Hi:
Been using a 300B PP amp for a number of years now (actually it's been in storage for a while) and considering making some upgrades and possible changes. I have two options that I am considering. Currently I am using something very similar to Option 1, but without the current sources. Using the 40 uF cap to cathode in each stage with a single bias resistor in the tail. Also been AC heated with less than 1 mV hum at the speaker, but going to go DC heating as long as the chassis is getting rewired.
Ultimately, I am trying to determine the optimal way to eliminate the first IT, if there is a better way to do it. I have summarized my thoughts below, feel free to chime in. I know it may not be to everyone's taste, with a lot of iron, but these are the two options I am currently considering. If it makes you feel any better, I am relocating the XLR input to the secondary of the input xfmr, using it as a balanced grid choke. RCA will be wired to the primary in standard fashion.
Option 1:
Benefits-
a. 150V and 300V supplies currently implemented with CCS-fed 0D3 shunt regulators, so are extremely quiet power supplies with a pretty glow.
b. First stage fully differential with a single adjustment, set and forget.
c. 46 is an excellent driver for 300B, and seems to have no issues with the IT, but is a little pricey.
Deficits-
a. First stage IT adds some phase and gain shifts at the higher frequencies. A bifilar IT may help in this regard, but testing will be at a future date. Asking a lot from an input stage.
b. 46 stage will require the extra DC heat circuitry under the hood, with possible contamination from stray magnetic fields.
c. Driver stage has 40uF cap, which would be nice to eliminate. Could put another CCS in the tail, going fully differential on the driver stage, not sure if this would be an improvement or not?
d. Negative DC supply required.
Option 2
Benefits-
a. No negative supplies needed.
b. First stage should be very linear with CCS load.
c. DC coupling exhibits no blocking distortion.
d. 6W6 has 1.6K Rp, so should be an excellent driver for 300B.
e. I like octals
f. Tubes are cheap and easily hand selected by test. If I get bored I can always swap in some overpriced 6SN7.
g. AC heating for first two stages is much simpler, but xfmrs need to be added to the signal chassis.
Deficits-
a. Not sure how stable and reliable the bias conditions will be. The CCS are adjustable, so in theory I should be able to closely dial in the 150V to produce good current match in 6W6 stage (place small R in cathodes).
b. Sad to lose the 46 DHT as a driver, but it just doesn't bias well in this topology.
c. 375V power is unregulated with over 500mV ripple. Not an issue for the first stage, shouldn't be a huge issue for the PSRR of the driver stage, but a tiny bit does seem to get through in my bench test.
d. Would like more than 150V on the plate of 6N8S, but there's only so much I can do with the existing 375V supply while retaining DC coupling.
Been using a 300B PP amp for a number of years now (actually it's been in storage for a while) and considering making some upgrades and possible changes. I have two options that I am considering. Currently I am using something very similar to Option 1, but without the current sources. Using the 40 uF cap to cathode in each stage with a single bias resistor in the tail. Also been AC heated with less than 1 mV hum at the speaker, but going to go DC heating as long as the chassis is getting rewired.
Ultimately, I am trying to determine the optimal way to eliminate the first IT, if there is a better way to do it. I have summarized my thoughts below, feel free to chime in. I know it may not be to everyone's taste, with a lot of iron, but these are the two options I am currently considering. If it makes you feel any better, I am relocating the XLR input to the secondary of the input xfmr, using it as a balanced grid choke. RCA will be wired to the primary in standard fashion.
Option 1:
Benefits-
a. 150V and 300V supplies currently implemented with CCS-fed 0D3 shunt regulators, so are extremely quiet power supplies with a pretty glow.
b. First stage fully differential with a single adjustment, set and forget.
c. 46 is an excellent driver for 300B, and seems to have no issues with the IT, but is a little pricey.
Deficits-
a. First stage IT adds some phase and gain shifts at the higher frequencies. A bifilar IT may help in this regard, but testing will be at a future date. Asking a lot from an input stage.
b. 46 stage will require the extra DC heat circuitry under the hood, with possible contamination from stray magnetic fields.
c. Driver stage has 40uF cap, which would be nice to eliminate. Could put another CCS in the tail, going fully differential on the driver stage, not sure if this would be an improvement or not?
d. Negative DC supply required.
Option 2
Benefits-
a. No negative supplies needed.
b. First stage should be very linear with CCS load.
c. DC coupling exhibits no blocking distortion.
d. 6W6 has 1.6K Rp, so should be an excellent driver for 300B.
e. I like octals
f. Tubes are cheap and easily hand selected by test. If I get bored I can always swap in some overpriced 6SN7.
g. AC heating for first two stages is much simpler, but xfmrs need to be added to the signal chassis.
Deficits-
a. Not sure how stable and reliable the bias conditions will be. The CCS are adjustable, so in theory I should be able to closely dial in the 150V to produce good current match in 6W6 stage (place small R in cathodes).
b. Sad to lose the 46 DHT as a driver, but it just doesn't bias well in this topology.
c. 375V power is unregulated with over 500mV ripple. Not an issue for the first stage, shouldn't be a huge issue for the PSRR of the driver stage, but a tiny bit does seem to get through in my bench test.
d. Would like more than 150V on the plate of 6N8S, but there's only so much I can do with the existing 375V supply while retaining DC coupling.
I read about regulated supplies and ccs bad for IT coupled stages. Why complicate everything, if you don't use feedback just an IT. IT provide already a good deal of filtering.
See this: a parallel 300b or single tube driven with transformers all the best from Lundhal. The only thing better I could imagine is a 3 x 300B parallel but the transformer ''gap'' would start to compromise the sound which nullify the sound quality gain obtained from a lower output impedance.
http://www.audiodesignguide.com/my/pse2.html
See this: a parallel 300b or single tube driven with transformers all the best from Lundhal. The only thing better I could imagine is a 3 x 300B parallel but the transformer ''gap'' would start to compromise the sound which nullify the sound quality gain obtained from a lower output impedance.
http://www.audiodesignguide.com/my/pse2.html
I'm in the same "why complicate everything?", but frequently is possible to be over-simple. I'm curios where you read that CCS and regulated supplies (I guess you think about series-shunt regulators) destroy sound on IT coupled Differential amplifier stages. My listening and reading show me oposite: Lynn Olson's and Kevin Carter designs for example.
zigzagflux:
I will try Option 2. Salas SSHV2 regulator Simplistic Mosfet HV Shunt Regs maybe is good option for you.
I guess power supply regulation was not a big factor for me, other than to point out when I did use it, it provided excellent results. Altering the regulation is not really an option in my mind, as long as the bias conditions allow it. Altering the regulation in any way complicates matters, a gas tube is about as simple as it gets.
The issue is the topology differences between Option 1 and Option 2. Lordoff, I have been leaning towards Option 2 for a number of years, but haven't found the perfect design that fits into my existing 375V power supply. This one is the best I have found so far.
The issue is the topology differences between Option 1 and Option 2. Lordoff, I have been leaning towards Option 2 for a number of years, but haven't found the perfect design that fits into my existing 375V power supply. This one is the best I have found so far.
A) in my experience bifilar winding is not a good option for center tapped input transformers. It can be good for SE inputs or interstage, if you look for good quality at low cost.
B) DC heater supply can be a challenge. IMHO a CLC filter is a very rewarding option providing you have room and budget for a good quality inductor
Any interstage drive can drive the grids positive, which is a very good thing, imo. I'd not skimp on the swing of the driver given the IT being there, cap coupled might be a different thing. Two xfmrs will generally sound very different than a direct coupled two stage input/driver.
Don't count on an IT having anything like ideal response, that's the first thing. The two stage input circuit likely will vary quite a bit in terms of subjective sound depending on the parts, tubes and operating points, as well as the way it is loaded.
I don't like cathode bias in general, Ymmv. I think you need to build more than one amp?
_-_-
Don't count on an IT having anything like ideal response, that's the first thing. The two stage input circuit likely will vary quite a bit in terms of subjective sound depending on the parts, tubes and operating points, as well as the way it is loaded.
I don't like cathode bias in general, Ymmv. I think you need to build more than one amp?
_-_-
The reason I was asking is that if you want to drive the grids of the 300Bs positive at all, you might want to avoid a CCS tail for the driver stage. Push-pull operation might be desirable rather than differential operation there. You don't want the fact that one side sees a difficult load (grid current in the output tube) to cause distortion in the other side.
Now, if you were to go with the differential driver, it may still be okay since the other tube might be cut off when the one requires grid current but you would want to work things out and make sure before you build it. Maybe you already have.
Last edited:
Agreed, I probably did not clarify enough. Input transformer is shielded and dual bobbin, whereas the first and second stage IT will be bifilar. (Or just the driver stage if I elect to remove the first stage IT) With that in mind I am hoping the IT does an even better job than the layered winding IT's I am currently using. They certainly work up to around 18k, but tend to get pretty wobbly after that. Always looking to improve. My experience with bifilar signal output transformers has been very astounding. When you don't need a low winding/winding capacitance, they're the way to go. This will be my first go at bifilar interstage.
Thanks, that's the kind of feedback I have been interested in hearing. I was not certain how a fully differential stage would respond to the load of 300B grids as we near 0V. It would be easy enough to put in the cathode resistor with a 40uF cap from B+ to cathode. Is that the optimal driver stage you are referring to?
I have not built the output stage as being fed from a differential-fed IT. It's a bit of work on a small bench to temp some of these test circuits up. The closer I get to deciding on a path before proceeding the fewer gray hairs I get.
Yeah, changing the cathode junction to a voltage source (at AC) by bypassing the CCS or a resistor would do it.
well don't use CCS for output stages and be careful about how the current and voltage are regulated for feeding stages. For a single ended amplifier you want to aim for a low output resistance B+, a simple way of doing this is to have a massive capacitor and a very efficient choke.
I like option 1, but as an alternative, how about an option 3 where you DC couple the driver tube to differential source/emitter followers. Then use a step up transformer for the second stage gain (LL1689 in reverse works well with the ability to balance gain v load).
For option 2, I like a CCS in the tail and resistor loads as it regulates the grids of the second stage (so long as the tubes stay reasonably matched). I especially then like replacing the triodes with pentodes and hooking up a servo to the screen grids to keep the bias current through the two tubes exactly the same.
Is that an issue with the transformer coupling?
For option 2, I like a CCS in the tail and resistor loads as it regulates the grids of the second stage (so long as the tubes stay reasonably matched). I especially then like replacing the triodes with pentodes and hooking up a servo to the screen grids to keep the bias current through the two tubes exactly the same.
Is that an issue with the transformer coupling?
Because of the current source in the tail, any change in current on one side caused by a sudden change of load (grid current from the output stage) has to cause a corresponding change in current on the other side.
Like I said, it might not be an issue if the other tube is fully cut off by the time a tube hits grid current.
However, we should stop and ask ourselves at this point, "What is it that an unbypassed CCS tail gets us?" It gets us nearly perfectly complementary outputs as long as the loads are identical. If we put a voltage source on the cathodes, what do we lose? Well, signal balance will be at the mercy of any mismatch in mu of your two driver tubes.
Looking at your schematic, your loads for your differential stage are the two halves of a transformer. How exact is the balance between the two halves of the transformer? If it is not more perfect than the matching between your tubes, a differential stage doesn't give you any advantage anyway, even at low signal levels.
With a voltage source on the cathode, one side has no effect on the other; they act independently. This can introduce a bit of imbalance if the gains of the two driver tubes are not well matched. I just made an amp that required a lot of voltage drive for the output tubes (560Vpk-pk) and for various reasons I could not do a differential stage. I put a pot in the plate loads of the input stage so that I could null out any differences in gain in the following stage. It worked pretty well, I got excellent open-loop performance form the amp, in fact it was so good that I decided to leave it that way.
Now that solution won't work if your input stage is transformer-loaded but you might want to look at it anyway: Tube Amps with a Twist: A Unity Coupled KT88 Amp with Plitron Transformers
But the nature of the transformer is to provide balance. Whatever happens on the secondary is reflected symmetrically. It is not two transformers on one core, it is one transformer -- you can rearrange the secondary as in the attached to force the issue.
It gets not having a capacitor in the signal current loop. I'd rather give up grid current to eliminate the cap. Desiging for where we listen is more important that designing for worst case scenarios.
If they are Lundahls, I'd assume they are better than the tube.
I was commenting on option #2. My point was that the sum of currents in the two driver tubes always has to be constant with the CCS tail, so if one tube has to work harder, the other is forced to work less, no matter what. This has some negative consequences in some situations, and this is one of them.
If the stage is push-pull (voltage source in cathode), one tube can work harder while the other continues to work the same as it was meant to, since the two tubes are working more independently. It is simply the difference between differential operation and push-pull operation. The sum of the currents in the two tubes in push-pull can vary to drive the non-linear load of the grids. In differential operation, the sum must always be constant.
I have youngish kids in the house, so I frequently listen to my system cranked to 11, so "where we listen" is kind of an individual thing. I am not going to presume to know where "where we listen" is for zzf.
It is good that you have opinions and design philosophies. I wasn't telling zzf what to do, merely pointing out what he was potentially giving up. The choice is ultimately up to the builder how he wants to build things. I'm not going to start pushing my own design philosophies on others, but if I were building this amp, I would do something completely different than both of these options. When I last developed similar voltage swings in a driver, I used a 6BL7 biased with LEDs so I wouldn't have a cap, but I had CCS plate loads instead of the transformer. LEDs don't seem to be a good option here due to the desire to direct couple, unless zzf doesn't mind a huge string of LEDs. I personally don't have a lot of heartburn over caps myself, but to each his own.
Could be, but I doubt they are as well balanced as .1% resistors we could use in a conventional resistor-loaded differential stage. I'm just saying that without quantifying any of these things it is hard to definitively compare them.
I don't know about that. The CCS and B+ set DC conditions. They are effectively ignored for AC conditions if the CCS has good performance. So I have a series circuit of two triodes that drive the two ends of the IT primary as if it were one single winding.
This is like the power company providing me with split phase 120/240. They have a single primary on the HV side, and a center tapped secondary. I can apply load to only one of the 120V windings, and it is transformed to the entire HV winding. I can have a combination of 240 loads and unbalanced 120V loads, and they will refer to the primary winding as a perfect line/line load.
Assuming, of course, we have a common core.
So I think differential is the optimal way to drive the transformer as long as I have suitable output impedance of the driver stage. The transformer resolves the imbalance, helping optimize performance of the differential stage. So the only issue in my mind becomes the difference between output impedance of a differential stage to that of a push pull stage. Design it right, with capability of grid current load, and it should work.
They currently are, but that is where I am considering a bifilar IT to replace them (LL1692A). They perform well, but I am interested to see if better is possible. Will be a month or two before they arrive.
Thanks, I have. The input tranny is actually only going to be used for RCA inputs. My limited experience with cascodes has not been favorable, so I plan on staying away for now. I think I just need to make up some circuits and break out the sound card, prove to myself which works better for heavy/imbalanced load (within reason of course).
You are right now that I think about it. Sorry for my confusion. The only point at which it will matter is if the driver stage has to operate in AB to supply the grid current. If operated only in class A, it should be fine. Your driver load line is going to be a horizontal line with sharp turns upward and downward at low and high voltages respectively due to the non-linear grid load. If you try to exceed twice the idle current in the differential circuit, you will clip while the push-pull would keep going.
It might actually be beneficial to have some sort of upper limit on grid current. I don't recall ever seeing grid current curves for a 300B so I don't know what is reasonable to expect.
It might actually be beneficial to have some sort of upper limit on grid current. I don't recall ever seeing grid current curves for a 300B so I don't know what is reasonable to expect.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.