[L0rdGwyn]

Im wondering if there isn't a supply chain issue on lundahls side of things. The whole world right now is a bit bottle necked.

I wouldn't be surprised, I'm sure Lundahl and Kevin are working on it, so I am just patiently waiting. Obviously the world is a mess right now, my fancy transformers arriving late is not a big deal.

 

[L0rdGwyn]

I've learned that K&K has the transformers, however they were part of a very large shipment from Lundahl and it is still being sorted through, so sounds like next week is the week. Edit: apparently one of my transformers was mistakenly sent to someone else, so no idea.

In the mean time, I have continued to work on my 45 parafeed design. I have taken a dive into pentode CCS devices as alterntaive to SS cascode CCS implementations. I have what I think are some decent draft schematics at this point. Here is the power supply:

• Given both input and output stages are CCS loaded, the PS caps are not in the signal path directly, so 600V Nichicon electrolytics will be used. What a relief after using Clarity Caps in the last build, sheesh...
• Single rail given CCS, not very concerned about cross-talk
• SS rectified, ultra-fast switching soft recovery Vishay HEXFRED diodes
• Using Lundahl chokes again, LL1685 at 13H for the 45 node, LL1668 (which is typically used as a anode choke) for the 6J5 node. Overkill given the CCS load of the 6J5, but will ensure unmeasurable ripple on the input

Here is the amplifier schematic:

• 6J5 biased at 200V / 8mA, with adequate headroom for the CCS (10V dropout), could push the voltage bit higher too. Cathode resistor is not bypassed. CCS will dissipate ~2W, K&K kit comes with a 3W heat sink
• Here is where things get a little interesting...the 45 CCS load is an EL83 pentode, more on that below
• 45 biased at 180V / 32mA, with enough headroom on the B+ to account for the EL83 dropout voltage
• Given the lower performance of the EL83 CCS, will likely bypass the 45 cathode resistor
• Rod Coleman filament regulators will be used for the 45, as well as his raw DC supply
• The parafeed capacitor value will need to be determined experimentally. There are ways to calculate the value using a anode choke load, not so for the CCS. I am going to take the advice of Paul Birkeland from Bottlehead and pick up a swatch of Solen caps, clip them into the circuit, take measurements (and listen) to determine the value. Then if I want to upgrade to a fancy PIO type, I can do so confidently (they are expensive)
• Output transformer is the Sowter 8983. This is a full-sized parafeed OPT with 8ohm and 16ohm taps (25:1 and 17:1 winding ratios, respectively). TBD on using resistors in parallel with the output

Okay so advantages to the EL83 CCS...like I said, this pentode can dissipate 9W on the plate maximum and get that heat out of the chassis. I would be lying to you if I said there wasn't an aesthetic draw as well, hey what can I say, I am a tube junkie!!! These tubes are very cheap, come in lots of nice flavors (Mullard/Philips/Valvo, Telefunken, and more). That's about it.

Disadvantages to the EL83 CCS, there are quite a few: dropout voltage is ~60-80V, I have accounted for this in my power supply and then some. Heaters of the EL83 will need to be elevated. The internal resistance of the EL83 in this setup is something like 260K. On the AC impedance side, there is an issue: the 73.6K g2 dropping resistor and 1uF bypass cap act as a shunt to the CCS, so they will largely determine the AC impedance to the 45 output signal in parallel with the OPT primary. Now something like a 75K AC impedance from a CCS is hardly the end of the world, it will still function like a high-inductance anode choke, just not as good as the gazillion ohm impedance offered by a SS solution.

So yes, quite a few disadvantages. I may play around with using a hybrid pentode/MOSFET cascode CCS as well, with the MOSFET setting the current. Regardless, I want to experiment with this type of CCS - worst case scenario, I need to make some changes down the road. If I have issues with the EL83 for any reason, my back up is the EL822 pentode, which can dissipate 12W on the plate and has similarly flat grid curves. Here are the EL83's, I will be fixing the bias on the -3V line:



So that is the design as it stands right now. With those components, I have been doing mockups of the chassis, 10.5" x 12" x 4.25", here is the top plate with the Sowter transformers at 90 degree angles to each other:



And here is the interior layout, left the transformers in for screw-hole considerations, all subject to change of course (note the dimensions of the mains are different, mine will fall in one of two different sizes, TBD).



Alright, that is the 6J5/45 all CCS loaded parafeed design for now. Decent draft I think, but may change based on feedback I receive across multiple forums.

If anyone has any questions or would like me to explain anything in non-technical terms, please feel free to ask, I am more than happy to.

 

[Tjj226 Angel]

Just a heads up, since everything is using a CCS, your second choke can easily be replaced by a resistor for a CRC filter. A resistor will remove some thermal load off your preamp CCS and it will be drastically cheaper.

I would also skip the hexfred diodes. They are great for when you are dealing with a few amps of current (filament supplies), but they are just stupid overkill for B+ supplies. I would use the ever popular UF4007 and just buy like 20 of them so that you will have a decent little supply.

I would change the 56uf capacitor to another 100uf capacitor. When you use a SS rectifier, you capacitor input is not limited. So you might as well buy a 3rd of your 100uf caps and keep things simple.

Lastly if you are going to get a sowter custom PSU transformer, get a center tapped transformer and use a full wave rectifier. Just remember to double your voltage. In this case you would need 700v CT.

 

[L0rdGwyn]

All good points as usual T. I figured the second choke was overkill, feel better not using it hearing it from someone else, I'll switch to an RC filter, will save me space inside along with the cash. Changing from a 56uF to 100uF is no problem either, I'll look into using a center-tapped and full-wave setup.

While perusing the annals of early 2000s audio forum posts, I found some giant pentodes that could possibly be used in extremely high-wattage CCS', the 803 and 804. The 803 can take 1600V and 125W on the plate problem is they are direct heated with 10V / 5A thoriated tungsten filaments. Never said they were perfect! 804 also directly heated,1500V 50W on the plate, filaments at 7.5V / 3A. People were talking about them in the context of DC-coupled amplifiers, no idea how practical it actually is, but would love to see it, just don't touch the top cap.

http://www.r-type.org/pdfs/803.pdf
http://www.r-type.org/pdfs/804.pdf

 

[Tjj226 Angel]

All good points as usual T. I figured the second choke was overkill, feel better not using it hearing it from someone else, I'll switch to an RC filter, will save me space inside along with the cash. Changing from a 56uF to 100uF is no problem either, I'll look into using a center-tapped and full-wave setup.

While perusing the annals of early 2000s audio forum posts, I found some giant pentodes that could possibly be used in extremely high-wattage CCS', the 803 and 804. The 803 can take 1600V and 125W on the plate problem is they are direct heated with 10V / 5A thoriated tungsten filaments. Never said they were perfect! 804 also directly heated,1500V 50W on the plate, filaments at 7.5V / 3A. People were talking about them in the context of DC-coupled amplifiers, no idea how practical it actually is, but would love to see it, just don't touch the top cap.

http://www.r-type.org/pdfs/803.pdf
http://www.r-type.org/pdfs/804.pdf

Yeahhhh, those pentodes are far from ideal.

Here is my thinking. Pentodes are a great idea if you can find a small 9 pin tube that can handle everything you need it to do for less than 40 bucks a pair. Beyond that, you get to a break even point where a SS CCS with a heat sink is just as expensive.

United SiC has a 1200v 33A part for 12 bucks. The thermal pads are a dollar, and I have found some heat sinks from ebay for about 20 bucks. So all in all by the time you add screws and everything else, you are looking at 50ish bucks for both channels. Maybe 70 bucks if you wanted two smaller individual heat sinks for left and right channel.

That being said, tubes do have a very clear electrical advantage over SS. They are incredibly stable when it comes to temperature drift.

However....I made a crappy test amp using a single 6E5P tube in triode at 40ma. I was dumping about 8 watts into each CCS and I had both CCSs bolted to the top plate (so 16 watts total into the top plate). The top plate gets pretty warm (130F or so), but it works and seems pretty darn stable over time. Sooooooo shrug? Maybe the jfets are more stable than one would guess?

When I stop coughing my lungs up, my next experiment is to try the jfets I have with a heat sink to see how stable it is and just refining the implementation overall.

 

[L0rdGwyn]

Yeahhhh, those pentodes are far from ideal.

Here is my thinking. Pentodes are a great idea if you can find a small 9 pin tube that can handle everything you need it to do for less than 40 bucks a pair. Beyond that, you get to a break even point where a SS CCS with a heat sink is just as expensive.

United SiC has a 1200v 33A part for 12 bucks. The thermal pads are a dollar, and I have found some heat sinks from ebay for about 20 bucks. So all in all by the time you add screws and everything else, you are looking at 50ish bucks for both channels. Maybe 70 bucks if you wanted two smaller individual heat sinks for left and right channel.

That being said, tubes do have a very clear electrical advantage over SS. They are incredibly stable when it comes to temperature drift.

However....I made a crappy test amp using a single 6E5P tube in triode at 40ma. I was dumping about 8 watts into each CCS and I had both CCSs bolted to the top plate (so 16 watts total into the top plate). The top plate gets pretty warm (130F or so), but it works and seems pretty darn stable over time. Sooooooo shrug? Maybe the jfets are more stable than one would guess?

When I stop coughing my lungs up, my next experiment is to try the jfets I have with a heat sink to see how stable it is and just refining the implementation overall.

Haha yes, far from ideal is probably an understatement, but amusing for me to think about. The EL83 are inexpensive ($10-15 each), and the Russian equivalent is dirt cheap. They are pretty nice looking to boot (which might only matter to me). Will be interested to hear the results of your experiment, the heat sink will be a fin type on the exterior? Was looking into this for my own design before switching to pentode. Hope you feel better.

 

[Zachik]

They are pretty nice looking to boot (which might only matter to me)

Everyone on this (and Glenn's) thread love the looks of tubes! Definitely not just you...

 

[L0rdGwyn]

FINALLY! They have arrived. Off to wire these up and get them into the amp. Will then do some confirmatory voltage checks, then hopefully audio bliss (assuming there are no hiccups, you never know).

Will post in a few hours with updates.

 

[L0rdGwyn]

Holy crap...

Okay first things first, got the transformers in the amp. Below is the secondary wiring, 470ohm resistor across the output to protect the transformer from open circuits.



Had a few other things to finish up, nothing major. Here is the final ciruict.



After I finished wiring, I eyeballed the whole circuit, reflowed a few solder joints, then brought the circuit up on a variac and checked voltages. Here is what I got for the two channels using the GEC U18/20:

MH4 cathode: 2.44V / 2.47V
MH4 plate to cathode: 174.4V / 184.4V
B+ top of OPT: 312V / 312V
6A5G cathode: 44.6V / 45.5V
6A5G Plate to cathode: 247.6V / 251.5V

All of these values are within 1-2% of my targets, I was absoltuely dumbfounded that I nailed them so accurately. That crazy Georg Ohm knew a thing or two. Difference between the MH4 plate voltages is tube variation I am sure.

Well then there was only one more thing to do...have a listen? NO! Take measurements of course. Here is the frequency response, down ~2dB at 20kHz, might investigate that, but essentially flat from 20Hz - 20kHz



I am witholding FFT measurements until the amp settles down and have some time to mess around with different tube combinations. Here are some beauty shots of the final product:



So how does it sound...with all of the bias suppression I can muster, it sounds incredible. I braced myself for bad sound (expect the worst, hope for the best kind of thing), but it didn't come. The soundstage is absolutely expansive, the bass is deep, full, and tight, treble is detailed, airy, and not the least bit strident. Maybe I shouldn't be, but I am still a bit shocked it all worked out! It is a keeper.

I have been listening for about an hour straight, the panel with some of the dropping resistors on it does get warm, but so far I think it is manageable, haven't taken any surface temperature measurements yet. Will be adding some ventilation in the near future.

I guess that is all for now, just going to keep on listening have quite a few different 4V drivers to try out (REN904, MH4, E424N, A4110, NR52, K1694), a few different rectifiers to try, and different makes of the 6A5G, but probably won't get to it tonight, might try one other combination. At some point I have to get this baby in my stereo, that will be its final destination, but enjoying headphone listening for now.

So overall a rousing success I would say. I will have more impressions and measurements to come I am sure (need to get a square wave generator...).

Big thank yous to @Tjj226 Angel , @2359glenn , and @A2029 for the input along the way. This is the hardest DIY project I have ever taken on, learned an absolute ton in the process. I'd imagine the first one is the hardest, should be smoother sailing from now on.

 

[L0rdGwyn]

Here are some 1930s gold Telefunken REN904 in the amp. These are some rare tubes, was able to make a NOS pair from two different sources. They sound so good! I apologize in advance if you see me posting elsewhere about this amp, I'm going to have to take a personal victory lap, sorry.

 

[leftside]

Congrats. You have every right to feel proud.

 

[mordy]

Here are some 1930s gold Telefunken REN904 in the amp. These are some rare tubes, was able to make a NOS pair from two different sources. They sound so good! I apologize in advance if you see me posting elsewhere about this amp, I'm going to have to take a personal victory lap, sorry.

Congratulations on finishing your special amp! Very impressive!
Are you able to describe the sound versus the GOTL?

 

[audiofest2018]

Here are some 1930s gold Telefunken REN904 in the amp. These are some rare tubes, was able to make a NOS pair from two different sources. They sound so good! I apologize in advance if you see me posting elsewhere about this amp, I'm going to have to take a personal victory lap, sorry.

Since you mention about a personal victory lap, how about a victory tour so we can get a good taste!? Just kidding! Congratulation.

 

[L0rdGwyn]

Congratulations on finishing your special amp! Very impressive!
Are you able to describe the sound versus the GOTL?

Thanks, mordy! I think it is still too early to say, the GOTL is such a changeling too, but at some point soon I will listen to them side-by-side, probably this weekend.

Since you mention about a personal victory lap, how about a victory tour so we can get a good taste!? Just kidding! Congratulation.

Never! But thank you I am hoping to make it to ZMFestivus again this year though, if so I will bring this amp along and my other one if it is done by then.

 

[maxpudding]

This is exciting!

Congratulations on the completed build. The amp, from the way you describe its sound, is incredible. Hoping one day to get to listen to your amp