ECP Audio T4 Headphone Amplifier - Built and Sold by BeezarAudio

[tomb]

While I'm still rinsing the PCBs of flux residue, I've also begun work on the power transformer. These are the very last steps prior to final assembly with the casework.

The power transformer connections and mains voltage setting all use the Molex KK254 terminal system. These are convenient crimp-on terminals with a dizzying array of housing and header offerings. Dsavitsk and I originally thought that we would be able to order the transformers with the headers already installed. Unfortunately, that did not turn out to be the case. So, the original prototype and loaner T4 were built with the power transformer leads soldered into place and the voltage setting jumpers were spent leads soldered into the PCB pads.

There were two issues with this:

1. Soldered mains voltage setting jumpers do not allow for user-changed voltage settings.
2. Because the transformer is mounted to the case bottom, disassembly becomes very difficult if the leads are soldered to the PCB.

With #1, this is a missed opportunity for convenience, because the power transformers are spec'd for 220V-110V windings and the IEC inlet allows the use of all types of domestic and international cords/plugs. However, if the actual voltage setting on the PCB requires soldering skills, this severely limits the options to the customer. So, if a US customer eventually wants to sell his/her amp to an international customer, soldering skills are necessary - even more, sufficient soldering skills to remove existing solder and jumpers.

With #2, there are a total of 9 leads from the power transformer that attach to the PCB. When assembling or disassembling, the fact that all the leads are soldered to the PCB, yet the transformer is bolted to the case bottom, means that assembly/disassembly is awkward at best, potentially damaging to the transformer at most.

So, once I started building T4s for sale (after the two prototypes), I started researching the Molex KK terminal system. Unfortunately, as noted above, I've never encountered any other parts system that is as widespread and confusing as the Molex. I eventually settled on the KK 254 terminal system, which seemed to be the proper choice for 22 gauge wiring and a 2.54mm pitch (pin spacing on PCB). I used the headers with friction locks, housings with both locking ramps and rib "wings." All pins and crimp terminals are gold.

The next step was to acquire a crimper tool. Knowing that this would be used for hundreds of crimp processes, I wanted to get the best tool available. So, I started looking for the actual Molex-brand tool: about $350, everywhere I looked. Well, that ended that. So, I started searching elsewhere and finally settled on this IWISS tool from Amazon:

It's Chinese and those handles are basically completely styrene, not mixed plastic with a gray rubber grip. However, the crimping die seemed of good quality and it got pretty good reviews, so I gave it a shot. It seems to be working OK, although you have to hold it upside down to use it properly. I have thought about removing the dies and re-installing them reversed, but I haven't had the time to try that, yet.

Here's a pic of other tools I used for the Molex terminal system:

Besides the IWISS tool, there's a set of Ideal Stripmaster wire strippers, a Lowes wire cutter/stripper, and a Harbor Freight 98-cent smooth-jaw needle nose pliers. Not pictured is also an X-acto knife that I use for the crimp terminals - more on that later.

Here's a pic of all of the Molex KK254 system pieces that I purchased, organized in a Harbor Freight parts box:

And … here's a pic of the SumR toroid power transformer used for the T4:

The transformer comes just as you see pictured, with two rubber mounting pads and a steel rubber clamp washer. No mounting bolt, nut, or washer/lock washer was included.

For perspective, here's the entire arrangement laid out on one of my work tables:

The first step I'm going to do is to make more Molex shorting jumpers that are used to change the voltage setting. Here's an excerpt from the T4 manual:

So, these jumpers provide the mains voltage setting capability, as indicated. Two jumpers are supplied per T4, so any customer can switch back and forth between 110VAC to 220VA mains voltage.

There's probably some Molex shorting jumpers already made and available somewhere, but none of my searching turned anything up. It may be that they're not manufactured for the KK 254 series, but are available for smaller series such as those used on PC computers that we've all seen. Anyway, it's good practice to do with the crimpers, strippers, etc., before starting on the very costly power transformers.

For the ongoing six T4s under construction. I'll need 12 shorting jumpers made up and a couple of extras, just in case. We begin with the hookup wire. I've used John's Wire Shop (user: navships) on ebay for years. He sells great multi-stranded, Silver Plated Copper (SPC), Teflon-insulated wire. It comes in all different sorts of colors and he offers it in lengths that won't require a bank to purchase and a storage shed to store:

Above is a recent set I purchased of five different colors, 10 feet each. Three are multi-color and then two are purple and brown, respectively. Again, very convenient for hookup wire. The Teflon insulation is also practically melt-proof and does not shrink away from a joint under soldering, like cheap PVC-insulated hookup wire. The only drawback is that his wire is probably not appropriate for headphone cabling, since it's so stiff. Stiff is good for hookup wire, though - it stays where you put it and at the same time, doesn't compress or smash very easily. All in all - good stuff!

Also in that pic, you can see some of the shorting jumpers already made (3 of them) and then new, empty, two-pin housings (white things) and some gold crimp terminals at right above the brown wire.

What I'm going to do is cut the wire, strip the wire, crimp the terminals on the wire, then insert them into the housings. That's up next!

 

[paradoxper]

Uhm.

I want purple hookup wire! No, no green!

Or how about you special order some coquelicot to match the boards.

 

[tomb]

Making the Molex shorting jumpers helps me transition from thinking totally about soldering to thinking totally about not soldering, but making mechanical connections. The shorting jumpers are small, simple, and help me get into a rhythm in using the crimper and the Molex terminals and housings. There's a lot of things to remember if you don't do this every day:

1. How long to cut the wire
2. How much insulation to strip off.
3. How to place the crimp terminal in the crimper
4. How to insert the crimped terminal into the housing

It's better to re-learn all of that on a small, inexpensive shorting jumper - than hundreds of dollars of power transformers.

Pictured below is a KK 254 crimp terminal:

There are two lugs and then the bent, looped portion is what makes the terminal electrical connection. The stripped portion of wire must be long enough to insert in the lugs from the left, so that the wire extends slightly beyond the 2nd lug. The idea is that the lug on the right crimps around the stripped portion of the wire, while the first, larger lug crimps around the wire insulation. The combination makes an electrical/mechanical connection with the small lug, then a second mechanical-only connection with the large lug. It makes for a very stable, secure connection. However, it's very important to get the correct length of the stripped end. Looking at the ruler, this works out to about 1/8" (the ruler is showing 32nds on top).

I won't go into detail about the IWISS tool. If you are interested, you can refer to this review on Amazon that is actually a tutorial on using the tool. (One of the major complaints in the reviews is that the tool doesn't come with a manual.) Here's the link: How to Use the IWISS Ratcheting Crimping Tool. Suffice to say that placing the crimp terminal within the dies of the crimping tool, and inserting the wire, is critical. Thanks to the tool, though, the act of crimping does not. That is the major reason that you need the tool: to make the proper crimp, without too much pressure or too little. The tool works by squeezing to crimp, but at the point of greatest pressure, it ratchets back open automatically - making it impossible to crimp too hard or too soft.

Anyway … to make the shorting jumpers, I cut 1" wire lengths, then strip 1/8" off of each end. Then I crimp on the terminals. This is the result:

That is a perfect case of the proverbial, "one picture is worth a thousand words." I could describe how these crimp terminals attach to the wire, but seeing it above makes it perfectly clear: one lug is crimped onto the wire insulation, the other directly on to the wire.

We're not done yet, though. The way the crimp terminals work in the housings is that there is a small bent tab on the top of the terminal that "catches" a slot in the end of the housing. This allows you to push the terminal into the housing until this tab reaches the slot. The tab springs up into the slot, so that you can't pull the terminal back out. (Use a very tiny screwdriver tip to press the tab down while pulling the wire/terminal back out, if necessary.)

In working with the crimp terminals, that tab often gets bent back down, flush with the crimp terminal, so it won't "catch" when you push it into a housing. So, I take an X-acto knife to bend the tabs back up:

Finally, the smooth jaw needle nose pliers helps me to bend everything back into a straight line. Again, working with the IWISS crimping tool, the crimp terminal sometimes ends up getting bent and distorted. This will prevent a proper insertion into the housing. The wire and tops of the crimp terminal should be as straight as possible across the top (except the bent up tab):

The next step is to bend the wire in half and insert both crimp terminals into the housing:

You can easily see the slots, in the housing above, where the bent up tabs on the crimp terminals "catch." The bent up tab is open to the right, so inserting to the left means they go in, but they can't be pulled back out (unless you use that screwdriver tip to press down on the tab through the slot).

Once all of the above procedures are established (and re-remembered), it goes very fast. These were all done in a little less than an hour:

Next, I test the jumpers for continuity (zero resistance), just to make certain:

Sorry if that was excruciating details about a tiny connector. I thought it important, though, to explain the correct way to do it. Once I start on the transformers, all the rules go out the window.

 

[nwavesailor]

Has anyone used A2900tubes in the T4?

Tom is checking the data sheet to be sure there are no issues.

 

[Marutks]

I am using GEC A2900 tubes with the T4

 

[paradoxper]

I am using GEC A2900 tubes with the T4

Comparisons to stock or alternatives?

 

[Marutks]

I haven't noticed any sound differences from stock tubes.

They do Mullard flash:

 

[paradoxper]

I haven't noticed any sound differences from stock tubes.

They do Mullard flash:

Thank you.

 

[nwavesailor]

I haven't noticed any sound differences from stock tubes.

If you found no difference between the A2900 and stock tubes I may hold off and see how the tubes I already own work in the T4. I wonder if this is because the A2900 you are using are in a tube hybrid amp with the T4.

I have several 12AT7 to try with the T4:

French production Philips Miniwatt,
UK production Marconi B309 and Brimar CV4033
German Siemens 6201
US Tung Sol 12AT7

I also have several good pairs of 12AU7, but not sure how these will work with the aplification factor of 20.

The tube lore is that the A2900 is THE 12AT7 Holy Grail!

 

[tomb]

I haven't noticed any sound differences from stock tubes.

They do Mullard flash:

Ah yes - the Mullard flash! The Mullard CV4024 tubes that I recommend do that.

You guys have got me interested in the A2900, now. I had not heard of it until nwavesailor asked me about it. I will have to try a pair and report back.

 

[tomb]

So, here we go - the last thing before casework assembly!

As mentioned earlier, the rules for the Molex connectors go out the window when it comes to the power transformers. The reason is that even though the leads on the SumR toroids are 22ga, the insulation is 600V. So, the insulation is easily twice the diameter of the stranded wire bundle. A different technique with the Molex terminals and the crimper must be used.

Here's a pic of one of the power transformers with the leads trimmed:

I am not a magician. There's no way I would trim the leads on an expensive, custom-made power transformer without having developed a guide of some sort. What you see there is my cheat sheet for the entire T4 construction. On the left, I have all of the parts documented, along with their heights, and ordered in the sequence that they should be soldered into the PCB. Also in that list is the lead-bending lengths to be used with the red plastic lead bending guide (seen in the early posts) for resistors and diodes.

Here's the section on the PT:

It's quite a bit of information, but absolutely necessary to prevent ruining one of the power transformers. The first step, as mentioned above, is trimming the leads to those lengths in the far right column. Then the groupings of the windings guide me in what to use for the Molex housings:

• AA-BB-CC-DD-Shield all go into a single, 5-pin housing,
• HT-1 and HT-2 go into a 2-pin housing, and
• LT-1 and LT-2 go into a 2-pin housing.

Here's a pic (sorry - slightly out of focus) of a crimped transformer lead:

What's necessary is to strip the lead so that the stripped portion is long enough to accept both crimp clamps on the Molex crimp terminal. The 600V insulation is so thick, that it won't even fit into the crimp terminal's larger lug - believe me, I've tried. So, clamping on the insulation is not an option. Crimping on the wire with both lugs requires a double crimp, because the dies in the IWISS tool are dual-sized: one for the wire, one for the insulation. This actually works quite well. If you strip and crimp at the proper length, the insulation is sized just right to press-fit into the housing. So the overall effect is still quite robust and little stress is encountered by the wire ends.

Here's a pic with the primary leads (4 plus a shield) completed with the crimp terminals:

All that's left is to insert these into a 5-pin housing.

Once you get into a rhythm doing this, it goes very, very fast - faster than soldering, if you can believe it. Almost the same amount of time is taken in cutting/stripping/tinning the leads if you solder. Then you have to apply the solder and wait for the melting and cooling. None of the soldering time is needed with the crimp terminals. Here's a power transformer ready to plug into a PCB:

Note the extremely important order of the leads in the 5-pin housing:

1. Black
2. Brown
3. White
4. Orange
5. Green

It's also extremely important that Green (the shield) is at the top of the connector when plugged into the headers on the PCB! You can actually reverse the order Black to Orange or Orange to Black (same windings relationship, but reverse order), but Green has to be the first connection, always.

Being AC voltage and only two wire connections, the order with the HT leads and the LT leads makes no difference. Just make dang sure the red and yellow go in the LT housing and the blue and gray go into the HT housing! Mixing that up is a no-no!

A test fit into a PCB (no bottom plate to mount the transformer for the time being):

As stated, once in a rhythm with the cutters, strippers, and crimper, things go very, very fast. I finished the rest of the transformers in under a couple of hours:

Final casework assembly begins!

EDIT: Should've mentioned this earlier, but except for the voltage-setting jumpers, I use the Molex housings with rib wings and locking ramp. As also mentioned, it doesn't really matter except for the 5-pin housing. The rib wings and locking ramp on the 5-pin connector strongly encourages that the transformer primaries won't be reverse-connected.

 

[paradoxper]

Bro. Where is our daily update!!!

 

[tomb]

Bro. Where is our daily update!!!

It's coming. I'm going through a transition right now to the case assembly, gathering up all the materials, and fixing a few loose ends.

Dsavitsk shipped the wood to me yesterday. FedEx says it should arrive on Monday, so things are progressing well. I should have some more updates tomorrow.

 

[tomb]

Wow - haven't posted in a couple of days and this fell all the way off to the second page of posts. This is a very busy forum section. I know it doesn't compare to the headphone section, but still didn't know it was this busy.

Anyway, starting with the casework … the first thing that has to be addressed is the thermal bar technology:

\
What you see here are several items, all needed to begin the casework assembly. I joke about dsavitsk's casework design that they are puzzle boxes. This is not far from the truth, but the result is something magical. His casework often reminds me of Howard Hughes's H1 and the first flush rivet finish. If dsavitsk could get rid of all attachment devices, he'd do it. Meanwhile, everything is designed so that with the parts of the case most visible, it's difficult to see the screws, etc.

This can make assembly difficult, if not impossible - if you aren't careful and don't have a plan. However, after doing this for a while, I've developed a system. That system begins with attaching the completed PCB to the case bottom plate. The bottom plate is what you see most prominently in the pic above. All of those drilled holes are for three things:

1. Thermal bar system,
2. PCB standoffs, and
3. Brackets for the front plate, back plate, and sides.

To the left of the bottom plate is a sheet of Bergquist thermal pads. In most instances, these pads are sized for TO-220 power transistors and used between a heat sink and the transistor. The pads have an adhesive backing, are made of cloth fibers, and impregnated with a coating that flows as a highly viscous liquid/goo under heat. It maximizes the heat transfer without all the mess of a classic thermal grease.

Above the bottom plate, you can see several aluminum rectangular bars. These are the thermal bars used for heat transfer in the T4, one per amplifier. The white rectangles are 2mm thick ceramic insulators. At top right, there's a jar of classic thermal grease (still needed for other things) and a bottle of screw thread Loctite. Just to the right of the Loctite are vertical LED holders. One of these is used in soldering the primary LED to the PCB. It goes just behind the XLR jack and is used to illuminate the ECP Audio logo on the amplifier top plate. Finally, there are some machine screws and high-temp fiber washers. These are used to attach the thermal bar to the PCB.

The first step is to apply Loctite to all of the thermal bar holes. These holes are threaded for the screws used to attach the thermal bar to the PCB:

The Loctite flows into the holes and then the excess is wiped off. It's good or bad, depending on where you are in the assembly of the casework, but the Loctite can take 2-3 days to fully cure.

Once that's done, the Bergquist pads are peeled off the plastic sheet and applied to the ceramic insulators. The Bergquist pads are used here to remove concerns with the messiness of the thermal grease, which ends up everywhere, despite the strongest attempts to keep it controlled. The ceramic insulators are used for the primary interface between the PCB and the thermal bar. The heat sink pads on the PCB are actually flowing with 250+ Volts and any chance of errant thermal grease in this interface needs to be removed. Thus, the use of the Bergquist pads:

Once those are applied, the next step is apply thermal grease to the back side of the ceramic insulators and place them in their proper positions on the thermal bar:

You have to be careful here, because those ceramic insulators are "floating" on grease on that thermal bar. Any slight touch and not only will they move, but you'll have white thermal grease mess everywhere!

The very next step is to place the thermal bar in the middle of the table (carefully!) and then hover the PCB over the bar until you get the holes lined up. We've already installed the standoffs under the tube sockets, IEC inlet, and choke. Because of the fact that the thermal bar with ceramic insulators is exactly the same height as the standoffs, you can place the PCB over the thermal bar/ceramic insulators so that the holes match up exactly and everything is stable. Then carefully, I screw in a couple of the thermal bar screws so that I know the bar won't move. Then the rest of the screws can be installed and torqued down. The Loctite will cure over the next several days and the connection will become quite secure.

The threaded holes in the thermal bar are through-holes, so they'll be used with the case bottom plate, just as if it was five more standoffs. Here's the thermal bar fully attached.

Looking at the top of the PCB, you can see the five screws and high-temp washers, all in a row:

You can now see how those pads on the PCB with all the tiny holes makes sense. Those pads are approximately the same size as the ceramic insulators underneath. Five pads on the PCB and five ceramic insulators, all screwed into a thick, solid aluminum "thermal bar" with five threaded holes.

If you look closely at the near tube socket and the transistor to its right, you can see what I mean about the white thermal grease - it gets everywhere, no matter how careful you are! (It was since cleaned up.)

The next step to complete mounting the PCB to the bottom plate is to attach the remaining standoffs. There are four more mounting holes on the PCB that were not attached to anything. Standoffs are attached and screwed in at these holes. One is different than the rest, however:

This standoff mounting hole serves as the primary ground connection between the PCB and the casework. You'll see later on that care has been taken to have the entire casework serve as the amplifier ground. This is done by actually masking the attachment points on the anodized case parts so that they remain uninsulated. Beginning with the pad at this standoff, the ground is contiguous through every metal part of the casework. To ensure that connection is made, a star washer is used here to ensure that the pad is cut into with the star washer. That ensures metal-to-metal contact among the screw, PCB, and standoff - then all the way through the rest of the metal casework.

All the standoffs are attached in this pic, ready for attachment to the bottom plate. Most important, additional thermal grease has been applied all along the bottom of the thermal bar. This ensures a good heat transfer between the thermal bar and the bottom plate:

And the bottom plate now completely attached:

Once the bottom plate is attached and the thermal bar screws are torqued in place, top and bottom, we can inspect the thermal bar interface between PCB and bottom plate:

And a top view of the entire PCB/bottom plate assembly, ready for the front plate, back plate and sides:

Front plate(s) and back plate are next!

 

[paradoxper]

Ah. Life. Feels back to normal.

Aside: I can't wait to see T8: off-board heatsink and tribulations there-in.

Keep it up, tomb!