[L0rdGwyn]

Just to see what all of the hype is about, this morning I did another experiment.

You might recall that I left the driver tube cathodes unbypassed in the 6A5 amp given the CCS load. I had some leftover electrolytics from the build, some non-polar Audio Note Kaisei, I thought I would clip in the caps, bypass the cathodes, and see how it affected the sound. Many tube DIYers go to great lengths to remove these electrolytics from the signal path, but how bad can it be really?

The alligator clips I'm using don't have teeth, so the caps can be very quickly removed for listening comparisons.



The difference is pretty BIG! With the cathodes bypassed, there is a collpase in the soundstage, instrument separation suffers, there is less definition in the bass, and the high end sounds more harsh. It isn't subtle whatsoever. The amp doesn't sound bad with the cathodes bypassed, but once you've heard them unbypassed, no way you could go back. Audio quality is all relative after all. I did do some FFT measurements, bypassing the cathode resistors doesn't seem to affect distortion in a measurable way.

So the hype is real, keeping electrolytics out of the signal path is the way to go. Luckily in the 45 parafeed amp, I will be leaving the 6J5 cathode resistors unbypassed, and I should be able to get away with leaving the 45 cathode resistors unbypassed as well using the "Western Electric" parafeed wiring configuration.



This setup could potentially inject some power supply noise onto the cathode, but seeing as I am back to using a MOSFET cascode CCS on the output tube (rather than the pentode CCS), I think I will get away with it. That would leave only two capacitors in the signal path, the interstage coupling caps and the parafeed caps, which will both be high quality film types.

Since I am waiting around for parts again, the gears have started turning on an all DHT two-stage amplifier. This is probably a 2021 project, so I won't talk about the details. Might be collecting some tubes for it, but I'll certainly be looking into ways to avoid bypass caps in this setup too. To be continued...

Another thing I did yesterday, which you might be able to see in my photo above, was rerouting the B+ wiring from the mains transformer to the rectifier. A made a novice mistake routing this wiring along the periphery of the chassis originally. Keeping this loop area small should help with power supply noise, and seems to have made a measurable difference. Hindsight is 20/20.

TTFN!

 

[Tjj226 Angel]

No major updates on the 45 amp, but did start some work on it.

Figured while I am waiting on the transformers and chassis, I would start putting together some of the PCBs, there will be eight boards in this amp. These are the 45 filament regulators, Rod Coleman's design, along with his raw DC supply. Final transistors aren't mounted yet, they will need heatsinking, waiting for the chassis to see how much slack I will need on the leads.



These will be fed by 6.3V 5A windings on the custom mains transformer. Dedicated filament transformers are often used in DHT builds (e.g., Hammond, AnTek) rather than running it off the HT transformer. I discussed this with Rod, the reason for his recommending dedicated filament transformers is a concern for regulator rectifier noise coupling into the HT winding of the mains transformer. But seeing as both amplifiication stages are going to be CCS loaded with >100dB PSRR, I'm running it off the mains without concern. Gotta love that CCS load! Will save me a big chunk of real estate.

The challenge of laying out this amp was balancing the size of the parafeed capacitors with my desire to keep the chassis somewhat compact. The other challenge was mounting power supply components - since the cores of the Sowter transformers lie flush with the top of the chassis, a significant amount of hardware mounting space on the top plate is lost. I could countersink so screws also lie flush with the top plate, but it would be an enormous PITA if I ever need to move/adjust a components position. I needed a solution to minimize the footprint of the power supply. I found these great capacitor clips from Cornell Dubilier, let you mount a 35mm cap on its side with a single 3mm screw! Most capacitor clamps take two minimum, often three. Saved me a lot of trouble and will keep the power supply footprint very small.



That's all for now, will do the CCS boards next. Chassis is already machined and being plated. When I get the transformers in two months or so, this amp will come together fast, should be a very quick build. The interesting part will be testing various parafeed cap values, taking measurements, listening...then deciding what the final caps will be when the capacitance is determined. Testing will be done with Solen caps, I'll have to see what will fit if I want to upgrade to something fancier. I've carved out space in the layout for some big-ass axial caps.

6A5 amp is still rocking my world, I am ordering a pair of Omega speakers to pair with it, I'll post some picks of the model I am going for once the order is finalized. They'll be around 97-98dB sensitive, very nice

Lol. You went from "meh, I might do a 45 amp later on" to "the chassis is done" pretty darn quickly.

I should find out who does your cassis. I need top plates, but I can't get to the CNC machine because of the pandemic.

 

[Tjj226 Angel]

TTFN!

Try LED or battery bias.

LED bias is super easy to implement, and depending on the circumstance, I like it more than an unbypassed cathode resistor.

It greatly lowers the impedance to ground like a bypass capacitor would, but it doesn't have weird frequency and timing issues that a cap has.

 

[L0rdGwyn]

Lol. You went from "meh, I might do a 45 amp later on" to "the chassis is done" pretty darn quickly.

I should find out who does your cassis. I need top plates, but I can't get to the CNC machine because of the pandemic.

Well, other than going to work, I have literally nothing else to do lol. It's taking so long to get transformers, had to decide if I was going to go for it well in advance. Won't get the 45 transformers until May or June. The chassis maker is Landfall Systems, Dave there is a great guy to work with, he does the machining and has the plating done at a separate shop.

 

[L0rdGwyn]

I was originally going to post this on the 2359glenn thread, but thought I should keep my DIY ramblings here...been doing a lot of reading into A2 designs.

Anyone have any experience with the Ampsandsound Kenzie? I've only heard the ZMF Pendant from them. The Kenzie uses the 1626 output tube, the reason I'm curious. These tubes are dirt cheap, plentiful, and look very nice!



Capable of being driven into A2 for better linearity, in a DC coupled design or using a FET source follower perhaps. Only a 5W max plate dissipation, but pretty intriguing for a budget headphone amp. Could go for a 125V 40mA A2 bias point with -7V on the grid, would just need a suitable 5K 40mA transformer. If using the LL2765 ($125 each), could do 165V 30mA, but linearity would suffer.

125Va / 40mA Ia / -7Vg


CCS load a driver that can swing >90Vpp with decent gain and a FET source follower and this hybrid would get you a little over 1W into a 5K primary. If you can find some reasonable transformers (or go with the LL2765), bet you could make a kickass headphone amp pretty inexpensively with a regulated supply. You can get a NOS pair of 1626 for like $30. This might go on my short list of things to experiment with.

 

[Xcalibur255]

Once you've built all the pie in the sky projects, it might be a rewarding pursuit to go the other direction and see how much performance you can wring out of a cheap and cheerful design. I think taking on both ends of the spectrum has a lot to teach.

 

[L0rdGwyn]

Absolutely, I'm starting to keep an eye out for more budget-friendly ideas, tubes that reasonably priced, available, and linear. Who knows, maybe if well-implemented, something I could build for others.

 

[Tjj226 Angel]

Once you've built all the pie in the sky projects, it might be a rewarding pursuit to go the other direction and see how much performance you can wring out of a cheap and cheerful design. I think taking on both ends of the spectrum has a lot to teach.

I have always wanted to do a youtube series on building your own hifi stereo system at costco.

Aluminum foil capacitors. PIckle jar vacuum tubes. ect.

I sort of had a theme where I would show people how to make hifi stuff in an apocalypse. Who knew that my idea would become so relevant?

 

[kvik]

The Darling Head Amp from Toolshed Amps also use 1626
https://www.toolshedamps.com/darling-stereo-desktopheadphone-amplifier

 

[L0rdGwyn]

The Darling Head Amp from Toolshed Amps also use 1626
https://www.toolshedamps.com/darling-stereo-desktopheadphone-amplifier

Hmmm that's interesting...not the first I've heard of the "Darling" 1626 amplifier. The original "Darling" was created by a guy named Bob Danielak in the late 1990s, he generated interest in this tube in the DIY community with his SET designs. In fact, if you Google 1626 tube amp, you'll get a ton of DIY hits for "Darling" variants. The Toolshed Darling seems to be descendant of his work, probably the only thing they have in common is they are both 1626 SET designs, class A1 I assume, seems to align with their power output specs. Looks like it uses an 84 rectifier and 12SL7 driver, DC on the heaters. Someone reviewed it on Head-Fi here, can check out the internals:

https://www.head-fi.org/showcase/toolshed-darling-1626-set-amplifier.23931/reviews

Looks like a lot of time and effort has gone into the chassis, not my style but the craftmanship is obvious! Pretty expensive though at $2600. For less than $1000, could build something pretty neat with these tubes I think. Haven't dived into any details, but like I said, maybe DC coupled or using an interstage FET buffer to drive the 1626 into A2. In A1, it is going to hit the knee of the grid curves very quickly on the negative peaks, but may not matter much with headphones. Could parallel a few of these and drive some sensitive speakers I think you might get better performance out of this tube in A2.

Pete Millett got 1.4W out of a single 1626 driving the grid with an opamp: http://www.pmillett.com/pwrop.htm

 

[L0rdGwyn]

Another interesting tube to design around might be the RFT EC360.



Not super available, but pretty cheap. Internal resistance from the spec sheet is 120ohm! Mu of 2.5. I'm a fan of the East German RFT brand.

Datasheet: http://tdsl.duncanamps.com/pdf/ec360.pdf

Here are some curves I was able to dig up:

 

[Tjj226 Angel]

I would look more at Shuguan and Sovtek for current production tubes that are cheap.

The Sovtek 6H30 might make an interesting little OTL amp. If you put both halves of the triode in parallel you could probably get decent output impedance.

 

[Xcalibur255]

Hmmm that's interesting...not the first I've heard of the "Darling" 1626 amplifier. The original "Darling" was created by a guy named Bob Danielak in the late 1990s, he generated interest in this tube in the DIY community with his SET designs. In fact, if you Google 1626 tube amp, you'll get a ton of DIY hits for "Darling" variants. The Toolshed Darling seems to be descendant of his work, probably the only thing they have in common is they are both 1626 SET designs, class A1 I assume, seems to align with their power output specs. Looks like it uses an 84 rectifier and 12SL7 driver, DC on the heaters. Someone reviewed it on Head-Fi here, can check out the internals:

https://www.head-fi.org/showcase/toolshed-darling-1626-set-amplifier.23931/reviews

Looks like a lot of time and effort has gone into the chassis, not my style but the craftmanship is obvious! Pretty expensive though at $2600. For less than $1000, could build something pretty neat with these tubes I think. Haven't dived into any details, but like I said, maybe DC coupled or using an interstage FET buffer to drive the 1626 into A2. In A1, it is going to hit the knee of the grid curves very quickly on the negative peaks, but may not matter much with headphones. Could parallel a few of these and drive some sensitive speakers I think you might get better performance out of this tube in A2.

Pete Millett got 1.4W out of a single 1626 driving the grid with an opamp: http://www.pmillett.com/pwrop.htm

In a simpler chassis those amps would actually be $1000 less I bet. That isn't a criticism of them at all, there is a place in the world for style and craftsmanship, but it does serve to highlight the advantage of DIY in terms of savings or re-directing the "bling" budget into more quality parts.

 

[L0rdGwyn]

In a simpler chassis those amps would actually be $1000 less I bet. That isn't a criticism of them at all, there is a place in the world for style and craftsmanship, but it does serve to highlight the advantage of DIY in terms of savings or re-directing the "bling" budget into more quality parts.

Agreed, has to be the most expensive component of that amp. The custom stuff I have done at Landfall is around $200 for machining, brushing, and anodizing. Not cheap, but I like the customizability, minimalist look, and quality. For me the bling is the tube loadout and iron, I prefer the other parts to be understated so as to give them the stage. DIY definitely has its advantages with being able to allocate funds where performance matters most, but I understand those guys need to make money to run a business, it's a high-margin low-volume industry.

 

[Xcalibur255]

I think finding Landfall back in 2012 has turned out to be the most meaningful contribution I've made to all of this. Being able to specify any X-Y-Z dimension is such a boon compared to the Chinese chassis products that are fixed dimension.