The official G9 help thread

[zebra50]

Very slight hum when gain fully cranked. I am using a Perusha case and Zebra50 mentioned something about this so will try what he suggests.

Yes, the paint is thick and insulating so you need to make sure the case connects electrically all round to complete your faraday cage. Have a look at my post a page or two earlier.

As a general point, I often troubleshoot hum issues in situ by using crocodile clip leads from one ground point to another. If the connection helps then I solder a thick wire from here to there. (But look out for those high voltage areas like the HT capacitors!) Use of too-skinny wire for grounding can be an issue.

DI on channel 1 doesn't pass any audio - channel 2 is fine. Need to look at this but otherwise I'm very happy.

Should be an obvious thing like crossed wires, bad solder joint or (less likely) a damaged socket.

 

[MeToo2]

DI on channel 1 doesn't pass any audio - channel 2 is fine. Need to look at this but otherwise I'm very happy.

Should indeed be very obvious if there is mic audio passing, as suggested: something like a dry joint, or header connections the wrong way round (switched versus straight through on the jack socket)

Thanks for supplying and sharing so much information and effort on this project. Jakob for the design. Gustav for the PCBs. Tat for the cases.

Here's some thoughts and hopefully constructive lessons learned after getting my first one running:

1) I had a problem with one HT supply (short near valve socket on channel 1). Fixed this, although it cost a 7812 due to a current surge. I started with building and testing the 3 power supplies all at once before starting on the audio. But with hindsight I would suggest starting with just the HT supply on its own, with nothing else on the board,  then the 12V heater supply, and finally the 48V, before moving onto the audio sections.

2) I made a dumb mistake with the orientation of one electrolytic capacitor on one 48V supply. Not sure how, as I was very careful and did the other board correctly. Mea Culpa. No damage and easy to find.

3) I had a problem with one of the switch boards. There was a short circuit solder trace hidden under the PCB header which lead to 10KHz oscillation at all gain levels. Was relatively easy to find where the problem was by removing valves to see which was oscillating, swapping out the switch board with the other channel to check if the problem moved or stayed with the channel (and it did swap channels) and then buzzing out the faulty switch board. Next time I'll check the gain boards before installing as it's relatively easy to check the correct resistances are being selected when rotating the switches. Once the short was located (it was well hidden) and then removed, everything worked fine.

4) If I was doing it again I'd probably mount the toroids on the side panel rather than the floor of the case (to stop the large bolt heads protruding below the chassis)

5) Good (star) grounding + screened cables on i/o connections seems to have worked. At least compared to domestic background levels, amp noise and hum seemed very low. Still got to measure the response properly.

6) Not entirely convinced about the layout of the switch board and PCB headers related to valve based projects compared to op amp based ones. Looking at it again it would seem better to run the DI completely separately via its own header, with the possibility of adding a strap so that you did not have to have any DI at all. The 47K gain pot is not PCB mounted which means more soldering. If you count the remaining wires I think you could fit them all into a 12 way ribbon cable (ground is duplicated), although again domestic IDT headers are also a pain. I don't really have a proper solution but I think this is worth some further thought. The "wire to board" puzzle seems to be a major part of this project. Especially around the 7812, things are very crowded if you want to fit in a decent sized heat sink. And the HT connector is physically too close to the 15V ac connector. Next project I'll do will be a point to point wired 660, so I can then compare how much effort the switch-PCB saved.

7) I'm feeling pretty good about myself for having built this, but I'm not convinced it was financially viable as a single project. A lot of DIY'ers have all sorts of additional bits and pieces lying around "for free." If you have to buy all of this stuff and tools just for one project then it can add up very quickly (especially in postage). If you also take into account there's likely to be no second hand resale value then you'll be very pushed to beat an off the shelf unit (if you can get an equivalent of course).

Anyway I think I've got the bug and I'll be building more projects.

The motivation is more for fun, and to be able make things that are simply not for sale, rather than for financial reasons.

Photos to follow.

 

[g9builder]

I agree about the cost. This was my first project and things did stack up as I was starting from scratch, but at the end of the day I am very pleased (and quite proud) of the finished product. The best preamp I had before building this was a SPL Goldmike and this thing blows it away. I like it so much I've aleady started ordering parts to build another!!! Should be cheaper thid time around too....!!

Thanks to everyone involved in this project.... Awesome

 

[gyraf]

Looking at it again it would seem better to run the DI completely separately via its own header,

Agreed - I'll be implementing this on a future update of the PCB..

Jakob E.

 

[beatnik]

is this transformer ok for the g9?

http://www.musikding.de/product_info.php/info/p1400_Toroidal-transformer-230V----270V--14V.html

 

[gyraf]

is this transformer ok for the g9?

http://www.musikding.de/product_info.php/info/p1400_Toroidal-transformer-230V----270V--14V.html

There is no simple way to tell. If you want to be sure, use the parts specified for the project.

Or try the above one out and let us know if it works.

edit: you'll probably get into trouble getting phantom power with that one.

Jakob E.

 

[beatnik]

can you explain? because it's 14 instead of 15V?

thanks

 

[MeToo2]

can you explain? because it's 14 instead of 15V?

thanks

It's only got a single 14V secondary.

The original design uses twin 15V secondaries. One for heaters and the other for 48V phantom +HT.
The heater supply takes a fair current, which may load the voltage significantly.

Also what is the 270V? peak or RMS? load or no load?

Your alternative might work, or it might not.



Here's photos of my build as promised. I didn't put them inline as it uses up forum space and bandwidth.

http://i958.photobucket.com/albums/ae69/MeToo2_Prodigy/DSC00514.jpg
http://i958.photobucket.com/albums/ae69/MeToo2_Prodigy/DSC00528.jpg

I think I'll move the power transformers to the side wall and back wall. It seems to make an improvement of about 6dB on noise and mains hum even though it looks crowded.

Experiment with ribbon cable to the sub board was a partial success. The IDT headers are a pain. Next time I might just solder the ribbon directly to the board.
http://i958.photobucket.com/albums/ae69/MeToo2_Prodigy/DSC00524.jpg

I added 1meg bleeder resistors on the HT capacitors for my own peace of mind.
http://i958.photobucket.com/albums/ae69/MeToo2_Prodigy/DSC00535.jpg

I also checked the absolute phase using a 'scope. Following the wiring of the schematic and Gustav's PCB, it came out non-inverting, as you'd expect.

 

[beatnik]

thanks!

i think i will use two transformers as specified in the project, there are rules for keeping the noise as low as possible?

 

[sirzodiac]

Hi all,

am I crazy or did I really spot a difference between the schematic and the PCB?
Have a close look around R11, according to Jakob's schematic:
Schematic:
- One side of R11 is connected to the rotary gain switch and to R13 (680R)
- The other side is connected to R12 (1M) and C7
PCB:
- One side of R11 is connected to the rotary gainn switch, R12 (1M) and R13 (680R)
- The other side is connected to C7

Does anyone agree? What could be the difference in behaviour of the circuit? What would be the right spot for that R12 shunt resistor, before or after R11? Or is there no importance at all?

Thanks in advance!

 

[MeToo2]

I've just completed my second G9 build (with Lundahl transformers). Worked first time  ;D

On the tests it seems to perform slightly better than my almost identical OEP based built (only component difference is the transformers)
Most noticeable on measured signal to noise ratio. e.g. 50Hz rejection is like 3dB or so better. -72dB versus -75dB relative to a non clipping 1KHz signal.

Question for you. How do you properly mount the OEP shielding cans?

Do you just slip them on physically (which is what I did)?

I've noticed that the upper portion of the shielding cans on only 1 out of 4 OEP transformers is well connected to ground.
The other three are fitted, but look to be floating wrt to chassis.

How do you ensure that they make good contact with the shield pin on the transformer and the lower portion of the can?

Is it a question of nipping the can physically to make a good contact?
Or do you have to solder the lower plate in place to the shielding pin on the transformer before mounting on the board?
And also solder the upper portion of the can to the lower portion to seal it?

[bah probably means quite an effort to to unmount everything to redo this]

 

[g9builder]

The later option is what I did.. solder the shilding pin to the lower plate and the upper secction to the lower plate...

 

[MeToo2]

The later option is what I did.. solder the shilding pin to the lower plate and the upper secction to the lower plate...

Thanks that's exactly what I've done now. Makes around 3-6dB of difference for 50Hz rejection (not surprisingly perhaps) and they now pretty much match the Lundahls within the limits of my rather poor test gear (yeah it's a sound card and it isn't fully impedance matched etc. etc.) Lundahls seem to be less load impedance sensitive. Every little helps getting that s/n figure up though.

I also got a reply from OEP too. Good customer service They suggested exactly the same. Solder the shield pin to the base plate at the shield pin. Then also solder the base plate to the top can at two places (next to the shield pin and in the middle at the diametrically opposite edge) See picture below.

Update: I've spent a couple of days now playing to see how to improve on some minor hum I had. I chased a lot of ghosts in this time, but I think it was well worthwhile.

Here's my tips on getting the S/N ratio up to well over 70dB and even possibly 80dB+ with a following wind (at least according to my crappy measurement kit anyway and of course ignoring those tube induced harmonics that we want.....).

Hope these help someone else.

1. Use screened cables for the sensitive wiring internally e.g. input and output to/from XLR to the PCB headers.
Also on the round trip to DI jack. (I had no oscillation issues without having to re-route significantly).

Make sure your grounding has no loops and goes to a common star point at pin 1 of the input XLR and it is well connected to the case (via a dedicated nut & bolt purely for grounding).

Ground the output XLR pin 1 too, and also connect the shield on the cable to the header K3/K103 (even though it isn't connected to anything at the other end on the board header).

2. The step up transformer was more of a mains noise source than the step down transformer.
Swapping their mounting position over helped a lot for very little effort. Rotating them also helped but not as much.
Keep power transformers are far away as possible from any other circuits. Physical separation is a lot cheaper than screening.

3. Scrape paint off the case around screw holes and used locking washers to ensure Faraday cage effect (thanks to Zebra50).

4. Twisted the wires to the output gain pot for the 2nd stage. This is also a relatively high impedance circuit on long wires.

5. Solder screening cans on OEP transformer properly and make sure they are well grounded.
(see above picture on how OEP recommends doing this)

6. Wait for the lift to stop before performing any measurements  :
Really, the heavy duty motor on the lift in my block produced more than 10dB of noise on my local mains.

Beware of refrigerators and other domestic appliances too.

7. Use more gain in the first stage if possible and then set the 2nd stage to around 70% of max

8. Use channel 1 rather than channel 2 for lower level signals.

I still have a difference between channels. Channel 1 is quieter than channel 2, but not by much.
I think that is simply due to physical proximity to the power supply and longer wires to the PCB and output connectors.

9. My Lundahl based build is clearly quieter and less sensitive to load than my OEP build even after all of this.
Whether they "sound better" is a different issue. Most people on the board seem to prefer the OEP sound. I dunno.

10. If you are running the OEPs into a modern (high impedance input) A/D converter, try terminating the output of the preamp (with around 600 ohm to 1K) to reduce hum.
Also when measuring make sure the source has a reasonably high impedance (of around 1-2K ohms) to avoid transformer ringing etc.
I got much cleaner test signals from a mic than a computer sound card.

Proper impedance matching can make another 5-6dB difference specifically for 50 & 100 Hz. Lundahls didn't seem to need this matching as much.

11. Don't use a cheap plastic 6.3mm - 3.5mm stereo jack adapter at the line input on the back of an Apple iMac. Use a solid metal one.
That can lead to a lot less frustration searching for mains noise that is nothing to do with your build  :-[

12. Turn down the sensitivity of your computer input card and use more gain on your G9.  I noticed that the noise spectrum was mainly 100Hz and its harmonics and was largely independent of gain, so I guess most seems to be leaking in either via the power supply or via the output and inter-connect cabling. Turning up the gain on the computer just increases the mains hum, whilst lowering your headroom and encouraging input clipping on the ADC or input amp. Old style valve circuits will drive +28dBU ~ ±15V so give them room to breathe.

13. Use the low cut filter in 'low' mode if you really don't need bass extension. Without the filter in place I measured the 3dB point at round 17Hz on OEPs and 24Hz on Lundahls and it was basically really flat down to around 33Hz which is around the lowest C on a piano (and lower than a standard E bass) so if you don't need it to record that low then filter it. I've heard reports that the bass response is solid and I can see why with that frequency response together with oodles of headroom.

Update March 7th. I've done some experimentation with MuMetal Ultraperm 80 sheet. It wasn't the most scientific test. I placed a cylinder around each of the transformers and checked on a spectrum analyzer whether it made any visible difference. As far as I could see any difference (if any) was negligible. Pretty much what I expected as it is probably saturating with such strong fields, but anyway I thought I'd give it a try. Probably better just to spend your money on decent interconnects and mic cables.

 

[sonolink]

Hi,
I would appreciate your feedback to help me choose my toroids for the G9 I am building. Which pair would you advise me to pick? I don't exactly understand the differences or conveniences between these:

http://es.farnell.com/multicomp/mcfe030-15/transformer-30va-2-x-15v/dp/9531726
http://es.farnell.com/multicomp/mcta030-15/transformer-30va-2-x-15v/dp/9530312
http://es.farnell.com/multicomp/mcfe050-12/transformer-50va-2-x-12v/dp/9531777
http://es.farnell.com/multicomp/mcta050-12/transformer-50va-2-x-12v/dp/9530363
http://es.farnell.com/multicomp/mcta050-15/transformer-50va-2-x-15v/dp/9530371

Thanks for your help and time

edit: a small extra question: a friend told me to use a Light Bulb Current Limiter (Variac) to test a newly built amp to protect from frying expensive parts in case something is wrong. Is this true at all and advisable? Thanks again

 

[Stagefright13]

MeToo2 did you actually listen to a WORKING G9? And use it on what it rules at? What other pre's did you compare it to? And are you a professional audio engineer?

To me this is a specialty preamp that has a sound that CANNOT be produced from anything else. I also use API style pres too... Arguably they don't produce proper waveforms either.

Actually neither produce the waveforms a scope would want to see. But sound is in the ears. And the G9 produces a special sound. And I personally love it.

John

 

[MeToo2]

MeToo2 did you actually listen to a WORKING G9? And use it on what it rules at? What other pre's did you compare it to? And are you a professional audio engineer?

To me this is a specialty preamp that has a sound that CANNOT be produced from anything else. I also use API style pres too... Arguably they don't produce proper waveforms either.

Actually neither produce the waveforms a scope would want to see. But sound is in the ears. And the G9 produces a special sound. And I personally love it.

John

No I did not listen to a WORKING G9. Not sure what you mean by this comment. I think my G9's were working just fine. For the record I was looking at a spectrum analyzer, not an oscilloscope. I'm not criticizing the design at all. I was not trying to reduce or modify any tube or transformer dynamics in any way. The harmonics added to the signal by the design were exactly as I'd expect (mainly 2nd 3rd & 5th) and are of course an integral part of the character. I don't count these as "noise" because we specifically want them. I was just trying to improve the general signal to noise ratio of my builds, especially noise due to 100Hz (diode rectifier switching of mains power) and its harmonics, which dominated the general noise floor. That's a comment I've seen from other builders that individual builds of the design can be sensitive to hum (also a comment from the designer, Jakob.)  I don't think it ever does any harm to get the general noise floor as low as possible, as long as it doesn't turn into just a numbers game like discussions on 192KHz 24 bit versus 96KHz 16 bit  [Don't get me started on that. Clock jitter is far more significant than word length at these levels.] Talking about trusting your ears: when I started out, the hum and buzz on my builds were much more noticeable than any hiss at high gain, so that's why I went for that. At the end of my trials and various experiments I could turn up all gain switches to max and I could barely hear any hum at all with zero signal even with max gain on all components in the chain, whilst the signal was obviously deafening at that point. That's the point at which I was happy and I stopped. I just thought I'd share those experiences with others on this board. Your mileage may vary on your own build. As for professional qualifications, I'm a chartered electronic engineer BEng CEng MIET if that makes any difference at all.

Hi,
I would appreciate your feedback to help me choose my toroids for the G9 I am building. Which pair would you advise me to pick? I don't exactly understand the differences or conveniences between these:

http://es.farnell.com/multicomp/mcfe030-15/transformer-30va-2-x-15v/dp/9531726
http://es.farnell.com/multicomp/mcta030-15/transformer-30va-2-x-15v/dp/9530312
http://es.farnell.com/multicomp/mcfe050-12/transformer-50va-2-x-12v/dp/9531777
http://es.farnell.com/multicomp/mcta050-12/transformer-50va-2-x-12v/dp/9530363
http://es.farnell.com/multicomp/mcta050-15/transformer-50va-2-x-15v/dp/9530371

Thanks for your help and time

edit: a small extra question: a friend told me to use a Light Bulb Current Limiter (Variac) to test a newly built amp to protect from frying expensive parts in case something is wrong. Is this true at all and advisable? Thanks again

Jakob specified the transformers as 30VA. He knows what he's talking about. So all of the ones you chose should work. The totally enclosed transformers do look pretty interesting (MCFE030/15). Unfortunately Farnell don't do a 30VA in 12V version (no MCFE030/12.) Check out the physical dimensions of these two: mcfe050-15 & mcfe050-12 to see if they'll fit easily in your case. They are quite a bit bigger physically. It's actually over specified for the power, but I do not think that larger transformers will do any harm because the supply is regulated later, so any over-voltage at low load will be compensated. Might also be worth a look and they only cost a few cents more. Otherwise the others should work.

Yes it can be a good idea to run say a 60W light bulb _in_series_ with your valve project when you first test your power supply.

 

[sonolink]

MeToo2,
thanks for your previous list of tips. I am quite amazed at the noise elevators can put in mains!! The mains in the studio I work should be isolated, but you never know!!

About the Toroids, I agree, it's a shame there isn't a 30VA/12V version. I needed to know if overpowering it at 50VA was ok. I think I'll definitely go with the encapsulated ones with probably a big nylon screw or similar. My OEPS arrived this morning and I've been through my shopping list 3 times already, checking all Lead Spacings and comparing the various BOMs (especially Skylar's) with Jakob's. I just need tubes, knobs and the switch.

I will build first the Variac and a Condenser bleeder. It's been about a couple month reading, planning and shopping, preparing everything.....now I can't wait to begin the battle  ;D

Thanks again for your time and help
Cheers

 

[ianmgull]

Hey all. I'm designing my front panel and ran into a snag. I have the 12 position Lorin rotary switches  for the gain swith. I know the G9 uses 11 steps so I'll have to adjust the switch.

Assuming the normal travel is 300 degrees, should the gain index now be 275 degrees? (12positions/300degee rotation= 25degrees per position). The only reason I'm not really sure is because on all the G9 pics I've seen it appears as though they spaced 11 positions out over 300 degrees.

Thanks

 

[chrispbass]

About the Toroids, I agree, it's a shame there isn't a 30VA/12V version. I needed to know if overpowering it at 50VA was ok.

Hi Sonolink, you'll be fine with 50VA tx's. The circuit will only pull the current it need's so using a bigger tx won't 'force-feed' it more juice. You could use a 2000VA tx and it would be ok, but obviously the further increase from spec increases the size/weight/cost/interference factor and makes it less practical.


Chris

 

[sonolink]

Thanks Chris. I have learnt this now and I'm most grateful about it
I couldn't find the a toroid that corresponded to Jakob's specs!!
So it's ok now. I have also learnt that Caps work pretty much in the same way +/-, meaning that if the BOM states a 47uF/63V cap, I can use a 47uF/100V, without going through 50 shops!!  ;D

Thanks for your tips and time.
Cheers
Sono

Edit: If I'd like to change the original 12V/50mA pilot light for a 4,4V/50mA would I be ok by putting an 88 Ohms resistor on the positive side of the light? (I'm guessing this using Ohm's Law V=IxR, of course but am unsure if it applies like that)
Thanks for any confirmation before I order