MILA-1 with 6CG7's Help

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

chrispsound

Well-known member
Joined
Sep 30, 2010
Messages
159
Location
Sacramento
I built a MILA-1 with 6CG7's and need some help.

Link to modified schematic with voltages:
http://chrispsound.com/wp-content/uploads/2012/12/MILA1-6cg7.jpg

Pic:
IMG_0203.jpg


I am having trouble understanding what is happening in the frequency response chart in RMAA.  All RMAA tests were done with a M-Audio 2626 line out to a -40db pad with 150ohm shunting resistor then to the mic input on the preamp and gain was set for -6db at the output.

This is what I get with RMAA:
2-6cg7-reg.png


Is that big boost in the low end normal?  Some of my other micpre's have a similar response but not with so much boost at 5 to 10Hz.  There is also that little boost in the high which actually might continue to rise but the free RMAA doesn't go higher than 20kHz.  I haven't done any listening tests yet and I am not sure if I am even there yet by the looks of the response chart.
Thanks and Happy Holidays, ChrisP


 
First of all, could you please use the provided zoom functionality of RMAA and show us the whole low end?

Also RMAA can run at 96khz if you want to see more of that high end, which is perfectly normal already by the way. But keep in mind running at 96khz the sound card input and output might have their own roll off above 20khz.

Then a quick test with a different size output cap might be revealing. Another one with the sound card line out to preamp DI input to rule out input transformer. Also can you measure any DC voltage at the preamp output? Also what termination does the output have? just transformer directly to soundcard? Non-terminated output transformers directly to very high impedance can result to strange responses. Wire a 600-2000 ohm resistor there for another quick test.
 
It is the negative feedback loop. The turnover frequency for 47K and 1uF is 3.4Hz so you would expect a 3dB rise in gain at that frequency. This is mitigated a bit by the reduction in open loop gain due to the 0.022uF and 1Meg at the input to the second 6CG7 which has a -3dB point at just over 7Hz and the cathode decoupling of the second 6CG7 of 470uF and 200 ohms which turns over at just under 17Hz.

Try disconnecting the NFB. The gain will go up but the low bump should disappear.

Cheers

Ian
 
Kingston, with the input transformer bypassed there is no change,  the output is not 600ohms terminated but I will experiment further.  Thanks for the RMAA tips, RMAA is so buggy sometimes I am afraid to change any settings once its working decently.

2-6cg7-help-results.png


Ian, I removed the feedback loop from the circuit and now have way too much gain but no low end boost,  actually it kind of clobbered the high and low end.  If I adjust the feedback dc blocking cap to a higher value will it move the low end boost far enough down that it wont matter or is there a better way of doing things to flatten out the response a bit.
 
The pot VR1 and R9 in parallel are about 90K.  C3 into this resistance has a high-pass turnover at 1.75Hz.

Since this is before the point where feedback is returned, wouldn't lowering C3 a bit reduce this peaking response by canceling some of the low end?  Based on the RMAA plot, a new turn over at 20Hz should help this.  The would mean lowering C3 from 1uF down to something around the 0.047uF (37Hz) to 0.082uF (21Hz) value.

In my other 6SN7 MILA-like thread, I high passed the same stage at 40Hz entering into the second tube stage so that the feedback cap could be lowered in value somewhat without disturbing the response of the overall system.
 
chrispsound said:
Ian, I removed the feedback loop from the circuit and now have way too much gain but no low end boost,  actually it kind of clobbered the high and low end.  If I adjust the feedback dc blocking cap to a higher value will it move the low end boost far enough down that it wont matter or is there a better way of doing things to flatten out the response a bit.

First of all, +6dB at 5hz is a non-issue on any real world material. Secondly, you can take the feedback directly from your 4.7uF output cap to take that bump even lower and as a bonus you save one cap. Third, do that feedback resistor as a rotary switch or a 470-1000k pot. Something like 10k/22k/47K/100K/220k/470k/inf steps. It's very handy to have control over distortion on the output stage, goes between about 0.3-2% THD depending on your overall levels of course.
 
Thanks Kingston, I took out the1uf and threw in a 4.7uf panasonic (small and handy)film and it brought it down 2db but it did bring distortion up a tiny bit.  I like the idea of saving a cap.  I did a mic check with a 57 and the sound is similar to a normal MILA-1 with a bit more backhair.  Love the feedback resistor switch idea and will begin implementing it immediately, if my wife doesn't kill me first(in-laws are over for the holidays).  Thanks everyone, any more ideas or improvements are always welcomed.  ChrisP
 
Looking at that schematic a bit more I notice R21 takes B+ down 80V. Perhaps it is intentional and you want a lot of PSU ripple smoothing at this point, but it also takes out more than fifth of the output stage headroom. R21 is another great place to experiment with distortion in this design. You've probably heard of starved plate, and this is exactly where you would do it with a reasonably big watt 10k rheostat for example.
 
Kingston said:
Looking at that schematic a bit more I notice R21 takes B+ down 80V. Perhaps it is intentional and you want a lot of PSU ripple smoothing at this point, but it also takes out more than fifth of the output stage headroom. R21 is another great place to experiment with distortion in this design. You've probably heard of starved plate, and this is exactly where you would do it with a reasonably big watt 10k rheostat for example.

I was going to put a 2.2k 2W at R21 but ran out of resistors, after Christmas I will be able to restock the cave and have the necessities needed for further experimentation.  Until then thanks Kingston, I appreciate your answers and ideas in this thread and all the other threads you post in.  ChrisP
 
chrispsound said:
Ian, I removed the feedback loop from the circuit and now have way too much gain but no low end boost,  actually it kind of clobbered the high and low end.

Reality check:  response without NFB is far from clobbered.  1 dB down at 20 and 12-15K?  Heaven help you if you study some other typical tube response charts.  Use your ears.  We have stupidly linear response covered with other types of circuit these days. 

It stands to reason with the original NFB method that raising the value of R18 can land you somewhere in between the two charts.  Gain will go up, along with a softening of the distortion knee. 

Ian is completely correct in assessment of the effect. 

Another idea:  for comparisons sake, try plate to plate NFB from V2 to V1B, resistive only, use a big rheostat to determine desired value on the fly.  Plate to plate NFB will also give you more V1B headroom due to boost in V1B B+.  If you want a reasonably direct comparison against the existing NFB and cap low resonance effect, dial in the same gain with plate to plate NFB, see what your subsonic response hump looks like, and if you feel the need then play with the size of C6 until happiest.  Then play with the NFB R again and see how it responds. 
 
chrispsound said:
Ian, I removed the feedback loop from the circuit and now have way too much gain but no low end boost,  actually it kind of clobbered the high and low end.  If I adjust the feedback dc blocking cap to a higher value will it move the low end boost far enough down that it wont matter or is there a better way of doing things to flatten out the response a bit.

Yes, that is basically what you would expect. If you increase the 1uF it will move the bump further downand it should be smaller

Cheers

Ian
 
emrr said:
chrispsound said:
Ian, I removed the feedback loop from the circuit and now have way too much gain but no low end boost,  actually it kind of clobbered the high and low end.

Reality check:  response without NFB is far from clobbered.  1 dB down at 20 and 12-15K?  Heaven help you if you study some other typical tube response charts.  Use your ears.  We have stupidly linear response covered with other types of circuit these days. 

It stands to reason with the original NFB method that raising the value of R18 can land you somewhere in between the two charts.  Gain will go up, along with a softening of the distortion knee. 

Ian is completely correct in assessment of the effect. 

Another idea:  for comparisons sake, try plate to plate NFB from V2 to V1B, resistive only, use a big rheostat to determine desired value on the fly.  Plate to plate NFB will also give you more V1B headroom due to boost in V1B B+.  If you want a reasonably direct comparison against the existing NFB and cap low resonance effect, dial in the same gain with plate to plate NFB, see what your subsonic response hump looks like, and if you feel the need then play with the size of C6 until happiest.  Then play with the NFB R again and see how it responds. 

You are right, I may have been a little dramatic in the use of the word clobbered and you are also right in the fact that I didn't use my ears to come to that assumption.  I am going to make it a point that for every sweep I do I will do a listening test as well.  I just recently set this up in my vocal booth/ closet and will start using it more with samples that I am familiar with since I have family camped out in the main recording room/ guest loft.
IMG_0204.jpg


As far as the plate to plate resistive feedback, I am short a rheostat and was wondering if you could give me a value to start with.  Thanks for your help and I greatly value your input.  BIG THANKS TO NEW YORK DAVE!!!!!! for designing such a wonderful platform for newbs like me to learn and tweak on, hours of enjoyment(although my wife doesn't think so)
 
You already have the 47K, so I'd try it.  A 500K pot would be ideal for experimentation here, I think. 
 
I increased C7 to 4.7uf and made a negative feedback switch with values going from 10k to 100k in approx. 20k steps(which I think sounds best in the 68k position).  I increased the output cap to 10uf and it only had minimal effects on the low end bump so I put the original in so I can use it on something else.  I also tried resistive plate feedback but implementing it didn't get rid of the low end bump and the preamp doesn't sound as good(to me) as with the original feedback network.  I terminated the output transformer with  620ohm's and that brought the bump down and brought the little high frequency bump down too, but even with a flatter frequency response I think it sounds better unterminated.  My soundcard has a line in impedance of >20k ohms balanced(according to the manual).  Kingston had mentioned that very high impedance soundcard inputs can effect the frequency response of the Edcor output transformer:

term-unterm.png


I am happy for now with my bump.  Next I am going to play around with an output attenuator and different terminations.  Thanks again  to everyone that has posted.  ChrisP
 
Would be possible to use the original "Range switch" implementation of the feedback network, with a 3 position switch?

I built a one bottle preamp, but I'd like to add a second stage with a 6cg7 I have lying around, but not loosing the Range switch.

I thought in re-build the one bottle according to original Mila Schematic, but changing the second tube for a 6cg7, taking into account the schematic posted at the beginning of this thread (take out R14 & R15, change value for R16, 17 & C5), but not so sure about the Feedback network switch, ¿Could I go with the original design or will it cause me troubles because the tube mixture?

I have a few more 6cg7's, maybe it's better to use a pair of them and follow the schematic in this thread, but I'd like to keep the Range switch or any other way to control the overall level, not only the Gain pot... I like the idea of changing the feedback resistor value and get different gain levels

Any tip or advice? I'm a noob when talking about tubes...

Thanks
 

Latest posts

Back
Top