Three boxes for your preamp? Those are rookie numbers, you need to pump those numbers up. If you give each channel an outboard power supply, that brings you to five boxes. Have a box per channel for attenuators and input selectors and a box per channel for volume and input control power supplies and you are at eight boxes for your preamp
I was sort of running into the same boat with my amp build. All the film power supply caps and large coupling caps take up a ton of room. I'm sure I'm not telling you something you don't already know because I have seen you stack boards vertically in your previous builds, but one of the things I did to keep the footprint smaller was to plan for populating both sides of the board. The best example I can give is this, a dual mono unregulated power supply board for a preamp. Beside the necessary four diodes and their switching noise caps and resistors, each channel has 9 Mundorf Mtube film power supply caps creating a CLCLC filter. 5 of these caps are on the top, and 4 are on the bottom along with a Mundorf Supreme on the top and one on the bottom. The footprint of the MTube caps are 30mm x 35mm each. Not small. So between both channels I have 18 of these caps, 4 Mundorf Supreme caps, 2 full rectifiers with caps and resistors, and 2 house ground circuits. This does take up more space vertically, but maybe this is an option since it sounds like the footprint space is at a premium. Sorry, a long winded way to get what I'm trying to say across but maybe this will help.
I decided to try a simpler PCB design before tackling the new project. So, while waiting for the new resistors I ordered to fix the bias on my Aikido ACF boards, I designed a new PCB that will fit in the same location as the Glassware PS-Tube and ACF boards. This new board combines the B+ tube rectifier and filter caps, DC filament supply, and a balanced differential line stage. The line stage circuit is from one of Broskie's TubeCad blogs. It doesn't look like he ever turned this into a Glassware kit, but the design looks sound and simulates well in LTSpice. I also added a muting relay so that I can wait until the decoupling caps have settled before unmuting. When checking my current ACF build, I noticed that there is a fairly significant turn-on transient while the decoupling caps charge up. While this is fairly well balanced between the two phases, I am worried that the DC offset could cause the input transformers on my 300B amps to become magnetized. I designed the board using EasyEDA and placed an order this morning. The board uses four layers, one of which is dedicated as a ground plane for the analog circuits. The power supply ground is on a separate net which terminates at a single location to prevent power supply ground currents from interfering with the audio signals. Here's the schematic for the board. And here's a 2d rendering. The 470 ohm resistors from HiFiCollective showed up in the mail this morning, so I'll try to get the current boards updated in the next couple days to see how this works. But assuming that this new board works well, I'll probably replace the existing boards with this new design.
Well, that didn't take you long at all! Did you even move anything since I saw the prototype with just the ratlines? Lol. It looks awesome though. Lots of space between things and a very logical layout. Curious to see how this sounds in comparison to the old version.
Hah... Well I did have to move a few parts and roasted a few others. Of course, I won't know if I did a decent job until I power one up.
It's gratifying to see others posting their exploits of a similar nature in this thread. Feel free to keep it coming! Jaytor, your board does look very nice indeed. Is this a unity gain circuit like the ACF (roughly) is?
Thanks. This board does have some gain - close to 16db. I found when I was testing the ACF that I was operating pretty close to max volume on the attenuator, so I think I'd probably prefer a little gain. 16db is probably higher than I'd like, but we'll see how it works.
I decided to try a simpler PCB design before tackling the new project. So, while waiting for the new resistors I ordered to fix the bias on my Aikido ACF boards, I designed a new PCB that will fit in the same location as the Glassware PS-Tube and ACF boards. This new board combines the B+ tube rectifier and filter caps, DC filament supply, and a balanced differential line stage. The line stage circuit is from one of Broskie's TubeCad blogs. It doesn't look like he ever turned this into a Glassware kit, but the design looks sound and simulates well in LTSpice. Do you have a link to the TubeCad article or the name of the TubeCad post. I always enjoy reading his stuff.
There are several posts that talk about differential and balanced circuits. If you do a google search for "site:tubecad.com differential aikido", you'll find a few different entries. The one I found the most useful is Balanced and More Balanced
I received the PCBs I ordered and the quality looks quite nice. Unfortunately, I didn't pay enough attention to the component footprint used for the octal tubes. Lesson learned: don't trust the EasyEDA footprints - particularly those submitted by random users. The holes are way to small and spaced too close to the center. I looked around for a socket that might fit but couldn't find anything. Oh well, I was thinking that I might want to redo these anyway since I kind-of rushed to get these done and didn't do as good a job as I should have with the routing. I've been working on a bunch of other boards over the past couple weeks and have learned more about how to best use EasyEDA.
These kind no good? CMCT-OCTAL: CMC Teflon Octal Chassis mount valve base | HIFICollective VT8-PTS: Belton Octal chassis mount valve base, with PCB pins | HIFICollective
Thanks for the suggestions, but these don't look like they will work. The second one looks like it might be the correct pin spacing, the the holes in the board are too small for the pins. I'm just going to redo the board since I think I can do a much better job at this point, and it wasn't that expensive. I'll chalk it up to a learning experience.
Thats a shame. Sorry to hear this. If you are going to redo the board, test fit every lead of every component before you reorder. A lesson I learned when I ordered my heater regulator boards. I found 2 different components that had a listed lead size and it didn't match the footprint on EasyEDA.
Nine weeks and four days after ordering these have turned up at my door. Charlie at Orso Audio was very good, with quick responses to my emails and he did make an effort to find the lowest shipping price he could. Interestingly enough the sender address on the package was actually ATC loudspeakers in Gloucestershire, UK and be the looks of it Charlie is an Analogue Design Engineer for them, with Orso being a bit of a side venture. Anyway, for anyone considering Orso Audio, I can recommend them. The remarkable length of time from order to receipt was not their fault. It was five weeks for the transformer winding, and an inexplicable 4 weeks shipping. While I'm here I should also mention the excellent service I received from Sonic Craft, even though I didn't really deserve it. Due to an apparent shortage for some values of Takman resistors at my usual outlets, I decided to give Sonic Craft a try, purchasing those and the big Mundorf Tubecaps. Being in a bit of a rush I created an account, misspelling my email address and at the same time noting down my new password incorrectly, meaning no reception of email messages and no way to access to my newly created account. Nor could I use the websites forgot password function, because those messages would also be sent to an email address that didn't exist. Sonic Craft kindly deleted my account so I could start again with the correct details - which I did - but at the same time I changed the quantities of the ordered items and the shipping method, when in hindsight they'd likely packaged up my order already. This became evident when the order was received safe & sound inside a USPS box placed inside a FedEx box. I am ashamed.
While the Lundahl mains transformer (which is designed for use in 300B amplifiers) is pretty impressive the LL2742 choke is, to me, the more interesting piece. Wound with two inductors on one core it can be wired in series (for the highest rated inductance) or parallel (doubling the current capability, but significantly reducing said inductance). A third style, also series, is achieved by connecting one coil to the positive power rail and the other to the negative facilitating an improvement in ripple rejection. In this setup the LL2742 is rated 42 Henries at 100mA with 2W dissipation. At first it appeared to me to be a common-mode choke when configured in this way. The Lundahl data sheet (below image) even lists this as a 'serial connection for improved common mode rejection'. Yet judging by the direction the coil polarities are wired I'm not sure this isn't actually differential-mode. The small amount of research I've done into common/differential noise has made me confused more than anything else, but whichever form it comes in there should at least be some beneficial noise reduction utilising the choke in this fashion.
Fully_articulated It seems we have very similar tastes. Both run alpha J on highly efficient speakers. And was looking to to build a sp14 (I know that's not what you ended up with but very similar typology) I was curious if you had any issues with too much gain and overall how did you like the pairing?
Hi Jacob, It's a little difficult to say, because most of my listening is done from computer to DAC where I use replay gain and room EQ adjustments that bring the volume level down about 10dB or more before it hits the preamp. So, in my use case a little gain is warranted. At 'regular' listening level this has me at about step 40 on my 63 step attenuator. Movies & CDs from my disc player anywhere from the 20th to 30th step. It's the same Khozmo volume control that the SP14 can be speced with. At one point, with my previous Paradigm speakers (about 90dB sensitivity) I converted the Aikido to no-gain, and found that with the Aleph J I could go right up to the 63rd step. Swapping in the First Watt F6 (that has lower gain than the Aleph) meant I didn't have enough. With the SP14 utilising 6SN7s, I would anticipate a similar amount of gain I get from the 6N6P, which I find quite reasonable. With the 6DJ8/6922 there was slightly too much, and the sound was a bit too clean & neutral. I like the pairing enough that I'm not going too far away from it with the new preamp. I'm still jealous of your Troels Gravesen speakers by the way, and think of them whenever I see your avatar.
thanks! i appreciate your thoughts! Still not ready to pull the trigger on that one. But it's something I've been mulling over. I could always use the left right balance potentiometers to lower gain. or just build myself in F4. Your build threads are great
Interesting, I currently run the similar Aleph 30 with 89dB W/m Paradigm and was thinking a First Watt would be a good way to get less gain, since I have too much with my DAC outputting 7V at full scale. Thanks for putting your history out there and sorry for the thread diversion.
The long wait for the transformers, then an additional period for the chassis, did a lot to kill the momentum on this project. That's a lot of empty time available for spotting all the flaws in the design and areas of poor construction. It has been difficult to regain my enthusiasm and continue, especially since the next step was a whole bunch wiring - not a favourite of mine. This is the RCA jack wiring, and it looks like I under-ordered the amount a bit as this is all I have left. Hopefully there's no accidents soldering them up. The power supply cabling goes all over the place, but I took my time making sure the lengths would be just right. This meant turning the board over a few times and, fully-loaded now, it creaks under the strain. I was glad to finally get it secured to the chassis base. There was quite a bit of bodge solder work done here, mostly after fighting to heat the pads surrounded by areas of copper pour. Some solder joins turned out to be more of the blobby mound variety. The Lundahl transformers are nicely proportioned and easily mounted (the holes are actually threaded!) but they are also hand-made in the best and worst ways. The winding connections are unmarked stubs, so a lot of datasheet checking is advisable, as well as confirming with a multimeter. I filed the stubs a bit to remove the varnish finish that had been sprayed on, not knowing if the soldering iron would burn it off or not, then wrapped the wire around them. All-in-all it came out pretty good I reckon. And the back-side, where the low-voltage connections are located. The three green wires are connected from the board grounds, one for each of the low-voltage power supplies located front & back, plus the main high voltage ground (which is also connected to signal ground over at the start of the CCDA circuit). These all run out to a 'house ground' or 'ground lift' circuit that I'm 90% sure I've wired up correctly. Initially I'd wanted to put this onto the PCB as well, but it just wasn't feasible. Next up is the mains wiring & signal connections. Then after that, more wiring!
