March 1

Streaming video is offline for a while, so if you're trying to access it, it won't appear. This is because I did a CPU upgrade on the server and the motherboard has bad capacitors. The capacitors were loading down the regulators due to their leakage, and the additional load of the Athlon CPU proved to be more than the regulators could handle. Hence, they went up in smoke.

I've ordered a replacement motherboard of an upgraded model using a compatible chipset. Hopefully, the operating system will be able to boot up on the new motherboard without major reinstallation required. The faster CPU was nice, while the board lasted--about 3 hours.

So if you notice broken links or anything else out of the ordinary, it's because I moved this site rather hastily to an offsite mirror server. Hopefully I'll have the replacement motherboard next week and have it all back up and running soon after.

March 2

I just completed modifications to the two bridge mono Hafler amps earlier this evening. These units had the optional bridge boards, which were needed when driven by single-ended sources like audiophile preamps. But since my DCX2496 has differential balanced outputs, I thought about whether there would be any advantages to driving the amps differentially direct off the DCX2496, rather than through the bridge board.

I came up with a couple of reasons to ditch the bridge boards: the first was that the bridge board would amplify input noise by 6dB. Using a differential feed would reduce the amp input gain, but also would have any noise riding on the cables cancelled due to common mode rejection. I would be utilizing an additional 6dB of the DCX2496 dynamic range. And finally, there would be fewer transistors in the signal path.

So to implement this change, I decided to remove the bridge boards and install TRS 1/4" jacks. Furthermore, I replaced the UTP cables with shielded cables to go from the input jack to the amp driver boards. Tip was wired to right channel, Ring was wired to left, and Sleeve is the return. After the amps were rewired, I made new cables to go from the DCX2496 to each of the monoblock Haflers. These cables have 6dB attenuators that I built, the purpose being to gain match the max clip level of the DCX2496 to the max clip level of the Hafler, thus making full use of the DCX2496 dynamic range. Part of the design change was to place the attenuators at the amplifier end of the cable, instead of where they were at the DCX2496 end. The former configuration would leave the length of cable at a higher impedance and prone to pickup of AC magnetic fields. By placing the attentuators at the end of the cable, the impedance in the cable is kept low, by direct connection to the source, and any noise that does get in is attenuated by the attenuator, which consists of two 2.51K metal film 1% resistors for each half of the differential signal.

In the midst of these modifications, I observed that one of the DH500s was idling hotter than the rest. I traced it to a bias adjustment on the left channel. Before adjusting this, I powered down the amp and got out my trusty inline ammeter and plugged the amp into that. I discovered that at idle, the amp was pulling 2.5 amps from the line. So I slowly tweaked the bias adjustment until the line consumption reached a dip. I verified the adjustment by running signals of various amplitudes through the amp. No problems there. Then I adjusted the input offset until the DC component in the output measured less than 1mV. The amp ran much cooler, and the current draw was down to 1.6A at idle. Good, I thought. This will save some electricity costs. I made a note to check the bias and idle current on the rest of the amps as I go through the modifications.

A couple of hours later, I had finished the two monoblock modifications. After blowing what little dust I could find out of the amp chassis, I buttoned them up and reinstalled them into the rack. The power up cycle was to be the true test. And I was delighted. These two amps had gone from being my two noisiest, to my two quietest. The buzz and 60Hz hum was gone. But I still had the advantage of bridging in mono mode for quadruple the per-channel power rating.

I am considering the shielded cable modification to the other two stereo mode Haflers, as well as moving the attenuators to the amp side. I realized that having a source impedance of 2500-ohms can play havoc with hum intrusion. But that's a project for another evening.

March 8

The server is back up and running with a faster CPU, faster system bus/motherboard and more and faster RAM. I had some strange things happen on this upgrade. The original motherboard couldn't take the load of a compatible, but upgraded CPU, apparently because of partially-shorted capacitors near the regulators. I ordered a similar, but upgraded motherboard on eBay and it came in Monday.

I got this motherboard up and running, but there are some oddities. For one thing, some components are out gassing and the odor is quite noticeable. It's becoming gradually less intense as the days pass, but it's enough to leave one wondering.

The next thing was that the Windows Media Encoder does weird things on this motherboard, mainly, encode streams that sound like they passed through a ring modulator. Stopping and starting the encoder numerous times can produce a good stream. There is some peculiar timing issue going on here. The stream seems to care about when the encoder was started--as in during which clock tick of some component in the system. If I started (by coincidence) at the precise right instant, the stream will sound like normal unprocessed music. I've some work to do to nail this one down.

The above problem led me to power off the server so as to do a cold boot. That revealed another quirk: the server would start up, produce one beep, then power down! That had me scratching my head all night Monday. At first, I thought it was the power supply, which was an upgraded unit from the original 300W PSU. I put in a huge premium PSU, one that I had pulled from my video editing workstation due to it's energy inefficiency, and put that in the server. Same problem! Powers up, one beep, then powers down.

I went through a painstaking series of elimination steps: removing the CPU, RAM and all component peripherals. The shutdown behavior persisted. Finally, in frustration, I put the original PSU back in and started it up--and it stayed up! Weird!

So the server has been reinstalled, and it is running. I just avoid powering it down! The smell persists, too, which is disconcerting. An odor similar to the one that the old motherboard gave out just before it went up in smoke, but not as pungent. It's not a PSU fault--all the voltages are fine, so I can only assume there is some component on the motherboard that gives off this odor.

After it was up and running, I upgraded to Server Extensions 2002, which adds Sharepoint and some new webbot functionality for nifty new features like the navigation bars.

March 20

Given the change of some drivers to higher compliancy types, I've felt the need to make some cabinet adjustments with regard to vent configuration. I built the mid-bass cabinets such that the internal volume can be adjusted via panels that swing in and out within the labyrinth of the vent. In light of this, I chose to adjust the vent for maximum volume available and step down the vent area by 10%. This brought Fc down about 6Hz, which is nice because it puts the corner at a musically-significant location--the start of the musical scale. A little bit of caulk and glue and all is well and getting more out of this cabinet configuration than before. It had been optimized for a EVM-18B prior to this. It now plays better with the EVX-180B.

I measured the exact Fc by placing two identical condenser mics, one in the vent and the other in front of the woofer dust dome, and connected their outputs directly to a dual-trace oscilloscope. As the corner frequency is approached, vent output will approach zero and below Fc, vent output will return, but increasingly out of phase as frequency decreases. This method lets me nail it to within 0.5Hz. I was able to spec both cabinets and align them for identical response.

While I was at it, I took care of another annoying problem: the tendency for the midrange cabinet to drift around on top of the mid-bass cabinet.. The solution was to insert 1/4" locking pins. I used some 1" long cut off sections of volume control shafts for the pins. 1/4" holes were drilled in the top of the mid-bass cabinet and 5/16" holes drilled in the bottom of the midrange cabinet. The pins were glued into the mid-bass cabinet and the midrange unit was set on top where it slides into the pin receptacles. Finally, a solution to a long-time annoyance.

The mid-bass cabinets play nicely with organ pedal tones now. As such, they contribute to the entire output of the system, even at very low frequencies. We're getting very close to the expected purchase date of the Bassmaxx ZR-18 drivers, and this will be a very exciting moment in history of this system. Meanwhile, tweaks are being done to optimize each cabinet.

March 29

Power line sags. That's what I'm working on this week. And it's looking to be a huge undertaking.

I made some changes to the programming of the loudspeaker management system, which allow the EVX-180Bs to be better utilized and balance the system load across all bass frequencies. However, this gave the balanced system the ability to pull more power from the line. I became acutely aware of this when a problem with audio muting occurred at high volume levels. One of the microprocessor-controlled EQs seems to be rebooting itself, due to line voltage sag.

This evening, I made some measurements with a voltmeter at various points along the AC circuit to the racks. It seems that I'm dropping 16 volts by the time the power gets to the rack and the amplifiers are placed under load. At idle, line = 116vac. Under load test, line = 100vac--not good!

Back at the breaker panel, I'm dropping 5 volts. The rest seems to be dropping over the 30' run of #6awg cable feeding the dedicated outlet that feeds the amp racks. I did some digging and tried hard to remember how I wired this 22 years ago, discovering that the #6 wire goes to a junction in the ceiling just above a conduit panel, where it connects to a #10awg BX cable taking from ceiling down to outlet, about 7' below the junction. Either this #10 wire is the culprit, or a combination of things. I also noted that the #6awg wire back at the breaker panel gets warm to the touch under test load conditions, so I think it's safe to believe that some voltage is getting dropped on this #6awg wire as well.

The prospect of replacing this cable is a daunting one because of the layers of sound and impact absorbing materials that are in the ceiling. First, there's an Armstrong drop ceiling. Above those panels are 6" of 24" wide fiberglass. Above that is 3/4" thick rigid high density fiberglass panels, nailed to the joists above. Above that is two 6" layers of fiberglass insulation. The cable traverses somewhere in this layer. Part of the difficulty is that the ceiling panels are jammed in so tight that they can't be lifted and slid to the side for access. The insulation is packed tightly behind them. I am not quite sure how I managed to put this ceiling together! I only recall that it was difficult. But the fact that the #6awg cable is inadequate. Dropping 16 volts is just not acceptable. Amplifiers need 120vac to achieve their full rated power. I have many projects to deal with, so this one will be "on my mind" for a while as I formulate a strategy for dealing with this very difficult wiring upgrade and the eye hazard of the fiberglass panels.