Recently Completed Stereomour - Hiss in Right Channel

[Montigne]

Bottlehead,

I recently completed my first amplifier kit, your Stereomour, and couldn't be happier. The quality, sound, and fun of it all for under $900 is very satisfying. You have made me a believer, and I truly feel that DIY vacuum tube-based Hi-Fi is the purest path to audio magic.

One problem: I have used it for about two weeks every night for a few hours and weekend days, and am getting a small hiss in the right channel only. I've tried swapping source component cables at the outputs, swapped the 2A3 tubes from L-R, and it's not the hum pots either. Not a buzz or hum, definitely a soft hiss. I'm using higher efficiency speakers and the noise is very subtle, but definitely audible when I get close to the driver and I'd like to have it perfect or remedied if possible. There's the usual turn on hum for a few seconds at power up that goes away, but I don't hear the hiss right away until the hum goes away and then the hiss kicks in like a switch. I am planning on getting another 12AT7 to swap and see if it's the NOS tube that came with the kit, but I wanted to get the question out into the forum first.

Any suggestions? Thanks in advance Bottleheaders.

[fullheadofnothing]

This could be many things. A different 12AT7 is certainly a good place to start.

How efficient are your speakers? You may simply be dealing with the noise floor. Can you hear the hiss with the input shorted? Is it volume dependent or not? What impedance setting did you use? What impedance are your speakers?

It seems that this noise is not audible from the listening seat. Is that correct? You will drive yourself insane tracking down noises you hear with your head pressed to a speaker, but they don't effect anything in the real world.

[Montigne]

Thank you for the reply. Yes, the hiss in the right channel is slightly audible from the listening position when the program material is soft or completely quiet. I know it'll drive me crazy trying to track it down, but I feel there could be a solution because it's something in the circuit making it manifest in the right channel only.

I've wired the OTs of the Stereomour kit to 16ohms, and my speakers are 16ohm as well. They are RCA LC-1Bs and are about 97db efficient I think. I tried a different NOS 12AT7 and also a new JJ tube, but the hiss still remained, leading be to believe it's source is in the circuit somewhere. I even tried shorting the inputs (which actually quieted them down vs. having a source cabled in), but it didn't't change the hiss. Conversely, the left channel is dead quiet.

Anything else I could try? Thanks for the feedback so far, I appreciate it.

[triode]

Hi
I have the same problem with Stereomour hiss in my horn speaker.
I changed the 431 in 12AT7 cathode with resistor and the hiss is gone

See toward the end of the thread : http://bottlehead.com/smf/index.php/topic,4521.0.html

Hope this helps

[Paul Joppa]

If this works, please post about it - I'm gathering data.

[Montigne]

Thank you for the suggestion, I had also read that thread but the swapping of parts and mentions of it affecting the frequency response of the amp deterred me honestly. If possible, I'd like to avoid deviating from the stock kit with modifications to try to solve this. The left channel is dead quiet, why would only the right channel exhibit this symptom? If I understand it correctly, the 431 is a shunt regulator that controls cathode bias voltage of the 12AT7 tube? I assume this is done in each channel, correct? Would trying a replacement 431 be worth it, or would I somehow have the same issue? I have read about the resistor replacement solution you said worked Triode, but what value of resistors should I use and how would I replace the three leads connection points on the 431 with just the two on a resistor?

Sorry if I'm asking redundant questions found in other threads, I'm still learning and really trying to understand. Thank you for the continued help.

[Paul Birkeland]

You could indeed just pop in a different TL431 and see if that helps.

-PB

[Paul Joppa]

Before I get into details, there is always the possibility of a poor connection, solder or tube socket or corroded tube pin.

That aside, PB is right. The reason I want more data is a suspicion that we have acquired some bad, possibly fake, 431s - there are a LOT of fake everythings coming out of [country names withheld], not just Rolexes. We seem to have had this problem only recently, or at least mostly.

A proper 431 should meet the specification. The most complete specification is that for the ON semiconductor product, which is what I used in the original design. It includes a very complete description of the noise character, which is better than most of these "adjustable Zener equivalent" chips - that's why we use that chip. You can swap the 431s to check - swap the connection from 12AT7 pins 3 and 8 to their respective kreg  connections, at one end or the other. If the noise moves then it's the 431.

If you don't want to unsolder the 431 and replace it (I wouldn't want to, my hands are too shaky these days) you can replace the chip with a resistor as described in the referenced thread. I'm not checking that thread while posting this, so let us know if it's unclear. The C4S current is about 3.8mA and the chip drops 2.5 volts, so the perfect resistor is about 660 ohms; 680 ohms is close enough and widely available. This will be quieter than even the best 431, if it is metal film or wirewound or bulk metal foil. It raises the driver's effective plate impedance by about mu times the resistance; mu is nominally 60 so about 55K total. The 2A3 Miller capacitance is about 70pF giving a corner frequency of 41kHz. At 20kHz that 1dB reduction, so probably not audible unless you have bat ears.

I think that same thread mentions substituting an LED for the resistor. The problem is, we don't know much about LEDs other than the one we use which is noted for both low noise and a low dynamic impedance. That LED, the HLMP-6000, runs about 1.57 volts, significantly less than the target 2.5 volts. It should work, we've used it in previous incarnations of this circuit, but I moved to 2.5 volts for a reason, I think it's a more reliably good sounding operating point, given the variety of 12AT7s out there. (This has to do with differences in grid-cathode contact potential, if anyone cares.) My compromise would be to use an HLMP-6000 in series with a 240 ohm resistor, which moves the corner frequency to over 100kHz.

[Montigne]

Paul,

Thank you for the detailed response. This weekend I'll begin by checking solder joints, but then will likely want to swap out the 431 if it's not a connection joint. Is it something I can order from Mouser, and if so, is there a specific model or part # I should be looking for? Or, can I order one from you guys? If that doesn't work I'll be dong the resistor substitutions if I can find a 660 ohm (or 680). If I do that, where would the two leads of the resistor go instead of the three of the 431, and what direction should the resistor be pointing?

Thanks.

[ffivaz]

That is good news. I have an audible hiss (from my seat, 3 meters away) in both channels since the beginning. I replaced the 12AT7, reducing the noise only a little. I also was reluctant to change the 431s with resistors since it would affect the sound of the amp. But given PJ's exaplanation, I'll give it a try.

[Paul Joppa]

The "previous referenced thread" is http://bottlehead.com/smf/index.php/topic,4521.15.html and my most current note is reply number 18.

I see there are no detailed descriptions, sorry! I'll try remotely (I don't have a Stereomour at hand).

You'll want to unscrew the PC board over the 12AT7 socket, first, so you can get access to the socket. This is going to be fussy work, sorry! In the link I mention 499 to 720 ohms as workable, with 660 the one that best matches the original design. I'll call it a 680 in the instructions below.

There is a black wire that goes from B3 to kreg on the board. Remove it at both ends (* see note). Do the same with the black wire from B8 to the other kreg.

Take a 680 ohm resistor  and solder it to B3 at one end, and terminal 6 at the other end. Yes I know, T6 is crowded, if T8 is easier then you can use it - 6, 8, and 10 are grounds.

Take another 680 ohm resistor and solder it from B8 to T10 (or T8 if that is easier)

*note that if the socket is impossible to get to, you can use the wires already soldered to B3 and B8, merely removing them from the kreg terminals on the PC board. Connect the resistor between the wire and T6, and the other between the other wire and T10. Rather than cut the wire to length, you can fold it to shorten it if that is easier - pinch a piece and twist the loop to minimize magnetic pickup. I'll try to do an ASCII picture here:

      _       pinch
      X
      X       twisted section
      X
______X_____ wire

This will require that you connect the resistor to the wire without a terminal to solder both of them to. Bend both 180 degrees to make a solid mechanical connection before soldering, and make an apologetic sacrifice to the solder gods for bad practice!

[Montigne]

Paul,

Thank you for the step-by-step response, this will be very helpful. I have a few more questions before I attempt the fix:

- I'd like to try replacing the 431 in the right channel first, so is there a specific part # I need to order one from Bottlehead or Mouser?
- I've also found 650 ohm metal film resistors, but what power rating, tolerance, and voltage rating should I get?
- in your previously posted instructions, are your steps of adding the resistors after removing the 431's, or do I leave those in and just bypass them with the resistors in your steps?

Thanks again.

[Paul Joppa]

Go to the Mouser site. It's under semiconductors > power > Power management > voltage references, then restrict the search to ON semiconductor. It's a TL431 in a TO-92 package. There are several variants, you want some in bulk (not on a tape or a reel of 2000 or an ammo box full of them...) and that are in stock. Other suppliers have similar search engines - now's a good time to learn to use them, they are quite handy.

I'm specifying ON because they have the best data sheet (you can download it from Mouser; the noise specification is in Figure 13 FYI) which is what I've used for all my designs to date. I believe it may have been a TI design originally, and they are a reliable manufacturer as well. Other variants may have different specs, I have not (obviously?) gone around reading and comparing all the available data sheets!

[Montigne]

Paul,

So I called last week and Eileen sent me three new 431s to try. I tried all three in the channel with the hiss, and it still remained. Also, I even tried one of these new 431s in the silent channel and guess what: hiss was introduced where it wasn't before. I took a closer look at the markings on each and found that the original two that came in my kit were different: fKB31 LM431 ACZ (silent channel) and WS TL431A 12427 (hiss channel). Also, the three that were sent as the replacements were all the WS TL431A 12427 as well. My conclusion: I agree with your previous assumption that these different 431s are the hiss problem possibly due to manufacturing.

My next step will be to bypass the 431s with resistors as instructed in your previous reply, but what tolerance, voltage, etc. of 680 ohm resistors should I get? What specs on those very resistors should I stick to to have a successful outcome?

Thanks, I feel we are close.

[Paul Joppa]

Wow! Thanks so much for doing such careful work and reporting it! Now we're getting somewhere ... :^)

For the 680 ohm replacement, the power is tiny so any size is fine. Use metal film for preference - I like the Dale RN series metal films but most brands are very quiet. A really good carbon film can be quiet but given the origin of this thread I would not use one to start with. Wirewound would be good as well; non-inductive or not should make no real difference in this application but they are likely to be bulky and inconvenient to install.