[tomb]

Nice review!

Have a question for Tom. Can the low setting on the amp be changed? My two cans are an Ether-2 and Abyss Diana. They have 16 ohm and 40 ohm loads accordingly. The Diana is a pretty tough load, but sounds great with the right amp.

Sorry, but no. The Z-switch switches between two different output windings in the Lundahl output transformers. One is optimized for 32 75 ohm loads, the other for 300 ohm loads. (Note that the actual output impedance is nowhere near 32 75 nor 300, though, it is much, much lower in both settings.)

Anyway, you'd have to rewind the output transformers to change either Z setting that's obtained through the Z-switch.

 

[Marutks]

One is optimized for 32 ohm loads, the other for 300 ohm loads.

Owner's manual says it is 75 and 300 ohms.

 

[adydula]

Tom,

Do you think with the 32 ohm position this would work well with a 16 ohm load like the persons concern is?

Alex

 

[Redwingnine]

Thanks, Tom. That's what I thought.

The website states the low side is 75 ohms, not 32 ohms. Can you clarify? Thanks!

 

[tomb]

Boy, that caused some consternation. Sorry, but I was going on memory and confusing with the Cinemags we used in the T3. The Lundahls are definitely optimized for 75 ohms on low Z, because of Focals.

 

[tomb]

This was further complicated and confusing because AtomicBob's tests (downloadable soon from Beezar Audio) were based on 32 ohms for his Low Z tests. I actually used 78 and 61, because of the resistors I had handy and also because there are a lot of headphones at 60 ohms (Sony V6, KSC-75, K701, etc.).

BTW, he measured the following for actual output impedance on the T4:
HiZ: 26 ohms
LoZ: 6.3 ohms

As I stated earlier, the actual output Z is much lower than the implied 75 ohm or 300 ohm setting for the Z-switch.

EDIT: Again, my apologies for the confusion.

 

[Redwingnine]

No worries. So, I have a pair of Dan Clark Ether-2 cans, which are a 16 ohm load. My other cans are Abyss Diana, which are a 40 ohm load, but only 91db/mw.

How well will the T4 drive these can, which are planar designs?

 

[tomb]

No worries. So, I have a pair of Dan Clark Ether-2 cans, which are a 16 ohm load. My other cans are Abyss Diana, which are a 40 ohm load, but only 91db/mw.

How well will the T4 drive these can, which are planar designs?

Low impedance is not as much an issue as poor sensitivity. A very poor sensitivity, low impedance planar headphone is probably not an optimum match with the T4.

 

[Redwingnine]

Thanks, Tom. I was afraid that was going to be the answer. The amp looks great, but does not appear to be a good match for the Diana cans.

 

[tomb]

Six T4 headphone amplifiers nearing completion:


The PCBs are complete; casework is next. Included in the scene is one of several COVID-19 masks that my wife has sewn lately - just so you all know that this is current.

If you will allow, I have some pics coming up next that detail the unique construction and parts that go into the ECP Audio/Beezar T4.

 

[tomb]

One of the first things to notice about the T4 design by dsavitsk are some unique features with the PCB:

The cutout for the Power Transformer is obvious, but this was done to avoid any cross-contamination of the 60Hz AC with the PCB. On other thing to notice about the top of the PCB (also on the bottom) are the heat sinking areas directly on the PCB. This is because the power transistors in the solid-state, fully differential buffer are SMD chips soldered directly to the PCB. However, despite the fact that they are small, SMD chips, they build up as much heat as many TO-220 transistors. You'll see later on how the heat transfer goes together, but there is a very thick, aluminum "heat bar" that is bolted to the bottom of the PCB and to the bottom of the casework. It attaches directly to these rectangular pads with ceramic insulators in-between. The ceramic insulators are needed, because there's close to 300V that pass through these pads. The numerous vias are used to fully integrate the sinking planes from the top of the PCB to the bottom of the PCB. Fairly unique, but it allows complete heat sinking through the entire casework, instead of bulky and somewhat inefficient (by comparison) PCB-mounted aluminum heat sinks.

Here's the PCB bottom:

Here you can see those heat-sinking planes duplicated on the bottom of the PCB, with the vias connecting from top to bottom. The other notable item is the complete absence of a ground plane. Usually, with most headphone amplifier designs that use a PCB, one surface of the PCB is used for the circuit traces and the other side is used for a ground plane that covers most of the area of the PCB. With the fully differential circuit beginning at the tubes and continuing to the output transformers, a common ground plane is impossible. However, there are many supporting parts of the circuit - power supply, tube heater supply, CCS circuits, and connectors - that utilize a common ground. That's the large trace pointed out above and the branches that follow.

 

[tomb]

The first step in populating the PCB begins with the lowest height part and ends with the tallest part. On the T4, that begins with the small diodes, followed by the tube plate resistors:

The small diodes are straightforward, you just need to make certain that the polarity is correct. As for the plate resistors, shown here as the large, reddish-tan, 75K resistors, there's a story behind those. During the T4 predecessor amplifier design by dsavitsk, the Torpedo III (or T3, for short), there was quite a discussion on another forum about the plate resistors. An individual noted that the plate resistors in a tube amplifier have an effect on the sound quality. As it turns out, he was right, but just not in his particular suggestion. We had originally used standard, higher-wattage metal film resistors in the T3, but this owner wanted to try Amtrans carbon-film resistors. They come with a unique construction and reputation that can be found on PartsConnexion here: https://www.partsconnexion.com/media/pdfs/Amtrans_amrg.pdf. After quite a bit of hoopla over them, I tried them … didn't like them. They actually dulled the response in the highs, making cymbal hits disappear in a fog. So, I dismissed the idea of changing the plate resistors from the metal films.

Until … a customer sent me a T3 for repair and asked me to install a different style of resistor at the plate resistor positions: the Texas Components Corporation TX2575 resistors. These were naked bulk metal foil resistors, now owned by Vishay (what electronic part isn't owned by them anymore?). Again PartsConnexion carries them under the new brand of Vishay here: https://www.partsconnexion.com/VISHAY-80844.html. When I installed them … Wow! They really made a difference: highs were cleaner and even more detailed than with the standard metal films. Unfortunately, as you can see, they are quite unorthodox and expensive, at almost $14 each. That was $56 for a single amp! That said, when dsavitsk designed the T4, I mentioned the TCC/Vishay resistors to him. His answer was the type of resistor shown above, with very similar qualities, but in a more orthodox construction, availability, and price. They are still expensive at $3-$4 each, but worth every penny on this amp. Please understand that differences such as these often go unnoticed in a typical amplifier. With the T4 and even the T3 before it, every little increase in part quality yields positive results, because the amplifier performance is at that level.

The actual resistors used in the T4 can be found here: https://www.mouser.com/datasheet/2/427/cpf-1762878.pdf. They make a difference and are very little different in performance from the TX2575 resistors described above. Here's a pic in mid assembly, with the plate resistors' leads bent and ready to insert in the PCB:

By the way, that little red triangle lead-bending tool on the left is worth its weight in gold, especially when assembling several amplifiers such as this, with varying lead lengths from part to part.

 

[tomb]

Next up are the SMD transistors, voltage regulators and MOSFETs. No big deal here, except that the assembly is typical SMD: liquid flux in a tube, small solder, small soldering iron tip, and tweezers to hold the parts:

1. Apply solder to a single pad at each part position.
2. Re-melt the solder with the soldering iron in one hand, while at the same time, holding the part with tweezers in your left.
3. Use the tweezer to move the SMD part into position, into the molten solder.
4. Remove the soldering iron while still holding the part with the tweezers.
5. Let everything cool, remove the tweezers and solder the other leads/etc. of the SMD part.

The tabs are particularly important here. One needs to be certain that the transistors are flat on the heat sink pad and that the transistor tab is soldered completely flush with a clean joint. Otherwise, you lose heat sinking capability.

Closeup (the SMD transistors on the right do not have their tabs soldered yet):

 

[tomb]

Ahh … this is the very, very boring part of populating a PCB - the rest of the resistors:

You can see the PCB flipped over here, so that I can solder all of the leads:

It's a lot of work and you don't really get a sense of having accomplished a lot, once completed. It's absolutely necessary, though, and probably because of my complaint - is often where focus is lost and the resistors get mixed up, causing untold grief in trying to troubleshoot once the amp is built. I keep all the resistors in their labeled bags and insert only one value in their proper positions at a time, to remove the chances of those errors.

Once again, dsavitsk surprised me. I actually replaced his basic resistors in the prototype with the familiar Vishay-Dale brown sausage-like resistors. The V-D resistors have been used by DIY-ers for at least 20 years, now. They are absolutely consistent in quality and manufacture and as metal films, are light years ahead of the old carbon comp resistors. They're hard to beat.

I had a small flirtation with PRP resistors, sold at PartsConnexion and many other places as a colorful (they're red) V-D replacement. Some people were excited about them around 10 years ago. I even had a line on a local distributor that was going to supply me in bulk with these. Unfortunately, when trying them I found that they were inconsistent: sometimes they were "blobs" instead of "sausages," and the red epoxy shell seemed to break and fragment very easily. You have to almost use a pair of pliers on a V-D resistor to break one. So, I was back to the dependable V-D resistors quickly.

So, I asked dsavitsk about the T4 - "Why do you keep using these cheap, colorful foreign resistors?" "Because they're better," he tells me. "Whaaaat?" I said. Sure enough, he got me again:
https://www.mouser.com/datasheet/2/427/mbxsma-1286643.pdf
Turns out they are German and started in Berlin as Beyschlag. Their power rating is better than the V-D RN55 resistors, so is their tolerance, and they're smaller. So, I switched, too:

Another shot in the middle of resistor work:

 

[tomb]

Thankfully, things start going faster (or at least the feeling that you're accomplishing something) after the TO-92 transistors. The first of the large parts, the tube sockets are next. With a tube amp, it's important to have the standoffs very, very close to the tube sockets. Or, better yet - directly under each tube socket. This is because the act of plugging a tube into a socket puts a lot of stress on the socket and the PCB. If it's not properly supported, bad things can happen.

Once again, Doug spec'd an improvement in tube sockets:
https://s3.amazonaws.com/tubedepot-...ached_files/00004280-VT9-ST-C1.pdf?1383248764
The Belton socket has a couple of advantages as a PCB-pin tube socket:

1. Fiberglass construction
2. Low profile

With #1, you get high-temperature structural stability, along with vibration resistance. You don't get any vibration resistance with a typical ceramic socket. Old NOS sockets made out of bakelite are not exactly vibration-resistant.

With vibration resistance, you minimize microphonics. Every tube is probably microphonic to some degree, simply from the mechanical construction involved. So even if it's not audible, it might be measurable and have an otherwise non-localized affect on the sound quality that could be audible.

With #2, every instance I've measured and tested where the lead lengths were increased on tubes, either from socket protectors, flying leads, or what-have-you, the longer signal path causes measurable increases in noise. So, low profile is good from that perspective.

From the perspective of using standoffs, low profile is bad. It means the standoff mounting screws below are captured by the tube sockets. IOW, if you don't put the screws in first - before soldering the tube sockets in place - you're in a world of hurt. I won't tell you what I went through with the first few T4s when I forgot this.

So with those screws highlighted above, here's a shot of the standoffs underneath the PCB:

And with the soldering done:

Top view of the sockets, along with the Molex plug-ins for the power transformer, the power inductor, and film filter caps also installed:

As you can see, there is a center hole in the sockets, so one is always able to tighten, loosen and remove the standoff. However, the screw stays secure under the socket, trapped by the socket pins.

Because the sockets undergo so much stress, I am careful to ensure that solder flows up the pin legs from the pads below. This ensures maximum strength with the socket's attachment to the PCB.