With a choke on the anode, it is going to steal voltage which could be used for the load instead. I really don't see an effective way to apply cross coupled feedback with cathode follower output. Having the feedback go to the local driver would really be a more elegant solution. If you run the output common-cathode, then cross-coupled feedback would be the right ticket.
Garter bias won't ensure that the cathode voltages are the same. They will be closer than they would be without it, but not an exact match. Also, having the transformer DC coupled between cathodes is going to ruin the affect of garter bias, unless the DCR is about 10x the value of the bias resistor. If the DCR is significantly lower than the bias resistors, it's going to behave similar to a single shared cathode resistor .
With the cathode follower output, you are probably going to need more supply voltage for the driver. The output gain is going to be about .4-.5 (perfect signal level for screen feedback, like UL). This means your going to need a peak voltage of around 60V(with bias of -30Vgk) to drive it to full power. I'd probably make a negative rail off the existing transformer (so you have and equal +/- voltage). You could then direct couple the the drivers screen grid to the output tubes cathode.
I'm not certain I agree with you on the garter not producing a matched voltage. If it creates a matched current then there has to be a matched voltage since it is going through a simple resistor. If the current is balanced to within 1mA, in my example the voltage imbalance would be about 0.7V which for most tranaformers would create little problem. However I take your point on it shorting out the Garter - so its a none runner.
The whole idea of getting the cathodes current and voltage balanced is a none runner really. Since there are three contol elements and what i am trying to achieve is to fix all three - its impossable.
The only solution would be to ground the cathode transformer, fix the anode voltage and then servo the grid bias. This may be a path worth investigating - and I suspect that Gary Pimm may have a suitable circuit.
On the point of the plate choke. I think you are a bit wrong. The plate choke will consume far less voltage than a voltage follower, and will be able to swing above +B which will allow for greater overall efficiency especially when entering class B - which is where the benefits of cross coupled feedback would come into play.
Research - research - research - O what fun.
Shoog
The whole idea of getting the cathodes current and voltage balanced is a none runner really. Since there are three contol elements and what i am trying to achieve is to fix all three - its impossable.
The only solution would be to ground the cathode transformer, fix the anode voltage and then servo the grid bias. This may be a path worth investigating - and I suspect that Gary Pimm may have a suitable circuit.
On the point of the plate choke. I think you are a bit wrong. The plate choke will consume far less voltage than a voltage follower, and will be able to swing above +B which will allow for greater overall efficiency especially when entering class B - which is where the benefits of cross coupled feedback would come into play.
Research - research - research - O what fun.
Shoog
It creates a closer-to, matched current and voltage, but they will only be an exact match if the tubes are. Think, if both voltages and currents are equal, both tubes will have the same grid voltage as well. This means they both would have identical Vgk.
Sorry, but it is true... that output, for the most part is a cathodyne phase splitter with mis-matched loads. For a given delta current, it will create voltage = I x R. The choke running with little shunt resistance will have a higher impedance than the transformers primary and hence create more voltage. However, the current drawn at the point of saturation will be much less (because the choke impedance is part of the loadline).
If you were to shunt the choke with a resistor, it would lower the amount of voltage dropped across the choke, but it is going to be stealing power from the load. Example, the primary impedance of the transformer is 1k. You shunt the choke with 1k. The tube will see a 2k load. They will see equal voltages in this case which means that resistor will get half the power and the other half will be distributed to the load. However, you are correct in saying the plate will swing above B+, but the cathode is no longer going to swing as far below ground potential, apposed to just putting a 2k load on the cathode.
Another example. You decide to shunt the choke with a low value resistor to minimize power losses to the output. We'll select 100R. You use a transformer impedance of 100R less in order to keep the same load impedance. The circuit has a delta I of 100mA, this would give 10V across the choke + shunt resistor. So, the 100R resistor will dissipate 1W which could have been going to the load.
Thanks for the input Jed,
I have come on a bit of a brick wall on this whole concept at the moment. The thing I will have to look into is self adjusting fixed bias - but there doesn't seem to be a lot of information out there so it obviously not a walk in the park to implement.
Unfortunately I have more pressing matters to attend to in the next fortnight, so i'm going to put this to bed for the moment. Its been a great brain workout and learning experience so far.
Cheers
Shoog
I have come on a bit of a brick wall on this whole concept at the moment. The thing I will have to look into is self adjusting fixed bias - but there doesn't seem to be a lot of information out there so it obviously not a walk in the park to implement.
Unfortunately I have more pressing matters to attend to in the next fortnight, so i'm going to put this to bed for the moment. Its been a great brain workout and learning experience so far.
Cheers
Shoog
That's the most important part. I've tried many ideas that didn't work, but a lot was learned from it. Some of which are concepts that can be applicable down the road.
BTW, The student edition of circuitmaker is a free download. It's a simulation program that is very intuitive (easy to use). The tube models it comes with to start are very limited, but I found a way to edit + add them. It is an extremely useful tool. You should give it a try.
P.S. The ECF80 is a hidden gem IMO. They aren't very linear in some configurations, but it always manages to sound good. I recently purchased a small quantity of the russian equivalents (6F1P). They are nice and dirt cheap.
I'm really happy with my ECF80 SLCF implementation. I am running the three tube SLCF at about 430V which gives each element a good 130V to play with - I'm also running them hard at 10mA. The front end has a 39K load so it gets the best out of it, and its great to have 8V of bias on the input. Really your right, it runs hard and plays hard - great tube. Its cheap enough to (shush don't tell anyone).
Shoog
Shoog
Digging up this old thread because I'm on a strikingly similar quest.
Shoog & Jeb, any tips on implementing the 6BN11 LTP with a 6AS7 PP?
I'm also working on a 6AS7 PP design, using a Garter Belt bias and toroidal outputs. Like Shoog, I'm hoping to keep this to two total tubes per channel (for nifty little monoblocks). Finding a suitable driver and phase splitter (in a single tube) has been a challenge until I stumbled on the mention of the 6BN11.
Any thoughts or tips either of you can share about the use of 6AS7+6BN11 or biasing the 6AS7 via a Garter Belt arrangement? I'd like to use a 230V:12V power transformer as the OPT (should reflect around 3k with 8 ohm speakers), so I'm curious as to how well the Garter Belt can match biases.
Shoog & Jeb, any tips on implementing the 6BN11 LTP with a 6AS7 PP?
I'm also working on a 6AS7 PP design, using a Garter Belt bias and toroidal outputs. Like Shoog, I'm hoping to keep this to two total tubes per channel (for nifty little monoblocks). Finding a suitable driver and phase splitter (in a single tube) has been a challenge until I stumbled on the mention of the 6BN11.
Any thoughts or tips either of you can share about the use of 6AS7+6BN11 or biasing the 6AS7 via a Garter Belt arrangement? I'd like to use a 230V:12V power transformer as the OPT (should reflect around 3k with 8 ohm speakers), so I'm curious as to how well the Garter Belt can match biases.
I'd suggest using to decent pair of Frame-grid, or even Special Quality (S.Q ) tubes ( B&W TV Tuner Types ) which have close tolerances to start with. Matching is a pain - but far better than using dual tubes and having to sift through a bunch of those to get balanced pair. Frame grid tubes are much closer in tolerance - easier to find a matched pair of triodes or pentodes.
As for compactrons - never used them - rarely see them in Australia. They would be the most likely source for a twin pentode having a 12 pin or similar socket; BUT as I understand, they are notoriously microphonic unlike frame grid tubes.
Last edited:
The amps working nicely using 6080's and 12L8GT as the driver - has been for about 2 years at this point.
Unfortunately I cannot comment on using the 6BN11 since I don't possess any. However designing a LTP driver using a dual pentode is no different to designing one with single pentodes with their cathodes tied together, matching is not really an issue (though they will generally be fairly well matched since they come from the same production run) as AC balance will be imposed rigorously by the LTP itself. Choose a stiff CCS for the tail. A simple bypassed resistor from B+ to the screens works well enough in this application. You will probably go AC coupled, though mine is DC coupled and works well. However DC coupling requires a negative rail and an input transformer, which is a lot considering the small gains for DC coupling.
When using the 6080 I have found that 100V and 100mA is a good operating point - this will bias them up to about -30V on the grid. This means that if you opt for Garter bias you will need a full 60V across the garter which is rather wasteful. I prefer to use a IXY chip to bias up my 6080's and employ cathode bypasses which link the 6080 cathodes together rather than to straight earth.
Expect this wonder to produce about 6watts power output. Why do we go to so much trouble for such an inefficient result - because we can and because a two tube PP amplifier is a cool concept.
Shoog
Unfortunately I cannot comment on using the 6BN11 since I don't possess any. However designing a LTP driver using a dual pentode is no different to designing one with single pentodes with their cathodes tied together, matching is not really an issue (though they will generally be fairly well matched since they come from the same production run) as AC balance will be imposed rigorously by the LTP itself. Choose a stiff CCS for the tail. A simple bypassed resistor from B+ to the screens works well enough in this application. You will probably go AC coupled, though mine is DC coupled and works well. However DC coupling requires a negative rail and an input transformer, which is a lot considering the small gains for DC coupling.
When using the 6080 I have found that 100V and 100mA is a good operating point - this will bias them up to about -30V on the grid. This means that if you opt for Garter bias you will need a full 60V across the garter which is rather wasteful. I prefer to use a IXY chip to bias up my 6080's and employ cathode bypasses which link the 6080 cathodes together rather than to straight earth.
Expect this wonder to produce about 6watts power output. Why do we go to so much trouble for such an inefficient result - because we can and because a two tube PP amplifier is a cool concept.
Shoog
Shoog-
Thanks for the input! I'm indeed thinking AC coupled LTP. I'm not too worried about burning up B+ in the Garter Bias because I'm hoping it means the power supply will be more simple:
I'm looking at 160V on the 6080 @ 63mA and 70V bias. If my thinking about the Garter Belt arrangement is correct, that will need 160V+70V+70V or a 300V B+. Suitable voltage for many drivers and would keep the PSU from becoming more complicated. Yes, that's a lot of big wattage resistors to dissipate, so I'll tweak to minimize once I can narrow down the front end.
I've been brushing up on my PP and LTP theory and will take a look at those qq and QQ's as potential drivers. The downside of the compactrons is availability of sockets and tubes. I'd like to design something with fairly easy to source parts (Antek toroids for all transformers, limited SS, fairly common tubes).
Thanks for the input! I'm indeed thinking AC coupled LTP. I'm not too worried about burning up B+ in the Garter Bias because I'm hoping it means the power supply will be more simple:
I'm looking at 160V on the 6080 @ 63mA and 70V bias. If my thinking about the Garter Belt arrangement is correct, that will need 160V+70V+70V or a 300V B+. Suitable voltage for many drivers and would keep the PSU from becoming more complicated. Yes, that's a lot of big wattage resistors to dissipate, so I'll tweak to minimize once I can narrow down the front end.
I've been brushing up on my PP and LTP theory and will take a look at those qq and QQ's as potential drivers. The downside of the compactrons is availability of sockets and tubes. I'd like to design something with fairly easy to source parts (Antek toroids for all transformers, limited SS, fairly common tubes).
Sodacose - your on the right track there - your suggested operating point looks fairly good and I see the advantage of having a more sensible +B for your front end. However you are still going to need a simple little -B supply for going under your LTP to give the CCS room to breath. Don't sweat it as this can usully be derived from a simple rectified heater supply or a tiny little stand alone transformer.
I saw some of the QQV02-6 on Ebay for £5.00 each combined with standard sockets this makes it an easy build out and cheap replacement into the forseeable future. The 12L8GT's I chose are not easy to come by and are fairly expensive to boot.
Broskie, of Tubecad, developed a transistor based garter bias which saves on one of the 70v, worth search for.
Shoog
I saw some of the QQV02-6 on Ebay for £5.00 each combined with standard sockets this makes it an easy build out and cheap replacement into the forseeable future. The 12L8GT's I chose are not easy to come by and are fairly expensive to boot.
Broskie, of Tubecad, developed a transistor based garter bias which saves on one of the 70v, worth search for.
Shoog
I second that !
The same tube from Philips is named QQE02-5, fully interchangeable with the Mullard QQV02-6.
They look perfect for that job !
Yves.
Thats what I used. Its not a great candidate since it has very low gm and is quite obscure making it expensive and difficult to get hold of.
--------
The only thing to watch out for with the QQ valves, especially the high gm ones, is that they are all designed for mhz duty and so need attention to detail in implementing. Grid and plate stoppers are a good idea or ferrite beads.
Shoog
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.