One of my colleagues...

[(*steve*)]

You may have noted I have occasionally referred to "one of my colleagues" in some of my posts. Unfortunately the person I most often referred to was suffering from some fairly nasty medical conditions which finally resulted in his death. I was charged by him (and later his family) to ensure his vast collection of electronics was put to some use. Alas, he has no relatives interested in electronics.

You may have noted that I am quite fond of purchasing the odd bit of dead test equipment to repair. Ross was more interested in building his tools. He often needed some gadget to make something easier or more efficient. He would often make several prototypes and go to a great deal of effort to get the device "just right". His skills with his lathe and other metal-working tools in his shed were formidable.

Fortunately he also kept lab notes (that I have managed to collect together) and also attached documentation to anything large enough to fit a label on.

I have ended up with a large number of electronic gadgets, some of which are not entirely obvious of purpose. I plan on a post or so on each one as I get around to photographing them.

 

[(*steve*)]

The "Low power load box".

This is fairly typical of the construction and documentation style.

Ross built a number of battery chargers and dischargers; I assume this came in handy while building these.







The base this is built on is steel. It gives it a bit of heft so it sits where you put it!

The standoffs are possibly teflon. Ross did like using teflon for all manner of things, although they also might be nylon.

 

[(*steve*)]

The other "Low power load box".

There were 2 of these, with different value resistors. This one has 2R2 steps rather than 27R.

It looked a little dusty before I cleaned it up. I don't think Ross had used this one in a while.



After cleaning up the top a bit...



Much nicer! And this one has a fuse too.



But no markings that the rear terminals are for your multimeter.

 

[davenn]

he did nice construction work

 

[(*steve*)]

Some "Positive displacement solder dispensers"

I introduced Ross to surface mount components. He was sold pretty much as soon as he realised he didn't have to drill so many holes in his home-made PCBs.

Fairly soon he decided that there had to be a better way of dispensing solder paste than squeezing it out of a syringe. His first designs were positive displacement pumps. This is essentially an automated syringe with a motor driving the plunger. The photo below shows four of his early designs that I managed to locate.



The top two designs are based on motors with worm drives, while the bottom two use stepper motors, the lower one fitted to a reduction gearbox (I think).

They all have a threaded rod and a threaded cylinder inside to convert the rotary motion into a linear motion to push the solder paste out. All of these designs showcase Ross' machining skills and are worthy of further investigation. I can't remember exactly how they were filled, but that may become obvious when I tear them down.

This design was ultimately a dead end as Ross discovered that any trapped air in the solder paste would get compressed and then slowly release more solder paste after the plunger action was stopped. In addition, these were somewhat difficult to fill with solder. You'll also notice that these were designed to use blunt needles, and these also proved to be a problem. These problems were fixed in his later designs.

More to come...

 

[AnalogKid]

Seriously impressive guy ...

 

[BobK]

My fingers are demanding one of those! (They can get pretty sore squeezing those syringes.)

Bob

 

[(*steve*)]

My fingers are demanding one of those! (They can get pretty sore squeezing those syringes.)

Wait until you see the pneumatic dispensers! Whilst he made a number of electronically operated versions, Ross also designed a model that is manually operated, works really well, and can *almost* be made at home from a few cheap Chinese products costing a handful of dollars.

 

[(*steve*)]

I was hoping to get some more detailed photos of those positive displacement solder dispensers, however today got eaten up and so will tomorrow...

Here are some gadgets Ross made to easily dispense small quantities of solder paste from the large syringe-type containers.



On the top and left are the devices for displacing the plunger. On the bottom right are the various adapters for syringes. The one in the middle contains an adapter to allow the receiving syringe to be filled from the bottom to minimise trapping of air. What isn't shown here are the luer-lock to luer-lock adapters that permit filling of a syringe from the narrow end. Also not shown is the adapter for the solder syringes that don't have a shaft on the plunger.

Ross used these adapters initially to fill his positive displacement pumps, and later for the pneumatic dispensers.

 

[(*steve*)]

A "Low Current In-circuit Monitor"

Ross built a couple of these. At this point I'm not sure if there were design improvements or if he just needed more than one.



As usual, the documentation is printed on the outside. The instructions don't specify the power supply voltage, but as Ross was keen on using 12V lead-acid batteries, I think that it's a safe bet.





It's odd that the accuracy falls off at high currents. Why is that?

Power supply and meter connections are on the side (sorry about the focus).



Hmmmm... 0.24Ω resistance is a bit odd. If this contains a current sense resistor, 0.1Ω plus a bit for lead resistance etc would make more sense.

I also noticed there's a bit of a rattle inside...



And look at that extention to the trimpot's shaft. Looks like a quick job on the lathe.

But Surprise! Is that a transformer? An inductor? Is this for AC measurement? It didn't seem that way???

Let's take a closer look:



Aha! Ross has very careful taken a slice out of a ferrite core and inserted a hall-effect sensor.

It stands to reason then, that the current measurements are taken with no common electrical connection to the circuit under test. I imagine that the series inductance is probably fairly high though.

This also explains the drop in accuracy at higher currents. I imagine that the core is saturating and the Hall effect sensor is no longer seeing a linear rise in flux with current. The other explanation might be that there are non-linearities in the hall sensor, but they're generally pretty linear.

The circuit uses

• 1 x LM324
• 1 x 78L05
• 6 x resistors 3 x 10kΩ, 2 x 120kΩ, 1 x 62kΩ
  
• 1 x 10 turn 5kΩ trimpot
  
• 1 x hall effect device (permanently epoxied in place)

I'll disassemble this further, but I suspect that the 5V regulator provides a stable voltage to the Hall sensor, at least some of those resistors are there (in conjunction with 1/4 of the LM324) to provide a 0V rail.

There seems to be no internal calibration, so I assume that the calibration was done by varying the number of turns on the coil (unless there's some very fortuitously selected resistors in there).

An image of this board from the top will not be very revealing, so here are some from an angle





And one from the bottom that is fairly perpendicular.



Uninsulated links on the bottom side! (and there's one on the top side too)

The shaft added to the trimpot is worth a closer look.



I can't be sure if that is teflon tape added to give a tighter fit or to make it turn smoothly against the body of the trimpot. I've tried to gently remove the shaft, but I don't want to damage it. It may be epoxied in place.

I'll have to measure the resistance of this trimpot because the added shaft hides the markings.

And oh dear, that's not a great solder joint there :-(

I don't have time to do the reverse engineering right now, maybe later (later has arrived).



The points labelled A, B, and C are the connection to the hall effect sensor. After labeling everything else up, it's pretty clear they are +5V, gnd, and output.

And here it is:



As suspected, there is no adjustment for gain, so the calibration must have been by adding and subtracting windings from the core until it was close enough!

It's also interesting that R6 and R7 are in parallel, and R5 has a different value. Why wasn't a single resistor used instead of R6/R7 having the same value as R5? There is room on the board for a resistor in parallel with R5. It's a mystery.

It's also worth pointing out that there's obviously something going on with the feedback path for U1. It's all been linked together, but there's a number of holes here. Perhaps Ross was considering some amount of adjustable gain here?

As I was going through Ross' components, I noted that he had taken a number of 1% resistors and matched them. It's quite likely that R1 and R2 here are matched to give an accurate gain without needing adjustment.

Another interesting thing is that this circuit is upside-down, using the output of the hall effect sensor as the reference voltage. As drawn, if an increasing current causes an increase in the output voltage of the sensor, the output voltage would fall. The easy fix is to reverse the sense of the winding (effectively reversing the hall effect sensor). I wonder, however, if this contributes to the reduction in accuracy as current rises?

Time to power it up, and... nothing!

It turns out that the last thing I would normally check was the problem (and in a blow to Murphy, I checked it first). The on/off switch was open circuit. Since I really don't need to turn it off the easiest solution was to just solder both wires to the one terminal.

And it works!

Setting zero is remarkably tricky. Given that my meter reads 0.01mV (10μV) I guess that's not surprising. However the larger problem is drift. As it warms up it drifts a couple of mV and needs quite a bit of adjustment to keep it near zero. After about 10 minutes that settles down too.

How accurate is it? See below. From 10mA to 800mA the accuracy is within 2%. The error at 30mA is possibly due to a failure to check the zero. And the loss of accuracy (possibly due to saturation of the core) starts at about 600mA.



The next thing is to measure the inductance seen by the circuit. (time passes...)

I can't help myself. The inductance of the input is 671μH with a Q of 15 at 1kHz. That's a lot of inductance, so a circuit like this is really only suited to slowly changing DC values.

edit 1: updated values and a better copper side photo
edit 2: annotated copper layer
edit 3: added the schematic.
edit 4: test results
edit 5: I can't help myself (inductance)

 

[(*steve*)]

On with the show... Here is the first of Ross' displacement pumps. I'm pretty sure that this was a very early model because I recognise some of the design issues Ross spoke to me about, and which his later designs addressed.



No real attempt has been made to mount a controller on this.



Common to many of these early designs is a cover to protect the tip (and maybe the operator). In this case the cover is threaded and screws on to the brass tip.



A wingnut attached to a short section of threaded rod serves to hold the tip in place. Once removed you can see (or perhaps you can't) that it presses against a further threaded section of the tip.



Unscrewing the tip reveals the plunger inside the hollow tip. The threaded section is a separate piece. I don't know if this was threaded first then drilled out. The remaining material is quite thin. On further examination, I believe the longer "tip" section only extends a short way into the threaded section (at least, it is clear it doesn't extend all the way through).




The shaft of the tip section is clearly brass tube, and the very end of the tip is some fine machining work. This dispenser has been soldered into the end of the tube.



End on, we can see that Ross has filed down the business end to give it a bit of a bevel. This step was clearly not done on the lathe and I expect it was a modification to enable the dispenser to be positioned closer to the pad when dispensing solder.

When filling this dispenser, Ross would insert a long needle into the brass tube and squeeze out solder so it trapped a minimum amount of air.



I haven't disassembled this end, but the design is reasonably clear.

The stepper motor has a shaft extending out of both ends of the motor and Ross took advantage of this by attaching a "knob" to the rear. This allows the plunger to be manually positioned.

The other shaft is connected to a threaded rod. Inside the main body of the dispenser is the plunger mechanism. One end is threaded to allow it to move up and down, with a rod projecting through a slot to prevent it from turning. A the other end of the internal mechanism the brass "plunger" is attached.

When looking at this design you might wonder why a syringe isn't used -- and since solder paste is available in pre-filled syringes, this seems like a sensible choice. Ross indicated to me when I asked this question that the syringes were quite heavy, that he used only a very small amount of solder at one sitting, and the larger diameter of the syringe body required much finer control of the plunger.

In this particular design Ross discovered a number of other problems, notably that the stepper motor didn't have sufficient torque to reliably push the solder out of the fine tip. In early use, he would add liquid flux to the solder paste to reduce the viscosity. Later designs of this style differ in two main areas, the most obvious being his trials of different motors and mechanisms to increase torque.

Another aspect of later designs is the evolution of the dispenser tip. Ross quickly discovered that the design used in this dispenser was hard to fill, required a lot of cleaning and these were not simple processes.

When I describe some of his later designs I will highlight the steps he took and the further problems he came up against.

 

[(*steve*)]

Here is another version of the solder pump, showing several improvements and also highlighting one of the major flaws.



This version has two major improvements. Firstly it uses a more powerful geared motor; secondly Ross and I had found a source of (and the existence of) blunt needles. Ross had experimented with using regular hypodermic needles, cutting the ends off, however this was less than optimum.

The more geared motor helped with the problem of the stepper motor missing steps while dispensing, and the use of a replaceable (and cheap) blunt needle provided both for easier cleaning and the ability to change dispenser size as required.

Another improvement is the attachment of the plunger "slider" (the part that prevents the plunger from rotating). This is screwed into place in this version, allowing easier removal.

Finally, there is a dispenser switch located near the tip allowing the operator to trigger the dispensing with the hand holding the tool.



This closeup of the tip probably doesn't show anything obvious, but what stands out to me is the silvery colour of the brass rod where it is pushed into the needle. This is obviously solder, and the presence here may appear to be benign, but it is not so elsewhere...



When trying to remove the tip, I found it was retained by the plunger shaft. This shaft should be a close fit inside the dispenser reservoir, but should slide in and out easily. If you look closely where the plunger is inserted into the tip structure, you should be able to make out a small amount of solder paste.

This demonstrates one of the problems with this design. If solder paste is left for long enough it turns solid. It certainly has the propensity to glue pieces of brass to each other. In fact, in this dispenser, it has also glued the needle to the brass shaft.

I suspect that this dispenser was left partially filled with solder paste, and that has completely gummed it up.

Many of those reading this series of posts may have wondered how these pumps prevented solder paste from pushing up between the plunger and the reservoir, and how this didn't cause problems. Well, it does and it did! The size of the dispenser nozzle is significantly larger than the gap around the plunger, and the beads of solder in the paste are surprisingly large. However the flux that suspends the solder (I guess solder paste is technically a slurry) is somewhat free to separate under pressure and move up between the plunger and the reservoir. The flux itself is somewhat volatile and eventually becomes quite sticky. This provides a secondary method of gumming up the pump.

By this stage, Ross was also making the entire pump easier to disassemble, with parts previously epoxied in place held by grub screws, or threaded. This was probably so he could refine the design more easily, but it also makes it far easier to show you the component pieces.



The top section is largely intact because I don't think there's a lot that can't be seen without further disassembly. Let me know if you disagree. You can now see the threaded rod inside the body of the pump. I believe this is a bolt that has been modified by Ross. The switch is epoxied in place. The geared motor is held in place by a couple of other carefully machined parts.

The mounting point for the tip is now itself held in place with a screw. This permits the plunger assembly to be removed entirely for cleaning or redesign.

I was successful removing the blunt needle from the tip. Next to it I have placed another blunt needle with a significantly larger inner diameter (not that it's easy to make out).

However I was not able to remove the plunger from the tip. You may notice a few gouge marks! I have reattached it to the threaded adapter connecting it to the rest of the plunger assembly. The brass rod was epoxied in place -- this epoxy broke when attempting to remove the tip from the body for the previous photos. To the immediate right is the last piece of the plunger assemble. It is threaded at both ends, allowing the turning bolt to translate to a linear motion. Not shown is the teflon slider that prevents this part from rotating.

Immediately below the plunger and tip assembly is a tool for cleaning the reservoir. I had hoped to use this to tap out the plunger, but that isn't possible yet.

At the bottom is the cover for the tip.



The button requires only a slight force to operate. The resistor is a pull (up or down) so the switch connections are always in a relatively low impedance state.



Whilst it may look like the tip has been turned from a single piece of brass, it is at least 3 pieces. Running from end to end is a piece of narrow tube. At the body end, this is inserted into a larger piece of brass that is clamped to the body of the dispenser. At the needle end, a sleeve has been placed over the tube to adapt it to a standard needle.



The adapter to the motor consists of a disk with an offset hole for the body (matching the offset output of the gearbox). The connection is made to existing mounting points on the gearbox with spacers to allow access to the interface between the motor and the threaded rod.



The disk is fixed to the body using a pair of grub screws. The motor is connected to the threaded rod, also held in place with a grub screw.

I believe (but I'm not certain) that Ross turned down the head of some sort of bolt rather than making a threaded rod. If Ross had made the threaded tod, I am almost certain he would have made the adapter from aluminium. Given that there is surface rust on this part, I'm pretty certain a galvanised bolt has been used.

A minor detail is the connection of the wires shown here. Ross was very fond of taking apart cheap DB25 connectors for the male and female solder cup pins. These made excellent gold plated connections. Here you can see three of these assemblies. Covered in shrink-wrap they allow for a simple attachment point.

I don't have a photo of it, but the motor has a diode wired directly across the pins. This indicates that the motor was only ever turned (by the controller) in a single direction. This unit required the operator to manually wind the plunger back into the body after each refill.

 

[(*steve*)]

A "Zener tester"/

This is one of the few odd bits of Ross' gear that has a label, but no instructions!



The banana plugs fit into a meter, the crocodile clips obviously connect to the device under test, and the banana jacks on the left to a high voltage source??? And then press to test? If it's that simple then it doesn't really need instructions...



Can it be as simple as this?



Surely not! Let's look inside...



OK, it's clearly more complex than I first thought. Here's the board:



and a flipped copper side



Let's see if I can reverse engineer this...



Hmmmm... Is that a blocking oscillator? It's odd that Ross takes his high voltage output with respect to the positive rail rather than the negative. I wonder why?

If this runs from 12V, it might draw fairly significant current, and the output voltage is unregulated. Without a zener, I wonder what it would reach, and what current will flow through the zener?

I'm going to have to try this out.

 

[BobK]

Not sure how, but I would guess that the output current stays in a narrow range.

Bob

 

[(*steve*)]

I've just measured the input vs zener current for a selection of zener voltages. The output capacitor is rated for 63V, so I used a string of zeners that added up to about this and measured as around 65V. I had a meter in series with the zeners, so the zener voltage read included the burden voltage of the meter.



Yeah, I used 6V. The first zener I measured was a 27V zener, and seeing the input current rise to 300mA was a little scary. I did test the device at 12V for the 65V zener(s) and the input current more than doubled from 185mA to 425mA.

This circuit is more than capable of exceeding the output capacitor voltage. Also, after using the device, the output voltage falls quite slowly. You probably want to be careful of what you do with the probes while there's still 60V across them.

With these things in mind, and having used this a couple of times, I would add the following improvements to it:

1. A diode in series with the input power. It's currently not protected,
2. Either a much higher voltage output capacitor OR an appropriate zener across the output (OR both) so nothing explodes.
3. Change the SPST switch for a SPDT switch and use the NC contacts to place a resistor across the output.

There are still a couple of mysteries in this circuit.

The first is the origin of the transformer. I suspect it was salvaged from something, and is possibly an output transformer with the primary (in this circuit) being the secondary used to drive a speaker. Given the significant current draw, this would explain the survival of the transformer!

The second mystery is the zener diode protecting the BE junction of the BD139. It appears to be connected around the wrong way if the zener is supposed to protect the transistor by breaking down before the transistor BE junction does. As it is, the reverse voltage is clipped to -0.7V and the forward voltage will never rise much above this either. Did Ross put this in the wrong way? Did he discover it worked better with a lower reverse voltage and used a zener (for no obvious reason)? If I ever find this circuit in his notebooks I will possibly have a clue, but for now it remains unknown.

 

[duke37]

Does zener D1 limit the base voltage swing and so limit the output voltage?

 

[(*steve*)]

Versatile Bipolar Transistor DC Beta Tester

This device was a featured by Ross' family at his funeral. I think it was an excellent selection by them. They have recently passed it on to me and I thought it would be the next thing to show here.



Clearly built to be versatile, this allows the selection from a wide range of base currents (from 4uA to a little over 200mA), and the ability to select collectors loads from about 3Ω to 5kΩ.



The front face allows the connection of transistors with leads in one of several configurations and spacings. Again, this is designed to be powered from a lead acid battery, and Ross kindly details the other equipment you'll need.



More details are provided on one side of the unit.



And the practical operation, including the calculations you need to make, are shown on the other side.



The rear provides access for power and the connection of the meter probes. I believe the meter probes are across the collector load, and you're measuring the voltage across the load.



Here's one of the little adapters for measuring small signal transistors.



It plugs in to the terminal block at the front left and gives you three lead layouts. I didn't find a surface mount adapter, and I don't know if Ross made one. My feeling that he used this for larger devices.



As an example of those "larger devices", he made a TO-3 adapter.




Rather than buying a socket, Ross manufactured his own.



This has the advantage of adding some heatsinking to the device under test.



And here it is, positioned ready for use.



The construction is typically tidy.



And, naturally, it's on a home-made PCB. You can see a number of bare wire jumpers on the copper side. The aluminium strip is a heatsink for the transistor that I believe is used in the constant current source for the base current.

I haven't had a chance to try this out yet, but that's next on my list. Maybe I can compare the results to a number of devices I have which also test transistors.

 

[(*steve*)]

I think I promised I would test a transistor using Ross's tester.

This a BC547 that was lying on my bench



The datasheet tells me that it has a hfe of between 110 and 800 with an Ic of 2mA and a Vce of 5V. Some datasheets go further and give ranges (or typical values) of hfe at other collector currents, but this one didn't. However it did have a graph indicating the relationship between Vce and Ic for various base currents.

While this doesn't show hfe, you could use it to calculate typical hfe at various collector currents for a given Vce.

The first tester I used was this one:



It gave me the pinout (CBE), indicated the HFE is 364 at an Ic of 2.5mA. It also told me that Vbe is 0.76V at 4.64mA.

It's a great tool for a whole number of reasons, but it only gives characteristics at a single point, and sometimes it chooses weird points!

The next test was done using what I believe to be an early version of the component testers available on eBay for a couple of bucks.



This again gave me the pinout, and indicated the hfe is 366. That's all the information this tester presented.

A couple of multimeters gave very similar readings around hfe = 400.



These are all horrid to use. You have to get the leads into tiny holes, and get them to make contact. That last multimeter (a teeny tiny $5 meter) required I hold the transistor in place (aaagh!)

One thing you do get to see on these that you don't on the others is that the hfe is strongly temperature dependent. In all cases, the reading started pretty close to 400 and steadily dropped as the transistor warmed up.

The next test was made using a tester that Ross constructed from an article in the December 1990 issue of Electronics Australia. This has Ross' almost trademark instructions on three sides. The only thing it didn't specify was the operating voltage. Ross has placed two banana jacks on the side. My guess was that this operated from 9V, and that was confirmed from the magazine article. That's one piece of information Ross left off the documentation!



With a full scale of 500, this is reading about 400. I didn't notice it drop, but the movement of the needle would have been only very slight.

Finally I pulled out the unit I described in the previous post.

This takes quite a while to use. In my case, it produced a page of readings (52 points) that I entered into a spreadsheet to perform the math and graph the results.

The first graph is just Ib vs calculated hfe for all points:



The X axis is base current in uA, the Y axis is calculated hfe.

The highest hfe calculated was 384. This occurred at a base current of 54uA and a Vce of about 11V.

With a bit of manipulation you can produce your own version of the graph from the datasheet


This shows the relationship between Vce and Ic for various values of Ib.

With another fairly simple manipulation we can show how hfe varies with Vce and Ib



Or how hfe varies with Ic and Vbe



Depending on what you need to know about your transistor, you can certainly manipulate the readings to see what is going on.

This takes quite a bit of time, and I'm certain I wouldn't want to do it for a batch of transistors!

However I could also use the device, without any calculations, to match transistors for Ic at a given Ib and collector load.

Because the readings are all discrete and manually taken, the graphs don't go to zero. For that you need a curve tracer. And while a curve tracer sounds really cool, it may come at the cost of getting actual numerical results.

 

[(*steve*)]

There were a few things that Ross had in almost every room of his house. A hacksaw, a magnifier, and a magnified headset.

Here is an example of one of his headsets.



It's a normal headset sold by the electronics shops here. They originally had a an incandescent torch bulb powered by a pair of AA cells on each side.

Ross removed the incandescent bulbs and battery holders, replacing them with LEDs and a module to drive them.

I believe this was a relatively early design. It simply uses a $1 buck converter from China and a couple of lithium ion batteries.



This particular model has 4 white LEDs, however there was at least one with red LEDs (used in his darkroom).



Slightly thinning hair required the addition of something to keep the cold metal away...



A closeup of the circuit board. The interesting thing here is how the wires are attached.

Ross found that DB25 connectors were a great source of individual connectors. These wires are soldered to some female sockets and then covered in heatshrink. They then connect quite firmly to PCB stakes.

This circuit doesn't prevent you from over discharging the batteries. Later versions corrected this fault.

Ross also made a number of chargers for lithium batteries. His family kept one of his headsets and I made sure they had the appropriate chargers to go with them. Needless to say, the chargers were all documented so they should have no trouble keeping things working.

I have my own version of this.



My version isn't as nice. I didn't bother to remove the battery holders and lamps from the sides (the covers remove themselves). I also left the sticker on the top, so you can see where it came from. And I used cable ties to hold the LED lamp in place.

The weight of the lamp on mine pulls the headset down. Ross was able to tilt his back with less problems (although having a lot more hair, it doesn't work as well for me). The lamp I have (a whole $5 investment) switches off-high-low-flashing-off, so there's a lot of button pushing when I need to turn it off.

 

[73's de Edd]

1 . . . .A hacksaw, in almost every room of his house.

? ? ?

2. . . . .
There are still a couple of mysteries in this circuit.
The second mystery is the zener diode protecting the BE junction of the BD139.

The zener is there to start clipping the positive node of the feedback sine signal and thereby
starting regulation of the power supplys output voltage at that threshold.