Mcandmars ramblings \ build thread...

[mcandmar]

I figured i would document the modifications i have done to lower the background hum of the amp. The motivation for all this was to lower the noise floor enough for my MS-Pros which are a variation of the Grado RS1 (32ohm) and rather sensitive.  To give you an idea of the noise level i am chasing the amp is almost silent with Sennheiser HD650's, you really have to concentrate to hear it.  But with Grado 80i, MS1, 225i it becomes more and more noticeable, and with the MS-Pros it becomes distracting during quite passages and between songs.  Output is set for the 4ohm tap, unbalanced.

Initial build and photo gallery can be found here --> http://bottlehead.com/smf/index.php?topic=5252.msg50648#msg50648

Part 1: Glowing bottles.  The general consensus is that there is little to no difference between tubes from different manufacturers.  Its pretty safe to say tube rolling isn't a thing with 6DN7's. There is in my opinion one or two exceptions to that rule being the Sylvania Black Plates, and Tung-Sol tubes.  Sylvanias can occasionally be found re branded under a few other labels such as International Servicemaster, Realistic International, Fivre, and Mullard.

So far i have collected sets from GE, RCA, Zenith, Fivre, National Electronics, International Servicemaster, Mullard, Tung-Sol, and Sylvania. I would go so far as to say 9/10 tubes are the same tube regardless of the manufactures stamp. The Sylvania however are clearly different in their construction, easily recognized by their cut away top support disk.

There are a few key differences in tubes so i have attached a few images to illustrate the point. Most are regular grey plates however some are black plates, and some have one grey and one black plate. And then the same three variations can be found in regular and coin base packages.  You may occasionally come across tubes with an exposed heater winding on the top which are wired in series instead of in parallel, as shown in the Zenith tube below. Overall my personal preference is for the Black plate Sylvanias.

Another alternative is the 6FJ7 tube which is electrically identical to the 6DN7 but packaged in a 12pin Compactron base vs the regular 8pin Octal, as shown in the last picture. A full write up on the adapter can be found here --> http://bottlehead.com/smf/index.php?topic=5811.0

[mcandmar]

Part 2: Heater supply cap and resistor.

-Capacitor: As discussed further on in the thread doubling the capacity of the smoothing capacitor serves to half the ripple on the supply but has little to no audible or measurable effect on the noise levels of the amp. The capacitor i have used is a 22000uf Panasonic TS similar to the rest of the power supply caps, and physically the same size.

10000uf 10v - Lelon REA 85c 20% H010(M) 35mm x 18mm x 7.5mm - 2430mA
22000uf 16v - Panasonic TS-H 85c AECOS1CP223CA - 25mm x 35mm x 7.5mm - 6100ma 

- .1r Resistor: My hatred for those ugly wire wound cement blocks was my motivation here to source a Dale for no other reason than its not a cement block.  There are two versions of these being inductive and non-inductive, i opted for the non-inductive version.

- 100r Resistors: As suggested by PJ connecting two 100r resistors in series between the + and - terminals and grounding the center point serves to balance out the voltages on the DC side of the heater circuit.  This simple modification seems to have a considerable effect in reducing background hum and is highly recommend if you are having issues. Connect one 100r resistor between terminal C1 to C3, and C2 to C3 as shown in the attached pic.

[mcandmar]

Part 3: Voltage doubler diode upgrade.  Replacing the two UF4007's with Cree CSD01060A as used in the Mainline.

I then took a snap shot of the noise at the headphone socket before/after and the difference was so small i couldn't see any difference between them. They work, and presumably better, but i really cant quantify it. As they say in motorsport, every little bit helps.

[mcandmar]

Part 4: Headphone jack resistors / Power supply chokes,

As suggest by Doc.B adding two 120ohm resistors in series with the headphone jack greatly reduced the background noise/hum on the amp making it almost silent. Refer to heater supply 100r resistors above.

Installing the two C3X chokes in place of the 680ohm resistors in the main power supply removed the remaining 10% of background hum.  Having listened to the amp at length with various headphones i cannot hear a thing, just a lovely black background which makes a world of difference to the clarity/detail. Really happy with the result

Specs for the C3X are 10Henries, 50mADC, 500ohms. They both measured ~470ohms on my meter however the power supply voltages measured before/after were all within three volts of each other.

As for physical placement, it would be possible to locate them to the side of the tube sockets on standoffs but i wanted to leave a bit of room around that area to replace the two capacitors in the future.  The only other place i could find was the front edge of the chassis so they sit below the headphone socket and power switch.  Obviously there are no issues with hum from their placement.

For the bracket i used a piece of steel strip i had, crudely bent to shape in a vice to fit around the various components and held in place by the volume pot.  I lucked out as i had a washer under the volume pot to space the knob flush with the top plate that happened to be exactly the same thickness as the steel strip so it all just fell into place.  Gave the bracket a quick sand/spray with clear coat to prevent rust and called it a day.

[Doc B.]

What is the vertical scale set to in your photo of the oscilloscope trace?

[Paul Birkeland]

If you absolutely must lower the noise floor even more, you can add 6dB of global feedback.  It'll basically f#ck the sonic signature of the amplifier, but you'll accomplish your goal. 

Swapping this part out for that part isn't going to make much of a difference.  The wirewound resistor is the power supply is 0.1 Ohms and way, way out of the signal path, so mucking around with that part may appease your eyes, but it is otherwise kinda pointless.  I would not at all expect the power supply chokes to lower the noise floor either. 

FWIW, I'd put the UF-4007's back in.  The Crees don't always do well being wired in a PTP circuit, as they have that metal tab that can cause quite some problems.

[Paul Joppa]

You can go larger with the heater power capacitor. I suggest you first just parallel the ones you have on hand, to determine how important that is - doubling the capacitance should halve the ripple if it is entirely due to the heaters. If that helps, then larger caps are readily available; a 10v rating is sufficient.

Do be aware that as the tube settles in, the getter will reduce the gas in the envelope and that will reduce heater-to-cathode leakage - this is one mechanism by which letting the tubes settle may improve the situation.

[mcandmar]

What is the vertical scale set to in your photo of the oscilloscope trace?

5mv/div

[mcandmar]

FWIW, I'd put the UF-4007's back in.  The Crees don't always do well being wired in a PTP circuit, as they have that metal tab that can cause quite some problems.

Don't worry the live backing tab was one of the first things i noticed in the datasheet, i have them well protected with heatshrink and tie wrapped to the plastic standoff for the C4S.

[mcandmar]

You can go larger with the heater power capacitor. I suggest you first just parallel the ones you have on hand, to determine how important that is - doubling the capacitance should halve the ripple if it is entirely due to the heaters. If that helps, then larger caps are readily available; a 10v rating is sufficient.

I will try that and see what effect it has, thanks Paul.

[Doc B.]

Btw, that trace is with the volume turned all the way down, yes? Knowing that is a 5mV/div scale, it looks like you have about 1mV pk-pk of overriding RF hash, probably from the probe cable, that is thickening the trace of the 100hz ripple. If you allow for that we are looking at about 1-1.5mv pk-pk of ripple, which translates to about .3-.5mV rms of ripple. So you are pretty much on spec for that amp.

I think the best solution for you is to use the 120 ohm series resistors at the headphone output that we have always recommended for lower impedance headphones with this amp. It will improve your signal to noise ratio.

[mcandmar]

Btw, that trace is with the volume turned all the way down, yes?

Yes volume all the way down, but the volume level doesn't have any effect on the background noise at all.  Source and headphones were also disconnected to eliminate them as sources of interference.

If you allow for that we are looking at about 1-1.5mv pk-pk of ripple, which translates to about .3-.5mV rms of ripple. So you are pretty much on spec for that amp.

Good to know, i had a feeling i was at the normal level of the amp/tubes. I am just looking to try and improve things as much as possible, open to any suggestions.

I think the best solution for you is to use the 120 ohm series resistors at the headphone output that we have always recommended for lower impedance headphones with this amp. It will improve your signal to noise ratio.

I will certainly give that a try, the wiring around the two headphone sockets is rather congested so i have to come up with an elegant way to fit them in.

[Doc B.]

Since the headphones were unplugged that is an unloaded noise floor reading. If the headphones were plugged in it would be even lower, so the figure is definitely in the nominal range.

If you want even more quiet background for headphones, buy a Mainline. There are limits to what one can do to make a cost effective amp that will play both speakers and headphones, and Mainline does not suffer from the same constraints.

But if you just put the 120 ohm resistors in I think you will appreciate the lower noise.

[mcandmar]

Resistors on order, will report back when i get them fitted.

I would love a Mainline but the S.E.X. kit was already a stretch for me, i was researching the Crack kit when i discovered it and got sucked in. I don't regret it at all though as i have come to the conclusion i'm a Grado man, i just cant get past that layer of disconnect with the Sennheisers so the S.E.X. amp was the right choice for me. I am really enjoying it as it does pair well with the Grados, if i can just knock the edge off the background noise it would be perfect.

In the meantime i'm going to continue enjoying it and tinker as i go, the experimentation and learning curve is all part of the fun for me. Next big purchase will be a decent DAC in the new year so i'm eagerly waiting to see what you guys come up with..

[Loquah]

Forgive me if I missed it, but have you installed the C4S?

I might be making this up, but I believe it helped the noise floor on my S.E.X.