WhitakerAudio
 
Trouble Log

This page focuses on the in-service reliability of the vacuum tube amplifiers and other products described in The TAB Guide to Vacuum Tube Audio and on this site.

Vacuum tubes

Summary Report

It is instructive to examine the failures in any system in order to improve the product and extend reliability. For that reason, the following log documents any failures experienced during normal use of the products described in the book and on this web site.

• First generation stereo preamplifier (described in the book): Placed into service February 2010. No failures.

• Second generation stereo preamplifier (described on this site): Placed into service November 2011. No failures.

• First generation 20 W stereo amplifier (described in the book): Placed into service March 2010. No failures. This amplifier was decommissioned in February 2014.

• Second generation 20 W stereo amplifier (described on this site): Placed into service December 2011. One failure—capacitor C9 in the bias power supply failed in a short circuit. Collateral damage was done to R17 and D2. Following an examination of the failure and parts specified, no changes were deemed necessary. Return to service one week later.

• First generation 50 W stereo amplifier (described in the book): Placed into service April 2010. No failures. This amplifier was decommissioned in November 2014.

• Second generation 40 W stereo amplifier (described on this site): Placed into service February 2012. One failure—the 6U8A input tube failed and was replaced. Returned to service one day later.

• 60 W stereo power amplifier (described on this site). Placed into service June 2013. No failures.

• AM/FM stereo tuner (described on this site): Placed into service August 2012. No failures.

• 20 W stereo integrated amp (described on this site): Placed into service October 2012. No failures.

As with any consumer device, the amount of usage each product has experienced varies. Usage of these products can be reasonably described as equivalent to that of most any other audio product in a home.


Circuit Board Issue

Aside from the issues detailed above, there have been no additional failures from the projects covered in the book and on this site as of April 2015. There was a failure observed during tests on a new amplifier under development, however, that is worth reporting. As with any new circuit implementation, there is a troubleshooting period where problems (or potential problems) are identified and corrected. This is the reason, naturally, that building the first unit is always more time-consuming and expensive than building the second.

The downside to a problem is obvious—the system does not work or is otherwise impaired. The upside (or at least the potential upside) is to learn something from the failure that permits a more reliable system to be built—either from scratch or as a modification to an existing system.

Several months ago I began building a new amplifier. Based largely on the 60 W Stereo Amplifier detailed on this site, the new amplifier uses a different 6U8A input configuration and a considerably enhanced power supply. Operationally, the power supply changes are the core of the upgrades.

For the 60 W amplifier, a capacitor-input filter was used that developed sufficient B+ voltage to achieve 30 W per channel from the circuit. For any capacitor-input filter, the regulation from light load (idling condition) to full load (rated output) is poor compared to a choke-input filter. For the new amplifier, a choke-input design was used. In order to achieve the needed voltage for the output tubes, a Hammond #382X power transformer was used, which provides 1,000 V ac center tapped at 283 mA for the plate supply. Using a conventional full-wave rectifier circuit (center tap grounded) feeding a 5U4 tube, a calculated B+ voltage of 450 V dc was obtained at the output of the filter (the actual voltage was less due to losses).

The rectifier circuit was implemented on a printed wiring board (PWB), as shown on the right.

The amplifier was successfully built and ran on the bench and in a test environment for more than a month without issues.

Rectifier circuit board

Quite unexpectedly, one connection pad on the rectifier board failed, as shown on the right. The connection point is for the input to the choke ("L51in"). This was quite a surprise since the same board has been used in other amplifiers running higher voltages for a longer period of time without any issues. Even under no-load conditions (all tubes removed except the rectifier), the maximum dc voltage at L51in is no higher than 470 V. For the 40 W Stereo Amplifier described on this site, the no-load output voltage on the same terminal is in excess of 560 V.

The first explanations that come to mind include a flaw in the circuit board itself, a voltage surge on the ac input, or some type of debris on or adjacent to the terminal itself. The last potential explanation may have been the cause, since the failure occurred a few minutes after service was performed on the amplifier on an unrelated board. No other electronic devices on at the time experienced a problem, which makes a transient disturbance on the ac line less likely. (It cannot be ruled out, however.)

Board failure

A close-up of the connection point in question on an unused board is shown at right. The board uses a solid ground plane on the component side of the board. The boards are manufactured double-sided with all holes plated-through. The laminate is 0.062" FR-4 epoxy glass with 1/2 ounce copper. FR-4 is a composite material composed of woven fiberglass cloth with an epoxy resin binder that is flame resistant (self-extinguishing). The top and bottom solder mask layers are green LPI (Liquid Photoimageable). A white silkscreen is printed on the top side of the board showing component outlines and text. The board finish is the industry standard SMOBC (solder mask over bare copper). The surface finish has two essential functions: 1) to protect the exposed copper circuitry, and 2) to provide a solderable surface when assembling (soldering) components to the PWB. The maximum operating temperature of the board is 125 degrees C. The default clearance around traces and pads is 0.05 inches, the maximum available with the ExpressPCB software.

Focusing on the L51in terminal, the white outline is somewhat deceptive in that the silkscreen box is actually inside the cut-back area of the ground plane. As such, there is more insulating distance than might at first appear. The insulation distance from each terminal to the ground plane is the same in the horizontal and vertical directions (although it may not appear to be so from the photo).

Close up of rectifier board

In order to determine the practical operating stand-off voltage of the regulator circuit board, as built, an experiment was run. As shown in the photo on the right, two high-voltage power supplies—each capable of producing 400 V dc—were connected in series to provide 0 to 800 V dc. The negative lead was connected to ground plane of the board that failed (shown above). The positive lead was connected to the L51out terminal on the board that failed. This terminal is directly adjacent to the L51in terminal, where the short-circuit occurred. A voltmeter was connected to the board ground plane and the L51out terminal so the actual voltage applied to the board could be monitored.

The goal here was to identify the maximum voltage that the circuit board could withstand before a flash-over failure occurred. Because the typical operating voltage at L51in (the terminal that failed) is 400 V dc, a test voltage of twice that (800 V dc) was deemed to be sufficient to confirm that the original design parameters were sufficient for reliable operation.

With the connections made, the voltage was ramped up while the current (as show on the power supplies) was monitored. The voltage was successfully brought up to the maximum output of the supplies (800 V dc) with no observable problems on the board, and zero current as measured at the supplies.

So, with a minimum 2X operating margin, the most likely cause of the failure was determined to be one of the following (in order of likelihood):

1) Conductive debris on or near the terminal resulting from maintenance work performed shortly before the problem occurred.
2) A transient disturbance on the ac line, with the voltage surge acting in conjunction with the choke (5 H) in such a way to produce a voltage that exceeded the breakdown voltage of the PWB.
3) A manufacturing defect on the board itself.

Based on these conclusions, a plan could be devised to prevent a similar problem from occurring in the future.

Test setup for board stand off voltage

If cause #1 led to the failure, the solution is simple—check for and remove any loose debris (clipped leads, etc.) in the chassis after any maintenance. Turn the amplifier in various positions to remove anything that might be lodged in a corner.

If cause #2 led to the failure, the solution is a little more complex. Transient disturbances are a fact of life. The amplifier already includes an ac line filter that will attenuate noise on the line, and to a lesser extent small transient disturbances. Metal oxide varistors are commonly available for protecting ac circuits. A varistor such as the EPCOS B72220S131K101 (130 V rms, 8 kA) should be quite sufficient. Because such devices are inexpensive, including one across the ac input to the amplifier (after the circuit breaker) is easily done. Such a device will be included on all amplifiers going forward.

Because the terminal on the PWB adjacent to the one that failed could hold off at least 800 V dc, a manufacturing defect is unlikely. However, as a general rule, I will begin applying conformal coating (see the section on Special-Purpose Chemicals on the Workbench page) to terminals operating at 400 V or greater that are adjacent to a ground plane. One product that works well for this is the MG Chemicals 419C-55ML Acrylic Conformal Coating (2 oz bottle).

Closeup of test set

So, it is expected that this in-service failure will lead to more reliable systems in the future. One of the greatest challenges in engineering and design is anticipating things that might go wrong. Once you knave an idea of what might fail, it is then possible to take steps to mitigate the possibility of failure in the field.

Circuit Board Issue Update

It has been more than a year since the problem documented above occurred. The amplifier has been in use ever since without issues.

Following up on the voltage standoff concern discussed in the previous post, I came upon a newsletter article from ExpressPCB. The article discussed the maximum standoff voltage for typical circuit boards. A portion of the article is abstracted below.

"The dielectric material of our 2-sided PCBs should handle up to around 50 kV with respect to dielectric breakdown between layers. FR-4 material will absorb moisture out of the air but the solder-mask does a good job of limiting that through the largest surface areas, leaving only the boards edges. This absorption of moisture around the edges typically has very little effect on the dielectric breakdown voltage and may lower the 50 kV number to 45 kV (to be safe). The main issue where temperature and humidity might come into play is arcing between adjacent current-carrying circuits. With PCBs incorporating a solder mask, there are typically no issues with temperature or humidity from a voltage standpoint; the dielectric breakdown of our solder-mask has been tested to 2800 V/mil of thickness. Our typical thickness in and around the conductors is about 1 – 1.2 mils. This would allow for about 2500 – 3000 V before there might be any breakdown of this layer. For PCBs without solder mask, it is theoretically possible for an arc to occur between adjacent copper traces in high temperature, high humidity if the conditions and voltage are just right to overcome the breakdown voltage of the surrounding air.

The main concern of bare PCBs in a humid environment would be oxidization of the exposed conductors. Immersion silver finish has been known to oxidize if stored without environmental protection. We recommend that bare immersion silver boards be treated similarly to moisture sensitive components, sealing any opened bags, using desiccant, and protecting from UV light exposure. Boards shipped with the tin lead finish are not nearly as susceptible to these environmental concerns and can typically be stored for a year without issues. Once a PCB with solder-mask is assembled, the solder typically coats all of these exposed areas and the oxidation of solder is very low."

To join the ExpressPCB mailing list see www.expresspcb.com.