repairing a Herrmans H-track dynamo taillight

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
An opportunity came up to help a friend who had a malfunctioning dynamo taillight on his hands. The story was that the standlight wasn't working. My guess was that the supercap might have been overstressed or possibly fatigued its leads. I asked him to see if he could open it up and take a few photos of the circuit board, just to get an idea of the damage or the cause of the failure. Also, it was a way to find out if it could even be opened up without destroying the light!

The photos showed three 0805 surface mount resistors that were quite brown and toasty, with some damage to the nearby portion of the circuit board. It didn't look too horrible. Seemed like it could be repaired, so I asked him to send it to me.

here are the photos that he sent.....

40320792011_b8e1109559_c_d.jpg


25449707857_54fc84419b_c_d.jpg



While waiting for it, I sketched out a schematic of what I thought the circuit might be. The three 120 ohm resistors (marked "121") suggested that one was in series with each LED to limit the current. The three 560 ohm resistors (marked "561") seemed like they would be used in series with the LEDs when the supercap discharged, providing the standlight function. U1 was a three terminal device. My guess was that it was a zener diode, with R6 being a series resistor. The zener voltage would then be fed into the base of a npn transistor, whose collector would be connected to the rectified dynamo voltage, and the emitter would be connected to the supercap. Basically, it was just a zener diode with the transistor acting as a current amplifier.


The Initial Inspection


After a handful of days, a small package appeared in my mailbox. Inside was a Herrman's "h-track dynamo standlicht". This is a fairly low profile taillight designed to fit to two holes spaced 50mm, typically on a rear rack. The light has 3 LEDs, one of which is aimed to the rear. The other two feed into a light pipe (i.e. clear plastic channel) that spreads light around the circumference of the light, making it visible from all directions.

25449963317_91d10c7b03_z_d.jpg


26449016518_beaccf9d22_z_d.jpg


A look at the board (which lifts right out of the housing) only shows the damage that the early photos showed; R1, R2 and R3 were browned rather badly. (see photo below)

25449702547_1b2ac9412b_c_d.jpg


One big concern was whether the supercap was good. A visual inspection didn't show any bulging, which was good. No signs of broken leads on the supercap either (also good).

26449013178_5f6458983b_z_d.jpg


The board turned out to be single sided; i.e. it only had traces on one side. This is the least expensive way to make a board, and it makes it much easier to reverse engineer. A close look at the board reveals most of the traces. A multimeter lets you find the rest of the connections, especially those circuit traces that run underneath the components or are rather narrow and hard to see.

39609723944_2f09069bcc_c_d.jpg


When the board had been checked and corrections made to my initial guess at the schematic, the only real place that I was wrong was about the two transistors. I had assumed that the two were used in parallel to charge the supercap. Instead, one transistor was used to provide a fairly steady 5V to each string of 120 ohm resistor and LED. Not a bad way to design it.

39424230675_5e65443516_c_d.jpg


(come back tomorrow-ish for the next installment in this exciting/interesting/tedious tale of technology!)
 

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
Part 2: Troubleshooting


The first step was to hook up the power supply and gradually apply power to the light's terminals.
It was a bit of a relief to see the LEDs start to glow with just 2V applied. A bit of probing around the board didn't show anything odd, except at Q1. The Vbe was close to zero, which seemed enough reason to pull it off the board and replace it with an MMBT3904.
With it off the board, the original Q1 was measured with the meter's diode check function. The Vbc was 0.7V and Vbe was 0.73V. Both of these seem high for the low current sourced by the meter.
In any case, the light seemed to work well with the new transistor for Q1.

A brief note from the end of the project:
In retrospect, I realize that I never did figure out what was causing the standlight to not function. It seems like Q1 was shorted when it let high voltages reach the 120 ohm resistors. If it was still acting as a short when the standlight was supposed to function, then could it divert the supercap current through the 120 ohm resistors instead of allowing it all to flow through the LEDs?? Perhaps, but even if Q1 was a short, how would the current get back to ground? The only path is through the zener, which isn't going to conduct much current when the voltage source is 5V or less.
Maybe Q1 just wasn't conducting much, and the only current illuminating the LEDs was coming from Q2, which kept the supercap from charging very much??
I still have the transistor, in case there is something else that I can check.
Okay... back to the troubleshooting:

A check of the voltage reference showed that it was a 5.6V zener diode. It charged the supercap to 5.0V at nominal input voltages.

With the light operating normally again, there was still the question about why the resistors appeared overheated and why Q1 was damaged. With a bit over 7VDC applied, and 99mA drawn by the light, a check of the temperatures of some components showed that they were fairly warm, but not excessively so.
R2, one of the 120 ohm resistors, was 50 deg C.
Q1 was 45 deg C.

Part of the troubleshooting or failure analysis process is to try to figure out what caused the failure. The damage to the three 120 ohm resistors suggest that they were exposed to too much current. This is partly due to the failure of Q1, which presumably also died due to excess power dissipation. Q1 is supposed to keep the top of the resistors at 5V. If Q1 failed, it could allow the rectified dynamo voltage to reach the resistors.

The cause of a failure of Q1 was likely a high dynamo voltage. Many dynamo headlights are designed to protect the taillight from high voltages. It turned out that this taillight was being used with a k-Lite light. There's not a lot of info on the web, but it is advertised as a 1300 lumen light, and has 3 LEDs. It's hard to know what the light does, but the lumen rating suggests that it is drawing roughly 10 watts from the dynamo. This suggests that the dynamo voltage would have to be quite a bit higher than the typical 6 or 7Vrms. Even if the lumen rating isn't accurate, the use of 3 LEDs in a small package suggests that the LEDs are wired in series, resulting in roughly 9Vrms out of the dynamo.
 

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
Part 3: Making Q1 more tolerant of high dynamo voltages


In light of the possible high voltage induced failure, it seemed appropriate to measure component temperatures at dynamo voltages higher than 7V. A quick check with a dynamo voltage of 10V resulted in Q1 reaching 86 deg C (in open air). This didn't seem like a great temperature for reliability, so I modified the design to replace Q1 with a PZTA42 mounted to a small piece of copper-clad circuit board. The PZTA42 is a sot-223 package, which when mounted to a modest bit of board, should have much better power dissipation than a tiny sot-23.

40321307901_7dba0a2fb1_c_d.jpg


40321304601_ecde7b8ecd_c_d.jpg



In open air, this new configuration resulted in the transistor temperature only reaching 36 deg C with a dynamo voltage of 10V. With 15V applied, it reached 54 deg C.
The photo below shows an initial temperature measurement where a binder clip held the thermocouple probe to the board.

40321301461_efe761e663_c_d.jpg


the next photo shows some copper wire tacked onto the board to provide a smaller method of holding the probe to the board.

39423703865_9a4cc22a00_c_d.jpg


While testing at 15V, I noticed that the taillight was drawing 150mA instead of the nominal 100mA, and the supercap charge voltage rose from 5.0V to 5.4V. Neither of these seemed desireable.

Before addressing this new issue, I put the circuit board into the housing to measure the transistor temperature. At 10V, it reached 49 deg C, and at 15V, it reached 77 deg C. Not great, but probably as good as can be done in a closed plastic housing.
 
Last edited:

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
Part 4: improving the voltage reference circuit

Going back to the issue of high current draw and high supercap voltage when the dynamo voltage is high... it looks like the problem is the zener diode circuit. In these cases, the excess current is going through R6, the 150 ohm series resistor, and the zener diode. It's quite normal for the zener voltage to go up when the zener current increases.
The quick solution to this would be to replace the zener diode with a precision voltage reference, such as the LM431. It is rated to operate at currents up to 100mA, so it should be able to handle this. It is also capable of regulating at fairly low currents, so it should allow R6 to be increased in value, resulting in less current draw at high voltages.

A small circuit board was built for the LM431 and the two resistors that it requires for feedback. The resistor values were calculated to produce the same 5.6V as the zener diode.... and then scaled a bit to match the resistor values of what I had in my parts box.
R6 was changed to 350 ohms, to reduce the current to the LM431.

39611419734_223c1ee8d0_o_d.jpg


Tacking the LM431 board to the light's circuit board, some quick measurements were taken. The numbers looked pretty good, although the reference voltage was a bit low at 6V in.
Dropping R6 to 220 ohms brought the reference voltage back up to where it had been with the zener diode at low input voltages, but still reduced the current draw at high input voltages. The supercap voltage at 7V input was 4.95V and only rose to 5.07V at 14V input (with 125mA current draw for the light).

39611419504_284b0fd16c_c_d.jpg



This all seemed suitable, so the board and wires were RTV'ed in place.

39611412944_1d6e8ef88e_c_d.jpg


To verify that the RTV was going to tolerate the moderately high temperatures, the board was put back in the housing and run at high input voltage for a few hours.

25451824537_3308507328_c_d.jpg



edit: oops... forgot to include the updated schematic:

38515171520_0204260f0a_c_d.jpg
 
Last edited:

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
Part 5: replacing the 120 ohm resistors

With this done, it was time to order some new 120 ohm resistors and replace the damaged (but still functioning) ones. When removing the old resistors, one of the pads came off of the board. The exposed board was black, and showed the woven fiber glass fibers. Apparently the epoxy in the board had all boiled away, or simply burned away! Considering how hot the epoxy must have been, it's amazing that the resistors still worked.

In this photo, the three resistors have been removed:
26451294638_8a41cac4b1_c_d.jpg



a closer shot, where you can see the damage to the board below the missing pad for R1:
38514184320_e5d382daa6_c_d.jpg



To compensate for the missing pad on R1, a jumper wire was run from one side of R1 to R4.

38514182080_3f85fc38f0_c_d.jpg


Some RTV was applied to secure R1 and the jumper wire to the board.
 

Steve K

Flashlight Enthusiast
Joined
Jun 10, 2002
Messages
2,786
Location
Peoria, IL
Part 6: Reassembly

A quick check showed that everything was working, so it was time to put everything back together! When installing the board back in the housing, a bit of RTV was used to attach the board, especially the section of copper-clad board that was added for the new Q1. A little RTV was applied at the terminals to help seal against moisture entry. Some RTV was applied under the supercap, just to provide a little extra support. It probably isn't needed, but shouldn't hurt.


40323354721_3e8932c7f9_c_d.jpg



40278073742_c2bb03d714_c_d.jpg



After allowing 24 hours for this to fully cure, a bead of RTV was applied to the circumference of the rear of the housing, and the lens/reflector portion was joined to it. While this was curing, I applied about 7V to the light, just to ensure that it was still working (because weird stuff can happen, and I didn't want to ship it back to the owner and have it not work).


40278068082_a4aeb233d2_c_d.jpg



39425667855_d66653d7d6_c_d.jpg



That wraps up the project!

The light has been returned to the owner and will hopefully provide reliable service.
 
Top