• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Two driver options for 300B push pull

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hard to say. I looked at 807, 811A, TT21(KT88), and 211 grid current curves. and they are all over the map. 211 can get +50V or so with only 10mA. 807 requires up to 75mA for +30V and TT21 is up to almost 90mA. 811A was 50ish mA for +30V. I think that higher gm grids, being typically closely spaced to cathodes, pull more current. 300B is pretty low gm so you may be fine, or you may run out of steam before the 300Bs reach saturation.

It sounds like you are happy with the current power output without grid current, though so maybe this is a non-issue for you. It might be worth just trying it with and without a cap bypassing the CCS and watching on a scope to see if clipping is less abrupt with the cap.
 
Update:

I have been doing a boatload of mods and testing using the existing chassis of the amp that will be rebuilt. I have managed to get ridiculously low distortion (at least compared to any other build I have made) with a topology that is kind of a hybrid between the two original ideas.

I did some testing with the Monolith IT-02, which is a custom PP gapped interstage. Very pleased with the results, so will be using one per channel. Managed to get rid of the first IT, and the input transformer (as long as I use balanced from preamp). So reducing the amount of iron, but continuing to use one for the 300B grid drive, is a win in my book.

So I have nearly finalized the design. I can share a schematic and FFT shortly, but I wanted to resolve one last issue.

Attached are two options for the first stage, one with resistor and one with LED. I wasn't finding a significant difference in the measurements, so I am wondering on a theoretical basis what is the fundamental difference between these two topologies? If I have differential in, differential out, does it really matter if I have a lower AC impedance in the tail?

Ignore any small differences in grid/plate voltage.
Thanks.
 

Attachments

  • diff-amp.png
    diff-amp.png
    68.1 KB · Views: 791
The way you are using them, there is very little difference between the two.

If you had a bit of a mismatch in gain between the two triodes in the tube, you would have a (tiny) difference in behavior between the two circuits, but not much. The load you are driving is pretty straightforward. The way you are driving it is straightforward as well.

The difference between the two circuits will become more apparent if you try to drive a load that requires more than twice the idle current at its peaks or if you had gross mismatch between tubes. You could play with a simulator if you want to see this in action.

A voltage source at the cathodes isolates the two sides to that they function independently and form a push-pull circuit. The defining feature of a push-pull circuit being that you can deliver more than twice the idle current to the load if necessary. This circuit is most sensitive to gain mismatch between the two triodes.

A current source at the cathodes makes a differential circuit. Balance is strictly enforced. One side is the exact compliment of the other at all times as long as loads are equal. You can only deliver twice the idle current to the load at most at any instant. If you try to do more, things tend to get ugly.

A resistor is somewhere in between a voltage and current source so you get a mix of the two effects.
 
Yes, that makes sense and intuitively I was thinking the LED's would get me closer to a push pull rather than differential. Since grid current of the next stage is effectively an impossibility, I have no concerns about drive capability.

I will stick with the resistor to K.I.S.S.
 
Zigzagflux,

What was your final version / measured and listening results finally ? I enjoy reading your initial development thread with comments from Lynn and others...I am currently intending one myself and not sure which path to follow (karna vs. Kevin)...

Your experience is highly appreciated...
 
Blitz,

I have attached the final version without power supply details. All direct heated triodes are heated with DC current sources, which provides a really low noise floor.

Major lessons I have learned:

1. Avoid a transformer between the input stage and driver stage. No matter what kind of transformer you install and no matter what topology (SE/PP), you will have a bear of a time getting distortion as low as with a cap-coupled design. If one is that averse to capacitors, then just accept the tradeoffs. I haven't found the teflons to be all that horrible. Distortion on the first stage of this design is 0.01% at every freq and every power from 0.1W-25W output at the speaker. Good FFT spectrum, too. It just does not change. Really hard to complain about that, capacitors or not.

2. Get the driver stage running warm, as much current as you can confidently provide. Matching tubes certainly helps here, but doesn't make or break the sound quality subjectively. It does begin to run out of gas above 15W, but for my system that is really doggone loud, and the speakers are probably distorting more than the amp anyway.

3. The output stage is consistently the weak point until you begin to require grid current (well above 15W), and even then the 46 does pretty well. Your only option would be to select a different driver or output tube, and this is not worth it to me when I listen below 3W. Removing the CCS in the driver stage helps driving grid current, but did not provide as low distortion at 1-3W levels.

4. Interstage transformer made a big difference, and the bifilar Monolith was the mac daddy.

5. This build ain't cheap.

Subjectively, I won't go into much rosy language and claim perfection has been reached. But this is likely my last amp, it is that satisfying. Any others I build will be for fun or for friends, but I do not plan on attempting to improve this a third time. Everything is just 'right' about it. It's not gushy and smooth, it's just correct and strong. Drives my Ariel speakers with ease (high passed actively at 120Hz). Zero noise, like absolutely nothing coming out with the ear right at the cone. Can listen for hours without fatigue, which is really important to me.

I have made numerous PP amps, and only one 300B SE (which I still have), and this one easily takes them all. The SE is definitely more smooth; you can tell it 'adds' something to the music, albeit the addition is quite pleasant. The SE also runs out of gas much more quickly than the PP, so depends on how loud you listen and what you want to hear (or experience). I appreciate them both. The SE and this PP (and only this PP) are the only two amps I have been able to listen to for extended periods without fatigue. That is one of the best ways I know of to judge an amp; it's too difficult to compare A/B. You either like it or you don't.
 

Attachments

  • 300bPP.pdf
    15.8 KB · Views: 441
I will provide you with some backround information what I try to achieve, but I have the impression that your preferences match very much mine....for the musical abilities of a system. I want resolution and the right tone/ authority, so that ideally you have the musicians in front of you, but the resolution is so good, that you can see their idea of interpretation and stop thinking about hifi and features.

To get there, I have build a pair of line sources with 12 Fountek Ft-2 ribbons and 18 4"Eton Kevlar drivers per side. 95db or more. But my Welbourne labs SE with 300blxs / 20Watts SE could notdrive them. My El34-PPUL drives them like a charm, even when they run sometimes only at 320V/35mA (happened as an accident when I played around with different tube rectifiers and had a too big voltage drop).

So, I guess 15W of real class A Triode in PP should be good enough, but need to try.

I got some time ago many of the ingredients: Monolith IT like you, OPt from Tribute (from a different project with 2k7 as I intended to go with EML 52), Some lundahl anode chokes, coleman regs, big, heavy power transformers from AE...

So far I thought to use Kevin Carter's input/driver section, so 6sn7 directly coupled to 6bx7 with IT to the 300B. Followed by the Karna-Output-Section. And test than different output tubes like 814a, 300b, emlkr52v2 or as well triode-strapped kt150 etc...OPT from Amplimo are there in 5k8, 2k7 from Tribute with AM core at 2k7...but if there is no other choice, I will order another pair of monolitz opt...

I thought after reading your report from 2009 and Lynn's article aboit the Karna and Thoms Mayer's design (vinylsavor): Maybe I want to upgrade the 6bx7 to a 45,46 or 801a (of which I have a nice inventory I bought 15 years ago)

So, first stage would have been Anode Choke like here http://kandkaudio.com/images/PPP_6550_triode_amp.pdf

...and ccs at the cathodes...

...but you went classical RC, which surprises me a bit as I would have thought a no coupling-cap-design would have benefits...?

Best Regards

ps. may I ask for your opinion on IT in line stages like the Raven or the designs of Thomas Mayer (vinylsavor) ?
 
Last edited:
I only have 1K grid stoppers on the SN7 stage, just didn't show them on the schematic. Given the very short and carefully trained wiring from the previous stage and the gm of the DHT's, there simply isn't a need for grid stoppers there. But you are free to use them wherever you like, of course.

I hope you will post your results of what you find with choke coupling. On paper, it is a very promising idea, as it eliminates capacitors, DC couples, and provides extra signal swing above that possible with CCS for a given B+. If you absolutely need the gain, it is a great solution. However, I will predict that you will not be able to approach the level of distortion of the CCS design when you look at all signal levels from 0 to full amp output, and also over the full frequency range of interest. Strays will begin to eat at performance, and the first stage triode selection becomes really important; not all tubes are up to that task, including the SN7 IMO.

I did in fact discuss this design with Kevin Carter a number of years ago when I was considering what topology to go with. What also pulled me away from the choke design was the DC coupling issue. I was now constrained to hand pick triodes (in both input and driver stage) that exhibited the right bias conditions for the task, in order to ensure relatively close current sharing. This is easy with the SN7 as I have over 100 of them, but I am not overflowing with 46's, and the ones I do have are not consistent. Then there is the question of bias drift over time as the tubes age. Tradeoffs, again. The cap design eliminates these concerns, and allows me to select tubes with the right match of gain/transconductance, for better distortion performance.

Final thought on cap coupling; it's worst application is in stages prone to blocking, when for example the load is prone to grid current. In the SN7/46 stage I have applied, there is zero chance of grid current. Both driver stage and output stage are completely clipped, and yet the input stage is still happily providing perfect output. So why the concern about capacitors there? I have none. Although I took the 'safe' approach and used a decent Russian teflon, I believe when properly applied (and this example is proper application), capacitors are largely invisible. I could throw in a V-cap or a $2 Wima film/foil, and you likely won't hear a difference. If I tried to cap couple to the 300B grids, like is commonly found, I agree that this would be a bad decision with audible results.

I completely agree that 15W of PP amp is enough to drive your speakers; mine are only 92dB, and there is ample headroom available.

If you want some opinions on output transformers, PM me instead. I have experience that goes against one of the popular mfgrs, and I would rather not start a firestorm on the forum and distract from what is otherwise a decent thread. I have already taken a huge risk by posting an opinion on capacitors...

I might have an opinion on Thomas Mayer's designs if I actually saw a schematic. I enjoy his blog, and am seriously impressed with his mechanical design, stealing some of his layout ideas. But I have yet to see a schematic or measurement of any of those amps. Maybe I missed something in the forum, but I don't believe it's available on his website. Just pictures and generalized descriptions; that doesn't help me.

Regarding the preamp stage, I can share my schematic if interested, but it is essentially a single 12B4A PP stage with Lundahl LL1689AM output transformer. My entire system is balanced throughout. People might cringe at the triode choice, but the low Rp and low gain is perfect in my application (don't need gain from my source, just volume control and drive capability). Crazy-low noise floor, with distortion better than the amp and good-looking FFT.
 
Yes, please do share the preamp schematic...I am building currently a new one as my DAC-output stage...so far I used the unbalancer from Broskie, but I want to go fully balnced now as well as my DAC is a Sabre 9018 which outputs a differential signal. Currently I use basically a LTP with a CCS on the cathode and a Lundahl 1692a as line-output with 4p1l in triode strapped....build finished last weekend and next weekend I will start to losten to it....
 
Attached. The Behringer is gutted of the output stage, and the transformers are nickel core. Direct from the DAC output pins to the transformer primary (NO resistors are needed!) and the the secondary is left floating to the XLR output jacks. As balanced as it gets.

The input is USB to SPDIF into the Behringer. Some day I might pursue i2s, but that isn't my biggest concern at this time. Benefit is I get customized EQ and crossover, without having to go A/D then back to D/A. All processing in the digital domain, and the source material is digital. No LP's for me, I had enough of that in my teens.

The preamp has a 6-channel ganged Goldpoint attenuator, which is the '50k' flags in the preamp schematic. So I get 6 channels of audio, 4 are tube for stereo mids/highs, and the 5th and 6th channels are for subwoofer using the SSM2019. Good application for bass IMO. Don't need tube there; my subwoofer amp is AB transistor anyway. With the volume control in this location, the burden on the Behringer DACs is reasonable, and they also run full scale output 100% of the time. Analog volume control is the way to go unless you have a whole lotta bits to spare.

That 4.7 uF cap in the output stage is a Wima film/foil, which does really well there. Also, I have an amorphous LL1689 stereo pair and a silicon steel LL1689 stereo pair, so can compare the differences in running the tweeters vs the midrange with amorphous. TBH I haven't played around with that yet; my Ariels are using passive crossovers per Lynn Olson design.
 

Attachments

  • IMG_3014.jpg
    IMG_3014.jpg
    993.9 KB · Views: 605
  • Schematic1.pdf
    19 KB · Views: 276
  • Schematic2.pdf
    16.8 KB · Views: 244
Did you ever build the version with the triode 6W6's? Just wonder how that worked if you did.
Thanks,
John

Actually, my formerly-referenced 300B SE amp is using the 6W6 in triode mode. It does quite well, methinks. Running SN7 SE into 6W6 SE into 300B SE. Would be an excellent replacement for the 46 in the PP amp, but perhaps a little more distortion. Rp is nice and low with the 6W6, not to be dissed. Really depends on whether you fall into the "DHT-only" crowd (not me) or if you respect the freedom in going AC heated and keeping it simple. I have a boatload of General Electric 6W6, kind of saving those for some creative project someday. Maybe a two stage PP flea amp?
 
Zigzagflux, so you preferred to have a Raven-Type of Outputstage over a true differential circuit with the ccs at the cathodes...I guess that you had very good reasons for that ? Did you compare the two settings ?

Currently I am running just a classic LCLC-Psu...but I like the glow of theshunt reg tubes....how much do they contribute to the overall quality of your amp ?

Best Regards
 
I did quite a bit of testing in comparing the two topologies, and really strained over the end choice to be honest. I found the differences to be very subtle, this is not a night-and-day difference.

I think there are two rules of thumb that one can use in choosing the optimal topology. When you are trying to drive a heavy burden, the RC design can provide some benefit in providing a path for unbalanced currents. For example, the SN7 to 46 stage is a fixed burden that does not change over the operating conditions of the amp (output power). So a CCS/diff stage I found to be optimal. The output stage, however, with the speaker load, benefits from the RC design, as that load is anything but predictable.

Second, I found that when using a push/pull to single output winding transformer, there was nothing gained in going CCS/diff. This would apply to the output transformer of the preamp. I wasn't really measuring a difference; it was a toss up. I chose RC there to accommodate line driving capability; the burden is not necessarily known, but depends on what you are driving and how far away it is.

The one stage that falls in neither camp of these rules of thumb is the driver stage. Being push/pull to push/pull, one could argue the CCS/diff is better than the RC. It measured that way at low driving voltages. However, when you began to approach some 300B grid current, the RC design began to perform better. So which is more important to you, performance at 1W or 15W? I chose 1W, for better or worse.

The gas regulators are used all over my amps. They are used in the preamp and first two stages of the power amp. They make a big difference in the noise floor; as the amp is warming up, you can hear the AC hum as the gas regulators are off and the supplies are rising. As soon as the regulators fire, the amp goes inky black silent. I would guess this carries the greatest benefit on the preamp and power amp driver stage. In theory, it should not matter to the SN7 stage given the CCS locations.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.