Original UA console - 100D preamp, EQ, 101D program amp

[bockaudio]

When tracking and mixing sources direct to tape, the operating levels are likely to be much lower, especially as many of these boards were used with ribbon mics.

conjecture.

 

[emrr]

Well, levels were expected to be much lower.  -22 dBm max input was very much a standard until about 1970. 

 

[ruffrecords]

When tracking and mixing sources direct to tape, the operating levels are likely to be much lower, especially as many of these boards were used with ribbon mics.

conjecture.

Of course, why else would I say 'likely'. There is plenty of anecdotal evidence to support the conjecture -  'Temples of Music' for example.

Cheers

Ian

 

[Bill Wilson]

The 100D preamps used what looks to be a UTC ouncer series input transformer; they couldonly take a maximum input signal of -10dbv. The later 1008 used a heftier input transformer & would likely take somewhere around +10dbv before core saturation. AC on filaments was very common in thru most of the 60's. The trick was to not use the center tap on the 6.3 volt filament winding but to use a 10 watt wirewound pot 150 ohms or so across the filament leads w the wiper tied to ground. You then tuned out any hum with the pot. The filament wiring in a preamp used twisted 20 ga. wire that was laid on the chassis; with this system ac on filaments caused no problems.

Abbey Road seemed to infer that 12AX, 12Au, 12AY7's as somewhat less than the best. I have a class A preamp using  the industrial version of 12AX7 (7025), very flat & smooth w very low distortion & a lot quieter than a pentode type. Back then the only thing a 12AT7 was used for was as a mixer oscillator in an FM receiver.

 

[emrr]

The UA docs show custom part numbers for the iron.  The input is definitely larger than UTC O series, looks closer to what Langevin used at the time.  At any rate the transformer easily takes more level than the following actives can take, which is normal for products of the era. 

In my personal experience, I've yet to hear a 12xx7 based preamp that could compete in sonic terms with earlier tube formats.  All of the commercially offered professional units using 12AX7's had very high NFB involved, and I tend to think the use of the 12AX7 was intertwined with the desired NFB quantity and target final gain figure.  As always, more than one way to skin a cat.    I like the 12AU7 fine, but I like similar octals better in a listening comparison. 

 

[Kingston]

In my personal experience, I've yet to hear a 12xx7 based preamp that could compete in sonic terms with earlier tube formats.   All of the commercially offered professional units using 12AX7's had very high NFB involved, and I tend to think the use of the 12AX7 was intertwined with the desired NFB quantity and target final gain figure.   As always, more than one way to skin a cat.    I like the 12AU7 fine, but I like similar octals better in a listening comparison.

I've heard you say this many times and it is an interesting subject. That's why I used their octal alternatives in my latest project. Few parts missing so no listening tests yet.

 

[abbey road d enfer]

Abbey Road seemed to infer that 12AX, 12Au, 12AY7's as somewhat less than the best.

No, you misunderstand me. It was just a pun in the direction of these audiophools for whom a tube with less than 100 years pedigree isn't worth a wink. 
I know I should put one or two of these emoticons so everybody would know they have to take it with a large pinch of salt. I hope with time everybody knows it's a given - Abbey is always in jokey mode.

I have a class A preamp using the industrial version of 12AX7 (7025), very flat & smooth w very low distortion

As in any electronic piece, components are half the story, design is the other half. I'd rather have a well designed piece using pedestrian 12AX7 than a sloppy design with whatever fancy tube.

& a lot quieter than a pentode type.

I must admit I don't get the use of a pentode in a professional mic pre, because it's bound to be noisier. I understand the reasons for using it in a Victrola, because it had enough gain to push the grid of the output tube, and noise & THD were not a paramount concern. I understand that the designers of the EF86 tried to improve most of the non-Johnson noise sources, with increased vibration-induced protection and extra shielding, but I have a feeling that the best performance is achieved by connecting it as a triode.

Back then the only thing a 12AT7 was used for was as a mixer oscillator in an FM receiver.

Aint' it what it was intended for?

 

[emrr]

I'd rather have a well designed piece using pedestrian 12AX7 than a sloppy design with whatever fancy tube.

True.  My observations stick to truly professional dedicated equipment of yesteryear, and ignores non-sense like repurposed Ampex tape machine electronics and Altec PA gear.  

& a lot quieter than a pentode type.

I must admit I don't get the use of a pentode in a professional mic pre, because it's bound to be noisier.

More gain, so you can use more NFB in most cases.  In 40 dB preamps, RCA went from using two triode connected pentodes with no NFB, to two pentodes with more than 20 dB NFB.   In cases with front end pentodes not in a NFB loop, it's almost always budget equipment, or portable equipment.

 

[abbey road d enfer]

I'd rather have a well designed piece using pedestrian 12AX7 than a sloppy design with whatever fancy tube.

True.  My observations stick to truly professional dedicated equipment of yesteryear, and ignores non-sense like repurposed Ampex tape machine electronics and Altec PA gear.

Just a coupla days ago, I've been offered a 601 "rack" for $300. I looked at the schem and thought it was not good enough, would have needed a complete  redesign. Plus it didn't have the input xfmr...

More gain, so you can use more NFB in most cases.  In 40 dB preamps, RCA went from using two triode connected pentodes with no NFB, to two pentodes with more than 20 dB NFB.

I get this, but still it doesn't make much sense in terms of ein. Maybe noise was not considered as important as frequency response or THD...or was it the novelty factor of NFB?

 

[emrr]

Yeah, an Ampex 601 might be a fun color box for some, but it's not a real classic preamp design by any stretch of the imagination. 

More gain, so you can use more NFB in most cases.  In 40 dB preamps, RCA went from using two triode connected pentodes with no NFB, to two pentodes with more than 20 dB NFB.

I get this, but still it doesn't make much sense in terms of ein. Maybe noise was not considered as important as frequency response or THD...or was it the novelty factor of NFB?

I can't speak with any authority there, but would assume, comparing the following:

2 triode preamp with loop NFB, 40 dB total gain
pentode/triode with loop NFB, 40 dB total gain

...that any noise differences would come out in the wash when looking at the varying amounts of NFB to hit the 40 dB mark.  I may be totally wrong; would be interested to see a walk through of the numbers there. 

I would think response and THD, along with consistent gain figures, would rank over importance of noise in any well designed preamp of the era. 

To rewind elsewhere, Ian/ruffrecords did some interesting tests with 6SN7s and 6CG7s, and collected enough data to (maybe) start looking for trends.  I think it's on his website. 

 

[ruffrecords]

I can't speak with any authority there, but would assume, comparing the following:

2 triode preamp with loop NFB, 40 dB total gain
pentode/triode with loop NFB, 40 dB total gain

...that any noise differences would come out in the wash when looking at the varying amounts of NFB to hit the 40 dB mark.   I may be totally wrong; would be interested to see a walk through of the numbers there.

It is reasonable to assume a pentode plus triode will have more open loop gain than a triode plus triode. The pentode/triode combo will therefore have more NFB to for the same 40dB closed loop gain. Some noise sources, notably power supply noise, will be reduce by the extra feedback. NFB however will not reduce the noise introduced by the first stage  so if this is a pentode, the final amplifier will be noisier than a pair of triodes. Maybe the designers of early mic pres were more concerned about noise entering from the HT than the intrinsic noise of the first stage.
   

To rewind elsewhere, Ian/ruffrecords did some interesting tests with 6SN7s and 6CG7s, and collected enough data to (maybe) start looking for trends.   I think it's on his website. 

The trouble with global tube NFB is that because of the difference in dc levels between the output and the input there is (almost) always either a capacitor or transformer in the feedback loop. In both cases this means the amount of NFB reduces at low frequency extremes which means there is a substantial peak in the amplifier response at some very low frequency. Now I have absolutely no evidence that this is a 'bad thing' but I am pretty sure it is not a 'good thing'. So when I got interested in tube mic pre design I wanted to find a circuit that gave a useful gain with low distortion without using global NFB. The solution I came up with depends on using tubes with good inherent linearity in a topology that does not compromise that linearity i.e  6SN7/6CG7/6FQ7 mu follower.  The 6CG7 is much more convenient to use than the 6SN7 so I did some tests to establish that a 6CG7 is just a 6SN7 in a smaller bottle, then a lot of distortion tests on a lot of 6CG7s.

Te results are available here:

http://www.ianbell.ukfsn.org/data/mufollowerdistortion.pdf

http://www.ianbell.ukfsn.org/data/6CG7mufollowerdistortion.pdf

Cheers

Ian

 

[abbey road d enfer]

Very informative. Thanks!

 

[emrr]

I might assume the initial transformer step-up gain was considered sufficient to overcome first stage noise.  I use this sort of antique pentode gear on a regular basis, and haven't experienced noise issues. 

So loop NFB returned to 1st stage cathode does nothing to alleviate noise from that 1st stage? 

How about the occasional cases where NFB is returned to 1st stage grid through the input transformer secondary? 


I looked at the 6CG7 test data again, and noticed the comment about anode size appearing to relate to distortion level.  This corroborates other anecdotal commentaries I've seen over the years. 

 

[ruffrecords]

I might assume the initial transformer step-up gain was considered sufficient to overcome first stage noise.   I use this sort of antique pentode gear on a regular basis, and haven't experienced noise issues.  

My own measurememnts indicate that a transformer was not only necessary but essential to ensure a reasonably decent noise figure even with triodes. Many noise issues do not become apparent because input levels are high enough and gains are low enough that the noise is not audible. Personally I have found few if any situations where incredible noise performance in a mic pre is necessary. If you want real sensitivity you use a condenser mic which provides a relatively high output and its noise plus ambient acoustic noise will mask any noise in the mic pre. For high input levels noise is not a problem. The worst case would be using a low sensitivity mic like a ribbon to record bird song; then you would hear the mic pre noise.

So loop NFB returned to 1st stage cathode does nothing to alleviate noise from that 1st stage?  

Nope. Most good NFB text books make this point. Noise introduced by subsequent stages is reduced but if the first stage has sufficient gain then the noise of latter stages is irrelevant.

How about the occasional cases where NFB is returned to 1st stage grid through the input transformer secondary?  

Ditto.

I looked at the 6CG7 test data again, and noticed the comment about anode size appearing to relate to distortion level.  This corroborates other anecdotal commentaries I've seen over the years.  

Do you have links to any of this anecdotal evidence? I am always on the look out for corroborating data.

Cheers

Ian

 

[abbey road d enfer]

I might assume the initial transformer step-up gain was considered sufficient to overcome first stage noise.   I use this sort of antique pentode gear on a regular basis, and haven't experienced noise issues. 

My own measurememnts indicate that it was not only necessary but essential to ensure a reasonably decent noise figure even with triodes. Many noise issues do not become apparent because input levels are high enough and gains are low enough that the noise is not audible. Personally I have found few if any situations where incredible noise performance in a mic pre is necessary. If you want real sensitivity you use a condenser mic which provides a relatively high output and its noise plus ambient acoustic noise will mask any noise in the mic pre. For high input levels noise is not a problem. The worst case would be using a low sensitivity mic like a ribbon to record bird song; then you would hear the mic pre noise.

I've been more than once in a situation where preamp noise was a limitation; recording double-bass with a ribbon mic (chinese junk) and fiddle with an M201, in both cases with an ISA430MkII. I must say that I'm anally obsessive about noise. Recording the same double-bass with an AT4060 or a Sigma active ribbon mic, the same fiddle with an AT450 and the same preamp was as noiseless as could be. Sometimes you want the signature of a particular mic on a source that is not adequate...

 

[ruffrecords]

I've been more than once in a situation where preamp noise was a limitation; recording double-bass with a ribbon mic (chinese junk) and fiddle with an M201, in both cases with an ISA430MkII. I must say that I'm anally obsessive about noise. Recording the same double-bass with an AT4060 or a Sigma active ribbon mic, the same fiddle with an AT450 and the same preamp was as noiseless as could be. Sometimes you want the signature of a particular mic on a source that is not adequate...

I agree, ribbons and quiet sources are the problem area. Rather than expect a general purpose preamp to perform extremely well in all possible circumstances I would be inclined to have a specially designed pre-amp for such situations.

That said, today's semiconductor pre-amps are about as noiseless as you can get and if the mic is low output then you are up against the Johnson noise of the mic itself and there's nothing you can do about that.

Cheers

Ian

 

[emrr]

I'll dig for some more commentaries concerning anode dimensions.  

It's pretty rare at my rock and roll studio to have sources so quiet as to reveal preamp noise.  Even most of the acoustic acts are really R&R, with singers who do just fine with a ribbon into a 40 dB preamp.  My usual acoustic guitar condenser usually only needs 20 dB, so a 40 dB preamp with a pad on the front end can fit the bill.    For vocals, using an SM7 with a 75 dB Gates tube unit, I find typical gain settings from 63-75 dB, and so long as 75 dB setting gives a reasonable output level, noise is sufficiently masked.   If output is not reasonable, noise is revealed, so it's a case of proper 'windowing' of source, mic, and preamp.  

I did a session for a friend I'd been in a rock band with, he's now transitioned to classical guitar.  I used a MS Beyerdynamic ribbon pair about 2.5 feet out, and ambient noise dominates.  

Some NFB reading to revisit.   Pentodes: I do have an RCA BA-3 which hasn't been put through the paces as preamp much yet, and will pay attention.   I have removed the initial voltage divider in mine.  I have already suspected since it was intended as a program amp that noise might not be as good, given the expectation of being driven with program levels.  

http://www.waltzingbear.com/Schematics/RCA/BA-3C.JPG

I mentioned elsewhere modifying a BA-11A to have variable NFB gain control, and what the acceptable plotted frequency extremes seemed to be.  In practice I've found that all obvious noise is regulated by the NFB quantity, and you don't get as much additional useful gain as the frequency response experiments suggested.  Increasing NFB increasingly vanquishes any noise one might detect. 

 

[ruffrecords]

I'll dig for some more commentaries concerning anode dimensions.  

Thanks, I appreciate that.

Some NFB reading to revisit.   Pentodes: I do have an RCA BA-3 which hasn't been put through the paces as preamp much yet, and will pay attention.   I have removed the initial voltage divider in mine.  I have already suspected since it was intended as a program amp that noise might not be as good, given the expectation of being driven with program levels.  

Yes, even with 10 sources on a passive bus the attenuation will only be 20dB and that input transformer looks like it has around a 1:8 ratio i.e 18dB gain so the input to the first pentode (which will dominate the noise) could be around -2dBu so pentode noise will not be an issue.

 In practice I've found that all obvious noise is regulated by the NFB quantity,

It is bound to because it reduces the closed loop gain. The (first) tube produces just as much noise, its just not amplified as much. The signal to noise is not made better though because the signal is reduced as much as the noise is.

Cheers

Ian

 

[PRR]

> Increasing NFB increasingly vanquishes any noise one might detect.

Only (as Ian says) by reducing gain.

Get Abbey's bassist and junk ribbon. Rig your variable NFB amp into another gain control and turn them opposite so that the peak sounds do NOT change with variable NFB. Compare noise levels.

At best, hiss is constant at all NFB-gain settings. (In practice, total hiss might rise at very low gain, high NFB, because output and NFB loop hiss are constant and input hiss is gained less.)

The NFB, however injected, is ultimately *compared* across the amp input, the grid-cathode junction. The NFB can not tell the difference between input signal and grid-cathode self-hiss.

Also input offset voltage in DC-coupled amplifiers. I had a long discussion with someone using erratic-bias input devices (which "work in SPICE") and assuming that huge NFB would drive-down DC error. That's why opamps have an input offset spec: no amount of NFB will reduce offset below the input junction's error.

Pentodes seem to be a path to higher (more fadish, also more constant gain with tube age) NFB with fewer tubes. The assumption seems to be that a sufficiently high input turns-ratio would overwhelm pentode partition hiss. There may be synergy with better dynamic mikes in that era: no longer scratching for the last dB of hiss as with older ribbon mikes. There's also a reduction in tube-types needed on-site: many small pentodes can be wired as fine triodes or as fairly beefy power drivers.

BA-3C seems to be an all-purpose medium-level amp. The input pot is not optimal for very low-level work, unless there are two custom tapers. The 6V6(?) output is more than a mike-amp would normally sport; eight of these in a knee-cubby would melt your pants. Of course many non-optimal things got done to fill a contract with what the factory had handy.

> the amount of NFB reduces at low frequency extremes which means there is a substantial peak in the amplifier response at some very low frequency.

Many ways to fix this. Often not pretty.

The BA-3C has a dedicated NFB winding, a nice start. Assume that T-2 is "audio reasonable", not extended sub-sonic response. Make C-6 5uFd-50uFd and C-7 1uFd-10uFd. Response is dominated by T-2 to well below 1Hz, 50:1 of excess gain can be blown-off without significant phase-shift. However leakage of such large caps ('specially in broadcast-reliability use) will be an issue. (In another module, RCA used 1uFd to block 200V then a second 1uFd to block the possible leakage of the first.)

I was raised on high-NFB thinking. I'm seeing that when you spend more time wrestling with NFB issues than with your audio, you may be missing the point. And one of the most realistic amps I ever heard had about zero NFB (not even triode-action).

In days before NFb, some PA amps delibrately used a C into an L (often driver load) to bump-up a few-dB@50cps, compensate the several part-dB@50cps losses in the other stages. Today such tricks get our panties in a wad.

 

[emrr]

plot with BA-3 response, relevant to PRR's comments about the DC end of things.

Thread with other RCA details and NFB experiments