Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
Category: Science
If you measure something often enough, it becomes science
Mary finished out the National Bike Challenge with a rank of 3353 of 47 k riders, which, by my reckoning, is wonderfully good. She’s #1 in the Poughkeepsie area (admittedly, of only eight riders), with the second-place rider at 90% of her point score.
She did it by riding on her usual missions, along our usual routes, around the usual obstacles:
NYS Rt 376 at Westview Terrace
Her bike odometer recently rolled past 20 k miles; at least one battery change stole a pile o’ miles from her total, so the bike has accumulated more than that.
As the song goes, my gal is red hot… in the best way!
Dell built the GX270 I’m repurposing back in 2004, early on in the capacitor plague years, but only one of the system board caps showed signs of leakage:
Capacitor plague – 2004 Dell Edition
While I was harvesting some of the connectors, it occurred to me that those powdered iron inductors might make good current sensors, as they’re already wound with heavy gauge copper wires.
I picked an inductor with enough turns and, although slitting didn’t pose much of a problem, the saw did make a mess of the turns adjacent to the cut:
Powdered iron toroid – slitting
Iron powder has more magnetic remnance than ferrite, to the extent that iron swarf clogged the gap. After the first pass, I ran the slit toroid through the degausser to shake it clean and see what damage had been done. It looked OK, so I realigned it on the saw blade and continued the mission, with all the dust vanishing into the vacuum cleaner’s snout.
Removing the damaged sections left 22 turns. For comparison, I converted the 56 turn ferrite toroid into a 25 turn model by paralleling two 25 turn sections:
Slit toroids – iron – ferrite
The enamel wire on the iron toroid measures 40 mil diameter, close enough to 18 AWG.
Paralleling two 24 AWG windings on the ferrite toroid produces twice the copper area of a single winding, so the resistance is the same as a single 21 AWG winding (3 AWG steps = factor of two area change). That’s three steps smaller than the 18 AWG on the iron toroid, so the resistance is a factor of two larger than the heavier wire.
The paralleled winding has the advantage of reducing the power dissipation required to produce the same magnetic flux density, without the difficulty of winding heavier wire. That may not actually matter, given the relatively low currents required by the motor in normal operation.
Wedging a Hall sensor into the gaps and stepping the current produced two useful graphs:
Iron and ferrite toroids – Hall sensor output
The iron toroid has lower permittivity (less flux density for a given magnetizing force), which means the full-scale range exceeds 3 A and the useful range up to 1 A covers only 300 mV.
The last point on the ferrite curve shows the Hall sensor output saturating just over 4 V, with 1.5 V of range.
The slope, in mV/A
Powdered iron: 340
Ferrite: 540
Boosting the slope of the powdered iron by 25/22 gives 386 mV/A, so the iron permeability really is 70% of the ferrite. That’s modulo the gap size, of course, which surely differs by enough to throw out all the significant digits.
Obviously, an op amp circuit to remove the offset and rescale the output to 0-5 V will be in order.
The previous graph for the ferrite toroid with the complete 56 turn winding shows, as expected, about twice the output of this 25 turn version:
FT82-43 – 56 turns – 24 AWG
The linear part of that line is 1375 mV/A, although I can’t vouch that the data came from the same Hall effect sensor. Scaling it by 25/56 gives 613 mV/A, suggesting it’s not the same sensor.
Having developed an emotional attachment to the ferrite toroid, I’ll use it in the first pass of the current feedback circuit. If the motor need a bit less sensitivity or lower resistance, the powdered iron toroid looks like a winner.
Memo to self: Always degauss iron toroids before slitting!
The desiccant definitely lasts longer during the winter, even though the dehumidifier fights the basement air to a standstill around 55%RH during the summer.
Each desiccant bag contains 500 g of silica gel and the most recent one adsorbed 73 g of water.
A Squidwrench Weekly Doings being useful for short-attention-span projects, I measured the DC current gain for all five ET227 transistors. The test conditions fall far below the ET227’s 1 kV / 100 A ratings, but they’re roughly what the sewing machine motor controller calls for.
The transistors don’t even begin to turn on until IB gets over about 50 mA, because there’s a 13 Ω shunt resistor (as measured, for either polarity) between the base and emitter terminal:
Fuji ET227 – equivalent circuit
In the ET227’s normal use, that resistor dumps the Miller effect charge injected from the collector (with the intent of improving the switching time), but you must ram nearly 70 mA into the resistor to get 900 mV at the base, so the actual transistor base current isn’t all that high for low collector currents. But you measure gain by dividing goes-outa by goes-inta, so that’s what I’ll do.
The ET227 needs something like IB = 30 A to switch 100 A at the collector, so a few dozen mA into that resistor rounds off to zilch for its usual driver circuit. FWIW, with IB = 30 A, VBE tops out at 2 V: the resistor carries 150 mA and dissipates 300 mW.
Anyhow, randomly labeling the transistors from A (on the heatsink) through E, then hitching them up to a 1.8 A bench supply with a 33 Ω resistor to the base terminal provided some readings at single-digit collector voltages.
For IB = 72 mA:
IB
IC
hFE
A
72
490
6.8
B
73
540
7.4
C
74
480
6.5
D
75
440
5.9
E
76
520
6.8
For IB = 108 mA, with one bumped-knob outlier:
IB
IC
hFE
A
108
1220
11.3
B
101
1190
11.8
C
108
1280
11.9
D
108
1170
10.8
E
108
1320
12.2
Although the gain around 1 A comes out slightly higher than while running the motor, it’s in the same ballpark. This is not a high-gain device: it’ll need a driver after the optoisolator to squeeze enough current through the collector.
Eks tried to unload a huge old Tek transistor curve tracer on me that would be ideal for this sort of thing. I’m still not tempted…
I’d have trouble faking this with a straight face:
FT82-43 – 56 turns – 24 AWG
That’s measured with the 56 turn winding connected directly to a bench power supply, cranking up the current, taking the reading, and turning the current back down again, so as to avoid cooking the poor thing inside its PLA armor:
FT82-43 toroid – mounted
The “49E” sensor came from one of the bags of eBay fallout. They saturate around 4.25 V; the outputs above 4 V lose their linearity due to the sensor, not ferrite saturation.
The original calculations guesstimates suggested 25 turns would produce full scale at 5 A, so 56 turns should top out at 2.2 A. Frankly, given all the imponderables in this lashup, a factor of two seems pretty close.
Offsetting the output by -1 A would yield a 2 A range that’s just about exactly right. Unfortunately, some fiddling about with neodymium magnets suggests that you (well, I) can’t stuff enough opposing field into the slit without saturating (some part of) the ferrite core, reducing the permeability, and blowing all the assumptions.
So that suggests a buck winding, obviously with more turns to allow less current for the same magnetizing force. Wrapping 110 turns reduces the buck current to 500 mA and assuming a bit over an inch/turn requires 10 feet, which is nearly 1 Ω of 30 AWG wire: the buck current dumps another 250 mW into (a somewhat larger version of) that PLA armor.
Or just throw away half of the Hall effect sensor range and use an op amp along the lines of the LED current sensor.
I didn’t bring the HDR-AS30V camera along on the Hudson River ride, simply because each battery lasts about 1.5 hr in 1920×1080 @ 60 fps mode and I wasn’t up to replacing batteries during the ride, then charging all three every evening. Obviously, the camera wasn’t intended for that use case.
Somewhat surprisingly, the Wasabi batteries deliver the same continuous run time as the Sony battery: 1:30 vs 1:33. I used 250 mA for those discharge curves, but I think something around 500 mA would better match the camera load.
I’m sorely tempted to drill a hole in the camera’s case and wire in a honkin’ big prismatic lithium cell.
The motor winding resistance limits the peak current to about 200 V / 40 Ω = 5 A, in the absence of the transistor current limiter, and, if it gets above that, things have gone very, very wrong. Mostly, I expect currents under 1 A and it may be useful to reduce the full scale appropriately.
The cheap eBay “SS49” Hall effect sensors I’m using produce anywhere between 0.9 and 1.8 mV/G; I’ll use 1.4 mV/G, which is at least close to the original Honeywell spec. That allows a bit over ±1000 G around the sensor’s VCC/2 bias within its output voltage range (the original datasheet says minimum ±650 G), so I’ll use B = 1000 G as the maximum magnetic flux density. The overall calibration will be output voltage / input current and I’m not above doing a one-off calibration run and baking the constant into the firmware.
The effective mean path length turns out to be a useful value for a slit toroid:
effective MPL = (toroid MPL - air gap length) + (µ · air gap length)
The SS49 style sensor spec says they’re 1.6 mm thick, and the saw-cut gaps run a bit more, but 1.5 mm will be close enough for now.
The relation between all those values:
B = 0.4 π µ NI / (effective MPL)
Solving for NI:
NI = B · (eff MPL) / (0.4 π µ)
Solving for N:
N = B · (eff MPL) / (0.4 π µ I)
You always round up the result for N, because fractional turns aren’t a thing you can do with a toroid.
The saturation flux density seems to be measured at H = 10 Oe, but that applies to the intact toroids. The air gap dramatically reduces the effective µ, so you must apply a higher H to get the same B in the ferrite at saturation. At least, I think that’s the way it should work.
H = 0.4 π NI / (geometric MPL)
Then:
FT50-61: H = 58 Oe
FT82-43: H = 30 Oe
I’m surely missing some second-order effect that invalidates all those numbers.
Figuring the wire size for the windings:
FT50:
ID = 0.281 inch
Circumference = 0.882 inch
28 turns → wire OD = 0.882/28 = 31 mil
20 AWG without insulation
FT82:
ID = 0.520 inch
Circumference = 1.63 inch
25 turns → wire OD = 1.63/25 = 65 mil
14 AWG without insulation
Of course, the wire needs insulation, but, even so, the FT82 allows a more rational wire size.
Page 4.12 of the writeup from Magnetics Inc has equations and a helpful chart. They suggest water cooling a diamond-bonded wheel during the slitting operation; my slapdash technique worked only because I took candy-ass cuts.
The Smell of Molten Projects in the Morning 