Modified RCA BA21 quartet

[blue_luke]

Hi,  Being snowed in and not having much to do, I decided to start this project that I have been putting off for far too long now.

I happen to have 4 sets of original input/output transformers to build a 4 channels preamp based on the RCA BA21
Now I am not a newbie, but tubes are still somewhat mysterious to me... ??? I guess I can say that I know enough to be dangerous!!

Here is a reduced version of the schema by removing the input socket, the metering switch and all those dividing networks for the metering. I have replaced those dividers with single resistances of equivalent values.

Now the circuit is much more simple to follow!!
The BA21 has about 40db of gain and is rated flat +/- 0.5db from 50-15Kcycles 

OK, now here is some mods I would like to try with this. Once done, it probably won't be an RCA BA21, so I decided to call it a Blue_Luke BL-21a.

Here is the schematic I came up with.

For the time being I am concentrating on the output section. You will notice that I changed the output tube for a tesla ECC99, in the hope of reducing distortion. Pots P2 is adjusted to get a null at V2 cathodes and pot V1 is adjusted with a THD meter.

The 12DW7 is one half a 12AX7 and one half a 12AU7. I am trying to get a bit more gain here to use ribbon microphones. I am contemplating the idea of adding DC bias to the grids of V1a and b... Not sure...

What do you think of all this??  ;D

Meanwhile, yesterday I started building the chassis... I proceeded at a fairly good clip so I should have pictures to show soon...
I intend to build a channel as the original, and then fudge around with another channel to see how it goes and have a reference.
I am really open to any comments, advise etc... I am not too sure my calculations for the resistors are correct. :
Luc

 

[emrr]

Build and listen to the stock circuit first.

I despise the 12AX7 in a mic preamp.  Personal problem.  Never heard one I preferred.   The gain is dictated by the feedback path more than the tube type.  There's probably no gain advantage to be had with a 12DW7, at all, though I like the use of 12AU7 as driver. 

Your transformer impedances are wrong, if you don't know it.  More like 50K input sec and 30K output pri.   The stock transformers don't require a zobel network; have you seen reason to use one?

Are you sure the ECC99 won't eat the output transformer with your plan?  In other words, have you compared current draw through it?  If the ECC99 is running low current, will it have any advantage?  That thing sure has way more potential dissipation than that output is rated to handle.  I suspect there is no significant distortion to be reduced via tube change.  The feedback is high and the transformer distortion is probably the dominant quantity.    Why the 10K grid resistors on the ECC99?  Recommended with that tube?  Cathode bypass cap on ECC99 will probably have to be a lot larger to be as flat as the stock circuit without bypass. 

Why the 2M references to B+?   Balance control you've not implemented would be pot in cathode of ECC99 for current balance in output.

The additional filter cap stages look like overkill.  Be sure to compare with and without.  I'd not be surprised to find them pointless. 

You'd have to say a lot to convince me that DC bias would have any practical advantage over cathode bias. 

P1 variable 10K; gain control will be in competition with the feedback loop, and they will somewhat counteract one another in negative ways.  I'd do it in the feedback loop.   I don't know, but suspect use of P1 imbalances the balance of phase splitter V1B.

I'd play with the stock circuit via smaller changes and variable feedback options for gain control, if you've not. 

Good luck!

 

[blue_luke]

Hi emrr! Thank you for your reply. That is exactly the kind of forum interaction I am looking for
I need to learn, and I am here for that.

I am not after changing everything just for the sake of modifications. Each mods has to bring something to the outcome. And your suggestion to build it as close to the original is very sound indeed! After all, who am I to dispute the design decisions of the original RCA engineers! These guys realy knonwed what they were doing.
And like I said, My status as a tube based equipment design is strongly in the "knows enough to be dangerous"  
Having said that....

Your transformer impedances are wrong, if you don't know it.  More like 50K input sec and 30K output pri.   The stock transformers don't require a zobel network; have you seen reason to use one?

I stand corrected here, I have not checked my transformers and yes I don't know it. This was based on a post I have seen somewhere about these, promise I will get a bit more rigorous and make a better assessment of them...

Actually, I wanted to take your post on a point to point basis, but I find that you gave me quite a lot to think about. So it seems I have a bit of homework to do.
Meanwhile, the chassis, with wathever version, will be quite the same. So I am forging ahead with the construction of the chassis and the power supply unit.
I will keep you posted and thanks a lot,
Luc

 

[emrr]

Hi Luc,
I'm posing a number of questions that I do not immediately know the answer too either; mainly to see if you have specific reasons, and what they are based on.  It is an interesting redesign, and you should probably follow through with it and see what you learn (and can teach about it).  The chassis should pose no problems for tinkering and redesign.  I think there are some things you will abandon after you try it, but I cannot say for certain.  There will be those here who can give a fully theoretical comparison, while my observations are based more on a lot of hands on experience with these classic type circuits, and the general trends and reasoning behind the original designs.

If I were doing a redesign, the first thing I'd do is switch to octal tubes.  I think they most always sound better than 9-pins, and you'll likely get lower distortion from that alone. 

 

[blue_luke]

If I were doing a redesign, the first thing I'd do is switch to octal tubes.  I think they most always sound better than 9-pins, and you'll likely get lower distortion from that alone.

Hmmm?  : What tube would you suggest?
By the way, I digged a bit further the issues about using a ECC99. First thing I realize is that this tubes will need 400ma for the heater. My reasoning in using this tube comes from this article.
Although I did not really think of the output transformer rating per se, I did not think about using the ECC99 at its full 5 watts power rating! We are driving a +4db standard broadcast ouput here.

My reasoning behind the use of a 12DW7 is that we all know the 12AU7 makes a very clean phase splitter. But the reason for the 12AX7 equivalent section is to get more gain frok the preamp. 40db's is fine with condenser mics, but I have a nice vintage RCA BX44 junior that I would use more often with a tube pre-amp.

Is the gain simply R11 / R1 ratio, notwithstanding the gain of transformers ? In this case, a 3 to one gain plus the gain of the output secondary feedback winding. I will have to test more these transformers.

Luc

 

[analag]

Why go through all that work to use a higher value cathode resistor on the output tube and then turn around and cancel the benefit with all those caps.

 

[emrr]

I'd build it with 6SN7's, or 6C5's if I was free to choose.  Be interesting to see how close the gain comes out with same feedback network. I don't want to do the math, nor do I have it in my head.    Probably very close.    I'd also not be surprised if using a pair of 12AU7's with suitable changes sounded better than the 12AY7.    A 12AX7 in practice will give no more gain than the 12AY7.  The feedback will compensate and cancel most of the difference.  Make the gain changes in the feedback path with the best sounding (to you) tubes, rather than worry about the theoretical tube gain, which is calculated with an (impossible) near-infinite load value.  Look around and you'll find detailed explanations of real world tube gain.  Author PRR most likely. 

Actually, if I were free to choose tubes, I'd not build a preamp with a tube phase splitter.  They rarely sound quite right to me.

The charts from the article aren't loading for me.  I'm making the assumption that the ECC99 is overkill here, and will not be operating at it's best point with this output transformer.  Being a mini power tube, it's distortion will probably actually be higher with the light load presented by the output transformer.  Generally, power tubes distortion is lowest at rated load, and rises with increased load.

The stock gain of 40 dB takes into account 20 dB of feedback, according to the manual.  This means that if the feedback is removed entirely, the gain is 60 dB.  60 dB may sound terrible; at very least it will be much more euphonic (distorted).  That amount is controlled by r11, with r12 and c5 providing high frequency compensation.  Change R11 to change gain, be aware of what r12/c5 do with any changes.  There will be an acceptable range of change, which could be a rheostat.  I've made these sorts of changes with other RCA preamps, but not this one. 

 

[blue_luke]

My head is spinning now!  Fun with electronics where you can't decide what to do...

Anyway, I did some more work on the chassis yesterday, should be finished by today.
Then I will build the PSU section and test the input transformers to see if the zobel network is required.

But I still did not get answers to my question about using a switching power supply for the filaments.
My concern is about getting high switching frequencies beat with audio, creating IM distortion.
If going trough all the efforts end up sounding like an AM radio then I will give up the idea

Also I had this question why it seems everybody uses 6volts on the heaters @300ma instead of 12V @ 150ma (typical) if the heaters are floated at say 35-50 Vdc above ground?
If I remember well, that's how it was made in the RCA power supply.
This supply had a 300Vdc regulated output and used a separate transformer for the heaters that was referenced to a pin on the power supply.

Meanwhile, I will follow your suggestion and build at least one channel of preamp stock and fudge around with it.
Maybe I could try also Jakob's site for the  "DISA 91B21" in the schematic section.
The green ancestor's??  But on to the BA 21 first!  ;D

Luc

 

[emrr]

Hi, I missed the part about switching filament supply.  If you have one, try it.  Add additional filtering if needed.  If you don't have one, seems overly complex.  A standard filament transformer with the biased center tap works fine.  Don't chase noise you haven't heard yet!  ;D

The Zobel; the preamp is flat, so.....is it needed?  There are a lot of people out there who claim to hear and dislike the 'sound' of a zobel.  I never think about them, and have never added one to a vintage preamp.  I'm happy to ignore that problem, if it's even there.  I do notice many modern transformers seem much more sensitive to loading, and do seem to require them.  I don't know what the difference in design would be, to cause the differences.    Many older transformers seem designed for a direct grid connection, most new ones for a specific load network.    Some older companies designed units in both loaded and unloaded versions. 

 

[blue_luke]

About the switching PSU...
I just decided against it, figuring that the power transformer I have will do.
It is an EDCOR I had made for me for a tube microphone power supply. 250VAC CT @100ma and 25V ct @2Amps. No this is not for the sake of grossly overbuilding, the units can power 6 mono mikes and a stereo one, all with remote selectable pattern. The filaments use a constant current source.
If I use the filaments at the 12,6 volt rating then 8 tubes X150ma = 1,2 A with some reserve.
Is there a point in going with DC filaments, or floating the AC at, say, 35-50Vdc gives the same results? ???
Whatever, another issue to ponder... then I can also use this secondary with a voltage doubler to make the phantom power supply, 4 channels = 80ma about....

Many older transformers seem designed for a direct grid connection, most new ones for a specific load network.    Some older companies designed units in both loaded and unloaded versions.

True, but now I am curious! Thanks ERRR, the more we discuss this, the more you are convincing me to stick to the original, but I will try out things nevertheless, this how we learn hey! (I am Canadian, and a french one at that!!  hee hee!)   ;D ;D

Now I am looking where I can put a volume control... Obvious place seems to be between the input stage and the phase splitter, 1 meg pot to ground after the capacitor, and then the wiper to the grid of second stage. Any comments?
Also, while we are at it, I would like to put a high-pass filter, so I guess the best place is also between the first and second stage, by switching different values of coupling caps should do the trick, 6db/oct.
I have some microphones that I can probably use as seismic sensor. Last year we recoded an organ concert in a church during a snow storm and you can hear the rumble of the snow removal teams on the street, afterward we hear distincly a Honda with those coffe can type muffler passing by!  :-\
So a selectable 80/160 hz filter would have helped quite a lot here...
Luc

 

[blue_luke]

Just designed the power supply section.
There is nothing new with it and I have tried this topology numerous times to know that it works great. 
Just never tried all of this together!

Here it is

The high voltage secondary is used full wave. Should give about 335V after first filter caps.
I am using this cap value simply because I happen to have them on hand
My rule of thumb to evaluate filter capacitor value is to allow 2,2uF per mA of current. An old article from Don Lancaster I read in popular electronics in the 70's... never forgot it!
The transfo is rated 100mA, so 150uF is somewhat short, and 300 is a bit overkill but like I said, I have them!!

The 10nF caps across each diodes of the bridge are there to minimize switching noise. I am a bit cautious when it comes to RF and other noise filtering. All my preamps are used on stage eventually, near dimmer packs, RF sources such as cell phones etc... and very often poor electrical installations, especially in churches where I record choirs and organs. I have seen often old plug without a ground prong!! Not even legal anymore, I can moan and cry  :'( , but in the end, I have to deal with those situations the best I can anyway. So UPS, EMI filters and a lenght of #12 wire with a clip at one end that I can use as ground, connected to a cold water pipe.

I have not specified the value of zener Z1. Should be around 285Volts in order to get about 280Vdc at the ouput. I am not after a super stiff regulation here, just a way to drop the raw voltage. Resistor R3 will slowly charge C9, thus smoothly ramping the final B+ to the output filter capacitors.
The diodes around the mosfet are there to protect it by ensuring that the gate will always be a few volts below the source.The diodes gives a positive path for all these filters to discharge.
The mosfet will have to be mounted on a heatsink. (A 40some volts to drop @100Ma... 'bout 4Watts :)

R3/R4 forms a voltage divider to create a DC bias for the tubes heater filaments. C24 smooths it out.

The heaters are placed equally on each sides of the tap. Altough I mentionned that this transfo had a 25V output, I just found it is 24V ct @ 2 Amps. Will do just fine I am sure.

This winding also feeds a voltage doubler isolated with capacitors (C16/C17) so now I can reference this section to ground. Simple filtration and a Texas Instrument TL831 regulator, I'm lazy 
R9 is there just to discharge the section if no microphones are connected.

Some components values are not written yet, and some values are just approximation that will need to be either calculated more rigorously or experimented with.

 

[blue_luke]

Now I am looking where I can put a volume control... Obvious place seems to be between the input stage and the phase splitter

Silly me! Of course it can not be there! because it would be within the feedback loop! dooohhh!

Between the input transformer and the first grid. Nice place also to place a DI jack.

But I still think a good place for the bass rool off is between V1a and V1b

Luc

 

[KrIVIUM2323]

Hi Luc and EMRR. 

Luc said:

Now I am looking where I can put a volume control...

Why not using a passive constant impedance attenuator in Front of the input Trannie? And by the way applying the Phantom at the same stage.

I remember an other problem about the +48V with an older trannnies ( in a V72 Telefunken): it didn't like at all the phantom supply without some kinds of protection against it. Don't know about the rca's but we should take care...

 

[blue_luke]

Here is some XXX photos  ;D

The bare chassis A few more holes to go
The main chassis is a piece that came from an old tape recorder. It is 16 3/4" X 10" with a 1/2" lip bent all around it. It makes for quite a sturdy base frame. I use it with the lip upward.
The rear apron is made from a piece of extruded aluminium U channel, 1/2" X 1 1/2" That I get localy. I find this extrusion perfect for making cases 1 unit high. They also have a 3" high U channel that is perfect to make 2units high frame

Underneath  Missing a few switches... Gonna get'em today.
You can see the EMI filter/DIN power entry unit and the two big switches are for "power"  and "standby".
Admittedly too big, but hey! I had them!

Rear view with main components installed.
The small grey sub-chassis will contain the B+ section. Not to shield it from EMI but to protect my self against sudden large flow of electrons trough my body!  :-\
The 4 Neutrik mic inputs (the first one is crooked! darn!  ) The ouputs are on TRS jacks. The switches beside each jacks is to switch on or off a 600 ohms load on the output. This is very important as most modern electronically balanced line inputs are 10Kohms load, and that change the sound tremendously.
In the same vein, I am considering installing a selector switch on the input to select either a 150 or 600 ohms input impedance. I wil try it first to see if that change the sound on some of my microphones.

Luc

 

[Larrchild]

Assemble your power supply with long leads on the power transformer, and allow enough room for it to be mounted at an angle, as unsightly as that may look.. If you get induced hum fields, you can rotate it for a null.

 

[emrr]

Lots to address. 

Chassis looking good!  Looks almost like 4 of the original module footprint drowning in a pool.  ;D

The 600 ohm load may not be necessary here due to the high amount of negative feedback.  I have found similar modules just lose a touch of treble bandwidth and a very small bit of gain. 

I have never seen a vintage american transformer that had problems with phantom power application.  IMO that is a purely european problem. 

You cannot wisely put an attenuator other than a basic 20 dBish pad between a mic and an input transformer, due to loading reasons.  There must be thousands of pages of explanation to be found on the topic.    Put at least part of any gain control in the feedback loop.  I'll try not to say that again.  Attenuation the input trans is tricky due to lead length causing loss of highs, which can increase greatly after -6 dB.  DI insert also has the lead length issue.  Anything between input trans and 1st stage needs to be laid out as close as possible; already maybe too late for that.  Looking again you may be close enough to the input for this, but the run from the front panel to the 1st stage is of concern, as is the proximity of the power switches to the closest channel input.  May want another shield plate in there. 

Bass roll-off maybe with selection of cathode bypass sizes on V2, but it's more of a shelving cut with limited attenuation range.  I'd maybe try to put it in the feedback path.    I am not fully clear on this one point, but I think roll-off between v1a/v1b will also fight with the feedback loop.  If you go with variable feedback gain there may be unavoidable interactions at any place.    I've not played around with that before.  It depends on what your other recording tools are; I have serviceable software HPF, and would never consider including one in a circuit as simple as this.  Preset HPF selections are never exactly what you want either.

Filaments.  Your plan has no individual filament center tap reference to ground to cancel hum; one side of each filament is at the center tap, and the filaments each see 1/2 of the 24VAC wave.  At a glance it appears impossible to reference CT, which tells me you need to go DC with this transformer.  The bias plan won't help with hum, it just helps reduce other noise with AC filaments. 

B+; you don't really need regulation here since the load is essentially constant, and line fluctuations present no significant problem.  You would be fine with an old school unregulated B+ controlled with series dropping resistors.  Go wild and use a variable tap resistor to trim the drop.  But reg is fine too, if you wish to build it.  Regulated does present more opportunities for parts failure. 

I'm sure I've overlooked something, and said at least one wrong thing in haste....

 

[blue_luke]

Thanks emrr...

All your comments make sense and I find it so frustrating sometimes to know "just enough" to understand these thing but not enough to see them by myself.
I am definitely not at ease with tubes, but the learning is mighty interesting

The filter I envision are just that, 6db shelving at say 40-80-160hz, to remove the low mud. I have a C12 and some modified APEX460 that are real happy with infra bass!! I prefer filtering right at the mic, but these microphones don't have filters anyway.

I can do whatever I please with op-amps, including some realy nice preamps I built over the years and I am sort of blasé with them, but tubes, despite their deceptively simple circuits, are really different beasts!
If I count only the parts between input and output transformers, the whole thing works with only 4 triodes, 12 resistors and 5 capacitors!! 

It's not too late for anything, and my plans right now are to build the two first channel as per my reduced schematic, and keep one as reference and fudge with the other to find out about all these questions.
I don't really care if the front panel ends up looking like if it was shot with a machine gun. If all goes well I will order an engraved frontpanel later.

Also, I am mightily interested by the DISA pre amp on Jakob's site (see attachement) I think it's easier to modify and it could use the tranfos and the layout I have already. It already has a gain select switch and I guess roll-off filters could also be added, but I don't want to chase too many rabbits for now, I will stick with the RCA pre's and may start another thread depending on what I decide to do later....

 

[emrr]

I hadn't notice the gain method in the Disa; only other thing kind of like it that comes to mind is the McCurdy AU300.   You could use the BA-21A trans and modify the feedback path to use the tertiary winding.  Would be an interesting experiment to compare feedback methods.

Here's a possible problem  The BA-21A input may look terrible if used in a push-pull configuration.  You'd want to measure first. 

 

[blue_luke]

OK, I will go with dc on the filaments.
But this raise two questions....  ???

I have seen often where the 6,3Vac is simply rectified and filtered and then applied as is to the heaters.
In my case, I would use bridge with the center tap to ground and the DC from the + of the bridge, filter cap  etc... That will give me about 17 Vdc ( pocket calculator somewhere...) Isn't this too much for the filaments? I will need 150mA X 8 - 1,2Amp total. Should I use a regulator? Or the raw DC is fine?

Second question, if I use DC, should I still have to float it above ground or I can have one side of the heaters to ground along with the center tap of the transformer??

Luc

 

[emrr]

You need to get the filament voltage within spec.  Too high will not be good.  Look up filament tolerances.  Usually +/- 0.3V. Several volts under is purposefully done in some microphone designs, and generally only creates a longer heat time.  But not always.    Again, pretty much constant load, so use dropping resistor or regulator if you must.  I'd use big fat resistor to drop in filter path.  You need a ground reference because of that little spec called 'heater +/- voltage with respect to cathode'.    Bias unnecessary with DC.