DIY inline preamp (triton fethead)

[f0m3]

Hi,
in this thread about active ribbons the user whoops opened a triton audio fethead and posted the schematic.
Since i had some 2SK170BL laying around, i matched some resistors and put them on a stripboard. Please be aware of the bend transistor pins!
!!! This is the corrected version after Reply #1 !!!

The preamp is working but it has more noise than i expected. It might be a problem of my Audio Interface Phantom Power. I made a comparison with two MD 421, one connected by the triton and one directly to the interface. Phantom power is applied to both channels. The noise on the channel without the triton raises when p48 is activated so i guess its part of the noise problem. How could i easily improve the schematic with phantom power filtering? Or is it only possible on the injection side?

 

[Bo Deadly]

Why orient the FETs in opposite ways? I would think they should both be oriented so that the source pin is connected to 68R. You have Q2 that way but Q1 you have the source pin connected to OUT pin2. Usually FETs are symmetrical but I would correct it anyway.

Also, you need to explain how you're measuring noise any why precisely you think it's a problem. There is a noise floor. If you amplify it enough, you will hear it.

Try this: Attach a microphone and place it 1 meter away from an audio source (preferably a studio monitor playing something with high and low frequencies). Adjust the level on your audio interface so that the signal is loud and clear but not clipping at all. Then, and this is the critical test, turn off the music source and place the mic between some couch cushions. Then listen to the noise level. Is noise still a problem?

 

[Winston OBoogie]

Why orient the FETs in opposite ways? I would think they should both be oriented so that the source pin is connected to 68R. You have Q2 that way but Q1 you have the source pin connected to OUT pin2. Usually FETs are symmetrical but I would correct it anyway.

Not sure if I'm guilty, but I'll take the blame for that bit nevertheless. 
In the other thread I said that the K170 does indeed work either way.  However, as good design practice, both should be the same orientation to avoid confusion and, if there's no need for otherwise,  just use drain as drain etc.
Sorry if I wasn't clear enough.

 

[f0m3]

Well, that was my mistake. i turned one Transistor on my stripeboard again and the noise is way better now. Still not very good but okay Thanks for the help. I correct the layout!

 

[Winston OBoogie]

I'm assuming the P48 before the 6K81 supply resistors is 48V,  what's the voltage at the other end on the drains?

 

[pucho812]

have you considered replacing the 2 x 2sk170 with an lsk389?

 

[f0m3]

I'm assuming the P48 before the 6K81 supply resistors is 48V,  what's the voltage at the other end on the drains?

You really mean at the drains or at the fetheads output?

have you considered replacing the 2 x 2sk170 with an lsk389?

Have not yet but that might be an idea.

It seems like triton audio changed the internal design of the fethead some years ago. Now they claim to have 4 jfets inside but they didnt tell that some years ago. Also the external design of the fethead changed so maybe they improved the fethead by using 2 x 2 parallel sk170 or as mentioned 2 lsk389 in parallel to reduce noise as mentioned by

[..]
Doubling up on another pair of the same type J-Fet in parallel would lower noise by up to 3dB.
[..]

in the active ribbon mic thread.

 

[f0m3]

It looks like as if cloudlifter uses lsk389 for that purpose...
I hope linking to other forums is okay? https://forum.cockos.com/showpost.php?p=2097941&postcount=16

 

[rogs]

It seems like triton audio changed the internal design of the fethead some years ago. Now they claim to have 4 jfets inside but they didnt tell that some years ago......

Maybe more like the Alctron MA-1 ? .....

 

[Winston OBoogie]

I wrote:
"I'm assuming the P48 before the 6K81 supply resistors is 48V,  what's the voltage at the other end on the drains?"

You really mean at the drains or at the fetheads output?

The drains and output are one and the same, or should be. 
I was interested in whether the current could be raised in the J-Fets by lowering the value of the 68R source resistors.  This would lower noise a little and also bring the drain voltage down which may be advantageous. 
Gain would normally also be increased but I'm not sure it would register here as the shunting of the following amp across the output will be limiting the gain. 

 

[Winston OBoogie]

The schematic Rogs posted shows the 2nd pair of J-Fets connected as common gate to the first pair,  in other words, a cascode connection. 
Cascode can be useful here (as I also wrote in the other thread)  but I'm not sure that another pair of the same J-Fet  type are the best choice  as you generally want a bit more voltage across  the bottom Fets than will be provided with a K170.

Back when Toshibas were easily available without selling your kids, the usual choice for the cascode transistor was the 2SK246.  But you don't necessarily need anything so special here, just something which self biases with more gate to source voltage.  I think the old J110 would work and there'd be no noise penalty for using it if the lower parts were still  K170's or a single K389.

Edit*** P.S. The reason I was asking about the drain voltages was, somewhat, related to  why I also suggested cascode *may* have been used in the updated 4 X J-Fet version in the other thread.  Namely that the K170 can become sensitive to gate leakage noise if drain voltage is above 15V.

1/Doubling up on another pair of the same type J-Fet in parallel would lower noise by up to 3dB.

2/Another possibility would be a pair of suitable J-Fets to cascode the K170's.  Not so much for the decrease in capacitance and extended top end, nor for any elimination of distortion at high frequencies, because the source z is low here. 
But because the K170 is best limited to 15V or lower on its drain to prevent against any possible leakage on its gate. 

 

[f0m3]

I cleaned the stripboard with alcohol and the noise was much less than before.
I think for me its okay now.
One other thing is that the double-XLR housing is made from metal. I didn't connect it to gnd/shield on any side. Unfortunately they are "microphoning" quite loudly. Any tricks to prevent this ? (maybe fill with paraffin wax
as they do it with guitar pickups?)

 

[kato]

The schematic Rogs posted shows the 2nd pair of J-Fets connected as common gate to the first pair,  in other words, a cascode connection. 
Cascode can be useful here (as I also wrote in the other thread)  but I'm not sure that another pair of the same J-Fet  type are the best choice  as you generally want

a bit more voltage across  the bottom Fets

than will be provided with a K170.

...

Hi Winston, I'm interested in experimenting with this and PRR's circuit from the other thread. (Active ribbon thread.)

Which pair do you mean by "the bottom Fets?" <-- the ones that feed the 2SK170's, or the pair that drains into the output jack? Thanks for clarifying.

That jumper near the front of the circuit is a bit of a mystery as well. I'm wondering what they hoped to accomplish with it. Hmmm.

 

[psujok]

Hello !
I am building this cirquit (double 2xsk170 source follower), everything is clear to me but, I was wondering about verion which passes Phantom power, like Triton fethed Phantom. I have no idea how to aproach this problem without a tranformer. Maybe you guys have some ideas ?