The official G9 help thread

[redmojosteve]

Thanks MeToo2 for your patience!
I checked the diodes and they are all correct, D19 and D20 are isolated from the chassis too. The TL783 has a finned heatsink which is not bolted to the chassis. Therefore I put my spare transformer in (for my next G9) and I am getting a slightly lower current of 3.5A but the 15V transformer is still getting quite warm, although not as bad. Would very low ESR of C14 and C15 cause a problem by any chance?
Cheers
Steve

 

[MeToo2]

Thanks MeToo2 for your patience!
I checked the diodes and they are all correct, D19 and D20 are isolated from the chassis too. The TL783 has a finned heatsink which is not bolted to the chassis. Therefore I put my spare transformer in (for my next G9) and I am getting a slightly lower current of 3.5A but the 15V transformer is still getting quite warm, although not as bad. Would very low ESR of C14 and C15 cause a problem by any chance?
Cheers
Steve

Don't think so. How exactly are you measuring the 4.4A (12V transformer sample 1) or 3.5A (12V transformer sample 2) ? Is that 3.5A or 4.4A measured as RMS with an ammeter? I guess so, because you are getting substantial heating. It'd be normal to get a short peak current surge for a few seconds as you first switch on as the large reservoir caps charge up, and then a few mS per mains cycle as they top up, but not sustained current draw over a long period. Anyway your voltage drop across R33 seems to rule out anything to the right of this point, as the current draw of the HT of 20mA is pretty much nominal. That 3.5A or 4.4A has to be going somewhere. If it isn't going through R33 where is it going? 4.4A would translate to 160mA at 260V or >40W. All I can think of is into the chassis somehow. Seems unlikely two transformers would be faulty. Unless it's systematic and there's some sort of nasty eddy currents in the transformer cores. Do you have model numbers for the 12V transformers? Or maybe your diodes are breaking down at high voltage. Are they true 1N4007's? Have you taken them out and tested them individually?

 

[redmojosteve]

Hi again.
The current is measured as RMS using a multimeter. It's puzzling me and I completely agree that it has to be going to the chassis somehow. I checked the diodes before I put them in the board but it would seem a good idea to check them again. I presume they are true 1N4007's as that is how they are marked. Is there any way of telling the difference? I have used them on other builds without any problem but my "stock" are not all from the same source and I think this project may be more demanding. The transformer is a Nuvotem RS0050P1-2-012. I have used this make on other builds and they have been OK.
Cheers
Steve

 

[redmojosteve]

Just a quick addition to my last message. I removed the diodes D17 to D20 and even though they tested satisfactorily, I have replaced them. Although they were from the same pack, it turns out that at least 2 were 1N4004's so I thought this could have been a problem due to HV breakdown as suggested by MeToo2.  I know I should have checked! These have been replaced with 1N4007's, which hopefully are correctly marked, but frustratingly it hasn't made any difference.
I have also checked for any solder bridges, bits of stray wire etc just in case, but again there's nothing obvious.
Cheers. Steve

 

[Luax]

Hi Guys,

Recently I've completed my G9 build. The left channel was superb, but the right channel had some noises and a slight buzz.
I couldn't help not opening my G9 again and decided to place some alu to shield the circuit form the toroids.
I also decided to use the external PSU instead of the main board. One thing that I could get quite clear (after reading the schematic).
Is that on the external PSU there is bridge 1 and 2 but with what diodes do they correspond on the main board?
If somebody has any knowledge on the matter i'd love to hear it.

All the best!

 

[barthman.de]

I finished a G9 which I made "my own way" - some pics you can find here ... http://www.groupdiy.com/index.php?topic=48522.0

But I have some questions:

1) Hum 

I put a SM57 into the preamp, layed the sm57 under my pillow and set "Gain" and "Output" to maximum.

The noise floor:

Looking not bad but if I turn on my headphones I can hear clearly "eeeehhhhhhhhh". Any ideas to further reduce the 50 Hz hum? Or do I hear the fleas coughing?

2) Is there a difference between big caps or smaler caps with near the same value? I built in smaller 100 uF / 350V  (diameter 18mm, height 32 mm) at first and changed them later to 220uF / 350V (diameter 35mm, high 62 mm)? There was no difference in hum. Can I use the smaller ones solid without problems (the big ones need so much space inside the housing)?

 

[MeToo2]

I finished a G9 which I made "my own way" - some pics you can find here ... http://www.groupdiy.com/index.php?topic=48522.0

But I have some questions:

1) Hum 

Before you start chasing ghosts, I'd humbly suggest you check 4 things related to correct gain staging.

1) The G9 is a mic pre. I'd therefore suggest that you test it with a 200 ohm shielded resistive dummy load on the input (not a live mic, even if it has an output impedance of 200 ohms), and perhaps a 600 ohm shielded resistive dummy load on the output. Your A/D converters might only present a high impedance load (>10K)

2) What is the noise floor of your measuring set up (200 ohm shielded resistive dummy load connected directly into your converter via the same XLR test cables, without the G9 in place)? In other words, is the limit the G9, or the native noise in your test set up/ converters?

3) What is the calibration on your graph? Does -102dB correspond to -102dB relative to digital Full Scale Deflection (FSD) or relative to +4dB into 600 Ohm or what?

What signal level would 94 dBA SPL directed into your mic correspond to at the same gain setting at the output of your G9?

[an SM57 has a sensitivity or open circuit voltage of -56.0 dBV/Pa* (1.6 mV) *(1 Pa = 94 dBA SPL @ 1KHz) at the input to your G9. A G9 will generally produce around 56dB or so of gain when cranked up near max, but you should measure that. That'll give you around 0BV (1V) of signal. 94dBA SPL @ 1KHz is a pretty loud source but not excessively so]

4) How much signal voltage can your A/D converter take before it clips (0dB Digital FSD)?

So what signal can you aim to have at the input to your A/D whilst still allowing 20dB for dynamics and 6 dB of cushion to avoid digital artifacts? Is your G9 gain too high, so that the A/D is in danger of digital clipping? Do you need to turn the gain down?

Then you can work out your equivalent signal to noise ratio at the input and output under real use conditions, which is IMHO more important than the absolute level of the noise at the input or output, and something you can compare with other people's builds. Equivalent Input Noise (output noise + gain) should be around 120dB or even better, resulting in a real life s/n ratio of ±66dB at nominal signal level, whilst still allowing you to keep 20dB or more of headroom for dynamics/ peaks + 6dB cushion.

That'll tell you if you're listening to fleas coughing or not.

 

[barthman.de]

Before you start chasing ghosts, I'd humbly suggest you check 4 things related to correct gain staging.

1) The G9 is a mic pre. I'd therefore suggest that you test it with a 200 ohm shielded resistive dummy load on the input (not a live mic, even if it has an output impedance of 200 ohms), and perhaps a 600 ohm shielded resistive dummy load on the output. Your A/D converters might only present a high impedance load (>10K)

2) What is the noise floor of your measuring set up (200 ohm shielded resistive dummy load connected directly into your converter via the same XLR test cables, without the G9 in place)? In other words, is the limit the G9, or the native noise in your test set up/ converters?

3) What is the calibration on your graph? Does -102dB correspond to -102dB relative to digital Full Scale Deflection (FSD) or relative to +4dB into 600 Ohm or what?

What signal level would 94 dBA SPL directed into your mic correspond to at the same gain setting at the output of your G9?

[an SM57 has a sensitivity or open circuit voltage of -56.0 dBV/Pa* (1.6 mV) *(1 Pa = 94 dBA SPL @ 1KHz) at the input to your G9. A G9 will generally produce around 56dB or so of gain when cranked up near max, but you should measure that. That'll give you around 0BV (1V) of signal. 94dBA SPL @ 1KHz is a pretty loud source but not excessively so]

4) How much signal voltage can your A/D converter take before it clips (0dB Digital FSD)?

So what signal can you aim to have at the input to your A/D whilst still allowing 20dB for dynamics and 6 dB of cushion to avoid digital artifacts? Is your G9 gain too high, so that the A/D is in danger of digital clipping? Do you need to turn the gain down?

Then you can work out your equivalent signal to noise ratio at the input and output under real use conditions, which is IMHO more important than the absolute level of the noise at the input or output, and something you can compare with other people's builds. Equivalent Input Noise (output noise + gain) should be around 120dB or even better, resulting in a real life s/n ratio of ±66dB at nominal signal level, whilst still allowing you to keep 20dB or more of headroom for dynamics/ peaks + 6dB cushion.

That'll tell you if you're listening to fleas coughing or not.

I'll try to answer the questions but because I'm not a pro I have to tell more questions

1) OK. Can I put some 600 ohm resistors on input and output between hot and cold? Can I go after the resistor on output to my Metric Halo 2882 A/D (this is my only way to picture the frequency response via DAW)

2) The limit is the G9, here the response without the G9:

3) The calibration of the graph is dbFS. I zoomed in so you can't see the range above -18dBFS.

4) My Metric Halo 2882 takes +26 dBU. I drive the levels usually at -10 dBFS till -6 dBFS to reduce distortion and have enough headroom in the DAW. So if I understand you right 26dBU-20dBU-6dBU = 0 dBU = - 9 dBFS. So I can turn down the output gain on the G9 to get a peak level round about -9dBFS. So I have a lower noise floor on output.

Let's say after reducing the output gain to "8" I would get maximum noise peeks on output at 95 dBFS + 45 dB Gain = 140 dB = better than 120 - no problems.

Is this calculation right?   

 

[MeToo2]

I'll try to answer the questions but because I'm not a pro I have to tell more questions

1) OK. Can I put some 600 ohm resistors on input and output between hot and cold? Can I go after the resistor on output to my Metric Halo 2882 A/D (this is my only way to picture the frequency response via DAW)

yes, you can try that. Terminating 600 ohm busses is less common that it used to be. The Metric Halo 2882 apparently also has a soft switchable level input impedance, so I'm guessing you should set these to +4 level (for pro equipment).

2) The limit is the G9, here the response without the G9:

yep it's clearly higher, but that could just simply be due to the additional gain of the G9. If your G9 is providing 36dB of gain it ain't effectively any higher at all, just the whole reference is moved up (noise floor as well as signal). Which is why I suggest you look at signal to noise ration rather than absolute levels.

3) The calibration of the graph is dbFS. I zoomed in so you can't see the range above -18dBFS.

OK.

4) My Metric Halo 2882 takes +26 dBU. I drive the levels usually at -10 dBFS till -6 dBFS to reduce distortion and have enough headroom in the DAW. So if I understand you right 26dBU-20dBU-6dBU = 0 dBU = - 9 dBFS. So I can turn down the output gain on the G9 to get a peak level round about -9dBFS. So I have a lower noise floor on output.

Let's say after reducing the output gain to "8" I would get maximum noise peeks on output at 95 dBFS + 45 dB Gain = 140 dB = better than 120 - no problems.

Is this calculation right? 

I'm not convinced about that max +26dBU specification on the input. That must be with the 20dB pad engaged. 26dBU is 45 volts peak to peak, which no computer interface that I know can handle raw without clipping.  26dBu - 20dBu pad would give 4.5 volts at the analogue digital converter, which sounds much more normal. I believe placing an input on the Metric Halo 2882 in "+4dB professional" mode automatically engages the 20dB input pad (at least according to the manual).

I don't understand exactly how you get from 26dBU-20dBU-6dBU = 0 dBU [for RMS pink noise] to -9 dBFS [for peaks] in your calculation. I'd assume there'd be 20dB of dynamics on top of the -26dB RMS, which gets your peaks back to -6dBFS. Either way, this matches almost exactly with your current practice to "drive the [peak] levels usually at -10 dBFS till -6 dBFS" which sounds good.

But whatever. If when you do your real-world gain-staging you measure that your noise peaks are at -95dBFS and your signal peaks are at -9 to -10 dBFS without any clipping or distortion (on whatever nominal scale) you've got a signal to noise ratio of >> 70dB, which is pretty excellent IMHO. I found this article that gives quite good explanation of the relationships between levels through the system (although it does not address digital overs).

You could even possibly afford to allow yourself up to another 6dB of headroom [peaks at ± --16dBFS to -12dBFS] and you shouldn't hear the noise on a real recording. Bear in mind when you mix multiple tracks into an overall mix that uncorrelated noise adds up as power, whereas correlated signal adds up as voltage, so you effectively hear less background noise the more tracks you have in the mix (assuming you record one track at a time). Also I find digital clipping and overs are far more annoying than some low level hum. You can almost always use a high pass filter or an expander or gate in your DAW.

 

[barthman.de]

Thank you METOO2 this are the informations to getting better. Good article. I work more practice-oriented because recording music and building all around is hobby for me. So I have to learn the more physical details while building some equipment 

 

[kent]

Hello all.  It has been literally years since I've posted but time has finally allowed me to work on something new. Now that I'm stuck it's time to ask for some help. My G9 is put together but on first fire up my voltages are reading quite high (among other issues but I'll start there). I used 115+115:15+15 and 115+115:12+12 torroidial I got from parts express. The voltages straight off the transformers with a load are as follows - heater=21v, phantom=21v, HV=344v. Double and triple checked wiring and it seems fine. Any ideas?

 

[gyraf]

Voltages measured under load - tubes mounted, unit heated up (some 30 sec.)? In that case you got a wrong transformer for the first (15V).

Jakob E.

 

[kent]

Here is a pic. Everything seems to be right. 15v for the first Tx and 12v for the second. Should I be suspect of the factory specs?


G9 power transformers by kentpdog, on Flickr

Here is the wiring. Kind of hard to see but I figure it has to be correct or my readings would be even farther off.


G9 transformer wiring by kentpdog, on Flickr

 

[Maynard]

This topic never ends being interesting! Kent I know you have check wiring several times, anyway if I were you I would get the transformers out and test them separately just to make sure they'r not the problem.

And I got a new one, what could make R33 toasted as badly as to destroy it without affecting the rest of components? too much current drawn but why?  I am still trying to find out. Yesterday turn it on and leave it heting as I have done for the last months and it start smelling a little bit funny. The unit was on, evertything look good but no output at all.
Today I opened and find out R33 toasted but the rectifying diodes, TL783 and protection diode and zeners are ok 
I just took the toasted resistor out, replaced it, turn unit on, got a voltage drop of around 5V (just using one channel) and a steady plate voltage of 243V. Used the unit for a while without any problem. So I have been uneable to reproduce the problem and got no idea what caused it, some kind of short? bad state of the resistor? Oh yes, reservoir C14 positive lead conection was loosed (could this be part of the problem?)

If anyone can give me a hint I will appreciate. I Will let you know if I find the cause so might be helpful to someone.

Ragards

 

[kent]

So I finally got around to pulling the transformer and measuring it by itself without a load. I am reading 21v on each winding. Is that normal for an unloaded 15v output?

 

[MeToo2]

So I finally got around to pulling the transformer and measuring it by itself without a load. I am reading 21v on each winding. Is that normal for an unloaded 15v output?

Regulation depends on the spec of each particular transformer. Larger wattage transformers tend to regulate better.

Your transformers (Y236103 & Y236102 in your photo) should have a spec of 3% variation in voltage at normal input (115V parallel primaries as you're located in the US) and full load, and 13.7% regulation, which is pretty good.

So (no load - full load V)/ no load V = 13.7% or in other words a no load output voltage of around 17.4V for 115V parallel input.

21V looks too high to me. But check your mains voltage too....

 

[redmojosteve]

Hi again,
It's been a few days since I have been on the forum and I have checked my G9 but I am still getting a warm transformer (the 15V one). As I have a second PCB and all the components for anpther build I am going to get this one together and swap the boards over to see if it makes any difference. I'll obviously be super careful with component selection and mounting.
Cheers
Steve

 

[kent]

Ha!  Checked voltage out of the wall and got 155v!  Couldn't be right so I checked my meter and found the battery wire was frayed and shorting out on the board. I hate when you have to troubleshoot a problem to troubleshoot a problem!  In any case - back to testing and an update when I'm done.

 

[phelar]

Hi!
I just finished my g9 and it was up and running for like 5min.
After 2min the transformers, most likely the 2x15v, turned so warm that I could feel the heat on the casing. All tubes was glowing. There was also a 50hz hum from the case.
I tried a guitar through the instrument input, no sound, just something like a ground noise unaffected buy the gain. Turned it off, checked component placement, wiring....everything seemed ok. When I turned it on again it was all dead.
I checked the 2x15v transformer (Velleman 230v 30va 2x15v with 130 degrees C thermal fuse), measured resistance on the primary winding and there was nothing. So I thought the fuse was blown. The second transformer (2x12v) measured ok.

Ok, so. What to do now? First of all, is this a crappy choice of transformer..? (It`s kind of cheap I know. I don´t think this problem IS the transformer but anyways).
I have to replace the transformer and start measuring, I know that,  but I´m fishing for some theories for this transformer breakdown.

Does anyone got a clue?

 

[MeToo2]

Ok, so. What to do now? First of all, is this a crappy choice of transformer..? (It`s kind of cheap I know. I don´t think this problem IS the transformer but anyways).
I have to replace the transformer and start measuring, I know that,  but I´m fishing for some theories for this transformer breakdown.

Does anyone got a clue?

I would not just replace the transformer and start measuring. A transformer gets hot because of too much current flowing. The root cause of "too much current flowing" is generally something pretty simple, like a short circuit, wrong spec transformer, induction loop, mis-oriented diodes. The transformer itself is very unlikely to be the root cause. So I would start measuring, and only once I'd found all the likely fault(s) I'd replace the transformer. You already know there's something wrong...... You can then power on your circuit either using a variac (to limit input voltage to the transformer) or use the light bulb test to limit the current, but first without the 3 loads connected (heater, HT, P48). Then you can connect the loads one by one. My experience is the slower you proceed, the faster you'll solve your problem.