Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
Being that type of guy, a red LED glowing in the far corner of the basement attracts my attention:
Verizon FiOS – replace battery
Back in the day, Verizon didn’t make it obvious that the customer is responsible for replacing the battery keeping the ONT alive during power failures. I expect VZ would eventually let me know the battery was dead, remind me I was on the hook for the replacement, then offer to send a tech around with a Genuine VZ Battery to maintain reliable service.
It’s an ordinary 12 V 8 A·hr sealed lead acid battery and, yes, it’s been in there for a while:
Verizon FiOS – OEM battery date
However, being that type of guy, I just happen to have a box of not-dead-yet SLA batteries waiting for recycling:
In preparation for the next time a task puts my head in a dark place, I got a cheap headband LED light:
Headband LED – overview
Unlike most of the others you’ll find, this one has a pair of 18650 lithium cells in the box on the back of the headband:
Headband LED – isolated cell
Contrary to what you might think, the cells are in parallel, with shorting plates connecting the battery compartment terminals. This works well for perfectly matched cells, which is not what arrives in the package.
The 3200 mA·hr capacity claimed (in one line of the product description) doesn’t match the 2200 mA·hr capacity (claimed in another line and) printed on the cells. As expected, both claims far exceed the actual 1500 mA·hr measured capacity.
LED Headband Light – 2022-01-12
The 1 A load is somewhat more than the 800 mA I measured at full brightness, but makes for easy comparisons.
I think they put the cells in parallel to reach the claimed 4-6 hours of run time, but in practice the connection discharges the better cell to match the weaker one with no assurance of equal load sharing thereafter.
So I conjured an insulator from the Box o’ Retail Clamshells:
Headband LED – cell isolator
In the unlikely event my head must remain stuck in a dark spot for longer than one cell lasts, I can move the insulator to the dead cell and continue the mission. Charging alternate cells isn’t much of a burden, either.
For unknown reasons, the (anonymous) manufacturer soldered the LED package at a jaunty angle inside the frame:
Headband LED – SMD alignment
The lens pulls in-and-out to zoom the focus. The tightest setting (all the way out) projects a bright tilted square out in front, which is somewhat unsettling.
The whole affair cost less than a pair of known-good 18650 cells from a reputable supplier, so ya get what ya get.
The top step of a folding step stool we’ve been (ab)using forever finally wore out, mostly because it was covered in vinyl and intended as a seat. We always used it as a step, despite knowing you should never stand on the top rung of a ladder: “Do not stand on or above this level”.
I tossed the ripped vinyl and warped particle board, cut a random chunk of wood-textured paneling (which Came With The House™) to fit, match-drilled four holes, and it looks OK:
Folding step stool – reseated
The original seat / step / whatever used press-fit studs with a flat flange covered by the vinyl, but I just slammed 10-32 tee nuts into the paneling:
Folding step stool – tee nut installed
That’s a ring of low-strength threadlock around the inside of the nut; I do not expect the screws to come out ever again.
I cut the screws to length with a Dremel cutoff wheel using a slightly shortened tee nut as a fixture:
Folding step stool – screw shortening fixture
Not visible: the vacuum hose clamped to the vise sucking up all the abrasive + metal dust.
Good for an hour of Quality Shop Time™ on a cold winter morning!
The Tektronix AM503 manual specifies a Special Adapter to inject a signal directly into the input connector in place of the A6302 Hall probe:
Tektronix AM503 Special Adapter
The intricate Amphenol plug might still be available at some phenomenal cost, but I’m willing to just jam a pair of wires into the AM593 connector and be done with it.
I combined a pigtail BNC sporting a male connector, two 51 Ω resistors in parallel, two snippets of 18 AWG wire (an exact match for the 40 mil connector pins!) with the ends filed smooth, and some heatshrink tubing to make a roughly equivalent adapter:
Tek AM503 – Crude Special Adapter
Because the pigtail didn’t quite reach the function generator, I joined it to a longer cable with a BNC bullet, whereupon a slight tug ripped the guts out of the bullet:
BNC Bullet – failed
A closer look:
BNC Bullet – parts
The center hole comes into play with their equally craptastic BNC tee connectors.
Comparing this bullet with others from the same eBay lot shows the outer shell didn’t get quite enough crimp around the metal ring. Because it’s not an electrical connection, I eased some epoxy onto the internal shoulder where that ring seats, then slid the guts back in place.
So, despite it not showing any leakage or damage, I replaced C155:
Tek AM503 B075593 – C155
Which had stopped being a capacitor some time ago:
Tek AM503 B075593 – C155 measurement
I also replaced C165 with a newer capacitor.
Again, having the hood up, I pulled C452 and C462 from the ±19.3 V supplies:
Tek AM503 B075593 – C452 C462
Despite the 1987 date code, they seemed to be in fine shape, but I replaced them anyway. The new caps have a 50 V rating, not the original 63 V: only a factor of two headroom.
The four new capacitors in their new home:
Tek AM503 B075593 – replaced caps
The power supply voltages looked clean before and look clean now.
The AM503 still has the mysterious 4 MHz oscillation, so the capacitors weren’t the problem. Even though the amp is still sick, I feel better.
The test signal (yellow) comes from the scope’s calibrator output into a 2320 Ω resistor, so the AM503 calibration is about right: 0.6 mA ≅ 1.5 V/2320 Ω.
Just to maintain historical accuracy in the two AM503 amps in the TM502 mainframe on the Electronics Workbench, I transplanted the good (not noisy) OEM Tek Q230 (from SN B075593) into the previously noisy-and-offset-prone AM503, which now works fine. I now have a pair of works-pretty-good AM503 amps, one not-so-good AM503 in the to-be-fixed lookaside buffer, plus a defunct Q230 dual JFET.
That third amp (B075593, now with the NOS 2N5911) has a nasty noise problem:
Tek AM503 B075593 – SDS2304 cal – 1 mA-div
The barely visible yellow trace is the same calibrator signal as before, but the output is a howling 4.2 MHz (!) sine wave. The oscillation amplitude responds to the AM503 front panel gain control, making it possible to see what’s going on:
Tek AM503 B075593 – 4 MHz oscillation
Flipping the front panel switch to limit the AM503 bandwidth to 5 MHz shaves off the fur:
Tek AM503 B075593 – 4 MHz osc – 5 MHz BW
Disconnecting the probe or unplugging P220 kills the oscillation, as does setting the front panel switch to CAL/DC LEVEL, which means it’s an internal feedback problem.
It’s trivially easy to construct an amplifier circuit that becomes an oscillator at the slightest provocation, but this puppy had been working dependably for somebody else during the three decades (!) before I bought it and continued for a few years after that, so the overall circuit topology is known-good.
Shooting this one will require more pondering, as the obvious first step of replacing the power supply’s electrolytic caps had no effect.
The Smell of Molten Projects in the Morning 