My GSG Mods

[cddeluca]

I had earlier posted about severe hum in my GSG and eventually resolved it by replacing the JAN tubes with Russians, or so I thought:

http://www.diytube.com/phpBB2/viewtopic.php?t=5371

Running direct from a line input it was acceptable, but when I put a preamp back in it was horrible again.

I made a couple of more changes this morning and now I can claim unqualified success - I can put my head in the right speaker horn and hear nothing; in the left horn I hear a barely perceptible hiss. And no, I didn't ride a jack-hammer at work this week.

As to what ultimately made the difference, I don't know, but I suspect it was principally lowering the gain of the input stage. For anyone who cares, here's what I did (in order) after I replaced the JAN 6B4Gs:

1. Installed line filter (common mode chokes, full Monty)
2. Class X cap across the IEC terminals
3. Grounded output binding posts (a good safety idea anyway)
4. Two anti-parallel pairs of 1N4 diodes between the circuit ground and chassis ground
5. Changed the 6SL7 cathode resistors from 3K to 5.1K
6. Replaced the 6SL7 with a Sovtek 6SN7

When I first powered it up I was very disappointed - it was absolutely silent and I figured I'd buggered it up for fair. It was only when I put my head in the second horn (couldn't hear anything in the first one) that I realized it might just be that quiet.

Personally, unless you're running a GSG direct, I'd go with the 6SN7 anyway - there's still tons of gain, more than enough with a pre-amp.

[BruceAnderson]

My experience and results somewhat mirror this. I made the following mods:
1) A 1uf 50v ceramic cap in parallel with a 5uf pp caps was placed across each filament.
2) a .047 pp cap goes from each 6SL7 filament leg to the ground side of R5 and R6.
3) A 1 ohm 6W resister in parallel with back/back diodes go from J6 pin 3 to chassis ground.
4) I replaced R17a and b with a 2nd 1.5H choke.
5) 1 600V pp .1 uf cap goes from 6AX5 pin 3 to pin 5 (snubber).
6) Lastly, the following can only be done if using the SOVTEK 6B4G's but may have made the biggest single improvement: I placed one UF4007 from the 6AX5 pin 3 to a 100 ohm 10W resister terminating at J5 pin 1. The same is done from the 6AX5 pin 5. This increases B+ as measured at the 6B4G pin 3 to about 345 volts.
The result of these mods is a quiet amp with a 6SL7 front end and in my opinion, hands down, one of the finest amps I have ever heard. Thanks to all that made this wonderful amp available.

[Shannon Parks]

Thanks for posting your DIY mods, guys! I finally got a chance to spice the 6SN7 mod, and it looks OK (voltage gain-wise). You will need a preamp able to drive 2Vrms for max output. One can also remove the cathode resistor bypass caps from the stock design for a 6dB drop in gain - a pretty simple solution if it's a problem.

My rule of thumb with these amps is to get max output with 1Vrms, with the exception of my newest Budgie amp which will require 0.5Vrms for max output. As more of a work or dorm amp, it is more suited to being driven directly from an iPod or smartphone.

Shannon

[highflyin9]

I'd love to see photos of your modded PCB Bruce, if time permits

[BruceAnderson]

Frankly, I don't even own a digital camera but I will see what I can do. In the meantime I have done more "playing around" and I am so happy with the amp I can't stop from smiling along with the music.
I placed a 10K 2W resister in parallel with the 2.7Ks for a total cathode resistance of 825 ohns (down from 900).
I added a 100uf 450V cap in parallel with C8 This becomes possible since I no longer use the 6AX5-GT.
I added a .1uf Russian surplus teflon cap in parallel with the .22 PIO coupling cap (takes FOREVER to break in).
Most important, I tried a direct connection from the board ground to the chassis ground (bypassing my resister/diode bundle). Shannon was absolutely correct and I had just over thought the problem. This revision does indeed reduce hum even further.

[highflyin9]

Regarding the snubber Bruce. I was reading a little more about it as I was unfamiliar with the term and purpose. It seems caps are specifically made for this purpose, either "snubber" or "high pulse" caps and a high voltage is often recommended (from 600-2000V in various cases that I read about). It seems this helps remove switching noise from the rectifier. It looked like people also add diodes to a snubber circuit.

To be beneficial in the GSG circuit, only a 0.1 uF cap is connected from pin 3 to pin 5 on the 6AX5?

Just trying to learn more, many thanks for any educational tidbits :)

[cddeluca]

If you want to know more this book is invaluable:

http://www.amazon.com/Designing-Power-S ... amplifiers

[BruceAnderson]

The first time I saw the power supply snubber was in the Bottlehead forum. They used a slightly more complex circuit to keep diode rectification nasties from going back thru the transformer and into ground, etc. I later read that the single cap like I used gives almost as much benefit. Don't expect a night and day difference but since I was already in there and had the parts I figured I might as well do it. I actually used a 600 V .1 PP cap in parallel with a 1000 V .01 ceramic.
By the way, I looked at your website and am in awe of the beauty of your builds. Wish I had the skill to do that.

[highflyin9]

Thanks for the book recommendation cd and good info Bruce. Trust me when I appreciate the compliments Bruce, definitely no skill involved, just time and a love for the hobby

[kku]

Replacing R17a/b with another choke means resistance down to 50 ohms from 90 ohms. How does this affect B+ and anode voltage? Should this be a concern if using NOS 250V anode voltage rated 6B4G's?

Bruce, if I get this correctly, direct connection from board ground to chassis ground + diode/resistor from board ground to chassis ground; both in parallel to each other going from PCB ground to chassis ground?

[BruceAnderson]

I noticed very little change in B+ with two chokes. I was using NOS 6B4G at the time with the 825 cathode resistance. I sure would make measurements the first time you make the mods just to be sure.
As far as the ground, I tried the resistor/diode combo first and it made a small improvement. Later I bypassed that with a direct connection and a further small improvement was seen. The resister/diode no longer is in play when bypassed and could be removed. These are simple, easy revisions that can be reversed if you are not happy with the result.

[kku]

Thanks for explaining, I appreciate it.

[BruceAnderson]

Well, never say never but I believe I have finished my Get*Set*Go mods. The final revision had nothing to do with the circuitry, it was to deal with a raspy buzz coming from my power transformer. I suspect, given the number of people who claimed their amps were quiet when bone stock, that my transformer is noisier than normal. When lightly placing my finger tips on it I could feel a small vibration. I wouldn't think shaking the top plate is in any way optimal. I tried tightening everything down but that didn't seem to make much of a difference. Next I put soft vinyl grommets around each bolt and tightened everything down only to the point that the transformer body is slightly above the top plate. The top plate bolt holes were not enlarged and the grommets sit on the plate not through the holes. I made very certain that electrical continuity between the plate and the transformer (and the power cord ground) remains in place. That made all the difference. No more buzz. I'm extremely pleased with this amp.

[kku]

As recommended by the designer of the Tubelab SSE, when I added a 80uf motor run capacitor in parallel with the last electrolytic in the power supply of my Tubelab SSE, it was the best thing I ever did to that amp. It made it sound silky smooth and juicy, opening up upper frequencies even more and just more fuller notes in general. I'm thinking about doing something similar to my GSG. Thinking about installing a motor run 80uf cap in parallel with C9. And replacing C10 with a motor run 20-30uf cap. Now I'm not knowledgeable at all in electronics, so I had to ask here if doing such to GSG circuit will be OK to do?

[highflyin9]

I've definitely seen a number of tube amps use the large oil filled polypropylene caps in power supply filter applications. http://www.partsconnexion.com/capacitor_film_asc.html ASC in particular seems to be pretty common. The 120uF caps are not completely in parallel though, R17 sits between the positive legs (either a pair of resistors or a choke), so I don't know if you could have different values on each side. I'll have to leave that question to the experts