RCA BA21 build

[dmp]

I put in a 220uF that I had and measured the highest and lowest gain positions. Looks and sounds good. You can see a little low end boost at the low end but it is minimal with the 220uF, actually flattens the response like the C4+R14 compensation. Could do a switched R+C for each position to match the C to the R for frequency compensation through the whole gain range. At high gain, C would drop to about 0.047uF. But this seems like it will be good enough. Like the V76, a slight change in freq response through the gain range. I can work out the steps, probably using a 12 step switch (2.5dB/step).
The most NFB I can get is about 28 dB. Lowering the R any further towards zero and I get some weirdness but it maxes at ~29 dB, because of the tertiary stepdown. I see this in the equations. Lowest R I'll use is ~900 ohm. I'm getting good performance at max gain too, using zero NFB. I am not noticing a noise issue with AC filaments.

The orig layout doesn't seem up to it.

Maybe the wire routing of the heaters?

A 20 dB input pad in combination with this gain range should be good. Leaning toward not having an output attenuator.
I still have to build the psu in the case also to match the performance I currently getting with the heathkit IP17.

I also noted C4/R14 are reverse of drawn according to layout.

I noticed that too. Doesn't matter though, I don't think.

Always faster for me to measure gain steps on the bench with a decade box.

That's smart - I'll measure at least a few steps to make sure before building a switch. But my calcs show it is log scaled through most of the range until the ends, so straightforward. I don't have a good way to measure gain when the input is balanced. My sig generator & scope are unbalanced, and the lowest level I can send is about 60mVrms. Turning the pad on with the unbalanced connections doesn't work. I could build an unbalanced 20 dB attenuator. Or build a simple balancing interface for my scope with chips I suppose.
And then using my interface with RMAA to measure gain isn't precise just because I don't have a good readout. I'd like to figure out a better method. A simple thing I have never done. Maybe just need a better sig gen / scope as software on the computer to use with the interface

 

[emrr]

I'm sure the noise I see is the filament wiring being 'good enough' for 40dB. I think it's at least partially screened.

+18dB preamps like this, I rarely ever need an output attenuator, and I don't tend to operate 'in the red' enough to want it dedicated. I can always patch in a pad if needed for that.

 

[dmp]

This can drive a sine wave to a pretty impressive level.

Shorting or non-shorting switch for the gain? Have to order a grayhill 71 series. Shorting (MBB) would put two Rs in parallel, lowering the gain, while BBM would potentially go to no NFB, high gain for a instant.

 

[emrr]

Shorting. Lesser problem with momentary contact. I used a Lorlin just fine. Would have used a 100K pot if I coulda found one that would fit well.

I used a 330mfd for my series NFB cap. Probably what I had around.

 

[emrr]

My plots

 

[dmp]

Is that with the input connected as 600 ohm or 150?

 

[emrr]

150. Output into 10k

 

[dmp]

Interesting - I get a really different response at 150
I wonder if I have a mistake or if it's just variation in the input Tx?

 

[emrr]

That notch at the top to my understanding is the result of humbucking split windings and the way they act when used single ended. I guarantee if you used the input secondary in PP it would go away entirely. That notch seems to vary more than any other aspect when comparing ‘like’ transformers. All of it being past the quoted spec, I would assume considered an unimportant variation. Would like to know more. That notch can also be wildly different if you reversed the polarity of the sec, along with the top end response overall. If mine weren’t notched like that, you might see a similar top end rise. I have 2, measure roughly the same, but they are also consecutive serial #s.

 

[dmp]

All of it being past the quoted spec, I would assume considered an unimportant variation.

They say +/-1dB 15-15kHz, but the plots from the manual look much better than mine with -4.5dB at 15kHz for the 150 ohm. Yours are much better.
I wonder if it is my test setup? I am connecting my audio interface directly to the input. By output into 10k do you mean that is the input impedance of your interface or you have a 10k resistor strapped across the OPT? I would think the OPT should have a 600ohm resistor across it to have the right load presented to the push pull stage?

That notch can also be wildly different if you reversed the polarity of the sec, along with the top end response overall.

I tried that and saw a very small difference.

Do you know what the date code is on these transformers? The input I have is marked CV25

 

[emrr]

Mine into a 10K input Z. No resistor. I look at outputs under varying loads and assess need on an individual basis. Not that many things need a true 600, even when something close is needed usually 1k2-2k looks better.

 

[dmp]

Built a power supply board using a toroid PT (that is thin enough to fit in the 1U case, got this from a ebay seller).
Used three stages copying the BA-2, but with a diode bridge instead of a tube rectifier. With 6.8k resistors the voltage at the preamp is 282V. Used some caps I had on hand 47uF with a 100uF in the middle.
My build is drawing slightly less mA than the RCA spec so with fresh tubes, the B+ might drop a bit more, but wouldn't be a problem. V2 draws the bulk of the current and my test voltage is 0.9 vs 1.0
The heater voltage was high, 7.1v so 1 ohm / 3w resistors in series before the 100 ohm resistors to ground put the H+ at 6.1v

Tests were good. First I switched this on in the case still powering the preamp off the bench supply to make sure magnetic induced hum wouldn't be a problem, then fully switched over.
Curious to try elevating the heater with a voltage divider off the B+ to see if this will improve the noise at all. I can't remember if I tried this on a previous tube pre build. If it would improve the noise performance, I'm surprised RCA didn't do it on the preamps with a PSU, like the BA-2

 

[dmp]

Finished mounting everything and made up the front panel. Have a few tests to do now for the gain and overall performance. Front panel switches are impedance, pad, and phase.

 

[gary o]

Well the build looks lovely well done... I built the circuit many moons ago when I first started DIYing, I used a Sowter 9045 input tran and a OP tran brian Sowter sold me, I remember him explaining tran didnt have tertary winding so feedback would not work ,it did have a ton of gain in feedback less config too much for my soundcard at the time, I might rebuild it for bit of fun I have some nos RCA 12AY7s

 

[dmp]

Thanks Gary. Without the feedback it should be about 60 dB. Unless going for authentic, other cheaper tubes would work fine.
Finished it up last night. There is no hint of 60 Hz or 120 Hz in the noise output, so I didn't bother testing elevated heaters. But in the future I'll try this out. Gain goes from 30 dB to 60 dB with the 12 step switch. And the 150 ohm impedance adds another 6 dB.

 

[gary o]

Excellent job Dmp it looks great ! I fired mine up to compare sound to other pres I use , quite warm n fuzzy without feedback I like it, I usually like kit with 12ay7s I have a small stash so will use them.

 

[dmp]

Noise spectrum measured with RMAA (white). Gain set to 42dB (with the soundcard output down 42dB).
The soundcard alone for reference is green.
Not sure how this can be related to a EIN but seems quite good. Tube preamps with AC heaters can do really well - no hum issues.

 

[heatwalk]

What is the purpose / result of Neg Feedback via Tertiary vs local off of a plate?

 

[dmp]

What is the purpose / result of Neg Feedback via Tertiary vs local off of a plate?

It includes imperfections of the transformer in the negative feedback, compensating for them, while also isolating from the line out (if the feedback was from the secondary)

 

[dmp]

And I'm guessing here, but maybe RCA was trying to reduce the cost of the transformers going from the no-feedback of the early BA-2 and console amps with the larger can transformers (i.e. similar to UTC LS series) to these smaller square can transformers (i.e. similar to UTC A series), with 20 dB of negative feedback.
Though I haven't had time yet to really compare the sound of this preamp yet with the BA-2 I built with original transformers.