In a previous post I mentioned acquiring a pair of 808 tube amplifiers and over last two months I completely rebuilded the amplifiers in order to improve the sound quality and reduce the noise floor.
The final circuit (lower in this page) is inspired by Shishido circuits but it’s not the official Shishido 808 circuit..
The output section, the grid bias circuit is from this 808se circuit :
The driver stage is a variation of this shishido circuit :
When I got the amplifiers I was planning to get rid of the 6EM7 since the nature of the tube makes it quite hard to match both triodes to get similar gain between channels. I tried going back to 2 separate tubes and realized the 6EM7 is overall a truly fantastic tube and shopped around to find 3 good matched pairs (mostly the first triode need to be matched) so I will keep this configuration.
In the circuit, the plate resistor for the first triode can be replaced with a pair of lundahl LL1667 / 05mA for a nice sonic upgrade. You will have to position them well to prevent any noise pick up. I’ve experimented with 2 hammond 156C in serie too, it work well but is very noise sensitive. I will be returning to plate chokes on the first triode eventually.
I got the amplifiers with noisy edcor power transformers and one electron ubt-1 output transformers. The one electron didn’t provide enough primary inductance to have proper bass and their impedance wasn’t high enough. The edcor had too much voltage variation.
I’ve decided to go full hammond transformers in the rebuilding process, the power transformers are the 374BX, the outputs 1630SEA, the interstages 126B, chokes 159Q, biasing transformers 166E30. The 1630SEA completely changed the low end of the amplifiers and are very good. I know the previous generation had their flaws but I think hammond did well with the SEA serie. The 1630SEA are massives btw, you will need a large chassis to build all of this. It was the first time working with the hammond 126B and the measurements are really impressives. I didn’t need any modification of the loading at the output to damper some resonances as I had to do with lundahls in the past.. The PS choke is a little bit small but I have limited space. If you can, go with 10H 200mA or 8H 200mA or go with CLCLC. With my actual configuration, using good Solen PPE, I have more than enough filtration as it is. I brought the noise down enough to use the amplifiers directly connected to a pair of 112-113 db eff horns in an active system.
For the 808 filament I went with switching power supplies. First, I’m not interested in going AC with anything DHT to avoid any intermodulation distortion with the filament. Secondly, I’m not afraid of switching power supplies if they are properly built and shielded. A friend recommended it, I tried it since it was relatively cheap to experiment and the results are very good. One trick with this kind of switching is not to add any capacitor at the output and not trying to add any filtering at the input. The manufacturer did his job filtering and eliminating noise and it’s easy to mess this up. Also, get he shielded version of the switchings if you go that route, the model ending with -C are shielded. I went with a pair of Mean Well EPS-65-7.5-C with the faraday cage. The noise floor with theses switching is a lot lower than with a conventional rectifier capacitor solution that was implemented when I received the amplifiers. Having separate supplies for the heaters bring the opportunity to get a separate preheating mode for the 808.
For the power supply capacitors I went with Solen PPE serie of metalized polypropylene capacitors. Avoiding electrolytics, especially at the last stage of filtering is one simple way to improve the overall sound of an amplifier like that. Use the same PPE from Solen for C3 C4 C5 C6 C7 if you can. C6 can also be mundorf Mlytic or Elna cerafine serie. The rectifier is a 5R4WGb which is one of my favorite. I did chose the 374BX knowing I would use the potato masher with higher loss of voltage but you can always tweak the final voltage changing the size of C5 which is the first capacitor after the rectifier. The smaller C5 is, the more space you need around L1 to avoid any induction from the magnetic field of the choke.
You can use the amplifiers with or without feedback. I don’t use any feedback when using it to power my midrange horns but adding 3-8db of feedback when using them full range give really good results. If you add feedback you will probably need to put a small capacitor in parallel with R13, adjust the size looking at a 10khz square wave on an oscilloscope. Stay with polystyrene or silver mica for this capacitor if possible. Personally I added 2 switches on the back for the feedback so I can change from 0 – 3 – 6 db for NFB on the fly. The NFB resistor and capacitor is very dependant on the output transformer used. You can also make the feedback loop outside the output transformers with a shade style injection. With the 1630SEA I can include them in the NFB loop with a real load. I had some 30khz + resonnances with the ubt-1.
Noise can be an issue with the 808. First, the filament need to be clean at 7.5v 4A. You should go DC for this considering the the 30w it will pull. Secondly, the tube need to be far away from the power transformer and the PS input choke. This choke, been almost a choke input, will radiate quite a bit and proper positioning and spacing is important. I have small chassis so I had to add Mu Metal sheet between the 808 and my transformers to reduce the 60hz hum. Using proper grounding techniques you can bring the amplifiers under 1-2 mv of hum before feedback. The 6EM7 is not too finicky fortunately.
I tried different operating point and returned to 350v b+ and 120mA (about 40w on the plate) for the 808. It will give you around 8-8.5W output depending of your output transformer efficiency. Adjust the lm317 to get around +30Vdc (biased in A2) at the first startup and when the tubes are hot adjust to get 0.120vdc at the 0.1 ohms test point. With good 808 you will need around 26-28Vdc on the grid (the other side of the interstage) to get there. The 120mA is more important than the grid voltage when adjusting and let the amplifiers heat up before the final adjustment.
For the driving stage you can remove both cathode bypass capacitors if you want. It will be beneficial to remove both if you don’t use feedback since the overall gain will be on the high side. The low mu triode of the 6EM7 sound better with degenerative cathode. The high mu triode sound good with or without, try it for yourself, you can even add a switch for this.
With dozen of reconfiguration and lots of testing, this is the configuration that I prefer so far. I will be testing the lundahl plate choke again now that my noise problem is fixed. It should work well if you have proper spacing between the plate chokes and … every thing radiation magnetic or electrical field.
Getting good 808 can be complex and apparently they can get gassy. I was lucky and got two pairs and a spare that are all good. The tubes are very old and you will probably need to test them on the amplifier directly since not many tube tester will support them. Btw it is perfectly normal that the tantalum plate of the tube get cherry red. This plate is used as a getter for the tube so don’t panic when they heat up. In the picture, the GE 808 are installed in the amplifiers. The RCA are build a little differently inside but measurement pretty much the same as the GE. I couldn’t find any other maker for this tube.
The amplifiers sound really nice but are more on the open and dynamic side than the sweet one. They are not as accurate as my SETH PP 2A3 but have that midrange clarity of large DHT while not having the typical dark sound of low voltage 845 or some 300b. If you have heard 845 running 1000V + of B+ and also heard vintage single plate 2A3 … it’s kind of a mix of the two. It’s a vivid tube with good level of details. Considering there is no expensive iron in this build, the voltages are reasonable for the tube amp and it’s a A2 class of operation … I’m really happy with the results.
My final circuit and couple pictures of the monoblocs :
You can right click on the images and select “open in new tab” to see them full size.
I plan on replacing the front switches soon and do something with the side wood panels that need sanding and some finishing.
Another thing I will be experimenting with soon is converting the amplifiers for SV572-30. I have some pairs of those here and I always wanted to see what I could do with them and the conversion to these seems relatively easy to do been already in A2 with grid bias, same socket and similar requirement. We will see but I would like to be able to run both tubes with a set of switches and quick bias adjustment. The 808 are getting impossible to find but the SV572-30 can still be found.
If you have any question, let me know. You can find my direct email in the about section.