I have achieved my stretch goal of winding some torrid-base inductors and I would like to share those results. Also, as a result of a comment on last blog by andrewj (thanks!) concerning transient changes, I have done a bit more testing and sharing those results.
Using the my old Micrometals toroidal powder cores that I found in one of my may parts boxes, I set out to wind my own inductors. Having used the part number on the smallest of my toroids (marked A-050056-2), I found a cross reference into a currently available part (MP-050125-2). The part numbers are actually quite meaning. The Prefix (MP) identifies the material type, MPP Molypermalloy. The first three digits (050) define the OD in 100th inches (0.5 inches). The next three digits are the Reference Permeability (125µ) of the device (used to determine the inductance per turn) value (AL) which is 56 nH/N2
Given the formula Turns=√(Ldesired ) where Ldesired is in nH,
I calculated the number of turns for a 2.2uH and 3.3uH inductor. The results were 6.26 for a 2.2uH and 7.67 for a 3.3uH inductor, and I decided on the 3.3uH inductor. Unwinding one of my earlier attempts, I measured the wire length at ~6 inches. Using a wire gauge chart, I determined the DC resistance of my two potential magnet wire choices, either 22 or 26 gauge. The 22 gauge yielded ~8 mOhms, where the 26 gauge wa ~20 mOhms. I then checked the fit of the two wire choices into the holes on my PCB, while the 22 gauge wire was more of a 'forced' fit (much twisting and pressing to get the lead into the hole), I decided to go with it, as the DC resistance was better.
Here are some examples of my first attempts:
Using the stiffer 22 gauge wire, produced a pretty rough looking coil. I then remembered a blog posted by three-phase (3000A current clamp table and amplifier ), where jw0752 had posted a comment with some great advice:
"at first I tried to thread the individual winds but it was too difficult and I found that the bends and twists that I was putting in the wires made them look really bad. I solved this problem by prewinding the wire in a coil like you did on the PVC pipe. I then screwed the helix onto the toroidal transformer. Finally I went down the coil and tightened each wind snugly onto the original winding."
Using this technique, I would my turns onto a 3/16" aluminum tube, and then 'screwed' it onto my toroid. This produced a much nicer looking inductor. Here is a picture of the finished coil installed on one of my DC-DC boost converter boards:
Comparing my Hand Wound Inductor to Other Inductors (1.0uH to 10uH)
With the newly created inductor mounted to a fresh DC-DC Boost Regulator board, I fired up my test setup and did an abbreviated test sequence, collecting the same data points in my prior Inductor range experiment (5.0 Volts output at 2.5V, 3.0V and 4.0V input). I then plotted all of the data for the 6 inductor values (1.0uH, 2.2uH, 3.3uH, 4.7uH, 10.0uH, 3.3uH HW)
The Handwound 3.3uH Inductor seems to behave similar to the KEMET part. In Voltage Output vs. Output Current, the handwound 3.3uH inductor seemed to have a slightly higher output voltage, which could be contributed to the lower DC resistance value or even tolerance differences of the resistors that are used to set the output voltage. This could also have been caused by the slightly higher ripple voltage, likely caused by the lower DC resistance value.
In the Efficiency vs. Output Current, the handwound 3.3uH inductor seemed to have as good, if not slightly better in overall efficiency (especially above 0.5 A).
In the Ripple PKPK vs Output Current, the handwound 3.3uH inductor seemed to similar ripple values, although slightly higher in the lower currents.
Overall, it looks like handwound 3.3uH inductors performed close to the 3.3uH KEMET part (don't worry KEMET, I will not go into competition with you as my volume is very low, only two inductors over the weekend).
Looking into Transient Load Effects
Using the "Switching Mode" of my PLZ72W Electronic Load I was able to inject large output load changes on a periodic basis. I set the load up to switch between 100 mA and 500 mA and captured the Output Voltage changes. Here are some of the captured waveforms:
1.0uH - 4.0Vin - 5.0Vout - Switching between 100mA and 500mA
3.3uH - 4.0Vin - 5.0Vout - Switching between 100mA and 500mA
10uH - 4.0Vin - 5.0Vout - Switching between 100mA and 500mA
The transient did not seem to cause much of change in the output levels, although in does clearly show a shift in the ripple voltage pattern as the two different current level put the boost regulator into different pulsing modes. There does not seem to be much difference in the waveforms based on the inductor values. (Note: the DC LOAD has a BNC output for tracking the load current waveform, but I could not find any of my 50 Ohm BNC cables. Using my scope probe, resulted in a lot of noise, so I needed to resort to heavily filtering the load current command voltage, hence the slow ramp on current changes)
In these tests I was careful not to command to large of a current step, so not to put the regulator past the point where the output voltages began to hold off. Overall, I was impressed with the transient responses, but still a little unhappy with the amount of ripple that occurs at differing points in the output current levels.
This was a very fun and interesting learning experience. Thanks for the challenge!