I've finished the wiring from the Power-686 to the Modulus-686 and Guardian-86 modules, as well as the speaker outputs from the Modulus-686 to the Guardian-86.
I'm using the Power-686 Wiring Harness, so that I don't have to futz with purchasing and terminating their choice of connectors. The wire is all the same color, so to keep things clear I used a bit of heat shrink tubing to band the positive with red and negative with back (leaving the ground unbanded).
The connections to the Power-686 are a bit odd, as the big connector on the left is ++G, while the one on the right is G--. I guess that makes space for those smaller pluggable connectors in the middle, which are both +G-. Rather than utilizing those little connectors, I just used the main connectors to feed each Guardian-86.
This weekend I'll start wiring interconnects. I remembered last night that I need the Quasi-balanced only when using a pre-amp with RCAs. My Benchmark DACs have balanced XLR outputs, so I'll probably build XLR Male-to-Female first.
I'm using the Power-686 Wiring Harness, so that I don't have to futz with purchasing and terminating their choice of connectors. The wire is all the same color, so to keep things clear I used a bit of heat shrink tubing to band the positive with red and negative with back (leaving the ground unbanded).
The connections to the Power-686 are a bit odd, as the big connector on the left is ++G, while the one on the right is G--. I guess that makes space for those smaller pluggable connectors in the middle, which are both +G-. Rather than utilizing those little connectors, I just used the main connectors to feed each Guardian-86.
This weekend I'll start wiring interconnects. I remembered last night that I need the Quasi-balanced only when using a pre-amp with RCAs. My Benchmark DACs have balanced XLR outputs, so I'll probably build XLR Male-to-Female first.
Yes, the 3U/400mm Dissipante splits the side heatsinks front and back, making it less efficient than a single 4U/300mm heatsink that's usually recommended for the Modulus-686. However, there are a few reasons that I think it will probably be fine:
- When I attached the mounting rails to the heatsink halves, I added thermal compound between them and clamped them together. Therefore, it's not only that little heat spreader conducting, but also the the inner sides of the heatsink halves.
- I choose a transformer with 2x22VAC secondaries, less than the typical 2x25VAC that is often used. This should provide ±30VDC when my mains are at their maximum of 124VAC. The mains tend to vary between that and 120VAC, so the actual voltage will often be lower.
- All of my speakers are nominally 8 ohms, so current will be half what it would be with 4 ohm speakers.
- I'm not a volume freak, so the amp will usually not be pushed very hard.
The LM3886es are spaced evenly across the mounting bracket, so there isn't a strong need for a heat spreader. Basically you get three LM3886es on one heat sink and three on the other if you place the module so that it's roughly centred. It would be better if the three LM3886es of each MOD686 half were centred on each of the 200 mm wide heat sinks, but I don't want to lock my customers into using that chassis.
I do agree that the two-heatsink design of the 400 mm and 500 mm deep Dissipante series is the Achilles heel of those chassis. I prefer the 300 mm deep version for that reason, but fitting all the components within that chassis can be a challenge.
I love the wiring harnesses and bootlace ferules, by the way. Very nice job.
Tom
I do agree that the two-heatsink design of the 400 mm and 500 mm deep Dissipante series is the Achilles heel of those chassis. I prefer the 300 mm deep version for that reason, but fitting all the components within that chassis can be a challenge.
I love the wiring harnesses and bootlace ferules, by the way. Very nice job.
Tom
I was out for a week, so I had a pause in my build. This was actually a forced pause, because one of the Modulus-686 boards failed. The resistor R44 fried itself, and fortunately my trip was a visit to Calgary, where Tom Christiansen of Neurochrome lives. He was kind enough to repair the board in exchange for a pint of beer. Thanks again, Tom!
He told me that, "The only way for R44 to fry is if there’s a potential difference between XLR pin 1 and the ground connection of the power supply connector." Rather than reinstall the repaired Modulus-686 and cross my fingers, I decided to investigate this possibility.
I reviewed the wiring, performed a bunch of tests, and eventually discovered that the positive rail of that channel's Power-686 board was somehow connected to the chassis. Consequently, the +32VDC went through the chassis into the mains earth, to the power bar, back up to the Benchmark DAC1's chassis, into the XLR output's shield pin 1, and down to the Modulus-686. The 32V above the supply ground fried the resistor. I'm still not sure why this didn't blow the fuse (currently 10A). I have noticed the AnTek transformer humming a bit, so perhaps this short is causing a somewhat heavy load.
The Power-686 is connected to the chassis only via the stand-off screws. Installed in the chassis, there's definitely a connection between the positive rail and the chassis (including the standoff screws). I removed the Power-686, and there is no connection between the positive rail and the screw pads, which indicates there's likely no short within the PCB itself.
It looks like the snap-in capacitor leads haven't been trimmed, so they're a bit longish. Although they seem fine, perhaps my stand-offs are too short, and one of those positive leads is touching the case. I'll install taller ones and retest.
He told me that, "The only way for R44 to fry is if there’s a potential difference between XLR pin 1 and the ground connection of the power supply connector." Rather than reinstall the repaired Modulus-686 and cross my fingers, I decided to investigate this possibility.
I reviewed the wiring, performed a bunch of tests, and eventually discovered that the positive rail of that channel's Power-686 board was somehow connected to the chassis. Consequently, the +32VDC went through the chassis into the mains earth, to the power bar, back up to the Benchmark DAC1's chassis, into the XLR output's shield pin 1, and down to the Modulus-686. The 32V above the supply ground fried the resistor. I'm still not sure why this didn't blow the fuse (currently 10A). I have noticed the AnTek transformer humming a bit, so perhaps this short is causing a somewhat heavy load.
The Power-686 is connected to the chassis only via the stand-off screws. Installed in the chassis, there's definitely a connection between the positive rail and the chassis (including the standoff screws). I removed the Power-686, and there is no connection between the positive rail and the screw pads, which indicates there's likely no short within the PCB itself.
It looks like the snap-in capacitor leads haven't been trimmed, so they're a bit longish. Although they seem fine, perhaps my stand-offs are too short, and one of those positive leads is touching the case. I'll install taller ones and retest.
In this case it's a consistent short between the Power-686 positive rail and the chassis. We'll see what happens when I add taller stand-offs, and possibly add a mica sheet between the board and the chassis. I'm probably going to trim the leads as well, just to be doubly sure.
A connection from V+ or V- to chassis is guaranteed to blow R44.
R44 is the connection between the EMI filter ground and power ground. It connects to XLR pin 1, which connects to the chassis. So if you have a hot chassis you end up with a connection through the input wiring to ground. The weakest link there is R44. When everything is connected correctly R44 never carries any substantial current.
I normally recommend 10 mm standoffs. If you trim the leads on the power supply caps you can get down to about 6 mm. The mounting holes in the Power-686 board are intended for M3 machine screws. US #4 can be used as well. I have sized the pads around the mounting holes such that M3 hardware will never come into contact with any voltage. The pads are floating (i.e., not connected electrically).
One guy forced a US #6 (~3.5 mm in diameter) into the mounting holes and used a hex nut on the top of the board. The nut wore through the solder mask and shorted V+ or V- to ground.
With #4 or M3 hardware you will not need mica or any other insulating washers.
Tom
R44 is the connection between the EMI filter ground and power ground. It connects to XLR pin 1, which connects to the chassis. So if you have a hot chassis you end up with a connection through the input wiring to ground. The weakest link there is R44. When everything is connected correctly R44 never carries any substantial current.
I normally recommend 10 mm standoffs. If you trim the leads on the power supply caps you can get down to about 6 mm. The mounting holes in the Power-686 board are intended for M3 machine screws. US #4 can be used as well. I have sized the pads around the mounting holes such that M3 hardware will never come into contact with any voltage. The pads are floating (i.e., not connected electrically).
One guy forced a US #6 (~3.5 mm in diameter) into the mounting holes and used a hex nut on the top of the board. The nut wore through the solder mask and shorted V+ or V- to ground.
With #4 or M3 hardware you will not need mica or any other insulating washers.
Tom
While you have the board out it would be worthwhile to measure from each mounting hole to V+ and V-. If there is a connection there I definitely want to know about it. All my circuit boards are electrically tested by the manufacturer. If a defective board slipped through the test I'll need to go and beat them over the head with a 2x4.
Tom
Tom
Shall do!
I've also decided to change to nylon standoffs and screws, to completely prevent this from affecting my build. I'll use 10mm standoffs, as mentioned above.
I would not use nylon screws. They tend to stretch and get brittle over time.
The boards are designed for use with metal hardware and there is absolutely no issue using metal hardware as long as it doesn't extend past the mounting pad. I normally use 10 mm aluminum standoffs (4.5 mm hex). I use M3 screws from the top and bottom to attach the standoff to the chassis and the board to the standoff. Alternatively, I'll use a longer stainless M3 machine screw inserted from the bottom of the chassis, a nylon spacer, and a plain M3 hex nut (5.5 mm hex) on the top of the board. I've never had issues with any of these methods.
I much prefer the first method, i.e., using hex standoffs attached to the chassis with screws. It greatly simplifies the assembly work.
I bet one of the pins of the reservoir caps shorted to the chassis. You might find a small burn mark or weld on the chassis where it connected. There's no issue with this. I'm just pointing out that you could potentially find evidence that would shed some light on exactly what happened.
Tom
The boards are designed for use with metal hardware and there is absolutely no issue using metal hardware as long as it doesn't extend past the mounting pad. I normally use 10 mm aluminum standoffs (4.5 mm hex). I use M3 screws from the top and bottom to attach the standoff to the chassis and the board to the standoff. Alternatively, I'll use a longer stainless M3 machine screw inserted from the bottom of the chassis, a nylon spacer, and a plain M3 hex nut (5.5 mm hex) on the top of the board. I've never had issues with any of these methods.
I much prefer the first method, i.e., using hex standoffs attached to the chassis with screws. It greatly simplifies the assembly work.
I bet one of the pins of the reservoir caps shorted to the chassis. You might find a small burn mark or weld on the chassis where it connected. There's no issue with this. I'm just pointing out that you could potentially find evidence that would shed some light on exactly what happened.
Tom
Thanks for the guidance, Tom. I'll stick with the standoffs (I've got brass on hand) and metal M3 screws.
As for the evidence of contact, there's only a small spot with bare aluminum, that might be where one of the capacitor legs touched. Nothing looks charred or welded. I suspect the current was low enough that the damage wasn't that extreme (if in fact that's where the connection occurred).
I added the taller standoffs and installed the boards, but the short from the positive rail to the chassis remained, so it wasn't a capacitor lead touching the case. (I had already confirmed that they weren't long enough, even with the shorter standoffs.)
I removed the Power-686, but left the standoffs on. Poking around with my DMM, I noted that the short is happening through the standoff on the positive side of the board, on the input end where the transformer connects. Tom coincidentally asked about the screws in a private email, and I noticed that the "washer-head" was touching the heatsink for one of the rectifier diodes, and this heatsink happens to be connected to the positive power rail.
It's the only one of the four screws near the heatsinks that came into contact, but I'll change them all to a screw with a smaller head, and call the problem solved.
I removed the Power-686, but left the standoffs on. Poking around with my DMM, I noted that the short is happening through the standoff on the positive side of the board, on the input end where the transformer connects. Tom coincidentally asked about the screws in a private email, and I noticed that the "washer-head" was touching the heatsink for one of the rectifier diodes, and this heatsink happens to be connected to the positive power rail.
It's the only one of the four screws near the heatsinks that came into contact, but I'll change them all to a screw with a smaller head, and call the problem solved.
BTW, I thought I would mention that I've been sharing a portion of my exploits here on the forum for everyone's entertainment and illumination. Behind the scenes, Tom has been responding to frequent emails, helping me each step of the way. I'm quite blown away by his attentiveness. 
If I had been paying closer attention, I might have realized earlier that the screw head was encroaching on the heatsink.
If I had been paying closer attention, I might have realized earlier that the screw head was encroaching on the heatsink.