Making woodwork tools

This post may seem unorganized. It is not, though. It is ordered by time. It shows the process of converting two pieces of maple into a spokeshave, a mallet and a marking gauge.

First the pieces of maple needed to be shaped.

Rip sawing is hard work. I needed a small break.
Roughing out the mallet head shape
Some more cleanup of the mallet head
New tool. Significant improvement.

I could not finish the mallet yet, because I did not have a drill bit of the size needed for the handle hole.

I continued working on a spokeshave. I decided to try to make the plane iron myself. Technically, this is not woodwork of course. But there is a strong relationship with woodwork. And, more importantly, it fit my goal to make things I never made before.

The plane iron is made of DIN 1.2510 steel (equal to O-1 tool steel). I cut it from flat stock using a hacksaw. A cutout was made by drilling several holes and filing with a flat file (no round file available at that point). Finally, I ground the bevel using the DIY belt grinder.

I heat treated the steel in my small heat treatment oven.

  • Preheat to 600 °C
  • Heat up to 815 °C
  • Soak for 15 minutes
  • Quench in peanut oil
  • Scratch test (the top layer always fails because of decarburization, but below that ‘soft’ surface, there is a glass hard piece of steel)
  • Temper twice for 2 hours (kitchen oven)

I first tempered at 195 °C, but after spending a long time with sandpaper there was too little progress. Besides, the edge seemed to chip, so the steel was clearly too hard. I retempered at 240 °C and according to the curve, this should result in a hardness around 59-60 HRC.

DIY spokeshave iron while cooling down after last temper.

I followed Rex Krueger’s example to shape the spokeshave using saw, chisel, knife and sandpaper. And I used a bit brace to drill the hole. It appears to be easier to drill at a right angle with a bit brace (plus visual aid) than with an electric drill.

Shaping the spokeshave

In the meantime I decided to try saw filing. All saws had case hardened teeth, so I took a worn out Bahco saw that had several broken teeth. I tempered it at 300 °C to reduce the hardness of the teeth to approximately ‘spring steel’ and ground off the teeth. The original teeth had a strange shape, at least not the 60° angle needed to use a saw file.

Suggestion: use the belt grinder to achieve this (instead of a flat file), if you wish to see a nice fireworks show.

The spokeshave was finished with boiled linseed oil and wax. I felt it was a pity to remove the nice blue oxide layer of the plane iron, so I only worked on the essential areas: flattened the back and sharpened the cutting edge. This matches well with the overall rustic look of the tool.

Finished spokeshave.

I decided to try to make the marking gauge of an offcut, a semicircle section of maple tree trunk. So the end grain will be the marking gauge face. I had no idea how it will work out. But I liked the idea of trying to flatten the end grain freehand.

Please don’t look for what I was not looking for. I did not select this wood because of its beauty. This tree stood in our backyard. It shared part of its life history with me, which is why it has much more value for me than an anonymous piece of wood, beautiful though it be.

Planed end grain. Very smooth to the touch, even though you can see slight planing marks.

The saw was retoothed as a rip saw. First I destroyed a Stanley triangular file. Each corner of that file was dull after filing less then 10 teeth. Then I switched to a Bahco saw file. One corner of that file lasted the whole saw. So there seems to be a significant quality difference between these files, or maybe the corners of the Stanley file were not suited to file a saw. I cannot check this, because the corners are now flat…

So far the saw performs well. I sawed small pieces of fir, maple and multiplex with it.

Retoothed and set saw

Then I switched back to the marking gauge. The crack now extends to the square mortise, but it does not prevent the tool from working. It is better to have this marking gauge than to have no marking gauge at all.

The cutter was hardened the same time as the spokeshave chisel, but not retempered, so it is extremely hard. I ground the tip round on the belt grinder and finished it on sharpening stones. Maybe I’ll wind some thread around the slot that holds the cutter, for a stiffer connection.

I finished the body with boiled linseed oil, but kept the sticks and mortise bare, in order not to reduce the friction.

Almost finished marking gauge

Then I moved to a workbench. I decided to try to build a Roman workbench, like Christopher Schwarz (Lost Art Press) describes in his excellent book. By the way, this workbench project fits perfectly in this blog post about tools, since in my opinion a workbench is a tool as well: a device used to carry out a particular function. If you think an object cannot be a tool if you don’t hold it, my answer is that I plan to grip this workbench with my legs often enough.

I am making this workbench from fine sawn Douglas fir. It is a great planing exercise. And I slightly underestimate the physical effort involved, working on the floor (I am making a workbench because I don’t have one). It is a perfect upper-body workout.

Wood for Roman workbench: Douglas fir, naturally marked with Roman numbers
Another view of the same wood. Next to it some oak, dust collection systems and a worm composting bin
Waste bin filled with shavings

I glued the benchtop together using as many clamps as possible. Because of the size, the number of available clamps became less and less. For the last piece of wood, there were only two bar clamps and two quick clamps.

I did the glue-up in the house, because it was close to freezing in the ‘workshop’. Below the minimum temperature for Bison wood glue. The low relative humidity caused the wood to shrink rapidly, especially at the exposed end grain. After two days here were several split glue joints at the end of the benchtop. I did not care much about that. But there were also splits along the length of the benchtop… Fortunately, when drilling the holes, I found that these splits were only superficial (a few millimeters deep), so the joints should still be plenty strong. I’m planning to store the bench in the ‘workshop’ now to dry gently and take it only in the house for as long as needed.

Glued together, still 1 to go (on the right)

I ripped two 5×10 cm offcuts into four 5×5 cm pieces to make the legs. The retoothed saw worked very well for this. I cut a round tenon of 38 mm (the size of my largest speed drill bit). Then I used the spokeshave to make a smooth transition between square leg and round tenon.

Compared to maple, this Douglas fir is like cheese… And the spokeshave performs extremely well, besides being quite ergonomic.

Making the round leg tenons
Three legs done

I drilled four 38 mm holes in the benchtop, according to Christopher’s description (not using rake and splay angles, but a sightline and one angle reference). This proved to be a straightforward and predictable way to drill holes at an angle.

Holes drilled for the legs

I sawed oak wedges with 4 degree (inclusive) angle and used my to-be-mallet-head to drive them in as far as they would go.

Legs glued and wedged
Tenons trimmed flush. The piece of wood is to become the planing stop.
The other end of the Roman workbench, before planing the benchtop flat

Ergometer generator

This is an ergometer with a useful load. Most ergometers have a dump load (electrical or mechanical). Which is what I didn’t like…

I started the design at the end: the generator, which was the area with the most technical challenges.


The thing closest to a generator I had in the junk stack, was a universal motor, from a disassembled washing machine. These will work as DC generators. In its previous life, the motor had been wired as a series-wound DC motor. Luckily, the rotor (armature) and stator (field) wires are all accessible through 5 wires.

Typical connections of washing machine universal motor. Source: link

Two wires are for the rotor. They connect to the carbon brushes and will carry the DC output current.

Three wires are for the stator. They carry the field current. In the washing machine, for low speed operation the ‘many stator turns’ option was used, for high speed operation the ‘few stator turns’ option. There is a tap at roughly 1/3 of the stator turns. You can find out the wiring by measuring the resistance, though the resistance is quite low (of the order of a few Ohm).

Two wires are for the tachometer, which I am not using yet. I am planning to add an output current sensor and use an Arduino to calculate the average output power. I might add an RPM measurement in the future.

For use as a generator, I chose the highest-resistance path through the stator, which means the most turns of wire. This results in maximum magnetic field strength for the applied excitation current, so it reduces power loss in that area.


There needs to be a current in the stator to excite the generator. You can simply apply a DC voltage to the stator (field winding). A field current of the order of a few hundred milliamp is needed to run it as a generator. Be careful, it is a line-voltage motor, so can generate line-voltages of enough current to electrocute!
But my intention was to build a self-exciting generator to charge a 12V battery.

There is some residual magnetism in the iron, and in principle this allows the generator to self-excite. When the rotor rotates in the (very weak) residual magnetic field, it generates a (very small) voltage. If the rotor is connected to the stator, a current will flow, which increases the magnetic field strength, which increases the current, etcetera.

It was no option to connect the motor as a series-wound generator, because this will not self-excite without enough load. Because the generator will charge a battery through a diode, it will have an open-circuit load when output voltage is too low, so it can never self-excite. This was not the way to go.

Source: link

The other option was to connect it as a shunt-wound generator. The main issue is that the field windings have very low resistance (1 – 2 Ohm) in this motor, so they act almost as a short-circuit to the rotor. I tried operating it like this, but when speed goes high enough, the power consumption of the stator is so high that it blocks the rotor. That way, all power is consumed in the stator and no power remains for the useful load.

Source: link

I tried connecting the stator to the rotor through a resistance. It is called Rfc in below picture. (Ra just is the internal resistance of the armature.)

Source: link

A 10 Ohm resistance worked well. Higher, and there was no self-excitation. Lower, and it loaded down the generator too much. This allowed generating power and charging a battery through a diode. The diode is important, because the motor, well, is a motor and it will happily do its job and drain the battery. The issue was that the output voltage of 12 – 14 V (this is set by the battery) caused more than 10 Watt of power dissipation in the resistors. The total power rating of the resistors was 10 Watt, but still it started smelling bad. And it is waste of power, of course, because the actual excitation power required is less than 1 A through 2 Ohm, that is, less than 2 Watt.

Excitation controller

In the end I decided to build a generator-powered generator controller. It should do two things: limit the output voltage and limit the stator current.

Output voltage limiting is to prevent overcharging the battery, and to keep things safe.

Stator current limiting is to allow adjusting the ergometer load. Otherwise the controller would just try to charge the battery at the set voltage (13.8 V) and apply whatever field current needed to reach that voltage. With an empty battery (very low internal resistance), this would block the rotor, the voltage would drop and the controller would unpower itself.

I took this example ( of an AVR as a starting point and designed my own version. You can download my circuit design here:

It works like this:

  1. Start up
    The positive output of the rotor powers the VDD net. This net has some big capacitors (C2, C9, C12). I did not look in detail to the noise on this net, but it seemed a good idea to add some filtering, because this net supplies power to a switching mode controller (more details below).
    The rotor (armature) is shunt-connected to the stator by relay (K1). This is done through a 10 Ohm resistance (R14, R15), to allow self-excitation of the generator, while still building a voltage.
    The VDD net powers the +12V net (voltage is not precisely controlled) through a diode (D2). When the +12V net rises, an LED (D6) is lit, the +10V net rises, and a voltage regulator (U3) powers the relay (K1). The relay disconnects the shunt resistors and the controller is allowed to take over.
    The +10V net is controlled by a shunt regulator (D5).
  2. Current limiting
    The stator current is measured by a current sense amplifier (U1). When the current is lower than the threshold set by the potentiometer connected to J3, the comparator (U2B) floats its open collector output. A resistor (R9) pulls up the comparator output and the push-pull gate driver (Q1, Q2) switches the MOSFET (Q3) on. The push-pull driver is isolated from the +10V net by a low pass filter (R12, C5), to reduce noise reaching the comparator. The MOSFET gate has a small series resistor (R13) to prevent ringing.
    When the MOSFET is on, the stator coil is connected to VDD and the current will rise until the current rises above the threshold plus hysteresis set by R5.
    When the current exceeds the threshold, the comparator (U2B) pulls its output down and the MOSFET switches off. The stator coil current then ‘freewheels’ through the Schottky diodes (D3, D8) until the current falls below the threshold minus hysteresis set by R5.
    A small capacitor (C10) adds positive feedback to provide fast comparator switching.
  3. Voltage limiting
    The generator output current flows through high-current Schottky diodes (D1, D7) to the battery. When the battery is full, or no load is connected, the output voltage rises. The battery voltage is measured by a voltage divider (R2, R3).
    To prevent draining the battery when not generating power, I planned to connect the bottom leg of the voltage divider to a MOSFET (Q4). But I did not trust this solution, because the battery voltage would be applied to the negative input of the unpowered comparator (U2A). So I connected the upper side of the voltage divider to VDD and corrected the setting for the diode drop.
    When the output voltage exceeds the threshold set by the trimmer (RV1), the comparator (U2A) pulls its output down and the MOSFET switches off. The stator coil current then ‘freewheels’ until the output voltage falls below the threshold minus hysteresis set by R6.
    A small capacitor (C11) adds positive feedback to provide fast comparator switching.

To summarize, it is a switching mode stator current controller that has two constraints: the coil current and the output voltage.

The finished controller.
PCB bottom left: MOSFET, gate driver, comparator.
PCB top left: trimmer for setting output voltage limit, voltage regulator, power LED.
PCB top right: filter capacitors, voltage regulator, relay.
PCB bottom right: startup resistors, Schottky diodes (1 freewheeling, 1 output).

The ergometer

I built an ergometer with horizontal flywheel, like Philip Borg built (great source of information: It consists of multiplex, screws, glue, bike wheel, nylon rope, elastic rope, nuts, threaded rod, pieces of sheet steel, bearings and some 3D-printed pulleys.

An early version with polypropylene rope (breaks because it is not nearly as abrasion resistant as nylon), without a good handle bar, without moving seat. But it generates power.

More to come…


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.



The comet NEOWISE. There was a lot of light pollution.