G9 and its frequency response ?? (impedance matching?)

[bernbrue]

Hi,
I´m just about to finish my G9 with Beyer 1:5 input transformer (yes, these tiny little things) and Edcor XS1100 as output transformer. I did some measurements with RMAA. First graph is line input (with input transformers), second one is direct input (without input transformer). The beyers are doing a great job in this tube preamp. Jacob, thanks a lot!

regards
Bernd

 

[mattnj]

I've come to the conclusion that the transformers are causing the "problems" in frequency response. The 10 Hz HPF I don't find a problem with. I've been using the pre and I love it and it sounds great so I'm not too worried about it. I don't think that it is a design problem (which other people have said) and I'm pretty sure that that frequency response is pretty well accepted.

 

[stanz]

The 10 Hz HPF I don't find a problem with.

I wouldn't have a problem with that either. It is a 300Hz HPF that I would like to see moved down. This only really becomes an issue when I am using it for recording a bass. It is losing a significant portion of the bottom end. I can always EQ it in the mix.

Still looking for answers to 2 questions:

1) Is this the normal output curve for the pre? (in which case, I can move on and make it work knowing what its capabilities are, or design a mod or work around)

2) If not, has anyone figured out why this shows up for some but not others?

 

[Neeno]

Well, since Bernbrue has no roll-off at all and he is using the edcor XS1100 that's a lundahl 5402 substitute I think that the problem might be the input transformer.

The DI input is pretty flat even using the 5402 but the mic input isn't.

Could it be the input trannie secondary termination ?
Maybe it should be loaded with a zobel network to match the right impedence ?

Bermbrue, can you please give us some more information about the Beyer transformer ? I would like to know the model, the DCR and the impedence of the winding and the way it is connected in the G9.

Thank you guys.

 

[Neeno]

Sorry for the monologue... but I really would like to solve this roll-off problem or at least know what is causing it.

I did some more measures, not real frequency sweeps but since i can "measure" the bass roll-off on the output i tried to trace the signal to see where the "roll-off" happends.

it looks like the first gain stage is working really good, there is no roo-off from 20Hz to 20kHz measuring the signal at the output potentiometer.

Then i continued tracing it and it appears that the roll-off happends on the second gain stage, the SRPP.

I tried to increment the output cap value, adding a 4u7 in parallel with the actual cap mounted on the pcb, but the roll-off is still there...

Anyone can do the same measurement and confirm this ?

 

[noulou]

When I measured my OEP-equipped G9 I got similar low-end rolloff like the other users.

This was using the line-in which goes through the input transformer. Through the DI it was flat.

It certainly is VERY audible on my large speakers either playing mixes through it or amplifying microphone signals.

I usually plug the G9 when I consciously do not want all the low frequency information of the original signal.

Perhaps the OEP A262A3E has a too small core?

 

[Neeno]

I don't think it's the input transformer...
I have the same problems and I'm using lundahls.
The roll off happends after the SRPP stage, so probably it could be the output transformer or a design limitation.

Keep posting your experience with this preamp guys !

 

[jensenmann]

Bernd is in vacation. He should be back by the end of next week.

 

[stanz]

I dug in and followed the signal through the unit. The signal is fine all the way through, until after the output transformer. I now forget if the preamp had the output connected to nothing, or if it was connected to my Echo A/D converter.

What I also found, was that if I moved the G9 output connections to the mic input of my mixer, the bass was fairly nice. The signal from the mixer was not measured with a scope, but it sounded good, which works for me. My big issue/complaint was not being able to use the G9 as a DI for recording a bass. Before I was running straight into my Echo A/D interface. I still am not totally clear on the whole transformer/load/frequency-response theory, but I have heard enough to know that load will effect frequency response.

Is it "the output needs to have a small impedance and wants to see a high impedance at the next stage", or is it the other way around?

 

[Neeno]

A question for you Gyraf, and all the guys that are encountering troubles with the low end.

Looking at the schematic, especially on the srpp stage, i found that there are 2x10k resistors in parallel on the output trannie.

Actually i don't know if both are needed, so i tried removing both, and guess what... my g9 is finally linear down to 20Hz (-0.3dB)...

Then i suspected that at least one of the resistor should be the right loading for the trannie primary (that in this case is the secondary because is connected 2:1) and the preamp is still linear.

Can you guys try that and confirm ?

Gyraf: what is the right loading here ?
10k ? 2x10k in parallel ? neither ?

Thanks for your help guys !

 

[matta]

Hey Neeno,

I'm bumping this as well... hopefully Jakob will chime in at some point... I'm having the same issues.

Cheers

Matt

 

[riggler]

Bump for me too! I finish my G9 this week, so I might do the 'bass fix' straight away, if we get a solid answer here.

:wink:

 

[stanz]

Hoping to spur some movement toward firuring this out. Here is what I have compiled about what is going on.

The (2) 10K resistors at the output (R32 & R44) are in parallel. The question was raised earlier about which to remove, one or both. I have also seen a post about raising the coupling 4.7uF cap (C13) to 20uF to extend the low end knee. http://www.groupdiy.com/index.php?topic=16699

In looking at Jakob's description of the output:

The output trafo is wired 2:1 to get the 2400 Ohms Zout of the SRPP stage down to around 600 Ohms. The 2:1 ratio means that when using the OEP 1+1:2+2 transformer, it has to be reversed. I've had very good results using this output stage topology in all sorts of designs - it's simply my favourite tube circuit. Also used in the Gyraf Pultec amplifier.

The output of the transformer is taken to SW4, the phase reverse switch, that has 10K resistors across it to control transformer load when switching. And then to the output XLR.....

A 4.7uF cap with a 600 ohms gives a knee of 56Hz. In calculating what resistance gives you a knee of around 450Hz, I get about 75 ohms. A high pass filter with 75 ohms and a 20uF cap gives a knee of 100Hz. It seems that taking out the 10K resistors, or raising the coupling C13 to 20uF is basically going after the same goal of lowering the bass cut off.

I am not sure about the variables, as my design skills are minimal at best, although my troubleshooting skills are respectable. Are the 10K output resistors really needed? Seems the easier solution (if this is really fixing it). What is lowering the resistance at this point in the circuit? I measure 23 ohms off of pins 2 & 3 on the XLR output while power is off. Taking out the 10K resistors seems it would do the opposite of what is needed, you want to increase the resistance to lower the low end cut off. In looking at it from a troubleshooting perspective, what is really causing this to happen, or is this the designed specs for the G9? Are the common variables OEP output transformers and Gustav boards? Is there a common mistake in assembly here? The phase switch is right in the middle of all of this and one of the 10K resistors is on the main board, while the other is next to the phase sw on the front panel PCB.

Part of me wants to figure it out, and the other is very happy to get it working and move on. I am going to try taking out the resistors to see what happens as that is the quickest and easiest thing to try.

 

[rodabod]

I suspect Jakob maybe added the 10K||10K load so as to dampen the response of the OEP transformer (keep HF response flatter and avoid HF ringing). A line input might be as low as 10K input impedance, so the overall impedance reflected onto the secondary might be ~3K.

A 4.7uF cap with a 600 ohms gives a knee of 56Hz

The way I see it, the filter combination is not with 600 Ohms as this is the impedance looks like from the secondary. The cap is in series with the reflected impedance which might be 4* 3K = 12K. Looks like an ok load for the output stage too. Correct me if I'm wrong, but to me this doesn't seem to be a problem. But that said, if removing the 10K secondary load resistors is solving the problem, then it suggests there is a loading problem...

 

[riggler]

Well, we can't damage anything by jumping over those resistors, right?

Does anyone have a scope that can look for ringing or side effects of doing this? If I can get ahold of one I'll scope mine when it's finished.

 

[stanz]

That would be shorting the output, as both 10K resistors are across the output. You would want to make it an open, not a short.

I am convinced it is a load issue. I just finished up some tracking by running the G9 into my board's mic inputs, with the trim at minimum, and it sounds great. This includes running the bass direct into the G9. Two elec guitars and ran some drum samples through it, one drum in a stereo pair at a time. Kick sounds great. Also, by running it into the board, it gets just a little extra boost in volume which is just about right, as opposed to going direct into AD converters.

For those of you who are having low end issues, what are you sending the pre into?

 

[gyraf]

The 5KOhm (10K//10K) loading at the output is for controlling transformer resonances when using the output with high-impedan'ish type of loads.

The output circuit - 4u7 into 2K4 Ohms (600R*2:1sqr) gives a corner frequency of ca. 15Hz into a 600 Ohms load on the output. removing the 5K load should not do any significant difference - unless a wrong value or wrong marked resistor had been put in by mistake.

If you have lack of low end at your mic/line input - and not at your instrument input - you should investigate the source impedance you're driving the input transformer with..

Jakob E.

 

[rodabod]

Somebody with a 'scope or an audio measurement set needs to make some simple tests.

 

[stanz]

I did scope it earlier when I was having low end drop off, and yes it is there around 450Hz, and only after the output transformer. I feed it with a signal generator using a sine wave, and trying a range of signals from 50Hz to 800Hz, and looked at each gain stage. Nice signal all the way through. But I was going directly into my DA. I was looking for just a bit more gain, which is why I ran it into the first two channels, which have a pre amp with trim control.


I just looked up the input impedance of my Echo Layla.

XLR input is 1.5K ohms (ch 1 & 2)
Guitar input (DI) is 102K ohms (ch 1 & 2)
Line inputs are 10K ohms (ch 3 - 8)

Mic input on my board is 2.5K ohms

I confess that transformers are still a bit of a black hole of knowledge as far as impedance is concerned. I understand the effect it has on a signal though. Doing that little exercise shows where I went wrong. I had assumed ch 1 & 2 were the same as the other 6. Also, ch 1 & 2 have two paths in that are very different. I hadn't used the xlr inputs because it was too hot, even at minimum trim. Using the TRS input was the issue.

I didn't scope it after switching the output of the G9 to my board because that definitely solved the problem. Didn't need a scope to tell me the low end is there. The final test was some tracking I did that I mentioned yesterday. Sounds GREAT!! :grin:

It would be great to get some feedback from others having this problem as to where they are running the output of their G9 in order to help others to see down the road.

 

[Kingston]

Hey DJ, where is the bass?  ???


Anyway, I've just finished a G9 unit with Lundahls, and the bass roll off is is there, just as I had feared, and just as so many DIYers have heard in this thread. It's very audible lack of bottom end, and I'm surprised it has been pretty much shrugged off as "part of the sound" this far. Also visible on spectrum analysis with RME HDSP. No other preamp of mine has anything rolling off at quite this scale.

Was there ever a final solution to this, other than switching to better suited transformers? The discussion seems to have died off and I could not find more threads on this topic.

Oh, and removing the 10k parallel output resistors did not help, which they shouldn't have anyway.

Wish I had a scope to pin point the exact point of bass removal. It's not the input trafo (DI input gives equal response). I'll try unbalanced out tomorrow to rule out the output trafo.