My take on a G9 with trafoless input. What's the Phantom?

Help Support GroupDIY:

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
Greetings folks.
I decided to have a swing at the G9 project since there where so many information about it around, especially from the gyraf guide.
I went a little over this and etched my own boards based on my own designs and I am currently trying to implement an electronically balanced input instead of the transformer (which I don't own, hence the lack of comparing of the two).

Let me take you through this, so I can get to the point. The point being that this thing is not silent enough (I get 50Hz and 100Hz hum with their harmonics).

First off, You will notice a 3rd toroid which is one providing the balanced circuit voltage @30V approx. It's my attempt (that still lives in my troubled future) to implement a 1176 style compressor into this preamp, but please let's leave this out of THIS discussion. So yeah, the 30V PSU is taken from the 1176 Rev.G schematic.

About the grounding scheme, I wanted to have an as clean as possible wiring, so I decided to split the ground into two busses, being one for the PSU and one for the audio. So the audio ground is the big copper wire connecting the in-out xlrs and all preamp-gain-volume-output.trafo grounds are connected on it. The PSU ground is connected again with a solid copper wire to the IEC socket and all mains-psu-heater.common-smoothing.caps grounds are connected on it. The chassis is grounded on the input xlr but I have tried doing so with a screw next to the IEC socket instead, which made no difference.

All my signal cables are single-side shielded on a near ground connection and I tried to twist all my power wires with a ground wire to minimize inteference.
Also, all regulators are isolated with silicon pads from the chassis on which they are mounted for heat dissipation. This decision was made to save some space and it seems to be working but it makes the chassis way too hot on the spot so I am considering of getting a big heat sink. This move also lives in my troubled future unless I understand that that's where my problems lie.

Now to the noise issues. From a poor spectrum analysis (output xlr to my presonus interface with interface gain at minimum) from Studio One I see a high noise floor, a 50Hz spike and it's harmonics.
This is with AND without the unbalancing circuit in the front. I found out that if I ground the input (without the unbalancing circuit) the hum disappears.

Volume is at 0% Volume is at 50%


It took me sometime struggling with the TL783 (HT ripple was about 30mVpp) but I replaced it and the filter caps (I believe it was the 783 though) but it didn't make any difference noisewise (ripple is about 6-10mVpp now). So I figure it must be something else... I am struggling with this because I fear it must be my wiring that is off, or the grounding scheme, so I would love some feedback on this. The two ground busses idea came to me after I looked at many many designs of other tube amp enthusiasts and it seemed like a real solid solution. Also it makes sense to my inexperienced hollow skull.


I also fear that this is an issue that a transformer would potentially solve. Thing is, I am trying to save the money for a trafo and furthemore I am too interested in making this unbalancing circuit to work (which is not my design either, of course). The aforementioned circuit is taken from Moxtone's take on the 1176 Rev.G and is more or less the same as the original 1176. I took the liberty to step up the gain to much the 1:5 transformer's gain (and try to boost my signal over the noise).

So, any thoughts?

This whole process is for learning purposes, experience, suffering and desperation that may lead to the loathing of one's self.
Wish you all the best.



And here are some pictures of my unit at this moment. Unbalanced circuit is on the breadboard.
 

beatnik

Well-known member
Joined
Oct 18, 2009
Messages
920
Location
Dorset, UK
if there is a ground problem maybe try the aes recommended grounding scheme.
pin 1 of input and output xlr grounds directly at the chassis, you can use the ground tab on the connector making sure it has a good bond to chassis.
ground the safety earth of the iec connector on a dedicated bolt to the chassis.
for the remaining you can use the bus bar to connect all the audio 0v signals and return this on a 0v point on the power supply and then run a thick wire from the filter capacitors negative to the same bolt where the safety earth is.
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
if there is a ground problem maybe try the aes recommended grounding scheme.
pin 1 of input and output xlr grounds directly at the chassis, you can use the ground tab on the connector making sure it has a good bond to chassis.
ground the safety earth of the iec connector on a dedicated bolt to the chassis.
for the remaining you can use the bus bar to connect all the audio 0v signals and return this on a 0v point on the power supply and then run a thick wire from the filter capacitors negative to the same bolt where the safety earth is.
Allright, I tried this configuration and I didn't notice any noteable difference. The 100Hz and it's Harmonics is still there and present. I got a feeling that my TL783 is giving up because I see the High Tension regulating at 240 instead of 245 at this point. This is the main reason I might go for a new heat sink, maybe the chassis mount solution isn't enough and the regulator fails after some time.

Now, anyone has any tips on how to troubleshoot where my Hum is coming from? I really have a gut feeling that it is not due to ground connections, though I am not so very well acquainted with the results of a good/bad ground. I just know what I encounter.

Oh man, it's been a hell of a headache this thing.
 

shabtek

Well-known member
Joined
Jun 8, 2005
Messages
2,083
Location
midwest usa
good luck,
you may be experiencing a bit of 'false economy' especially if your time is of value, but this is a learning opportunity

the g9 with gustav boards is tricky enough to get right--one of the reasons there is so many pages of help threads

I did not see your electronic balancing technique --but get the unbalanced DI working before worrying about that.

it might be useful to build an unregulated simple power supply with only filament and B+ that can be connected at some distance, to rule out the etched power supply and possible magnetic radiation/interaction from torroids/enclosure

again I do not know how you are balancing but you are forfeiting the galvanic isolation afforded by a transformer which offers safety for users and connected equipmant.
 

beatnik

Well-known member
Joined
Oct 18, 2009
Messages
920
Location
Dorset, UK
+1 on unregulated ht and try moving the psu externally

are you using ac or dc supply for filaments ? try keep it simple with ac filaments and see how that goes. if dc filaments are not properly done they just make things worse.
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
For the shake of solving my hum problem, I have removed the balancing circuitry and grounded the preamp's input. By grounding the input a get a minimum noise floor and a little lower hum in decibels, though it still is profoundly present.

Now, my HT and filament PSUs are the original G9 ones. Filaments are 12V DC powered. So is the phantom power (that I still have not in use). I do not use any switches (line-phase etc.) nor the instrument jack input untill I solve the hum issue.

@shabtek I may be following your advice and build an outside PSU for HT and filaments, though if unregulated AND on a breadboard I am not optimistic on it being silent.
 
Last edited:

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
Status update:

My new theory blames the heaters supply for the hum noise and here is what lead me to this conclusion.

For this experiment the heater voltage emenates directly from my bench power supply ([email protected]). Supply is a Twintex TP-4303.
The HT+ is provided by the original power supply schematic that I conjured on breadboard in order to isolate it from the box (the toroids are out of the box as well) and get rid of the phantom/heaters section.


A. Bench PSU Heaters, HT  OFF.PNG. B.Bench PSU Heaters, HT on Bread board @ Vol.0%.PNG
The above pictures depict the output with (A) only heater voltage being present (HT+ is OFF) and (B) HT and heater supply are ON while volume is at 0%. With volume engaged at 50% the noise floor level sets at about -90dB which looks okay but sounds excessive.

Here you can listen to what it sounds like on 50% volume.
https://soundcloud.com/the-indecisive-few%2Fnoisefloor-bench-heaters%2Fs-pgGIaxKXV3V

My 12V heater supply (prior to this experiment!) is the original G9 power supply (as is the HT+ of course) so I guess it should work.
Now what could cause this supply to not work properly on my printed circuit board? Could the PCB design be "too lame"?

Before I assembled the unit inside it's box, I did some testing with the PSUs on some protoboards but each section was seperate (HT seperate than heaters and phantom seperate on it's own) so the design was really tight and they didn't interfere with each other in the way they would on a PCB all together. In that form, I still had the same high noise floor AND 100Hz hum that I have now, though I am not sure if it has increased or decreased in power. That's when I decided to build specified PCBs and put it inside a box with clean wiring, thinking it would solve the issue.

Does anyone have any thoughts on this? If you compare the image A to my first post's images you will notice the same kind of hum only extremely more intense. Seems like heater supply issue..
Any thoughts?

Thank all of you and Panadol's Extra for helping me clear the headache.
 

Attachments

  • 1b.jpg
    1b.jpg
    2 MB · Views: 17
  • 1a.jpg
    1a.jpg
    1.9 MB · Views: 17
Last edited:

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,964
Location
Norfolk - UK
If you hum is predominantly 100Hz then there is a good chance it comes from the HT. You say you have 10mV of hum after the TL782 which is far too much. The TL782 does need about 25V or so across it for it to be able to regulate properly so check the dc voltage and ripple going into the TL782 and the dc voltage coming out.

P.S I trust you do have one side of the heaters connected to analogue 0V?

Cheers

Ian
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
P.S I trust you do have one side of the heaters connected to analogue 0V?
You mean the common between the heaters (it's pin 9 for ECC82) right? Yes it is connected to ground.

@ruffrecords One more thing. When measuring the ripple should I be doing it under load or not? The aforementioned measurements where taken with tubes as load, but when I have the PSU disconnected? I use a 22kohm 5w resistor (it's the highest watt resistor I got ) or a [email protected] to keep current down.
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
Status Update #2:

This experiment was done with bench supply providing the Heaters voltage (as previously) and my original printed board that used to cause all the issues (instead of a breadboard as before) for the High Tension.
I went one step ahead and grounded the in and outs of the other supply sections (the +30/-10, +48 phantom and +12 Heater voltages) so that I get minimum interference from the components being idle.


A. Twintex Heaters, Printed board HT+, Vol0%.PNG B. Twintex Heaters, Printed board HT+, Vol50%.PNG
Picture A depicts the noise floor with Heaters and HT+ engaged at 0% volume while picture B in the same conditions while volume is at 50%.
I see no hum whatsoever, and mind you that since the voltages are supplied by outboard means , the wiring and especially grounding is not at all optimal, though I managed to keep audio grounds away from supply grounds.
I used many crocodile clips and flying wires.

Next up I'm wiring the same situation, only instead of bench for the heaters it's going to be the pcb unit.
In my opinion this experiment dictates that the heater supply is to be blamed, though I am not sure if there's some blame to be put on the transformer that generates the heater voltage.
Will be back. Thanks
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
I have resolved the hum issue in my opinion. I have eliminated the 100Hz hum by upgrading the HT smoothing caps to 330uF. It was actually THAT simple of a deal. Still have some 50Hz hum with harmonics. but I believe it is not tied to the preamp but to the mains (meaning my workshop wiring). May delve in that later.

Now that it's all working properly, I have an other issue to address with you guys.

As you see in the headline of this thread, I am working on a trafoless INPUT design of the G9 and below you can see the balanced input.

The Moxtone Input.PNG
It is a pretty standard design that I "stole" from Moxtone's 1176 compressor. I changed the R7 resistor to 22k to obtain a 1:2 gain increase. I considering maybe a 47k to take my signal a little more over the noise floor and have a better use of the lower values on the Gain pot, but I find my soundcard clipping pretty soon already so the 22k stays for now.

Now what concerns me is that I need to power a condenser mic through a phantom power. I decided to get rid of the switch relay and use a 2pdt toggle switch instead.

My first attempt is to just connect the phantom to the input with two 6k8 resistors:

Phantom #1.PNG
This way I find my output to drop a lot in volume and I believe that I need to seperate the phantom voltage from the NE5532.
*There is ~25V at the C1-R2 and C2-R3 connections without the phantom because of the opamp's supply.

An other way I am trying is this (which seems to work better but looks rather cheap):
Phantom #2.PNG
Does anyone have any thoughts on this?
First solution looks wrong, but the first solution looks cheap.
Maybe if the first solution is okay I can increase to 47k the R7 and make up for the lost gain?
I am concerned about the capacitors reverse voltage beeing too high though, is that a real issue? Seems like one.

Thank you all!
 

musipol

New Member
Joined
Jun 28, 2021
Messages
23
Location
Idaho
A small comment a month late. You are powering the 5532 from a single supply(Vcc) so you have to "bias" it at 1/2 Vcc to get a symmetrical swing + and -. How can you get that when you have isolated your R1,R6 divider from ground with that 220pf cap(C5)? I don't know, maybe I'm missing something someone else already said???
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
A small comment a month late. You are powering the 5532 from a single supply(Vcc) so you have to "bias" it at 1/2 Vcc to get a symmetrical swing + and -. How can you get that when you have isolated your R1,R6 divider from ground with that 220pf cap(C5)? I don't know, maybe I'm missing something someone else already said???
Truth be told, I am not aware of what's the C5's point in the circuit. It is taken from moxtone's take on the 1176 but modified for a little more gain. Would love if someone could explain.

The power is only positive though, there is no negative power on the 5532

I actually did the job by using non polarized electrolytic capacitors on both inputs.
 

MisterCMRR

Well-known member
Joined
Jul 1, 2019
Messages
141
Location
Ventura, CA
The circuit couldn't possible work with C5 connected as shown. R1 and R6 are a voltage divider to create at +15V "pseudo-ground" reference for the op-amps. Are you really intending to use this as a mic preamp? If not, why would you want phantom power. If used as a mic preamp, resistors R2 through R5 will make noise performance terrible!
 

moamps

Well-known member
Joined
Sep 24, 2005
Messages
1,914
Location
Croatia, HR
Moxtone is my site and here is the schematic mentioned here:

1633544471162.png

As can be seen, it is just a redrawn schematic of line input stage of later versions of the 1176 compressor, and there is no capacitor C5 220pF in it.
 

gyraf

Well-known member
Staff member
Moderator
Joined
Jun 4, 2004
Messages
10,385
Location
Aarhus, Denmark
..input impedance of 20K Ohms is far from optimal for mic input use.. Better use a dedicated SSM or THAT chip here..

/Jakob E.
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
Moxtone is my site and here is the schematic mentioned here:

View attachment 85321

As can be seen, it is just a redrawn schematic of line input stage of later versions of the 1176 compressor, and there is no capacitor C5 220pF in it.
Allright mr. Moamps, you are definitely right, the C5 in my schematics should depict the C35X 100nF in your's, though I did misplace and misslabel it. However, I didn't notice untill now.

Thanks for takin the time to correct me.

I used this schematic for creating the unbalanced signal to the preamp from the mic's balanced signal since that's what this does.
I chose it since it is perfectly validated since it is used in the original 1176 and in Moamps' version.
Now is it the perfect option for a mic pre? I don't know. I thought, if it's good enough for the 1176, it should be good enough for my pre. The 1176 wasn't designed to have a mic as an input? Not sure about that either.
 

moamps

Well-known member
Joined
Sep 24, 2005
Messages
1,914
Location
Croatia, HR
The 1176 has line input so it is not optimized for microphone connection. You need to find another schematic, although I don't really see the point of placing such a circuit in front of a tube preamp. This G9 design is good as it is.
 

Philip_BlueFX

Active member
Joined
Feb 27, 2020
Messages
27
Location
Ελλάδα
The 1176 has line input so it is not optimized for microphone connection. You need to find another schematic, although I don't really see the point of placing such a circuit in front of a tube preamp. This G9 design is good as it is.
I am trying to get rid of the transformer, it's my only reason :)
 
Top