3fA, with a max of +/- 25fA.

Let's take that in for a second. This part was available in the 1980s and was specified to allow, at room temperature, no more than about 500 electrons per second in or out of it's input terminals, and at 85C no more than about 4000.

These are about $25US each in single quantity from Mouser. Sometimes it does feel like we live in the future.

Devices like this have big datasheets. Getting the performance listed in the datasheet requires care and understanding the vagaries of working with extremely low currents. Some of the techniques used to characterize these devices are lovingly explained by the godfather of analog, Bob Pease, in his articles for Electronic Design called 'What's all this Teflon stuff anyhow?' and 'What's all this femtoampere stuff anyhow?'.

If you'd like to see how deep the rabbit hole goes then I'd recommend Analog Device's insane ADA4530. Their datasheets are always divine, and this one is no exception. This is often specified as a 20fA input bias current opamp, but that's over the entire -40C to 85C range at <50% relative humidity. At room temperature typical input bias currents of less than 1fA have been measured.
Most of us don't often use the word attoamps, but those AD folks are next level. One hundred attoamperes is about sixteen electrons per second.
One of these will set you back just over $33US.
I'll finish off with the smug first page graph from the datasheet.

Repairing an Energy Concepts Inc. 30820 Oscilloscope

2023-06-09T05:48:45Z

First things first, I'm going to put the relevant links at the top.

Iwatsu SS-5702 Service Manual locally hosted
2SK30ATM Datasheet via Octopart
uPA61A Datasheet locally hosted
LS844 Datasheet via Linear Systems

Disclaimer: All prices are from time of writing. If you're reading this in the future, good luck!

So I bought another oscilloscope. #

I was fantasizing about getting one of those nice Tektronix 600 series XY displays and poking through Ebay when I decided to try something that I'd done before. After some light swearing I found my way into the test equipment category I was aiming for and sorted by nearest distance. If I can find interesting test gear going for cheap that's within my gas money and time limit then I can sometimes save with a local pickup. There was someone nearby with an older analog 20MHz scope with an XY mode (with Z intensity) so for $10US I put in a bid. Two road trips later I had a new project, so I quickly swept all my old, half finished projects aside.

My first question was 'Who the hell is Energy Concepts Inc.?'
They're a technical education outfit specializing in electronics and industrial controls. Their website (with no HTTPS certificate? come on!) still uses the same logo. Further searching suggested that the scope was a rebranded Iwatsu SS-5702. This makes more sense, as it's easy to imagine them striking a deal and rebranding test gear to put together curriculum and equipment packages for schools and industry.
Iwatsu made good stuff, and they're still around and making test gear. They do have a page for manuals on their website, but at time of writing the links I've tested are broken. The version (un)available there is the SS-5702A, which is notably different and won't help us here.

The first thing I did was zero the controls. Analog scopes are an irresistible invitation to twiddle and click every setting on the machine. Next it's time to pop it open and make sure there's nothing obviously wrong. It was nice to see all the screws in place, that was a good sign. Nice and clean on the inside with no signs of service or trauma. The attenuator wafer switches looked pretty clean, low tarnish. I powered it up and connected a probe to the cal output. Everything seemed good with channel one after exercising some of the switches. Channel two was a different story. Something was heavily attenuating the signal, to the point it was barely visible. After a lot of tinkering it seemed that the signal was heavily attenuated and also acting like there was a parallel capacitance that brought the leading edge down. That's super weird.

I started by searching online to see if anyone else had issues with their scope front end. A fellow tinkerer who goes by ziplock9000 seemed to have had similar issues in late January of 2017. The collected readings suggested that tracing the signal in from the front end would be a good strategy and that I should hold the dual JFET and following buffer IC in particular suspect. Manuals and datasheets had been located but the links had gone off. Still, I had a good jumping off point. Time to check the schematic and see what the front end looks like.

The frontend topology is typical of an analog scope designed in the late 1970s. The input jack is connected to a coupling selector, the attenuator bank, the 'dominant impedance', the JFET buffer, and finally (for our purposes) the input preamp. There's a few other impedance and filtering components sprinkled in, as well as a dirty trick from the before times.

The coupling selector is straightforward, using a 47nF series capacitor for AC coupling and followed by a 22Ω resistor.
The attenuator bank selects four decades of compensated attenuation. We'll be using the lowest 5mV setting, so it's a 4pF capacitor in parallel with the signal and a 10Ω resistor in series.
The dominant impedance is the standard 1MΩ. The network of Rx09 and Cx12 exists to help tune the step response and limit current to Qx01 and Qx02 in case of a fault.
The dirty trick is using a 2SK30A JFET as a low leakage protection diode. I split discussion of this off to a separate post. Suffice it to say that the 2SK30 will clamp fault voltage very close to the -12VA rail, closer than a typical diode drop due to the heavy current limiting of Rx09.
Then we get to the dual JFET buffer, NEC's uPA61AM. For clarity, Rx11 is external to the JFET despite the schematic's implication. The first element in the JFET is acting as a source follower with the second element biasing the first at a constant current. Being a dual JFET the two share a die and this helps (dramatically) mitigate thermal drift. The same intrinsic diode from gate to drain that worked for the 2SK30 will work for positive going voltages here, clamping the input near to the drain voltage.
The signal then passes to the first preamp stage made from a HA1127 transistor array. This also converts the signal from single ended to differential, simplifying things for the CRT vertical driver.

The good news is that there are modern options for replacing a blown 2SK30 in this application. The FDH300 is less than $0.40US in single quantity from Mouser and should work just as well. That dual JFET and the HA1127 are a lot more of a concern.

After some quick poking around it seemed fairly obvious that I was dealing with a dead dual JFET. Desoldering and testing it verified that one side was bad with the gate testing against either source or drain at several thousand ohms. I did notice the use of a guard track around the input node but the entire area (heck, the entire board) was covered in flux residue so I'm not sure how much good that was doing. Further poking around suggested that the other components, particularly the 2SK30A, were likely in good shape. At very least nothing looked shorted or likely to cause further issues.

The search for a replacement begins!

I began by searching for the uPA61AM datasheet, but I couldn't find even a mention of it outside of the usual global sourcing sites except for the discussion of this particular scope. Eventually I found some basic specs from an old databook. Better than nothing, but hard to feel confident specifying a replacement without the gritty details. I contacted ziplock9000 and was graciously gifted their copy of the full datasheet from NEC. Now we're getting somewhere!

So what's critical about the parameters of this JFET buffer anyway? Fortunately JFETs are really good at being high impedance buffers and their notably terrible process variation is usually figured in to mass produced designs.

Packaging is important. I'd rather not try to dead bug a tiny IC if I don't have to.
On die Idss matching need not be stellar as the two sections are doing different things.
Getting the Vgs(off) (pinch off voltage) and Idss of the replacement close to the original will avoid the necessity of adjusting the circuit to bias properly.
On die thermal tracking is important, though I think the worst JFETs ever to share a die would proly track well enough for this job.
Transconductance is nice but it's all going in to linearization. Most any JFET would have 'enough'.
Bandwidth could be an issue so aiming for low input capacitance is important.
Noise in JFETs is usually pretty amazing but it's worth aiming for the same ballpark.

Time to see what's left in the world for dual JFETs. I used the venerable technique of hopping in Mouser, searching for JFET, dual, in stock, TO-71 or TO-78 case style, and sorting by price. This gave me twenty or so choices, which I must admit surprised the hell out of me.

A quick note about pinout. Dual JFETs in a can like the TO-71 (small) and TO-78 (big, same as uPA61A) have a standardized pinout. The datasheet of the uPA61A lays out what it looks like for an 8 pin version. Pin 8 is always at the index tab of the can, and, looking at the bottom of the can, pin 1 is the next pin clockwise from eight. Pins eight and four are used for substrate or case connections (substrate in the uPA61A). Six pin versions which lack pins 4 and 8 will omit them but they keep the pin spacing. Pin 1 remains in the same place, just clockwise of the tab.
This means that you must pay a bit of attention when replacing an 8 pin part with a 6 pin version. It should line up fairly easily, just don't get things rotated. If there's a silkscreen indicator of the tab on the can it'll be correct regardless of which version you use.
If all this is a bit confusing see the picture of the TO-71 6 pin installed below.

After poking around I had a strong candidate in Linear Integrated Systems' LS844. I guess LIS had started sourcing to Mouser recently and they were nearly half the price of the InterFET parts. They had better datasheets too.

Pinout was standard, minus the substrate pins (not important).
Idss and pinchoff ranged a bit higher.
Transconductance was very close.
Ciss was a bit higher, but should be manageable.
Noise was hard to read but looked very close. Given it improves with drain current and the replacement was likely to operate at a bit higher current it might even improve.
It's cheap at about $12US.

I thought I might make myself feel better to get an idea of what the bias currents might do, so I whipped up a quick sim of the buffer in Paul Falstad's circuit simulator. If you'd like to follow along here's a link. It's great for quick stuff but the models are generalized and not intended for great detail or accuracy.

Now I'm no engineer and I'm a pretty terrible circuit designer. I'm not one to get out the formulae and start making load lines unless there's no other choice. This is very much a repair technician's approach to the problem.

I gave a range that covered the worst case of both pinch off voltage and 'beta'. The beta calculation was sort of a hack of a hack as it was intended for BJTs, had a hack for MOSFETs, and I was trying to use it with JFETs. I found there was plenty of range in the DC BAL control to adjust for any combination. Worst case Id ranged from roughly 18mA down to a few μA but both extremes are pretty unlikely. The M variant of the uPA61A is binned for 2mA to 5mA Idss, and the LS844 was spec'ed for 1.5mA to 15mA with an average of 5mA.

I'm not giving a great measure of confidence to this simulation, but it calms my anxieties well in to 'it's worth a shot for $20 shipped' territory.

Naturally I placed an order. Mouser only had the smaller TO-71 in stock, which was fine for me, but Digikey had both. Digikey, generally speaking, tends to have a wider variety of parts and package types available. I stick with Mouser when I can as they were good to me when I was just a weird kid who wanted catalogs. Soon I had an LS844-TO-71-6L-BK in hand.

Rather than wait to measure anything I just soldered the thing in right away like a kid on Christmas. Not my proudest moment, but I made up for it by collecting some data. The first thing I noticed was that the coupling switch was acting dirty again, so that got another blast with quick dry contact cleaner and some exercise. Things were looking good right away after a bit of tinkering so I set about making adjustments.

For clarity's sake, channel 1 remained stock and the new part went in to channel 2. Henceforth I'll simply refer to each channel by number.

Not wanting to embark on a full suite of adjustment just yet I headed for the '-Vertical Deflection System-' section, page 42 in the pdf and 6-5 in the manual. I went through sections eight through eleven and that did a good job of nulling out the difference between the two channels so I stopped there. The x5 magnification control was notably less noisy on channel 2, so that was a welcome change. With everything looking good I fired up my signal generator and tested bandwidth. Both channels were well within eyeball range of a -3dB bandwidth of 20MHz, with channel 2 very slightly less attenuated. I'm unsure what parts actually limit the bandwidth though, so it could easily be variation in other parts accounting for the difference.

Now for the data. First I measured Rx11 on each channel. It's the 100Ω resistor visible just to the left of the dual JFET in the picture above. It's easy to reach and a nice round number so I can use it to find the operating current of the whole string.

	Old Channel 1	New Channel 2
Rx11 resistance	97.3	95.5
Voltage across Rx11	0.179	0.304
Id	1.840	3.183

Ok, so the new setup uses about 73% more current, that's no problem.
How about temperature? The new one feels a little warmer to my finger, but it might be the smaller package size. I'll use my standard method of mashing the junction bead of my multimeter thermocouple against the case with a pencil eraser and waiting until the reading stops changing.

(All temps in °C, in case you might think it a bit cold)

Minutes Elapsed	Old C1	New C2	Ambient
5	30.6	35.0	26.0
10	31.5	36.8	25.4
15	32.2	37.1	25.6
20	32.0	36.1	25.4
25	32.1	36.0	25.5
30	32.0	36.4	25.3

Four or five degrees difference doesn't leave me with any concerns as to long term longevity.

So it looks like problem solved. I'm glad there turned out to be a fairly cheap solution. If anything more pops up that's specific to this issue I'll add it on to this page. I'll be setting up a Device Notes page for the scope as well to be a more general repository.

Kicad Libraries

2021-08-07T05:48:45Z

Just a quick note.
Many years ago, in late May of 2015, I decided to use a new-ish feature of Kicad and set up a repository for my own libraries. I'd never mastered git (and I still haven't) so the repo just languished all this time. It may never see glory, but I've updated it with the most recent local backup I have and with any luck I'll be organizing and documenting things in the near future.

Here's the link if you think you'd find it useful.

User Symbols #

These are the schematic symbols for use in EESchema.
The FCE libraries contain my interperetation of common parts, just reduced in size where possible and in an (I hope) consistent style.
The library named with my amateur radio callsign is a collection of random stuff I've made footprints for over the years.
The accompanying *.dcm files contain the footprint metadata for the respective libraries.

User footprints #

These are the corresponding footprints for Pcbnew.
I believe all the footprints here match to provided symbols, but I need to verify this.

User 3D Models #

This directory is blank in my local copy so doesn't show up in git. I may try to practice Blender by generating some models of my own but it's likely that I'll be referencing what I have to existing models in the Kicad standard library for simplicity

User Templates #

This is here for my benefit. I can start a project with all the library and layout preferences set up how I like. Feel free to use it as an example.

I'd like to come up with a script to put all the footprint names in a text file, or maybe the readme, so they're easily keyword searchable. I'll come back and add that in here in the future if I can. It'll make finding these easier for people who might not want to make these footprints themselves.

Electromagnetic Microphone Part One

2018-07-04T12:00:00Z

Did you know that there is a wonderful world of annoying noise all around you?

What you hear in that video are oscillating electromagnetic fields. In some cases, like motor drive, they’re current pulses driving something at an audio frequency directly. Some noises are bursts of data happening so fast that the audible noise is caused by thousands of packets per second, each packet containing a large amount of it’s own ones and zeroes. A whole bus several wires wide might be a roaring waterfall of apparent stochastic noise, but the results are quite deterministic. Some of it, well, I have no idea. EMC is a whole industry with it’s own tricks and experts and ways-of-knowing.

What a mess! Government standards abound for reducing this noise to bearable levels, and it’s a good thing. Some of us remember losing TV reception when a nearby blender was making margaritas. If your aim is to sense this noise, for troubleshooting or for art, you can beat the system with the right sensor and enough amplification.

It’s worth knowing that the widgets you saw in the above video were a model of Elektrosluch made by the folks at LOM. They’ve been refining the design for that widget for at least four years and if you’d rather have something already built and proven then I’d suggest checking out their site. They’re a small batch manufacturer, and that takes a special kind of courage.

The Right Sensor #

So you want to listen to EM noise? You’re going to need an antenna. You want to do it at audio frequencies? You’re going to need an inductor.

Luckily, and rarely for electronics, the specs for our application are extremely loose. Our application is ‘wideband antenna’, and so we have very few requirements:

Open shape to facilitate wide angle reception – bar core, drum core – avoid toroid, assembled transformer
Ferrite material – high inductance per volume vs. laminated and powdered iron
Enough Inductance to provide good signal – more than 1uH, aim for 25-100uH

This will get the job done, but there’s lots of room for improvement and experimentation.

This is a good chance to salvage some old junk. Bar and drum core inductors are both found in high current power supplies like those used in desktop computers. They’re used to suppress noise on the low voltage output rails, and because there are very high currents involved they tend to use beefy wire. You can get a good deal more inductance by removing the thick wire and wrapping them with thinner lacquered magnet wire. Any comfortable size wire to work with will do. You can disassemble the switching transformer and use what’s in there. Try to keep it between ten and one hundred turns. Getting crazy with hundreds of turns of guitar pickup wire would add a lot of capacitance, though I’m not sure even that would matter.

Similarly, you could poach the bar antenna from an old radio. You could try using it as is with the existing Litz wire, or remove the Litz (spool it up for later use?) and wrap magnet wire directly on the bar. Ferrite bar stock can be scored or notched with a file (even a nail file) and it will break somewhat cleanly. This would allow several small inductors to be made out of one bar.

High frequency transformers like the switching transformers in those computer power supplies use ferrite as well. careful disassembly of the core will reveal a pair of E shaped pieces of ferrite that can be used as is. The thinner parts can also be snapped off to provide a short bar of ferrite suitable as a core.

If you don’t have an electronics boneyard to pick through or you just want an easy solution then it’s the internet to the rescue. Any through hole (not surface mount) 100uH drum core inductor should work, current rating and dimensions are unimportant. They look like this. A quick trip to our favorite auction site with the keywords “100uH inductor” shows several places to get a ten pack of what we need for less than $5US shipped, though a closer seller may charge more.

That’s your antenna done. If you have your own source of amplification like a microphone preamp then you can wire up your antenna as per the following drawing:

There’s no reference to ground required anywhere in the system, so this device is inherently balanced and can happily be wired directly to the differential inputs of a standard XLR or 1/4″ TRS jack. Don’t worry about trying to shield every last millimeter of the hot and cold wires, it’s not critical.

At first I thought there might be a danger of a phantom power system damaging itself on this, as the inductor presents a DC short across the + and – pins of the XLR. I’d forgotten that both pins are energized with phantom power and that the shield was the current return. This means that care must be taken with exposed connections at the inductor end of the cable if there is a chance of accidental activation of phantom power. If active it would cause 48VDC to show up on both pins of the inductor relative to the shield, which is often connected to a chassis or mains earth. A hand on the inductor and a hand on a metal case would make a circuit out of you. For safety’s sake it is critical to take care when experimenting, and if you’re satisfied that you have a working device then take care to heat shrink and insulate your connections properly. In the event that your amplifying device has no phantom power then you have nothing to worry about, though proper insulation is still recommended.

Finally, should you be lacking your own amplification, never fear. I’m working on a part two where I’ll make a simple circuit to bring the signal from our antenna up to line level.

Edit (2020AUG26): Part two was lost in the great cataclysm of 2020. I'll be rewriting it, but I've learned not to promise. Interested? Bother me with an email! Nothing motivates more quickly than knowing someone cares.

Meet the Zenith H845

2016-09-27T12:00:00Z

Note - This project writeup was undertaken in mid-May of 2021 and has been backdated to match the timestamps of the pictures documenting this work.

* See the Device Notes page for service info and picture gallery links.

Today we tell the story of a tabletop AM/FM radio from the mid-1960s. This guy was in good shape aside a fair bit of dust and hadn't seen service in the past. There are quite a few other tales of H845 repair out on the internet and for each story told I imagine a great many repairs are done in silence. It's good to see that a lot of these will be joining us in the future.
I have a soft spot for this one, I must say. It's evident that someone wanted to make an excellent radio and wasn't nipped too hard by bean counters. The tuning system has three permeability tuned inductors, there are four different negative tempco caps, the filaments are series string but there's a transformer for the two dial lights. You know, interesting stuff.

Cleaning #

This guy's looking pretty good. There is a gentle gradient of dust deposited on the speaker cones that came away with the gentle application of an acid brush and a vacuum. The chassis looked remarkable after a clean up with little or no visible corrosion aside some of the screws. The bottom plate kept the chassis interior completely dust free. Other writeups on the internet seem to be similar.
The dial cords didn't show any sign of slipping, but just to be sure I gave them a touch of wax. The rubber grommets and insulators on the tuning deck were still in fair shape, which is good as some would be challenging to replace.
Every so often I find strange things in a radio. I imagine that often it's the work of tiny people and their natural tendency to perform their generation's equivalent of putting a sandwich in the VCR. In this case I found a Christmas card sticker and a business card.

Tubes #

The tubes were checked to ensure they were appropriate and tested for basic emission and shorts. There were a few original Zenith tubes in there and a few replacements but all was as it should be.

Capacitors #

Capacitor replacement was straightforward. I decided to restuff the electrolytic can after seeing how it was mounted. Two of the four mounting tabs were soldered and two were installed dry. This allowed me to straighten the tabs holding the unsoldered sides and use a paint scraper and a hammer to shear the two tabs that were soldered in. This saves the time and potential collateral damage of fight to free the two soldered sides. The can is adequately attached by the two remaining tabs for all purposes short of rough shipping, which I can rule out in the case of this specific radio.
I still haven't mastered opening can caps without deforming the open end slightly. I think a jig would help and it's on my list if I'm faced with more than the occasional one off. The contents were allowed to finish drying and the can was gutted. I gave some thought to using the existing internal paper anchors with some heat shrink insulation but found that space was too tight. Care was taken to ensure that the terminals would connect internally to the proper values as this is an easy mistake to make, as is flipping the polarity accidentally.
The original common line for the can had been it's attachment to the chassis via the case and I had to do something to reestablish a solid connection. I could have soldered the remaining two tabs to the chassis, but rather than bust out the soldering gun I decided to try something else. I connected the internal negative common line to a piece of solid core wire scavenged from 14ga Romex. This was brought out of the former vent hole and firmly attached to a nearby terminal strip's chassis lead. I've a fondness for using NOS surplus PCB mount film caps as I've quite the stock and they're only a bit less convenient than their axial counterparts, so several were substituted where voltage ratings allowed in order to save the higher spec stuff for when it's necessary.

Power system #

The selenium rectifier was replaced with a single 1N4007 internal to the chassis. The resulting voltage was close enough to 140V to make me happy. The selenium rectifiers are so neat looking, but it only takes one smoking up a room for the smell to guarantee you never trust one again. Given the diode is hooked directly up to the unfiltered mains all the time it's wise to avoid using a diode with a lower voltage rating. Mains voltage quality can vary dramatically from place to place and is in most places subject to sharp transients in the hundreds of volts. A 1N4003 rated for 200V is more susceptible to failing short due to junction erosion from such transients than a 1N4007 rated at 1kV.
There's a dual ceramic capacitor (Sam's designated C43A and C43B) from each mains lead to the chassis that needed replacement. I used a pair of X1Y1 safety disc ceramics. The industrial rating may help protect that 1N4007 in the long run.
Fuses are a must for me. The simple solution was to fuse the incoming hot lead. I don't remember the value used, but it was likely 0.6A. The Sam's indicates 34W at 117VAC so I'd feel ok about anything in the 0.6A to 1A range.
*Note - The owner hasn't reported nuisance fuse breaks in the intervening five-ish years, so that's a sign that it wasn't too low a value.
Checking around for other safety issues I found something odd. The case of the 'Loudness'/power switch and the shaft of the band select switch are both bypassed by a 10n 1.6kV ceramic disc capacitor. The shaft is bypassed to ground, presumably to prevent it's acting like an antenna for RF. The case of the volume pot is bypassed to the low side of the mains directly, prior to the power switch. The case is otherwise floating, and the pot mounting and shaft are both in direct contact with the chassis. I verified that this is both in the Sam's schematic and present as original configuration in the radio. If anyone can shed light on this design choice please let me know.
The chassis has a bottom plate with plastic feet that in turn screw into the wood case. After some reflection it seems that they've made an early attempt at ensuring there's no metal that can be contacted on the exterior that isn't at very least at chassis potential, right down to the screws on the bottom. In their lust for self destruction a suitably motivated toddler might pry a knob off, but that's as close as they'll get. The back panel has a mains interlock and a polarized plug.
I'm normally not much of a fan of low side switching for mains, particularly without a safety earth. In this case, given the excellent (for the time) isolation and the owner's state of affairs I can live with leaving it as built.

LED supply #

There are many ways to approach LED conversion. In this case the owner wanted warm light with some provision for adjustment. I think I used an LM317T set up in constant current mode. The 330Ω fixed resistor is limiting the max current, the trimmer is from the ADJ pin to the OUT pin, and the diode and capacitor are used to keep things on the DC. I may have an opportunity to verify this at a later date.
There is a photo of two pieces of doorbell wire in the album. I found these looking for wire to hook up the LED lamps. Despite being friends all these years it seems that one color of the insulation has provoked tarnishing in the copper where the other hasn't. Curious.

LED casting #

The dial lamps on this radio are shrouded in a reflective cardboard that's designed to press fit in to round plastic holes in the front face. The light emitted is diffused and piped to illuminate the vertical tuning dial. I aimed for a mounting solution that was simple to replicate and entirely removable. I poked the leads of a pair of warm white LEDs into electrical tape and secured them at the bottom of an appropriately sized socket with the leads sticking out of the drive hole. I then added hot glue to the top of the socket and let it cool. The resulting 'bulb' was trimmed with a hobby knife to fit securely into the holes in the front face. I've used this method before where mounting LEDs in place of push fit lamps and it seems to work well. The LED leads were connected using Dupont connectors as shown in the supply test, allowing for easy removal and replacement if a different color is desired. Some tinting of the hot glue can be achieved, and foil tape (as in HVAC applications) can be used to guide reflection where necessary.

Conclusion #

This was a pretty straightforward recap and mod job. The owner reports good function and excellent weak signal reception from the first floor with a consumer 300Ω dipole. I had a chance to use some of my favorite weird solutions in this one. I really enjoyed exploring the weird tuning mechanism and design choices on this guy. I think there was a team out there determined to make a respectable consumer tube radio and show off a bit before transistors took over, and I think they did a great job.

Leader LBO-301 Oscilloscope

2015-09-17T12:00:00Z

A fun little guy.

The Leader LBO-301 Oscilloscope is a compact 3-inch scope featuring high sensitivity and wide bandwidth. It is engineered specially for servicing color TV and computer circuitry in the shop and in the field.

10mV/div max sensitivity
7MHz max bandwidth
Single channel with SO-239 UHF connector
Vectorscope mode!
Sawtooth sweep output!
Intensity modulation input!
Worldwide voltage friendly
Only slightly longer than an average toaster

So you see one of these sitting at your local hamfest/dumpster and you wonder, is it really worth bending down to pick it up?

Of course it is, but there are good reasons why.

If you’re into RF or building your own test gear then this guy is super neat. The sawtooth output and intensity input are perfect for turning this into the display for a spectrum analyzer. The sawtooth output is used to sweep an oscillator over a known frequency range in sync with the horizontal travel of the trace, and the intensity input can be used to create a marker.

The sawtooth signal can be scaled and fed into a varactor to sweep . . . well . . . anything, but the sweep oscillator of a spectrum analyzer comes to mind. Ten volts peak to peak is a nice chunky start, and it’s constant over all the ranges. All sorts of instruments have been built to make use of this signal but nowadays the trend is to trigger the instrument rather than have the instrument send a trigger. We adapt and overcome, but in this case it’s super easy and just sitting there waiting to be put to use.

Markers can be generated with intensity rather easily, either by a simple gate ala the HP3336 or a more complex circuit that will give you a bright dot on your trace. Multiple markers can occupy the same screen at different brightness levels, or made to blink or throb. Intensity voltage control will also allow you to render a greyscale raster, so you can watch an analog TV signal in monochrome on your three inch oscilloscope!

22SEP2015 Service Manual 5.6MiB PDF

Pictures to come as time permits.

Edit (2020AUG26): This fella has gone to live with a friend and may not return. I think my chance to take pictures has passed unless it returns to me someday.

Better than Skittles

2015-08-25T12:00:00Z

Do you know Morse code?

Wish I did. I’ve never put in the effort, but it’d be wicked handy. The gui on the things I build for myself would be a lot easier to code.

ASCII perhaps?

I always forget where to start the alphabet. I’m sure it’d stick if I did more firmware work. I also still mess up Commodore ASCII and ANSI ASCII, so punctuation is likely forever scrambled. I still sometimes use shift+2 for double quote.

How about resistor color codes?

Now you’re talking. After a seemingly interminable beating my mind did finally ossify around the simple table of numbers and colors that rules vintage electronics. I’ve used this to every advantage I’ve been able. Once you know an alternate representation of things the world takes on a different dimension. Knowing the resistor color code makes flags in particular a new experience. The temptation is to put the code in service elsewhere in your world where it’s strengths are valuable. The reality is that I’ll have to put a lot of energy into making this practice efficient, and that kind of ruins the point.

The resistor color code is nearing a century of age now. Care went in to it’s creation, as with most standards, but there are some more modern difficulties that complicate it’s widespread adoption as a general numeric code.

Start with black. Not a good place for zero. Many substrates are black already, especially after sufficient heating. A charred resistor may say that it’s resistance is zero . . . zero, times ten to the power of zero, but it’s usually closer to infinite. I’ll let the pedantic reflect on that assertion.

Brown is very close to black, sometimes deceptively. When heat attacks warm colors (heh) they tend to brown like a turkey, which complicates things. Also, have you ever seen a brown LED? Emissive brown is sort of a dull shade of orange/red. It’s a bit tricky to pull off without contrast or context, at least while avoiding ambiguity with it’s more saturated neighbors. This makes it ok on a flat panel TV when you’re obviously looking at trees and sky, but crummy when you want a three color LED to emit red, orange, brown, and yellow as distinct colors.

The bulk of the colors following are without serious issue. Let’s skip to the troublemakers at the end.

Grey? Eight is my favorite number too. Again we have a toasted white, and though a grey LED might be a bit easier to work around than brown it still has it’s issues.

Finally we have white at nine. If we are guaranteed to have a substrate that is neither black, nor white, nor brown, nor burnt, nor any other of the above colors, then we’ve avoided ambiguity. Due to white being a common substrate it’s a bit rare for marking media to have a ‘white’ color. Chalk, crayons, and paint (white out?) come to mind.

There are workarounds for some of these. Substrate colors like black and white could be outlined in their alternate for emphasis. Use of reflective substrates rather than emission helps. We also need a marking medium that can mark in all of the above colors. What a pain.

Despite all these disadvantages there are times when I find it useful to deploy the resistor color code as my marking of choice. Properly colored shrink wrap can be cut into bands to permanently number wire. There are products out there designed to treat to this, but those I’ve seen that are cheap are also poor fits for the task, i.e. clumsy with multi-digit numbers.

Often I don’t need to be able to relate a full decade of numeracy. Eight bits will get you to violet if you start at zero like a sane person, and that frees up grey and white allowing light colored substrates. That solves an awful lot of problems right there, and if you’re willing to work in octal you’re in luck.

Sometimes I have to mark something much smaller or more awkward than I could write a simple number on. In these cases it’s useful to be able to simply tag something with a color. Connector header pins and wires pulled from old ribbon cable are favorites. I’ve color coded my Bus Pirate pins in this manner, for example.

Enter the Sharpie. Now in more colors and sizes than ever, these permanent markers easily clean off of most solids with isopropyl alcohol. Long had I yearned for the mythical brown sharpie, thinking to myself that nobody would bother making a sharpie in brown. That would at least get me my number one, and I had a red one for two. Perhaps I could learn trinary? I have green and blue as well, but without those warm colors I’d be just making things up as I went along.

Then I saw this . . .

Available for a pittance at my local Wide Wide World of Walter. I looked online and found that the simplest way to find these is to search for ‘sharpie 8 count’.

Now I depart to see how useful these can be. I’ll try to visit back on occasion and add ideas as they arrive. If you have any ideas for me then please do send me an email and if they’re sane I’ll add them in. I intend to leave comments off until enough people badger me and make it worth fighting the spam.

Edit (2020AUG26): Still haven't used these in anger except to color code my resistor drawers. They will likely prove useful when the time comes to organize the house ethernet wiring. In encouraging others to pick up the resistor color code the Greek flag has proven childishly comic.

Getting Elsie running under Linux Mint 17 64bit

2014-10-29T12:00:00Z

This seems like a decent way to start out the bench notebook. I beat my head against something for several hours over three days and figured it out, and hopefully reasonable documentation and communication will prevail and spare someone else the grief.

Elsie is an excellent RF filter design program for Windows available over at Tonne Software. This is a go-to program for amateur radio folks and professionals working with analog RF filters.

Like most relatively undemanding windows programs I assumed it’d work just fine under Wine. I was only partly incorrect. I loaded up an existing filter and started plugging at all the tabs. Attempting a plot resulted in a crash.

Unhandled exception: page fault on read access to 0x00000000 in 32-bit code (0x7d8e17ff).

Ah crap. It may be advisable to start at the beginning in most things, but this appears to be . . . an exception.

Squinting at the backtrace like I knew what I was doing turned up some familiar faces. I recognised some of these DLLs, so it made sense that the last one before the page fault was my starting point. I’d never heard of windowscodecs before.

This sent me looking through Winetricks. Sure enough, windowscodecs is in there, but attempting to install it resulted in another crash.

Note: command 'taskset -c 0 wine wic_x86_enu.exe /passive' returned status 67.

Hmm. Poking around these bugs looking for ideas resulted in lots of ‘bug ignored as uninteresting’ style posts and dead ends. Advice in one caught my attention. Someone was insisting that a troubled party start a clean 32bit prefix before bug reporting.

What's a 'Prefix'? #

Wine uses the concept of a prefix to contain all the junk of a standard windows install like programs, registry, drive contents, and settings. The default is in a hidden folder in your home directory called .wine and may be revealed in Nautilus with Control-H.

I don’t have much in my wine prefix myself, and while it’s possible to have multiple prefixes I can’t say I have a good reason to struggle with sorting it out. I simply deleted my existing prefix. Renaming it might be a better idea for those with something to lose.

After that I opened a terminal window and ran the command I found in WineHQ's documentation.

WINEARCH=win32 WINEPREFIX=/path/to/wineprefix winecfg

Sweet. Now to try windowscodecs again. Open winetricks and:

Select the default wineprefix
Install a Windows DLL or component
windowscodecs

Installed! Ok, now to install Elsie and try a plot.

It works! I’ll be damned. That caught me by surprise.

I suspect the problem might be fixed in a 64bit prefix with the manual install of windowscodecs, but that’s a bit over my head.

I have a report from an Arch user that this method proved ineffective using Arch Linux x86_64, kernel version 3.17.1-1, and Wine 1.7.29. User reports a runtime error when attempting to install windowscodecs from winetricks. I couldn’t verify they were using a clean wine prefix, and they didn’t want to persist in troubleshooting as they already had working what they needed. Can’t fault them there.

If anyone has more to contribute please let me know and I’ll amend this as needed.