So close and yet so far.... The "Heavy" preamp mess continues

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
ruffrecords said:
Yes
You test it with no audio. Short the input to minimise noise. You are just looking at the twice mains frequency ripply. It will be a lot higher than any noise created by the mic pre itself.

I don't risk my scope by connecting it directly to a big dc voltage with a small ac signal on it. I use a 100nF/400V cap from the HT supply to a 100K resistor the other side of which is connected to ground and measure across the 100K

Cheers

Ian


Moving right along!  I added the additional decoupling you recommended using some parts I picked up this morning.  Decided to just plug a mic in, throw on the headphones, and give it a listen.

The hum is GONE!  Hallelujah!

Clean as a whistle.
 
CurtZHP said:
Moving right along!  I added the additional decoupling you recommended using some parts I picked up this morning.  Decided to just plug a mic in, throw on the headphones, and give it a listen.

The hum is GONE!  Hallelujah!

Clean as a whistle.

Now you understand why removing ripple from a power supply is so important for a mic pre. I realised this when I was at Neve and saw the 1000uF capacitors decoupling the supply to 1073 modules.

Cheers

Ian
 
ruffrecords said:
Now you understand why removing ripple from a power supply is so important for a mic pre. I realised this when I was at Neve and saw the 1000uF capacitors decoupling the supply to 1073 modules.

Cheers

Ian


Oh, yeah!
The learning never stops!

So, last night, I confirmed that the other channel works too.  Hooked it up identically to the first one.  The only problem I'm having now is a little bit of crosstalk between them.  When I connect a tone generator to Channel 1 and a headphone amp to Channel 2 I can hear the tone.  (Swept through a bunch of frequencies to make sure I wasn't just hearing some other grunge...)

I've isolated it to the first stage, upstream of the gain pot.  I also connected the generator downstream of the input transformer (basically right to the first stage grid) just to rule out any silliness there.

Looking at my wiring, each channel uses one half of a 12ax7 for the first gain stage, and then they each use one half of a 12au7 for the second stage, so the tubes are each carrying both channels.  Could that be a source of crosstalk?  (I tried briefly to find some crosstalk figures for that, but came up empty.  The one datasheet I have doesn't even mention it.)  Or, more likely, the outputs from the 12ax7 plates are presently running parallel with each other for about 3 inches, about a half inch apart.  And their plate supply wires are right next to each other.

I've already reworked the layout to get the inputs and grid resistors a lot closer to the tube, so I guess I'll need to get these first stage outputs away from each other too.



 
Physically separating the two channels as much as possible if obviously a good idea. There will be crosstalk inside the tube but it will be very frequency dependent. It will be due to the small capacitance between the two anodes. There is usually a figure for this in the data sheet.

Physically separating common current carrying conductors is also a good idea so take a separate 0V from each channel back to the PSU. Similarly have separate HT decoupling for each channel.

Cheers

Ian
 
ruffrecords said:
Physically separating the two channels as much as possible if obviously a good idea. There will be crosstalk inside the tube but it will be very frequency dependent. It will be due to the small capacitance between the two anodes. There is usually a figure for this in the data sheet.

Physically separating common current carrying conductors is also a good idea so take a separate 0V from each channel back to the PSU. Similarly have separate HT decoupling for each channel.

Cheers

Ian


Yes, I definitely have separate supply decoupling for each channel.

I came across a couple 12dw7 tubes recently.  Sure, I'd have to redesign the whole layout, but I would think that would certainly keeps both channels well away from each other if each had its very own tube from input to output.  :D
 
CurtZHP said:
Yes, I definitely have separate supply decoupling for each channel.

I came across a couple 12dw7 tubes recently.  Sure, I'd have to redesign the whole layout, but I would think that would certainly keeps both channels well away from each other if each had its very own tube from input to output.  :D

Yes, that would definitely help.

Cheers

Ian
 
CurtZHP said:
I guess the only concern would be that they are not as common as the others, so I'd have to make sure I keep spares on hand.

They used to be very rare but I think they are in production now so less common than they were. Not so expensive now so you can afford to keep a few spares.

Cheers


Ian

 
ruffrecords said:
They used to be very rare but I think they are in production now so less common than they were. Not so expensive now so you can afford to keep a few spares.

Cheers


Ian


I checked with The Tube Depot (where I usually shop...).  They've got them.  Pretty much same price as the others.  And for an extra few bucks, they'll match them.  I might have to give this some serious thought.  At least I've got the power supply sorted out.

:D
 
So, I took a little more time with it today to try to physically separate the first-stage outputs as well as their respective plate supplies.

Neither did a thing to improve the crosstalk.

Before I go through with doing similar with the cathode networks of either channel, is that even worth the trouble?  Or should I just assume that the crosstalk is confined to the tube itself?

 
CurtZHP said:
So, I took a little more time with it today to try to physically separate the first-stage outputs as well as their respective plate supplies.

Neither did a thing to improve the crosstalk.

Before I go through with doing similar with the cathode networks of either channel, is that even worth the trouble?  Or should I just assume that the crosstalk is confined to the tube itself?

Grounds are a prime candidate so I would say it is worth doing them. On the HT side, the individual decoupling helps improve crosstalk but there is none on the 0V so keeping them separate until the power supply is a good idea. Any wire common to both channels is a possible source of cross talk. You can either separate them until the power supply or run them in a real thick very low impedance bus bar. I use the copper from 15 amp mains cables for my 0V bus bars in point to point builds.

Also make sure that interstage signal connections are as short as possible and input and output signals are carried in screened cables.

it is just a question of chiselling away at it.

Cheers

Ian
 
ruffrecords said:
Grounds are a prime candidate so I would say it is worth doing them. On the HT side, the individual decoupling helps improve crosstalk but there is none on the 0V so keeping them separate until the power supply is a good idea. Any wire common to both channels is a possible source of cross talk. You can either separate them until the power supply or run them in a real thick very low impedance bus bar. I use the copper from 15 amp mains cables for my 0V bus bars in point to point builds.

Also make sure that interstage signal connections are as short as possible and input and output signals are carried in screened cables.

it is just a question of chiselling away at it.

Cheers

Ian

That's probably where I went wrong.  For the entire audio circuit, I basically have one ground for everything; so both cathode networks for both channels' first stages are connected to ground at nearly the same physical point.  And that same wire is handling ground for the output sides as well.  (I sort of did the buss bar thing, but with wire...)  The interconnect between stages is the gain pot, which is on the front panel, so of course the wires are going to be longer than the rest.  But they're not ridiculously long, and they are shielded.

I've got some 15A mains wire laying around.  I should try that.

But I'm seriously considering the whole 12dw7 idea.  Seems it may be a little more forgiving in a build where space will be at a premium.
 
CurtZHP said:
But I'm seriously considering the whole 12dw7 idea.  Seems it may be a little more forgiving in a build where space will be at a premium.

it would certainly make a compact layout easier and you know that any crosstalk will not be due to the tubes.

Cheers

ian
 
ruffrecords said:
it would certainly make a compact layout easier and you know that any crosstalk will not be due to the tubes.

Cheers

ian

It's got to be the tube at this point.  I've tried getting everything as far away from everything else as I can, and I even tried the fat copper wire buss bar idea you suggested.  Absolutely no change in the crosstalk.

I did notice that it seemed to get slightly louder as the frequency of the tone went up.  That would point to a capacitance related issue, right?
 
Did a little test this evening.  Found a spare 12AX7 and a tube socket.  Rigged it up so that one half of it replaced one half of the existing 12AX7 in the circuit.  So now each channel's input stage has its very own tube.  I figure, if the crosstalk goes away, then it was happening inside the tube.

Well, shoot that idea.  Still plenty of crosstalk.  It ain't the tube.  Must be my layout.  The good news it, the loose way I wired the spare tube allowed me to move it around a little and sort of find where things get bad.  It just so happened that I picked the channel whose 12AX7 stage is physically closer to the 12AU7.  Moving the wire for the plate supply/output closer and further away from its original position definitely changed the crosstalk level.  The area where it was running just happens to be near the cathode network for the second stage of the other channel.  Prime candidate for crosstalk.

The problem was, in order to get the crosstalk to go away almost completely, I had to practically move the spare tube to another ZIP code!  Must still be a wire or two somewhere in there that's not where it should be.

Anyway, I don't think swapping everything out for a pair of 12DW7's will necessarily solve this problem.  Maybe if I were to place them at opposite ends of the circuit board with their outputs facing away from each other?  While I was researching tube crosstalk, I came across one suggestion that involved switching the channels on the second stage around, so that whatever crosstalk happened would end up on the same channel.  ???  That sounds like a phase disaster waiting to happen.

Build two separate mono preamps in two separate cases?  :-\

 
Have you measured the level of cross-talk? If you feed a test tone into one input so its output is 0dBu, what is the level of this tone at the other output?  Try this at 1KHz and 10KHz. Just so we get an idea of what we are up against.

Cheers

Ian
 
ruffrecords said:
Have you measured the level of cross-talk? If you feed a test tone into one input so its output is 0dBu, what is the level of this tone at the other output?  Try this at 1KHz and 10KHz. Just so we get an idea of what we are up against.

Cheers

Ian

Haven't measured it.  All I know is that it's there as soon as I start turning up the gain on the other channel, which I'm listening to on headphones.  And the gain on the tone generator channel is all the way down.  That's why I figured it was being generated by the first stage (12AX7).

Let me see what I can come up with in terms of actual numbers.

For now, I'm blaming the wiring job.  I had the plate supplies/outputs of the first stages running right parallel with each other for a few inches, straight through the middle of the circuit.  And as I mentioned, the cathode (and ground) of at least one channel's second stage is right next to that.  Looked nice and neat on paper.  Not so much in actual practice.
 
The hard part is getting things laid out so nothing from "competing" stages are in parallel with each other, when the pins on the tube are that close together, and you don't have acres to work with.

I've got some spare time at work today, so I'm fooling around with a PCB layout program and the idea of using two 12DW7's.  Even then, it's hard keeping the output of one stage away from the input or ground of the other.  :eek:

(I know, I'm putting the cart before the horse.  But I had nothing better to do right now...)
 
Everything in the world crosstalks.

The BBC, on a network console which carries football on one path and bird-calls on another path, may want >60dB crosstalk. (Which may be hard to get at 15KHz; and may be inaudible.) When all inputs are basically going to the same tape, -40dB may be ample; and 15KHz xtalk may be smothered in cymbal glare.
 

Latest posts

Back
Top