Question about various topologies for tube preamp

[Matt C]

I'm starting to think about a tube-based mic preamp build for the studio, and I'm finding myself lost among the many options.  I'm wondering if anyone out there can comment on the different tonal characteristics of a few different preamp topologies.  Specifically I'm wondering about:

1. Output transformer primary in series with tube plate (as in RCA BA-2C)
2. Fully balanced, push-pull, I'm not sure the proper way to describe this one (as in Langevin 5116b)
3. Inductor as output plate load, capacitor coupled to output transformer (as in V72 or RCA OP6)
4. Plain old R-C coupling to output transformer
5. triodes vs. pentodes vs. pentodes wired as triodes

Ideally I'm looking for something that can provide plenty of clean(ish) gain, but also can be cranked up to get significant distortion.  That's what makes me think a single-ended design with no negative feedback, like the BA-2C, would be the best choice.  But, I'm hunting for opinions among those who have actually used this equipment. Anyone care to point me in a particular direction? Or am I overthinking it?

 

[ruffrecords]

I'm starting to think about a tube-based mic preamp build for the studio, and I'm finding myself lost among the many options.  I'm wondering if anyone out there can comment on the different tonal characteristics of a few different preamp topologies.  Specifically I'm wondering about:

1. Output transformer primary in series with tube plate (as in RCA BA-2C)

This means the transformer has to be able to carry dc. This generally reduces its effective inductance and hence bass performance can suffer. That is why these tend to be big boys. Advantage is tube plate can in theory swing up to twice the supply volts giving 6dB extra headroom.

2. Fully balanced, push-pull, I'm not sure the proper way to describe this one (as in Langevin 5116b)

Similar issues as single ended but both tubes can swing to twice supply volts so a theoretical 12dB extra headroom.  Also plate dc currents cancel out in transformer so inductance much less affected and bass repsonse therefore better.. Push pull cancels even harmonics so the tone is very different.

3. Inductor as output plate load, capacitor coupled to output transformer (as in V72 or RCA OP6)

Inductor plate load is identical to single ended really. I suppose you can optimise the inductor for its job. The transformer, having no dc through it, is easier to design.

4. Plain old R-C coupled output stage

Tubes have relatively high output impedance so you usually need a transformer at the output. RC coupling plus output transformer is what I use. Transformer design is much easier because there is no dc but you don't get the headroom advantages of the other methods. Common topology in tube mics. For a line level output (up to 20dBu into 600 ohms) you need to do something smart with the output stage.

5. triodes vs. pentodes vs. pentodes wired as triodes

Pentodes (wired as pentodes) are noisier than triodes which is important for first stage of mic pre. Pentodes wired as triodes act like triodes. Triodes produce mostly 2nd harmonic and some 3rd harmonic distortion before they clip. Pentodes (wired as pentodes) produce mainly odd harmonics. The output impedance of pentodes is much higher than the output impedance of triodes which can be an issue with output stages.

Ideally I'm looking for something that can provide plenty of clean(ish) gain, but also can be cranked up to get significant distortion.  That's what makes me think a single-ended design with no negative feedback, like the BA-2C, would be the best choice.  But, I'm hunting for opinions among those who have actually used this equipment. Anyone care to point me in a particular direction? Or am I overthinking it?

Negative feedback (NFB) will keep the gain clean for longer but once it clips it falls to pieces. However, in a good design this will not happen unit you reach about +26dBu and by then your soundcard has given up the ghost anyway. No NFB will distort more and earlier. In triodes, distortion is directly proportional to output level. So if you want a clean 0.1% at normal setting, you will need to crank up the level by 20dB to reach 1% distortion.

Some topologies distort more than others. A triode mu follower or CCS loaded stage will produce the least distortion. Using a regular plate resistor will increase it and using an SRPP topology will increase it even more. You can alter the plate voltage and/or the bias pint to change to amount of distortion and the rate at which it increases (Culture Vulture for example)

Cheers

Ian

 

[abbey road d enfer]

In addition to all Ian wrote, you may note that, although making the design of the OT easier, choke load makes the design of the choke almost as difficult as that of a SE transformer, because it has to pass DC and still have high inductance (in fact significantly higher because it in in parallels with the xfmr's primary inductance) and low stray capacitance.
Regarding triode vs. pentode the only benefit of the latter IMO is that it has much less Miller capacitance, which makes the design of the input xfmr easier; for the rest, they are noisier and distort more agressively. The Miller effect issue can be solved beautifully by using a cascode input stage.

 

[ruffrecords]

In addition to all Ian wrote, you may note that, although making the design of the OT easier, choke load makes the design of the choke almost as difficult as that of a SE transformer, because it has to pass DC and still have high inductance (in fact significantly higher because it in in parallels with the xfmr's primary inductance) and low stray capacitance.
Regarding triode vs. pentode the only benefit of the latter IMO is that it has much less Miller capacitance, which makes the design of the input xfmr easier; for the rest, they are noisier and distort more agressively. The Miller effect issue can be solved beautifully by using a cascode input stage.

Yes, I have never quite understood the attraction of a plate choke followed by a capacitor coupled output transformer. I cannot see an advantage.

A cascode does indeed overcome Miller capacitance problems but it does distort a lot more than a triode alone (but probably not as much as a pentode).

Cheers

Ian

 

[abbey road d enfer]

Yes, I have never quite understood the attraction of a plate choke followed by a capacitor coupled output transformer. I cannot see an advantage.

Neither can I; maybe PRR could chime a wise bell...

A cascode does indeed overcome Miller capacitance problems but it does distort a lot more than a triode alone (but probably not as much as a pentode).

Indeed; I would tend to use it with a reasonable amount of NFB. In fact, the bottom triode would distort less than the top one, that does the heavy lifting. For a given output level, I don't think there would be so much difference, but a cascode stage without NFB would probably operate at a hotter level than a basic common-cathode stage, just because the gain is much higher.

 

[Heikki]

Yes, I have never quite understood the attraction of a plate choke followed by a capacitor coupled output transformer. I cannot see an advantage.

Advantage for DIY might be lower price if cheap choke like hammond 156C is used as a plate choke. 156C and a ungapped output transformer  will probably be cheaper than a decent SE transformer

 

[abbey road d enfer]

Advantage for DIY might be lower price if cheap choke like hammond 156C is used as a plate choke.

Illusory, IMO. I've often seen this opinion sacralised by a comment that the choke is not in the signal path, which is utterly wrong. In practice, performance is that of the weakest link.

 

[Tubetec]

Ian has summed things up nicely there,

I tend to stick with single triode preamps, of relatively low gain  with a high value grid resistor ,you can always add an extra  gain stage at the input if more gain is required , to much gain is a pain when not needed,you either have to attenuate or use NFB to get things into the right range .Only time Ive used a pentode is in a baxandall tone control which is high nfb, german postal tube C3G was what I chose there.

I also usually keep my input transformer seperate from the pre ,that way its easy to select the transformer on the basis of what step up ratio/impedences are required. Housing the input transformer seperately reduces any chance of induced currents in the case of on board power supply it also allows free orientation to minimize hums,  its more a table top setup ,a convenience penalty for some maybe.

theres a pic of my E88C(eqiv WE417 ,EC86) based pre,Resistor load ,cap coupled from the anode to the output transformer,NFB from secondary to lower end of grid resistor ,pot to vary amount and gain. Battery bias on the cathode.Ptfe tube socket acoustically  isolated on rubber motor mounts ,and connected with litz cord.  power supply housed in an old pc smps box , regulated dc heater, 7805, two diodes in series with gnd pin to get 6.3v, Ht passively smoothed RC<RC<LC<LC, potted chokes, virtually all components scavenged/reclaimed , cost next to nothing apart from time.Noise/bandwidth/output drive capabillity slightly source and load dependant but well beyond whats required for most sources.

 

[Tubetec]

Innards pic

 

[Tubetec]

First attachment not found ,appologies,

 

[moamps]

Thread about parafeed:

https://groupdiy.com/index.php?topic=59853.0

 

[Tubetec]

Ive been wanting to try the choke load /paralell fed cap to output transformer for a while now , havent got around to it yet .
Was considering a large, maybe 50-100 henry C core choke, from J&K audio and a sowter 8940 output .
Would the choke load not give the same +6db output headroom as mentioned by Ian in the transformer loaded arrangement?
I remember reading about the old BBC tube consoles ,this choke /parallel fed  transformer arrangement was featured on the output of some of these units ,they seemed to be saying the gentle overload charachteristic was basically used instead of limiters to help prevent over modulation of the transmitters. 

 

[Tubetec]

Plenty of good stuff in the older post on parafeed. Could anyone explain how the winding geometry would differ between a regular mains choke ,and a choke specifically designed for plate load ?

 

[Heikki]

In practice, performance is that of the weakest link.

Who here measures the performance of the equipment they build anyway.

 

[Tubetec]

Let the ears be the final arbiter of quality,not the eyes .

 

[emrr]

Who here measures the performance of the equipment they build anyway.

Probably not many, many more who probably don't really understand what they've measured. 

 

[emrr]

Let the ears be the final arbiter of quality,not the eyes .

It is amazing what terrible response can sound fine if you never compare it against a known.  You can be 12 dB down at 50Hz and 20kHz and think it's all good.  If you don't believe this, it just means you haven't noticed it or experienced the phenomenon yet.  The brain adjusts and normalized very quickly. 

 

[Tubetec]

In saying what I said above ,of course I dont mean to underplay the importance of proper scientific measurements and testing .However with simple tube and transformer coupled circuits analising about  THD numbers doesnt give the full story, the balance between lower and higher order harmonics is more important.
I most cases a simple a-b test with a quality amplifier of known perfomance is going to reveal if any major anomalies .
I havent done any in depth measurements on the above pre yet ,due to the puny gapped output transformer I do of course expect a drop off in the bass,excluding sources with seismic low end ,it might not be a problem and could even be beneficial with certain things.Why bother with the extra complication of a  HPF when you could just pick a pre off the shelf with the requisite characteristics built in .

 

[Bo Deadly]

Negative feedback (NFB) will keep the gain clean for longer but once it clips it falls to pieces.

Why don't tube pres use the old fuzz pedal trick where the emitter / cathode resistor is actually a potentiometer with a large cap to ground on the wiper. So the pot bypasses the cathode more or less which adjusts the NFB and thus the gain. This is nice because at low gain the circuit is more linear / cleaner and at higher gain settings it doesn't "fall to pieces" so much (albeit at the expense of higher distortion).

And you can get a lot more gain if the second stage can be a little dirty. If the first stage uses a little NFB it could make 20dB no problem and the second stage with the pot cathode bypass trick could make another 30db and a good input transformer should be good for 10dB giving a total of 60dB.

But I know nothing of tube pres if you can't tell so maybe I'm just thinking-out-loud here.

 

[abbey road d enfer]

Plenty of good stuff in the older post on parafeed. Could anyone explain how the winding geometry would differ between a regular mains choke ,and a choke specifically designed for plate load ?

A mains choke is meant to attenuate the ripple of the rectified voltage, so it must optimize LF inductance.
A plate load choke is in parallels with the load AC-wise; there are three main consequences:
The LF response is not as good because the resulting inductance is lower.
Distortion due to iron saturation will be superimposed to the signal.
Stray capacitance acts as a treble bleed.
A plate load choke must have a very high inductance, higher than the transformer's inductance, it must operate at low induction in order to minimize saturation, and it must use a special winding technique in order to minimize stray capacitance.
In comparison, a mains choke is just basic winding.